PLI Programming Guide

Enterprise PL/I for z/OS
Programming Guide
Version 4 Release 2
GI11-9145-01

Programming Guide
Version 4 Release 2
GI11-9145-01
Second Edition (September 2011)

This edition applies to Version 4 Release 2 of Enterprise PL/I for z/OS, 5655-W67, and to any subsequent releases
until otherwise indicated in new editions or technical newsletters. Make sure you are using the correct edition for
the level of the product.
Order publications through your IBM representative or the IBM branch office serving your locality. Publications are
not stocked at the address below.
A form for readers' comments is provided at the back of this publication. If the form has been removed, address
your comments to:
IBM Corporation, Department H150/090
555 Bailey Ave
San Jose, CA, 95141-1099
United States of America
When you send information to IBM, you grant IBM a nonexclusive right to use or distribute the information in any
way it believes appropriate without incurring any obligation to you.
Copyright IBM Corporation 1999, 2011.
US Government Users Restricted Rights Use, duplication or disclosure restricted by GSA ADP Schedule Contract
with IBM Corp.
Note!
Before using this information and the product it supports, be sure to read the general
information under Notices on page 595.
Contents
Tables . . . . . . . . . . . . . . . ix
Figures . . . . . . . . . . . . . . . xi
Introduction . . . . . . . . . . . . xiii
About This Book . . . . . . . . . . . . xiii
Run-time environment for Enterprise PL/I for
z/OS . . . . . . . . . . . . . . . . xiii
Using your documentation . . . . . . . . . xiv
PL/I information . . . . . . . . . . . xiv
Notation conventions used in this book . . . . . xv
Conventions used . . . . . . . . . . . xv
How to read the syntax notation . . . . . . xv
How to read the notational symbols . . . . xviii
Summary of changes . . . . . . . . . . xviii
Enhancements in this release . . . . . . . xviii
Enhancements from V4R1 . . . . . . . . xx
Enhancements from V3R8 . . . . . . . . xxii
Enhancements from V3R7 . . . . . . . . xxiii
Enhancements from V3R6 . . . . . . . . xxiv
Enhancements from V3R5 . . . . . . . . xxv
Enhancements from V3R4 . . . . . . . . xxvi
Enhancements from V3R3 . . . . . . . xxviii
Enhancements from V3R2 . . . . . . . . xxix
Enhancements from V3R1 . . . . . . . . xxx
Enhancements from VisualAge PL/I . . . . xxxi
How to send your comments . . . . . . . xxxii
Accessibility . . . . . . . . . . . . . xxxiii
Interface information . . . . . . . . . xxxiii
Keyboard navigation . . . . . . . . . xxxiii
Accessibility of this information . . . . . xxxiii
IBM and accessibility . . . . . . . . . xxxiii
Part 1. Compiling your program . . . 1
Chapter 1. Using compiler options and
facilities. . . . . . . . . . . . . . . 5
Compile-time option descriptions . . . . . . . 5
AGGREGATE . . . . . . . . . . . . . 9
ARCH . . . . . . . . . . . . . . . 10
ATTRIBUTES . . . . . . . . . . . . . 12
BACKREG. . . . . . . . . . . . . . 13
BIFPREC . . . . . . . . . . . . . . 14
BLANK. . . . . . . . . . . . . . . 15
BLKOFF . . . . . . . . . . . . . . 16
CEESTART . . . . . . . . . . . . . 17
CHECK . . . . . . . . . . . . . . 18
CMPAT. . . . . . . . . . . . . . . 19
CODEPAGE . . . . . . . . . . . . . 21
COMMON . . . . . . . . . . . . . 22
COMPILE . . . . . . . . . . . . . . 23
COPYRIGHT . . . . . . . . . . . . . 24
CSECT . . . . . . . . . . . . . . . 25
CSECTCUT . . . . . . . . . . . . . 26
CURRENCY . . . . . . . . . . . . . 27
DBCS . . . . . . . . . . . . . . . 28
DD . . . . . . . . . . . . . . . . 29
DDSQL. . . . . . . . . . . . . . . 30
DECIMAL. . . . . . . . . . . . . . 31
DEFAULT . . . . . . . . . . . . . . 33
DEPRECATE . . . . . . . . . . . . . 42
DISPLAY . . . . . . . . . . . . . . 44
DLLINIT . . . . . . . . . . . . . . 45
EXIT. . . . . . . . . . . . . . . . 46
EXTRN. . . . . . . . . . . . . . . 47
FLAG . . . . . . . . . . . . . . . 48
FLOAT . . . . . . . . . . . . . . . 49
FLOATINMATH. . . . . . . . . . . . 52
GOFF . . . . . . . . . . . . . . . 53
GONUMBER . . . . . . . . . . . . . 54
GRAPHIC . . . . . . . . . . . . . . 55
HGPR . . . . . . . . . . . . . . . 56
IGNORE . . . . . . . . . . . . . . 57
INCAFTER . . . . . . . . . . . . . 58
INCDIR . . . . . . . . . . . . . . 59
INCPDS . . . . . . . . . . . . . . 60
INITAUTO . . . . . . . . . . . . . 61
INITBASED . . . . . . . . . . . . . 62
INITCTL . . . . . . . . . . . . . . 63
INITSTATIC . . . . . . . . . . . . . 64
INSOURCE . . . . . . . . . . . . . 65
INTERRUPT . . . . . . . . . . . . . 66
LANGLVL. . . . . . . . . . . . . . 67
LIMITS . . . . . . . . . . . . . . . 68
LINECOUNT. . . . . . . . . . . . . 70
LINEDIR . . . . . . . . . . . . . . 71
LIST. . . . . . . . . . . . . . . . 72
LISTVIEW. . . . . . . . . . . . . . 73
MACRO . . . . . . . . . . . . . . 75
MAP . . . . . . . . . . . . . . . 76
MARGINI . . . . . . . . . . . . . . 77
MARGINS. . . . . . . . . . . . . . 78
MAXGEN . . . . . . . . . . . . . . 80
MAXMEM. . . . . . . . . . . . . . 81
MAXMSG . . . . . . . . . . . . . . 82
MAXNEST . . . . . . . . . . . . . 83
MAXSTMT . . . . . . . . . . . . . 84
MAXTEMP . . . . . . . . . . . . . 85
MDECK . . . . . . . . . . . . . . 86
NAME . . . . . . . . . . . . . . . 87
NAMES . . . . . . . . . . . . . . 88
NATLANG . . . . . . . . . . . . . 89
NEST . . . . . . . . . . . . . . . 90
NOT. . . . . . . . . . . . . . . . 91
NUMBER . . . . . . . . . . . . . . 92
OBJECT . . . . . . . . . . . . . . 93
OFFSET . . . . . . . . . . . . . . 94
ONSNAP . . . . . . . . . . . . . . 95
OPTIMIZE. . . . . . . . . . . . . . 96
Copyright IBM Corp. 1999, 2011 iii
OPTIONS . . . . . . . . . . . . . . 98
OR . . . . . . . . . . . . . . . . 99
PP . . . . . . . . . . . . . . . . 100
PPCICS . . . . . . . . . . . . . . 101
PPINCLUDE . . . . . . . . . . . . 102
PPLIST . . . . . . . . . . . . . . 103
PPMACRO . . . . . . . . . . . . . 104
PPSQL . . . . . . . . . . . . . . 105
PPTRACE . . . . . . . . . . . . . 106
PRECTYPE . . . . . . . . . . . . . 107
PREFIX . . . . . . . . . . . . . . 108
PROCEED . . . . . . . . . . . . . 109
PROCESS. . . . . . . . . . . . . . 110
QUOTE . . . . . . . . . . . . . . 111
REDUCE . . . . . . . . . . . . . . 112
RENT . . . . . . . . . . . . . . . 113
RESEXP . . . . . . . . . . . . . . 115
RESPECT. . . . . . . . . . . . . . 116
RTCHECK . . . . . . . . . . . . . 117
RULES . . . . . . . . . . . . . . 118
SEMANTIC . . . . . . . . . . . . . 126
SERVICE . . . . . . . . . . . . . . 127
SOURCE . . . . . . . . . . . . . . 128
SPILL . . . . . . . . . . . . . . . 129
STATIC . . . . . . . . . . . . . . 130
STDSYS . . . . . . . . . . . . . . 131
STMT . . . . . . . . . . . . . . . 132
STORAGE . . . . . . . . . . . . . 133
STRINGOFGRAPHIC . . . . . . . . . 134
SYNTAX . . . . . . . . . . . . . . 135
SYSPARM . . . . . . . . . . . . . 136
SYSTEM . . . . . . . . . . . . . . 137
TERMINAL . . . . . . . . . . . . . 138
TEST . . . . . . . . . . . . . . . 139
UNROLL. . . . . . . . . . . . . . 142
USAGE . . . . . . . . . . . . . . 143
WIDECHAR. . . . . . . . . . . . . 144
WINDOW . . . . . . . . . . . . . 145
WRITABLE . . . . . . . . . . . . . 146
XINFO . . . . . . . . . . . . . . 148
XML . . . . . . . . . . . . . . . 151
XREF . . . . . . . . . . . . . . . 152
Blanks, comments and strings in options . . . . 153
Changing the default options . . . . . . . . 153
Specifying options in the %PROCESS or *PROCESS
statements . . . . . . . . . . . . . . 153
Using % statements . . . . . . . . . . . 154
Using the %INCLUDE statement . . . . . . 155
Using the compiler listing . . . . . . . . . 156
Heading information . . . . . . . . . . 156
Options used for compilation . . . . . . . 157
Preprocessor input . . . . . . . . . . 157
SOURCE program. . . . . . . . . . . 157
Statement nesting level . . . . . . . . . 157
ATTRIBUTE and cross-reference table . . . . 158
Aggregate length table . . . . . . . . . 159
Statement offset addresses . . . . . . . . 159
Storage offset listing . . . . . . . . . . 161
File reference table . . . . . . . . . . 162
Messages and return codes . . . . . . . . 163
Chapter 2. PL/I preprocessors . . . . 165
Include preprocessor . . . . . . . . . . . 166
Macro preprocessor . . . . . . . . . . . 167
Macro preprocessor options . . . . . . . 167
Macro preprocessor example . . . . . . . 174
SQL preprocessor . . . . . . . . . . . . 175
Programming and compilation considerations 175
SQL preprocessor options . . . . . . . . 177
Coding SQL statements in PL/I applications 195
General information on LOBs . . . . . . . 205
Suppressing SQL preprocessor messages . . . 205
CICS Preprocessor. . . . . . . . . . . . 205
CICS preprocessor options . . . . . . . . 206
Coding CICS statements in PL/I applications 206
Writing CICS transactions in PL/I . . . . . 207
Error-handling . . . . . . . . . . . . 207
Chapter 3. Using PL/I cataloged
procedures . . . . . . . . . . . . 209
IBM-supplied cataloged procedures . . . . . . 209
Compile only (IBMZC) . . . . . . . . . 210
Compile and bind (IBMZCB) . . . . . . . 211
Compile, bind, and run (IBMZCBG) . . . . . 213
Compile, prelink, and link-edit (IBMZCPL) . . 214
Compile, prelink, link-edit, and run
(IBMZCPLG) . . . . . . . . . . . . 216
Compile, prelink, load and run (IBMZCPG) . . 218
Invoking a cataloged procedure . . . . . . . 220
Specifying multiple invocations . . . . . . 220
Modifying the PL/I cataloged procedures . . . . 221
EXEC statement . . . . . . . . . . . 222
DD statement . . . . . . . . . . . . 222
Chapter 4. Compiling your program 225
Invoking the compiler under z/OS UNIX . . . . 225
Input files . . . . . . . . . . . . . 225
Specifying compile-time options under z/OS
UNIX . . . . . . . . . . . . . . . 226
-qoption_keyword. . . . . . . . . . . 226
Single and multiletter flags . . . . . . . . 226
Invoking the compiler under z/OS using JCL . . 227
EXEC statement . . . . . . . . . . . 227
DD statements for the standard data sets . . . 228
Listing (SYSPRINT) . . . . . . . . . . 229
Source Statement Library (SYSLIB) . . . . . 230
Specifying options. . . . . . . . . . . 230
Specifying options in the EXEC statement . . . 230
Specifying options in the EXEC statement using
an options file . . . . . . . . . . . . 231
Chapter 5. Link-editing and running 233
Link-edit considerations . . . . . . . . . . 233
Using the binder . . . . . . . . . . . 233
Using the prelinker . . . . . . . . . . 233
Using the ENTRY card . . . . . . . . . 234
Run-time considerations . . . . . . . . . . 234
Formatting conventions for PRINT files . . . 234
Changing the format on PRINT files. . . . . 234
Automatic prompting . . . . . . . . . 235
iv Enterprise PL/I for z/OS Programming Guide
Punctuating long input lines . . . . . . . 236
Punctuating GET LIST and GET DATA
statements . . . . . . . . . . . . . 236
ENDFILE. . . . . . . . . . . . . . 237
SYSPRINT considerations . . . . . . . . . 237
Using MSGFILE(SYSPRINT) . . . . . . . 239
Using FETCH in your routines . . . . . . . 239
Fetching Enterprise PL/I routines . . . . . 239
Fetching PL/I MAIN Routines . . . . . . 247
Fetching z/OS C routines . . . . . . . . 247
Fetching assembler routines . . . . . . . 248
Invoking MAIN under TSO/E. . . . . . . . 248
Invoking MAIN under z/OS UNIX . . . . . . 249
Part 2. Using I/O facilities . . . . . 251
Chapter 6. Using data sets and files 253
Allocating files . . . . . . . . . . . . . 253
Associating data sets with files under z/OS . . . 254
Associating several files with one data set . . . 256
Associating several data sets with one file . . . 256
Concatenating several data sets . . . . . . 257
Accessing HFS files under z/OS . . . . . . 257
Associating data sets with files under z/OS UNIX 258
Using environment variables . . . . . . . 259
Using the TITLE option of the OPEN statement 260
Attempting to use files not associated with data
sets. . . . . . . . . . . . . . . . 261
How PL/I finds data sets . . . . . . . . 261
Specifying characteristics using DD_DDNAME
environment variables . . . . . . . . . 261
Establishing data set characteristics . . . . . . 267
Blocks and records . . . . . . . . . . 267
Record formats . . . . . . . . . . . . 268
Data set organization. . . . . . . . . . 270
Labels . . . . . . . . . . . . . . . 271
Data Definition (DD) statement . . . . . . 271
Associating PL/I files with data sets . . . . 273
Specifying characteristics in the
ENVIRONMENT attribute . . . . . . . . 275
Chapter 7. Using libraries . . . . . . 285
Types of libraries . . . . . . . . . . . . 285
How to use a library . . . . . . . . . . 285
Creating a library . . . . . . . . . . . . 286
SPACE parameter . . . . . . . . . . . 286
Creating and updating a library member . . . . 287
Examples. . . . . . . . . . . . . . 287
Extracting information from a library directory . . 289
Chapter 8. Defining and using
consecutive data sets. . . . . . . . 291
Using stream-oriented data transmission . . . . 291
Defining files using stream I/O . . . . . . 291
Defining stream files using PL/I dynamic
allocation. . . . . . . . . . . . . . 292
Specifying ENVIRONMENT options . . . . 292
Creating a data set with stream I/O. . . . . 294
Accessing a data set with stream I/O . . . . 297
Using PRINT files with stream I/O . . . . . 299
Using SYSIN and SYSPRINT files . . . . . 303
Controlling input from the terminal . . . . . . 304
Format of data . . . . . . . . . . . . . 305
Stream and record files . . . . . . . . . . 306
Defining QSAM files using PL/I dynamic
allocation. . . . . . . . . . . . . . . 306
Capital and lowercase letters . . . . . . . . 307
End-of-file . . . . . . . . . . . . . . 307
COPY option of GET statement . . . . . . . 307
Chapter 9. Controlling output to the
terminal . . . . . . . . . . . . . . 309
Format of PRINT files . . . . . . . . . . 309
Output from the PUT EDIT command . . . . . 309
Chapter 10. Using record-oriented
data transmission . . . . . . . . . 311
Specifying record format . . . . . . . . . 312
Defining files using record I/O . . . . . . . 312
Specifying ENVIRONMENT options . . . . . 312
CONSECUTIVE . . . . . . . . . . . 312
ORGANIZATION(CONSECUTIVE) . . . . . 313
CTLASA|CTL360 . . . . . . . . . . . 313
LEAVE|REREAD . . . . . . . . . . . 314
Creating a data set with record I/O . . . . . . 315
Essential information . . . . . . . . . . 315
Accessing and updating a data set with record I/O 316
Example of consecutive data sets . . . . . . 317
Chapter 11. Defining and using
regional data sets . . . . . . . . . 321
Defining REGIONAL(1) data sets using PL/I
dynamic allocation . . . . . . . . . . . 323
Defining files for a regional data set . . . . . . 323
Using keys with REGIONAL data sets . . . . 324
Using REGIONAL(1) data sets . . . . . . . 324
Dummy Records . . . . . . . . . . . 325
Creating a REGIONAL(1) data set . . . . . 325
Accessing and updating a REGIONAL(1) data
set . . . . . . . . . . . . . . . . 326
Essential information for creating and accessing
regional data sets . . . . . . . . . . . . 329
Chapter 12. Defining and using VSAM
data sets . . . . . . . . . . . . . 333
Defining VSAM file using PL/I dynamic allocation 333
Using VSAM data sets . . . . . . . . . . 333
How to run a program with VSAM data sets 333
Pairing an Alternate Index Path with a File . . 334
VSAM organization . . . . . . . . . . . 334
Keys for VSAM data sets . . . . . . . . 336
Choosing a data set type . . . . . . . . 337
Defining files for VSAM data sets . . . . . . 339
Contents v
Performance options . . . . . . . . . . 342
Defining Files for Alternate Index Paths . . . . 342
Defining VSAM data sets . . . . . . . . . 343
Entry-sequenced data sets . . . . . . . . . 344
Loading an ESDS . . . . . . . . . . . 344
Using a SEQUENTIAL file to access an ESDS 345
Using Files Defined for non-VSAM Data Sets. . . 366
Using Shared Data Sets . . . . . . . . . 366
Part 3. Improving your program 367
Chapter 13. Improving performance 369
Selecting compiler options for optimal performance 369
OPTIMIZE . . . . . . . . . . . . . 369
GONUMBER . . . . . . . . . . . . 369
ARCH. . . . . . . . . . . . . . . 369
REDUCE . . . . . . . . . . . . . . 370
RULES . . . . . . . . . . . . . . 370
PREFIX . . . . . . . . . . . . . . 371
DEFAULT . . . . . . . . . . . . . 372
Summary of compiler options that improve
performance. . . . . . . . . . . . . 375
Coding for better performance . . . . . . . 375
DATA-directed input and output . . . . . . 375
Input-only parameters . . . . . . . . . 376
GOTO statements . . . . . . . . . . . 376
String assignments . . . . . . . . . . 376
Loop control variables . . . . . . . . . 377
PACKAGEs versus nested PROCEDUREs . . . 377
REDUCIBLE Functions . . . . . . . . . 378
DESCLOCATOR or DESCLIST . . . . . . 379
DEFINED versus UNION . . . . . . . . 379
Named constants versus static variables . . . 379
Avoiding calls to library routines . . . . . . 381
Preloading library routines . . . . . . . . 381
Part 4. Using interfaces to other
products . . . . . . . . . . . . . 383
Chapter 14. Using the Sort program 385
Preparing to use Sort . . . . . . . . . . . 385
Choosing the type of Sort . . . . . . . . 386
Specifying the sorting field . . . . . . . . 389
Specifying the records to be sorted . . . . . 390
Determining storage needed for Sort . . . . 391
Calling the Sort program . . . . . . . . . 392
Example 1 . . . . . . . . . . . . . 393
Example 2 . . . . . . . . . . . . . 394
Example 3 . . . . . . . . . . . . . 394
Example 4 . . . . . . . . . . . . . 394
Example 5 . . . . . . . . . . . . . 394
Determining whether the Sort was successful 394
Establishing data sets for Sort . . . . . . . 395
Sort data input and output . . . . . . . . . 396
Data input and output handling routines . . . . 396
E15Input handling routine (Sort Exit E15) . . 397
E35Output handling routine (Sort Exit E35) 400
Calling PLISRTA example . . . . . . . . 401
Calling PLISRTB example . . . . . . . . 402
Calling PLISRTC example . . . . . . . . 403
Calling PLISRTD example . . . . . . . . 404
Sorting variable-length records example . . . 406
Chapter 15. ILC with C . . . . . . . 409
Equivalent data types . . . . . . . . . . 409
Simple type equivalence. . . . . . . . . 409
Struct type equivalence . . . . . . . . . 410
Enum type equivalence . . . . . . . . . 410
File type equivalence . . . . . . . . . . 411
Using C functions . . . . . . . . . . . . 411
Matching simple parameter types . . . . . 412
Matching string parameter types . . . . . . 415
Functions returning ENTRYs . . . . . . . 416
Linkages . . . . . . . . . . . . . . . 417
Sharing output and input . . . . . . . . . 419
Sharing output . . . . . . . . . . . . 419
Sharing input . . . . . . . . . . . . 420
Using the ATTACH statement . . . . . . . 420
Redirecting C standard streams . . . . . . 420
Summary. . . . . . . . . . . . . . . 420
Chapter 16. Interfacing with Java . . . 423
Java Native Interface (JNI) . . . . . . . . . 423
Calling PL/I program from Java . . . . . . 424
JNI Sample Program #1 - 'Hello World' . . . . . 424
Writing Java Sample Program #1 . . . . . . 424
JNI Sample Program #2 - Passing a String . . . . 427
JNI Sample Program #3 - Passing an Integer . . . 432
JNI Sample Program #4 - Java Invocation API . . 436
Determining equivalent Java and PL/I data types 440
Part 5. Specialized programming
tasks . . . . . . . . . . . . . . 441
Chapter 17. Using the PLISAXA and
PLISAXB XML parsers . . . . . . . 443
Overview. . . . . . . . . . . . . . . 443
The PLISAXA built-in subroutine. . . . . . . 444
The PLISAXB built-in subroutine . . . . . . . 444
The SAX event structure. . . . . . . . . . 445
start_of_document. . . . . . . . . . . 445
version_information . . . . . . . . . . 445
encoding_declaration . . . . . . . . . . 446
standalone_declaration . . . . . . . . . 446
document_type_declaration. . . . . . . . 446
end_of_document . . . . . . . . . . . 446
start_of_element . . . . . . . . . . . 446
attribute_name . . . . . . . . . . . . 446
attribute_characters . . . . . . . . . . 446
attribute_predefined_reference. . . . . . . 447
attribute_character_reference . . . . . . . 447
end_of_element. . . . . . . . . . . . 447
start_of_CDATA_section. . . . . . . . . 447
end_of_CDATA_section . . . . . . . . . 447
content_characters . . . . . . . . . . . 447
vi Enterprise PL/I for z/OS Programming Guide
content_predefined_reference . . . . . . . 448
content_character_reference. . . . . . . . 448
processing_instruction . . . . . . . . . 448
comment . . . . . . . . . . . . . . 448
unknown_attribute_reference . . . . . . . 448
unknown_content_reference . . . . . . . 448
start_of_prefix_mapping. . . . . . . . . 448
end_of_prefix_mapping . . . . . . . . . 448
exception. . . . . . . . . . . . . . 448
Parameters to the event functions . . . . . 448
Coded character sets for XML documents . . . . 449
Supported EBCDIC code pages . . . . . . 450
Supported ASCII code pages . . . . . . . 450
Specifying the code page . . . . . . . . 450
Exceptions . . . . . . . . . . . . . . 451
Example . . . . . . . . . . . . . . . 452
Continuable exception codes . . . . . . . . 464
Terminating exception codes . . . . . . . . 467
Chapter 18. Using the PLISAXC and
PLISAXD XML parsers . . . . . . . 473
Overview. . . . . . . . . . . . . . . 473
The PLISAXC built-in subroutine. . . . . . . 474
The PLISAXD built-in subroutine. . . . . . . 474
end_of_document . . . . . . . . . . . 476
start_of_element . . . . . . . . . . . 476
attribute_name . . . . . . . . . . . . 476
end_of_element. . . . . . . . . . . . 476
comment . . . . . . . . . . . . . . 477
namespace_declare . . . . . . . . . . 477
end_of_input . . . . . . . . . . . . 477
unresolved_reference . . . . . . . . . . 478
exception. . . . . . . . . . . . . . 478
Differences in the events . . . . . . . . 479
Supported code pages . . . . . . . . . 481
Exceptions . . . . . . . . . . . . . . 482
Parsing XML documents with validation . . . . 482
XML schema . . . . . . . . . . . . 482
Creating an OSR . . . . . . . . . . . 483
Example with a simple document . . . . . . 484
Example of using the PLISAXC built-in
subroutine . . . . . . . . . . . . . 484
Example of using the PLISAXD built-in
subroutine . . . . . . . . . . . . . 494
Chapter 19. Using PLIDUMP . . . . . 507
PLIDUMP usage notes . . . . . . . . . . 508
Locating variables in the PLIDUMP output . . . 509
Locating AUTOMATIC variables . . . . . . 509
Locating STATIC variables . . . . . . . . 510
Locating CONTROLLED variables . . . . . 511
Saved Compilation Data. . . . . . . . . . 514
Copyright . . . . . . . . . . . . . 514
Timestamp . . . . . . . . . . . . . 514
Saved options string . . . . . . . . . . 515
Chapter 20. Interrupts and attention
processing . . . . . . . . . . . . 521
Using ATTENTION ON-units . . . . . . . . 522
Interaction with a debugging tool . . . . . . 522
Chapter 21. Using the
Checkpoint/Restart facility. . . . . . 523
Requesting a checkpoint record . . . . . . . 523
Defining the checkpoint data set . . . . . . 524
Requesting a restart . . . . . . . . . . . 525
Automatic restart after a system failure. . . . 525
Automatic restart within a program . . . . . 525
Getting a deferred restart . . . . . . . . 525
Modifying checkpoint/restart activity . . . . 526
Chapter 22. Using user exits . . . . . 527
Procedures performed by the compiler user exit 527
Activating the compiler user exit . . . . . . . 528
The IBM-supplied compiler exit, IBMUEXIT . . 528
Customizing the compiler user exit . . . . . 528
Modifying SYSUEXIT . . . . . . . . . 529
Writing your own compiler exit . . . . . . 529
Structure of global control blocks . . . . . . 529
Writing the initialization procedure . . . . . 530
Writing the message filtering procedure . . . 531
Writing the termination procedure . . . . . 533
Example of suppressing SQL messages . . . . . 533
Chapter 23. PL/I descriptors . . . . . 541
Passing an argument . . . . . . . . . . . 541
Argument passing by descriptor list . . . . . 541
Argument passing by descriptor-locator . . . 542
CMPAT(V*) descriptors . . . . . . . . . . 542
String descriptors . . . . . . . . . . . 542
Array descriptors . . . . . . . . . . . 543
CMPAT(LE) descriptors . . . . . . . . . . 544
Part 6. Appendixes . . . . . . . . 547
Appendix. SYSADATA message
information . . . . . . . . . . . . 549
Understanding the SYSADATA file . . . . . . 549
Summary Record . . . . . . . . . . . 550
Counter records . . . . . . . . . . . 551
Literal records . . . . . . . . . . . . 551
File records . . . . . . . . . . . . . 552
Message records . . . . . . . . . . . 552
Contents vii
Understanding SYSADATA symbol information 553
Ordinal type records . . . . . . . . . . 553
Ordinal element records . . . . . . . . . 554
Symbol records . . . . . . . . . . . . 555
Understanding SYSADATA syntax information . . 570
Source records . . . . . . . . . . . . 570
Token records . . . . . . . . . . . . 571
Syntax records . . . . . . . . . . . . 572
Notices . . . . . . . . . . . . . . 595
Trademarks . . . . . . . . . . . . . . 597
Bibliography. . . . . . . . . . . . 599
PL/I publications . . . . . . . . . . . . 599
Enterprise PL/I for z/OS . . . . . . . . 599
PL/I for MVS & VM . . . . . . . . . . 599
PL/I for AIX . . . . . . . . . . . . 599
Related publications . . . . . . . . . . . 599
DB2 UDB for z/OS . . . . . . . . . . 599
DFSORT . . . . . . . . . . . . . . 599
IMS/ESA. . . . . . . . . . . . . . 599
TXSeries for Multiplatforms . . . . . . . 600
z/Architecture . . . . . . . . . . . . 600
z/OS Language Environment . . . . . . . 600
z/OS MVS . . . . . . . . . . . . . 600
z/OS TSO/E . . . . . . . . . . . . 600
z/OS UNIX System Services . . . . . . . 600
Unicode and character representation . . . . 600
Glossary . . . . . . . . . . . . . 601
Index . . . . . . . . . . . . . . . 615
viii Enterprise PL/I for z/OS Programming Guide
Tables
1. How to use Enterprise PL/I publications xiv
2. How to use z/OS Language Environment
publications . . . . . . . . . . . . xiv
3. Compile-time options, abbreviations, and
IBM-supplied defaults . . . . . . . . . 5
4. SYSTEM option table . . . . . . . . . 137
5. Description of PL/I error codes and return
codes . . . . . . . . . . . . . . 163
6. Using the FLAG option to select the lowest
message severity listed . . . . . . . . 164
7. SQL preprocessor options and IBM-supplied
defaults . . . . . . . . . . . . . 177
8. SQL data types generated from PL/I
declarations . . . . . . . . . . . . 200
9. SQL data types generated from SQL TYPE
declarations . . . . . . . . . . . . 201
10. SQL data types mapped to PL/I declarations 202
11. SQL data types mapped to SQL TYPE
declarations . . . . . . . . . . . . 203
12. Compile-time option flags supported by
Enterprise PL/I under z/OS UNIX . . . . 227
13. Compiler standard data sets . . . . . . 228
14. Attributes of PL/I file declarations . . . . 275
15. A comparison of data set types available to
PL/I record I/O . . . . . . . . . . 282
16. Information required when creating a library 286
17. Statements and options allowed for creating
and accessing consecutive data sets . . . . 311
18. IBM machine code print control characters
(CTL360) . . . . . . . . . . . . . 314
19. Effect of LEAVE and REREAD Options 314
20. Creating a consecutive data set with record
I/O: essential parameters of the DD statement 315
21. Accessing a consecutive data set with record
I/O: essential parameters of the DD statement 317
22. Statements and options allowed for creating
and accessing regional data sets . . . . . 322
23. Creating a regional data set: essential
parameters of the DD statement . . . . . 330
24. DCB subparameters for a regional data set 330
25. Accessing a regional data set: essential
parameters of the DD statement . . . . . 331
26. Types and advantages of VSAM data sets 336
27. Processing Allowed on Alternate Index Paths 339
28. Statements and options allowed for loading
and accessing VSAM entry-sequenced data
sets . . . . . . . . . . . . . . . 344
and accessing VSAM indexed data sets . . . 347
and accessing VSAM relative-record data sets. 360
31. The entry points and arguments to PLISRTx
(x = A, B, C, or D) . . . . . . . . . . 392
32. C and PL/I Type Equivalents . . . . . . 409
33. Java Primitive Types and PL/I Native
Equivalents . . . . . . . . . . . . 440
34. Continuable Exceptions . . . . . . . . 464
35. Terminating Exceptions . . . . . . . . 467
Copyright IBM Corp. 1999, 2011 ix
x Enterprise PL/I for z/OS Programming Guide
Figures
1. Including source statements from a library 156
2. Finding statement number (compiler listing
example) . . . . . . . . . . . . . 160
3. Finding statement number (runtime message
example) . . . . . . . . . . . . . 161
4. Using the macro preprocessor to produce a
source deck . . . . . . . . . . . . 174
5. The PL/I declaration of SQLCA . . . . . 195
6. The PL/I declaration of an SQL descriptor
area. . . . . . . . . . . . . . . 196
7. Invoking a cataloged procedure . . . . . 210
8. Cataloged Procedure IBMZC . . . . . . 211
9. Cataloged Procedure IBMZCB . . . . . . 212
10. Cataloged Procedure IBMZCBG . . . . . 213
11. Cataloged Procedure IBMZCPL . . . . . 215
12. Cataloged Procedure IBMZCPLG . . . . . 217
13. Cataloged Procedure IBMZCPG . . . . . 219
14. Declaration of PLITABS . . . . . . . . 235
15. PAGELENGTH and PAGESIZE . . . . . 235
16. Sample JCL to compile, link, and invoke the
user exit . . . . . . . . . . . . . 241
17. Sample program to display program
arguments from the CPPL under TSO when
using SYSTEM(STD) option. . . . . . . 249
18. Sample program to display z/OS UNIX args
and environment variables . . . . . . . 250
19. Fixed-length records . . . . . . . . . 269
20. How the operating system completes the
DCB . . . . . . . . . . . . . . 274
21. Creating new libraries for compiled object
modules . . . . . . . . . . . . . 288
22. Placing a load module in an existing library 288
23. Creating a library member in a PL/I program 289
24. Updating a library member . . . . . . . 289
25. Creating a data set with stream-oriented data
transmission . . . . . . . . . . . . 296
26. Writing graphic data to a stream file . . . . 297
27. Accessing a data set with stream-oriented
data transmission . . . . . . . . . . 299
28. Creating a print file via stream data
transmission . . . . . . . . . . . . 302
29. PL/I structure PLITABS for modifying the
preset tab settings . . . . . . . . . . 303
30. American National Standard print and card
punch control characters (CTLASA) . . . . 313
31. Merge Sortcreating and accessing a
consecutive data set . . . . . . . . . 318
32. Printing record-oriented data transmission 320
33. Creating a REGIONAL(1) data set . . . . 326
34. Updating a REGIONAL(1) data set . . . . 328
35. VSAM data sets and allowed file attributes 338
36. Defining and loading an entry-sequenced
data set (ESDS) . . . . . . . . . . . 346
37. Updating an ESDS . . . . . . . . . . 347
38. Defining and loading a key-sequenced data
set (KSDS) . . . . . . . . . . . . 350
39. Updating a KSDS . . . . . . . . . . 352
40. Creating a Unique Key Alternate Index Path
for an ESDS . . . . . . . . . . . . 354
41. Creating a Nonunique Key Alternate Index
Path for an ESDS . . . . . . . . . . 355
42. Creating a unique Key Alternate Index Path
for a KSDS . . . . . . . . . . . . 356
43. Alternate Index Paths and Backward Reading
with an ESDS . . . . . . . . . . . 358
44. Using a Unique Alternate Index Path to
Access a KSDS . . . . . . . . . . . 360
45. Defining and loading a relative-record data
set (RRDS) . . . . . . . . . . . . 363
46. Updating an RRDS. . . . . . . . . . 365
47. Flow of control for Sort program . . . . . 388
48. Flowcharts for input and output handling
subroutines . . . . . . . . . . . . 398
49. Skeletal code for an input procedure 399
50. Skeletal code for an output handling
procedure. . . . . . . . . . . . . 400
51. PLISRTAsorting from input data set to
output data set . . . . . . . . . . . 401
52. PLISRTBsorting from input handling
routine to output data set . . . . . . . 402
53. PLISRTCsorting from input data set to
output handling routine . . . . . . . . 403
54. PLISRTDsorting from input handling
routine to output handling routine . . . . 404
55. Sorting varying-length records using input
and output handling routines . . . . . . 406
56. Simple type equivalence . . . . . . . . 409
57. Sample struct type equivalence . . . . . 410
58. Sample enum type equivalence . . . . . 411
59. Start of the C declaration for its FILE type 411
60. PL/I equivalent for a C file . . . . . . . 411
61. Sample code to use fopen and fread to dump
a file . . . . . . . . . . . . . . 412
62. Declarations for filedump program . . . . 412
63. C declaration of fread. . . . . . . . . 413
64. First incorrect declaration of fread . . . . 413
65. Second incorrect declaration of fread 413
66. Third incorrect declaration of fread . . . . 413
67. Code generated for RETURNS BYADDR 414
68. Correct declaration of fread . . . . . . . 414
69. Code generated for RETURNS BYVALUE 414
70. First incorrect declaration of fopen . . . . 415
71. Second incorrect declaration of fopen 415
72. Correct declaration of fopen . . . . . . 415
73. Optimal, correct declaration of fopen 415
74. Declaration of fclose . . . . . . . . . 416
75. Commands to compile and run filedump 416
76. Output of running filedump . . . . . . 416
77. Sample compare routine for C qsort function 416
78. Sample code to use C qsort function . . . . 417
79. Incorrect declaration of qsort . . . . . . 417
80. Correct declaration of qsort . . . . . . . 417
Copyright IBM Corp. 1999, 2011 xi
81. Code when parameters are BYADDR 418
82. Code when parameters are BYVALUE 419
83. Java Sample Program #2 - Passing a String 429
84. PL/I Sample Program #2 - Passing a String 431
85. Java Sample Program #3 - Passing an Integer 433
86. PL/I Sample Program #3 - Passing an Integer 435
87. Java Sample Program #4 - Receiving and
printing a String . . . . . . . . . . 436
88. PL/I Sample Program #4 - Calling the Java
Invocation API . . . . . . . . . . . 439
89. Sample XML document . . . . . . . . 445
90. PLISAXA coding example - type declarations 453
91. PLISAXA coding example - event structure 454
92. PLISAXA coding example - main routine 455
93. PLISAXA coding example - event routines 456
94. PLISAXA coding example - program output 464
95. Sample XML document . . . . . . . . 475
96. PLISAXC coding example - type declarations 485
97. PLISAXC coding example - event structure 486
98. PLISAXC coding example - main routine 487
99. PLISAXC coding example - event routines 488
100. PLISAXC coding example - program output 493
101. PLISAXD coding example - event routines 495
102. Output from PLISAXD sample. . . . . . 505
103. Example PL/I routine calling PLIDUMP 507
104. Declare for the saved options string . . . . 515
105. Using an ATTENTION ON-unit . . . . . 522
106. PL/I compiler user exit procedures . . . . 528
107. Example of an user exit input file . . . . . 529
108. Suppressing SQL messages . . . . . . . 534
109. Record types encoded as an ordinal value 550
110. Declare for the header part of a record 550
111. Declare for a summary record . . . . . . 551
112. Declare for a counter record . . . . . . 551
113. Declare for a literal record . . . . . . . 552
114. Declare for a file record . . . . . . . . 552
115. Declare for a message record . . . . . . 553
116. Declare for an ordinal type record . . . . 554
117. Declare for an ordinal element record 555
118. Symbol indices assigned to the elements of a
structure . . . . . . . . . . . . . 556
119. Data type of a variable . . . . . . . . 557
120. Declare for a symbol record. . . . . . . 558
121. Declare for xin_Bif_Kind . . . . . . . . 565
122. Declare for a source record . . . . . . . 571
123. Declare for a token record . . . . . . . 572
124. Declare for the token record kind . . . . . 572
125. Node indices assigned to the blocks in a
program . . . . . . . . . . . . . 573
126. Declare for a syntax record . . . . . . . 573
127. Declare for the syntax record kind . . . . 576
128. Node indices assigned to the syntax records
in a program. . . . . . . . . . . . 577
129. Declare for the expression kind . . . . . 578
130. Declare for the number kind . . . . . . 578
131. Declare for the lexeme kind. . . . . . . 579
132. Declare for the voc kind . . . . . . . . 580
xii Enterprise PL/I for z/OS Programming Guide
Introduction
About This Book . . . . . . . . . . . . xiii
Run-time environment for Enterprise PL/I for
z/OS . . . . . . . . . . . . . . . . xiii
Using your documentation . . . . . . . . . xiv
PL/I information . . . . . . . . . . . xiv
Language Environment information . . . . xiv
Notation conventions used in this book . . . . . xv
Conventions used . . . . . . . . . . . xv
How to read the syntax notation . . . . . . xv
How to read the notational symbols . . . . xviii
Example of notation . . . . . . . . . xviii
Summary of changes . . . . . . . . . . xviii
Enhancements in this release . . . . . . . xviii
Performance improvements . . . . . . xviii
Debugging improvements . . . . . . . xix
SQL support enhancements . . . . . . . xix
Usability enhancements . . . . . . . . xix
Debugging improvements . . . . . . . xx
Performance improvements . . . . . . . xx
Usability enhancements . . . . . . . . xx
Serviceability enhancements . . . . . . . xx
Performance improvements . . . . . . . xx
Usability enhancements . . . . . . . . xxi
Serviceability enhancements . . . . . . xxii
Enhancements from V3R8 . . . . . . . . xxii
Performance improvements . . . . . . xxii
Usability enhancements . . . . . . . . xxii
Enhancements from V3R7 . . . . . . . . xxiii
Debugging improvements . . . . . . . xxiii
Performance improvements . . . . . . xxiii
Usability enhancements . . . . . . . xxiii
Enhancements from V3R6 . . . . . . . . xxiv
DB2 V9 support . . . . . . . . . . xxiv
Debugging improvements . . . . . . . xxiv
Performance improvements . . . . . . xxv
Usability enhancements . . . . . . . . xxv
Enhancements from V3R5 . . . . . . . . xxv
Debugging improvements . . . . . . . xxv
Performance improvements . . . . . . xxvi
Usability enhancements. . . . . . . . xxvi
Enhancements from V3R4 . . . . . . . . xxvi
Migration enhancements . . . . . . . xxvi
Performance improvements . . . . . . xxvii
Usability enhancements . . . . . . . xxvii
Debugging improvements . . . . . . xxviii
Enhancements from V3R3 . . . . . . . xxviii
More XML support. . . . . . . . . xxviii
Improved performance . . . . . . . xxviii
Easier migration. . . . . . . . . . xxviii
Improved usability . . . . . . . . . xxviii
Improved debug support . . . . . . . xxix
Enhancements from V3R2 . . . . . . . . xxix
Improved performance . . . . . . . . xxix
Easier migration . . . . . . . . . . xxix
Improved usability . . . . . . . . . xxx
Enhancements from V3R1 . . . . . . . . xxx
Enhancements from VisualAge PL/I . . . . xxxi
How to send your comments . . . . . . . xxxii
Accessibility . . . . . . . . . . . . . xxxiii
Interface information . . . . . . . . . xxxiii
Keyboard navigation . . . . . . . . . xxxiii
Accessibility of this information . . . . . xxxiii
IBM and accessibility . . . . . . . . . xxxiii
About This Book
This book is for PL/I programmers and system programmers. It helps you
understand how to use Enterprise PL/I for z/OS in order to compile PL/I
programs. It also describes the operating system features that you might need to
optimize program performance or handle errors.
Important: Enterprise PL/I for z/OS will be referred to as Enterprise PL/I
throughout this book.
Run-time environment for Enterprise PL/I for z/OS
Enterprise PL/I uses Language Environment
as its run-time environment. It

conforms to Language Environment architecture and can share the run-time
environment with other Language Environment-conforming languages.
Language Environment provides a common set of run-time options and callable
services. It also improves interlanguage communication (ILC) between high-level
languages (HLL) and assembler by eliminating language-specific initialization and
termination on each ILC invocation.
Copyright IBM Corp. 1999, 2011 xiii
Using your documentation
The publications provided with Enterprise PL/I are designed to help you program
with PL/I. The publications provided with Language Environment are designed to
help you manage your run-time environment for applications generated with
Enterprise PL/I. Each publication helps you perform a different task.
The following tables show you how to use the publications you receive with
Enterprise PL/I and Language Environment. You'll want to know information
about both your compiler and run-time environment. For the complete titles and
order numbers of these and other related publications, see Bibliography on page
599.
PL/I information
Table 1. How to use Enterprise PL/I publications
To... Use...
Evaluate Enterprise PL/I Fact Sheet
Understand warranty information Licensed Programming Specifications
Plan for and install Enterprise PL/I Enterprise PL/I Program Directory
Understand compiler and run-time changes and
adapt programs to Enterprise PL/I and Language
Environment
Compiler and Run-Time Migration Guide
Prepare and test your programs and get details on
compiler options
Programming Guide
Get details on PL/I syntax and specifications of
language elements
Language Reference
Diagnose compiler problems and report them to
IBM
Diagnosis Guide
Get details on compile-time messages Compile-Time Messages and Codes
Language Environment information
Table 2. How to use z/OS Language Environment publications
To... Use...
Evaluate Language Environment Concepts Guide
Plan for Language Environment Concepts Guide
Run-Time Application Migration Guide
Install Language Environment on z/OS z/OS Program Directory
Customize Language Environment on z/OS Customization
Understand Language Environment program
models and concepts
Concepts Guide
Programming Guide
Find syntax for Language Environment run-time
options and callable services
Programming Reference
Develop applications that run with Language
Environment
Programming Guide and your language
Programming Guide
Debug applications that run with Language
Environment, get details on run-time messages,
diagnose problems with Language Environment
Debugging Guide and Run-Time Messages
Develop interlanguage communication (ILC)
applications
Writing Interlanguage Applications
Migrate applications to Language Environment Run-Time Application Migration Guide and the
migration guide for each Language
Environment-enabled language
xiv Enterprise PL/I for z/OS Programming Guide
Notation conventions used in this book
This book uses the conventions, diagramming techniques, and notation described
in Conventions used and How to read the notational symbols on page xviii to
illustrate PL/I and non-PL/I programming syntax.
Conventions used
Some of the programming syntax in this book uses type fonts to denote different
elements:
v Items shown in UPPERCASE letters indicate key elements that must be typed
exactly as shown.
v Items shown in lowercase letters indicate user-supplied variables for which you
must substitute appropriate names or values. The variables begin with a letter
and can include hyphens, numbers, or the underscore character (_).
v The term digit indicates that a digit (0 through 9) should be substituted.
v The term do-group indicates that a do-group should be substituted.
v Underlined items indicate default options.
v Examples are shown in monocase type.
v Unless otherwise indicated, separate repeatable items from each other by one or
more blanks.
Note: Any symbols shown that are not purely notational, as described in How to
read the notational symbols on page xviii, are part of the programming
syntax itself.
For an example of programming syntax that follows these conventions, see
Example of notation on page xviii.
How to read the syntax notation
The following rules apply to the syntax diagrams used in this book:
Arrow symbols
Read the syntax diagrams from left to right, from top to bottom, following
the path of the line.
Indicates the beginning of a statement.
Indicates that the statement syntax is continued on the next line.
Indicates that a statement is continued from the previous line.
Indicates the end of a statement.
Diagrams of syntactical units other than complete statements start with the
symbol and end with the symbol.
Conventions
v Keywords, their allowable synonyms, and reserved parameters, appear
in uppercase for MVS and OS/2
platforms, and lowercase for UNIX
platforms. These items must be entered exactly as shown.

v Variables appear in lowercase italics (for example, column-name). They
represent user-defined parameters or suboptions.
v When entering commands, separate parameters and keywords by at
least one blank if there is no intervening punctuation.
v Enter punctuation marks (slashes, commas, periods, parentheses,
quotation marks, equal signs) and numbers exactly as given.
Introduction xv
v Footnotes are shown by a number in parentheses, for example, (1).
v A symbol indicates one blank position.
Required items
Required items appear on the horizontal line (the main path).
REQUIRED_ITEM
Optional Items
Optional items appear below the main path.
REQUIRED_ITEM
optional_item
If an optional item appears above the main path, that item has no effect on
the execution of the statement and is used only for readability.
REQUIRED_ITEM
optional_item
Multiple required or optional items

If you can choose from two or more items, they appear vertically in a
stack. If you must choose one of the items, one item of the stack appears on
the main path.
REQUIRED_ITEM required_choice1
required_choice2
If choosing one of the items is optional, the entire stack appears below the
main path.
REQUIRED_ITEM
optional_choice1
optional_choice2
Repeatable items
An arrow returning to the left above the main line indicates that an item
can be repeated.
REQUIRED_ITEM

repeatable_item
If the repeat arrow contains a comma, you must separate repeated items
with a comma.
REQUIRED_ITEM

,
repeatable_item
A repeat arrow above a stack indicates that you can specify more than one
of the choices in the stack.
xvi Enterprise PL/I for z/OS Programming Guide
Default keywords
IBM-supplied default keywords appear above the main path, and the
remaining choices are shown below the main path. In the parameter list
following the syntax diagram, the default choices are underlined.
REQUIRED_ITEM
default_choice
optional_choice
optional_choice
Fragments
Sometimes a diagram must be split into fragments. The fragments are
represented by a letter or fragment name, set off like this: | A |. The
fragment follows the end of the main diagram. The following example
shows the use of a fragment.
STATEMENT item 1 item 2 A
A:
item 3
item 4
KEYWORD
item 5
item 6
Substitution-block
Sometimes a set of several parameters is represented by a
substitution-block such as <A>. For example, in the imaginary /VERB
command you could enter /VERB LINE 1, /VERB EITHER LINE 1, or /VERB
OR LINE 1.
/VERB
<A>
LINE line#
where <A> is:
EITHER
OR
Parameter endings
Parameters with number values end with the symbol '#', parameters that
are names end with 'name', and parameters that can be generic end with
'*'.
/MSVERIFY MSNAME msname
SYSID sysid#
The MSNAME keyword in the example supports a name value and the
SYSID keyword supports a number value.
Introduction xvii
How to read the notational symbols
Some of the programming syntax in this book is presented using notational
symbols. This is to maintain consistency with descriptions of the same syntax in
other IBM publications, or to allow the syntax to be shown on single lines within a
table or heading.
v Braces, { }, indicate a choice of entry. Unless an item is underlined, indicating a
default, or the items are enclosed in brackets, you must choose at least one of
the entries.
v Items separated by a single vertical bar, |, are alternative items. You can select
only one of the group of items separated by single vertical bars. (Double vertical
bars, ||, specify a concatenation operation, not alternative items. See the PL/I
Language Reference for more information on double vertical bars.)
v Anything enclosed in brackets, [ ], is optional. If the items are vertically stacked
within the brackets, you can specify only one item.
v An ellipsis, ..., indicates that multiple entries of the type immediately preceding
the ellipsis are allowed.
Example of notation
The following example of PL/I syntax illustrates the notational symbols described
In How to read the notational symbols:
DCL file-reference FILE STREAM
{INPUT | OUTPUT [PRINT]}
ENVIRONMENT(option ...);
Interpret this example as follows:
v You must spell and enter the first line as shown, except for file-reference, for
which you must substitute the name of the file you are referencing.
v In the second line, you can specify INPUT or OUTPUT, but not both. If you
specify OUTPUT, you can optionally specify PRINT as well. If you do not
specify either alternative, INPUT takes effect by default.
v You must enter and spell the last line as shown (including the parentheses and
semicolon), except for option ..., for which you must substitute one or more
options separated from each other by one or more blanks.
Summary of changes
Enhancements in this release
This release provides the following functional enhancements described in this and
the other IBM PL/I books.
Performance improvements
v The ARCH(9) option provides further exploitation of the IBM zEnterprise
196
(z196) System instructions, including the high-word, floating-point extension,
and population count facilities.
v The new UNROLL compiler option gives you control of loop unrolling.
v The compiler now generates inline code to resolve the ULENGTH and USUBSTR
built-in functions for character strings.
v The compiler now generates inline code for MEMINDEX(p, n, x) where x is
WCHAR(1).
v The compiler now generates inline code for STG(x) where x is BASED variable
using REFER and meets both of the following conditions:
All NONVARYING BIT in x are specified with the ALIGNED attribute.
xviii Enterprise PL/I for z/OS Programming Guide
All other elements in x are specified with UNALIGNED.
Debugging improvements
v The compiler now supports typed structures in Debug Tool.
SQL support enhancements
v The SQL Preprocessor has the following significant changes:
It fully and correctly supports block scoping.
The preprocessor load module is more than eight times smaller.
The preprocessor runs faster.
It supports the use of the SQL TYPE attribute anywhere you can specify a
PL/I data type.
It now supports the following compiler options when processing declarations
of host variables. It applies the defaults correctly and rejects unsuitable host
variables as appropriate.
- DEFAULT(ANS | IBM)
- DEFAULT(ASCII | EBCDIC)
- DEFAULT((NO)EVENDEC)
- DEFAULT((NON)NATIVE)
- DEFAUT(SHORT(HEX | IEEE))
- RULES((NO)LAXCTL)
It now correctly processes the PRECISION attribute.
It now recognizes the UNSIGNED and COMPLEX attributes and rejects the
use of these attributes in any host variable.
It now ensures that DSNHMLTR is declared in the outermost procedure that
contains codes that need DSNHMLTR.
It now handles packages correctly.
It now ensures that characters in the source code are printable when it
generates code to set the SQLAVDAID.
It no longer requires that indicator arrays have a lower bound of 1.
It now generates an SQL parameter list structure that has fewer unions, fewer
init clauses, and no additional declarations based on elements of the structure.
Usability enhancements
v The new PPLIST compiler option conditionally erases the part of the listing that
is generated by any preprocessor phase that produces no messages.
v The compiler issues better messages when a comma is missing in a structure
declaration.
v The compiler issues new messages when the source code contains invalid
shift-in and shift-out bytes.
v The compiler applies the NONASSIGNABLE attribute to any parameter that is
declared with the INONLY attribute; it flags any assignment to a parameter that
is declared as INONLY.
v Under RULES(NOLAXENTRY), the compiler does not flag names that start with
DSN.
v Under RULES(NOUNREF), the compiler does not flag names that start with
DSN or SQL.
v Under the new NOSELFASSIGN suboption of the RULES compiler option, the
compiler flags assignments of variables to themselves.
Introduction xix
v Under the new NOLAXRETURN suboption of the RULES compiler option, the
compiler generates codes to raise the ERROR condition when a RETURN
statement is used in some invalid ways.
Enhancements from V4R1
v Under TEST(SEPARATE), the compiler optionally places the statement number
table in the debug file and thus reduces the size of the generated object deck.
v Under TEST(SEPARATE), the compiler includes information identifying the
source lines for declarations, references, and assignments.
v Under TEST(SEPARATE), the compiler generates information to identify the
implicit locator reference when a BASED variable is based on the ADDR of an
array element or other complex references.
v The ARCH(9) option provides exploitation of the IBM zEnterprise System
instructions.
v If all the elements of structures using REFER are byte-aligned, the compiler
inlines code to resolve references to these elements rather than through library
calls.
v The new DEPRECATE compiler option flags variable names and included file
names that you want to deprecate.
v The new SEPARATE suboption of the GONUMBER option is provided to place
the generated statement number table in the separate debug file.
v The new NOGLOBALDO suboption of the RULES option flags all DO loop
control variables that are declared in the parent block.
v The new NOPADDING suboption of the RULES option flags all structures that
contain padding.
v The SQL preprocessor now supports the XREF option.
v The new PLISAXD built-in subroutine provides the ability to parse XML
documents with validation against a schema by using the XML System Services
parser.
Serviceability enhancements
v The use of either the GOSTMT option or the IMPRECISE option is now flagged
as unsupported.
v The compiler now always lists all SQL preprocessor options that are in effect.
v UVALID will now be inlined for for strings of length 256 or less.
v Under ARCH(7) and higher, the CU12, CU14, CU21, CU24, CU41, and CU42
instructions will be used to provide for fast conversions between UTF-8, UFT-16
and UTF-32.
xx Enterprise PL/I for z/OS Programming Guide
v Under ARCH(7) and higher, the TRTT, TROT, TRTO, and TROO instructions will
be used to provide for fast translations between one- and two-byte buffers.
v Assignments of like arrays of scalars will now be handled as storage copy
operations.
v All assignments of BIT VARYING to BIT VARYING will now be inlined.
v All assignments of byte-aligned BIT NONVARYING to BIT VARYING will now
be inlined.
v The ROUND and ROUNDDEC built-in functions will be inlined when the
argument to be rounded is DFP.
v To simplify the choosing of options for best performance,
the COMPACT option has been dropped
The default setting for DEFAULT(REORDER | ORDER) has been changed to
DEFAULT(REORDER)
the TUNE option has been dropped
v Detection of the dereferencing of null pointers exploits the new
compare-and-trap instruction under ARCH(8).
v The compiler has been built with ARCH(6) to improve its performance.
v The CICS preprocessor now supports block-scoping and consequently will add
the needed local CICS declares to all non-nested procedures.
v The SQL preprocessor now supports the PL/I rules for the scope of declarations
when resolving host variable references through the new SCOPE option.
NOSCOPE is the default for compatibility with previous releases.
v The MACRO preprocessor will now leave %include, %xinclude, %inscan, and
%xinscan statements in the compiler listing as comments.
v The MACRO preprocessor now provides via the %DO SKIP; statement an easy
and clear way to omit sections of code from a compile.
v The MACRO preprocessor now supports an option called NAMEPREFIX which
allows the user to force macro procedures and variables to start with a specified
character.
v The IGNORE compiler option provides the ability to suppress PUT FILE and/or
DISPLAY statements (either of which may have been used for debug purposes
but which should be compiled out of the production version).
v The NULLSTRPTR suboption of the DEFAULT compiler option provides user
control of whether sysnull or null is assigned to a pointer when the source in the
assignment is a null string.
v The new MAXGEN option specifies the maximum number of intermediate
language statements that should be generated for any one user statement and
will cause the compiler to flag any statement where this maximum is exceeded.
v The new ONSNAP option will allow the user to request the compiler to insert
an ON STRINGRANGE SNAP; or an ON STRINGSIZE SNAP; statement into the
prologue of a MAIN or FROMALIEN proc.
v The new SHORT suboption of the INITAUTO option will limit the INITAUTO
option so that it does not duplicate all of the runtime STORAGE option, but
does initialize variables that might be optimized to registers.
v The new RTCHECK option will generate code to test for the dereferencing of
null pointers.
v The compiler will now flag various statements that may be risky:
code where the result of a FIXED operation has a scale factor less than zero
Introduction xxi
ENTRYs used as functions but declared without the RETURNS attribute
parameters declared as BYVALUE when doing so is ill-advised, e.g. declaring
a FIXED DEC parameter BYVALUE
FIXED DECIMAL add and multiply operations that might raise
FIXEDOVERFLOW
v The RULES option has been expanded to allow more control over and flagging
of poor code:
NOPROCENDONLY will flag END statements for PROCs that don't name the
PROC they are closing
NOSTOP will flag the use of STOP and EXIT
NOLAXQUAL(STRICT) will flag variables not qualified with their level-1
name
NOLAXSCALE will flag declares of FIXED DEC(p,q) and FIXED BIN(p,q)
where q < 0 or p < q
NOGOTO(LOOSE) will allow GOTOs only if in the same block
More than one DELAY STATEMENTS can be concurrently executed in
different procedures.
Serviceability enhancements
v When the compiler cannot open a file, the compiler will now, if possible, also
include the related C runtime message in the message in the listing.
v If user code requires a DFP conversion at compile time but the compile is
running on a machine without DFP hardware, this error will be trapped and a
meaningful error issued.
v If the SQL preprocessor is invoked more than once without INCONLY as its
suboption, then the DBRM library created by the compiler will be empty, and
now an E-level message will be issued to warn the user about this problem.
v The ARCH(8) and TUNE(8) options provide exploitation of the z/HE
instructions.
v The HGPR option supports using 64-bit registers in 32-bit code.
v The GOFF option supports generation of GOFF objects.
v The PFPO instruction will be exploited in conversions between differing float
formats.
v SRSTU will be used in code generated for UTF-16 INDEX.
v Calls to null internal procedures will now be completely removed.
v PLISAXC provides for access to the XML System Services parser via a SAX
interface.
v The INCDIR option is supported under batch.
v The LISTVIEW option now provides the support previously provided by the
AFTERMACRO etc suboptions of TEST.
v The NOLAXENTRY suboption of the RULES option allows for the flagging of
unprototyped ENTRYs.
xxii Enterprise PL/I for z/OS Programming Guide
v The (NO)FOFLONMULT suboption of the DECIMAL option allows for control
of whether FOFL is raised in MULTIPLY of FIXED DECIMAL.
v The HEX and SUBSTR suboptions of the USAGE option provide more user
control of the behavior of the corresponding built-in functions.
v The DDSQL compiler option provides the ability to specify an alternate DD
name to be used for EXEC SQL INCLUDEs.
v The INCONLY suboption provided by the MACRO and SQL preprocessors can
request those preprocessors to perform only INCLUDEs.
v The integrated SQL preprocessor will now generate DB2 precompiler style
declares for all *LOB_FILE, *LOCATOR, ROWID, BINARY and VARBINARY
SQL types, in addition to the BLOB, CLOB and DBCLOB SQL types already
supported, when the LOB(DB2) SQL preprocessor option is selected.
v The TEST option has been enhanced so that users can choose to view the source
in the listing and in the Debug Tool source window as that source would appear
after a user-specified preprocessor had been run (or after all the preprocessors
had been run).
v The BASR instruction will now be used instead of the BALR instruction.
v The conversions of FIXED DEC with large precision to FLOAT will be inlined
and speeded up by the use of FIXED BIN(63) as an intermediary.
v The CHAR built-in when applied to CHAR expressions will now always be
inlined.
v The code generated for conversions of FIXED BIN(p,q) to unscaled FIXED DEC
has been significantly improved.
v TRTR will be used, under ARCH(7), for SEARCHR and VERIFYR in the same
situations where TRT would be used for SEARCH and VERIFY.
v UNPKU will be used to convert some PICTURE to WIDECHAR (rather than
making a library call).
v IEEE Decimal Floating-Point (DFP) is supported.
v The new MEMCONVERT built-in function will allow the user to convert
arbitrary lengths of data between arbitrary code pages.
v The new ONOFFSET built-in function will allow the user to have easy access to
another piece of information formerly available only in the runtime error
message or dump, namely the offset in the user procedure at which a condition
was raised.
v The new STACKADDR built-in function will return the address of the current
dynamic save area (register 13 on z/OS) and will make it easier for users to
write their own diagnostic code.
v The length of the mnemonic field in the assembler listing will be increased to
allow for better support of the new z/OS instructions that have long
mnemonics.
v More of the right margin will be used in the attributes, cross-reference and
message listings.
Introduction xxiii
v The CODEPAGE option will now accept 1026 (the Turkish code page) and 1155
(the 1026 code page plus the Euro symbol).
v The new MAXNEST option allows the user to flag excessive nesting of BEGIN,
DO, IF and PROC statements.
v Under the new (and non-default) suboption NOELSEIF of the RULES option, the
compiler will flag any ELSE statement that is immediately followed by an IF
statement and suggest that it be rewritten as a SELECT statement.
v Under the new (and non-default) suboption NOLAXSTG of the RULES option,
the compiler will flag declares where a variable A is declared as BASED on
ADDR(B) and STG(A) > STG(B) not only (as the compiler did before) when B is
AUTOMATIC, BASED or STATIC with constant extents but now also when B is
a parameter declared with constant extents.
v The new QUOTE option will allow the user to specify alternate code points for
the quote (") symbol since this symbol is not code-page invariant.
v The new XML compiler option can be used to specify that the tags in the output
of the XMLCHAR built-in function be either in all upper case or in the case in
which they were declared.
v For compilations that produce no messages, the compiler will now include a line
saying "no compiler messages" where the compiler messages would have been
listed.
v The MACRO preprocessor will support a new suboption that will allow the user
to specify whether it should process only %INCLUDE statements or whether it
should process all macro statements.
v The integrated SQL preprocessor will now generate DB2 precompiler style
declares for all *LOB_FILE, *LOCATOR, ROWID, BINARY and VARBINARY
SQL types, in addition to the BLOB, CLOB and DBCLOB SQL types already
supported, when the LOB(DB2) SQL preprocessor option is selected.
DB2 V9 support
v Support for STDSQL(YES/NO)
v Support for CREATE TRIGGER (aka multiple SQL statements)
v Support for FETCH CONTINUE
v Support for SQL style comments ('--') embedded in SQL statements
v Support for additional SQL TYPES including
SQL TYPE IS BLOB_FILE
SQL TYPE IS CLOB_FILE
SQL TYPE IS DBCLOB_FILE
SQL TYPE IS XML AS
SQL TYPE IS BIGINT
SQL TYPE IS BINARY
SQL TYPE IS VARBINARY
v The SQL preprocessor will also now list the DB2 coprocessor options
v Under TEST(NOSEPNAME), the name of the debug side file will not be saved in
the object deck.
xxiv Enterprise PL/I for z/OS Programming Guide
v Support, under ARCH(7), of the z/OS extended-immediate facility
v Exploitation of the CLCLU, MVCLU, PKA, TP and UNPKA instructions under
ARCH(6)
v Exploitation of the CVBG and CVDG instructions under ARCH(5)
v Expanded use of CLCLE
v Conversions involving DB2 date-time patterns are now inlined
v The ALLOCATION built-in function is now inlined
v Conversions with a drifting $ have been inlined
v Conditional code has been eliminated from assignments to PIC'(n)Z'
v Conversions from FIXED BIN to a PICTURE that specifies a scale factor have
been inlined
v Assignments to BIT variables that have inherited dimensions but which have
strides that are divisible by 8 will now be inlined
v The MAP output will now also include a list in order of storage offset (per
block) of the AUTOMATIC storage used by the block
v Conformance checking has been extended to include structures
v Listings will now include 7 columns for the line number in a file
v The THREADID built-in function is now supported under z/OS
v The PICSPEC built-in function is now supported
v The new CEESTART option allows you to position the CEESTART csect at the
start or end of the object deck
v The new PPCICS, PPMACRO and PPSQL options allow you to specify the
default options to be used by the corresponding preprocessor
v The ENVIRONMENT option is now included in the ATTRIBUTES listing
v The DISPLAY option now supports a suboption to allow different DESC codes
for DISPLAY with REPLY versus DISPLAY without REPLY
v The message flagging a semicolon in a comment will now include the line
number of the line where the semicolon appears
v Flag assignments that can change a REFER object
v Flag use of KEYED DIRECT files without a KEY/KEYFROM clause
v Flag use of PICTURE as loop control variables
v Under TEST(SEPARATE), the vast majority of debug information will be written
to a separate debug file
v AUTOMONITOR will include the target in assignments
v The AT ENTRY hook will now be placed after AUTOMATIC has been initialized,
thereby eliminating the need to step into the block before looking at any
variables
Introduction xxv
v Generation of branch-relative instructions so that the need for base registers and
transfer vectors will be significantly eliminated
v Support, under ARCH(6), of the z/OS long displacement facility
v Simple structures using REFER will be mapped inlined rather than via a library
call
v For structures using REFER still mapped via a library call, less code will be
generated if the REFER specifies the bound for an array of substructures
v Faster processing of duplicate INCLUDEs
v Conversions to PICTURE variables with an I or R in the last position will now
be inlined (such conversions had already been inlined when the last character
was a T)
v Conversions to PICTURE variables ending with one or more B's will now be
inlined if the corresponding picture without the B's would have been inlined
v Conversions to from CHARACTER to PICTURE variables consisting only of X's
will now be inlined
v All parts of the listing, etc will count the source file as file 0, the first include file
as file 1, the second (unique) include file as file 2, etc
v Conformance checking extended to include arrays
v Listings will include the build dates for any preprocessors invoked
v One-byte FIXED BINARY arguments can be suppressed for easier ILC with
COBOL
v Alternate DD names may be specified for SYSADATA, SYSXMLSD and
SYSDEBUG
v RULES(NOLAXMARGINS) tolerates, under XNUMERIC, sequence numbers
v RULES(NOUNREF) flags unreferenced AUTOMATIC variables
v If an assignment to a variable is done via library call, the message flagging the
library call will include the name of the target variable
v Flag one-time DO loops
v Flag labels used as arguments
v Flag ALLOCATE and FREE of non-PARAMETER CONTROLLED in
FETCHABLE if PRV used
v Flag DEFINED and BASED larger than their base even if the base is declared
later
v Flag implicit FIXED DEC to 8-byte integer conversions
Migration enhancements
v Support sharing of CONTROLLED with old code
v Improve default initialization
v Ease decimal precision specification in ADD, DIVIDE and MULTIPLY
v Support old semantics for STRING of GRAPHIC
v Support old semantics for the DEFAULT statement
v Flag declares with storage overlay
xxvi Enterprise PL/I for z/OS Programming Guide
v Lift restrictions on RETURN inside BEGIN
v Optionally flag semicolons in comments
v Support EXT STATIC initialized in assembler
v Flag invalid carriage control characters
v Flag more language misuse, especially with RETURN
v Support the REPLACEBY2 built-in function
v Optionally suppress FOFL on decimal assignments that would raise SIZE
v Flag more language handled differently than the old compiler
v Improve code generated for INDEX and TRANSLATE
v Inline more assignments to pictures
v Improve the code generated for conversions of CHARACTER to PICTURE when
the conversion would be done inline if the source were FIXED DEC
v Improve the code generated for packed decimal conversions
v Improve the code generated for some uses of REFER
v Inline compares of character strings of unknown length
v Reduce the amount of stack storage used for concatenates
v Inline more GET/PUT STRING EDIT statements
v Short-circuit more LE condition handling
v Inline more Or and And of BIN FIXED
v Inline SIGNED FIXED BIN(8) to ALIGNED BIT(8)
v Flag statements where the compiler generates a call to a library routine to map a
structure at run time
v Lessen the amount of I/O used to produce the listing
v Optionally provide offsets in the AGGREGATE listing in hex
v Support DEC(31) only when needed via the LIMITS(FIXEDDEC(15,31)) option
v Allow comments in options
v Optionally flag FIXED DEC declares with even precision
v Optionally flag DEC to DEC assignments that could raise SIZE
v Flag DEC/PIC to PIC assignments that could raise SIZE
v Support LIKE without INIT via the NOINIT attribute
v Ease includes from PDSs under z/OS UNIX
v Support the LOWERCASE, MACNAME, TRIM and LOWERCASE built-in
functions in the MACRO preprocessor
v Ease introduction of options via PTF
v Optionally disallow use of *PROCESS
v Optionally keep *PROCESS in MDECK
v Support one-time INCLUDE
v Support macro-determined INCLUDE name
v Support runtime string parameter checking
v Flag more possibly uninitialized variables
v Flag unusual compares that are likely to be coding errors
Introduction xxvii
v The output of the STORAGE option is now formatted more nicely, and the
output of the LIST option will now include the hex offset for each block from
the start of the compilation unit.
v Better support for overlay hooks
v Easier resolution of CONTROLLED variables in LE dumps
v Always include user specified options in the listing
This release also provides all of the functional enhancements offered in Enterprise
PL/I V3R3, including the following:
More XML support
The XMLCHAR built-in function will write XML with the names and values of the
elements of a referenced structure to a buffer and return the number of bytes
written. This XML can then be passed to other applications, including code using
the PL/I SAX parser, which want to consume it.
Improved performance
v The compilation time under OPT(2) will be significantly less than under
Enterprise PL/I V3R2, especially for large programs.
v The compiler now uses the ED and EDMK instructions for inlined numeric
conversions to PICTURE and CHARACTER. This results in faster, shorter code
sequences and also in faster compilations.
v The compiler now generates better code for string comparisons. This also results
in faster, shorter code sequences.
v The compiler now generates shorter, faster code for conversion from FIXED
DECIMAL to PICTURE with trailing overpunch characters.
v The ARCH and TUNE compiler options now accept 5 as a valid sub-option.
Under ARCH(5), the compiler will generate, when appropriate, some new
z/Architecture instructions such as NILL, NILH, OILL, OILH, LLILL, and
LLILH.
Easier migration
v The new BIFPREC compiler option controls the precision of the FIXED BIN
result returned by various built-in functions and thus provides for better
compatibility with the OS PL/I compiler.
v The new BACKREG compiler option controls which register the compiler uses as
the backchain register and thus allows for easier mixing of old and new object
code.
v The new RESEXP compiler option controls the evaluation of restricted
expressions in code, and thus provides for better compatibility with the OS PL/I
compiler.
v The new BLKOFF compiler option provides for controlling the way offsets in the
compiler's pseudo-assembler listing are calculated.
v The STORAGE compiler option causes the compiler to produce, as part of the
listing, a summary, similar to that produced by the OS PL/I compiler, of the
storage used by each procedure and begin-block.
Improved usability
v The new LAXDEF suboption of the RULES compiler option allows the use of
so-called illegal defining without having the compiler generate E-level messages.
xxviii Enterprise PL/I for z/OS Programming Guide
v The new FLOATINMATH compiler option offers easier control of the precision
with which math functions are evaluated.
v The new MEMINDEX, MEMSEARCH(R) and MEMVERIFY(R) built-in functions
provide the ability to search strings larger than 32K.
v The new ROUTCDE and DESC suboptions of the DISPLAY(WTO) compiler
option offers control of the corresponding elements of the WTO.
v The compiler now stores in each object a short string that will be in storage
even when the associated code runs and that records all the options used to
produce that object. This allows various tools to produce better diagnostics.
v The compiler now issues messages identifying more of the places where
statements have been merged or deleted.
v The PLIDUMP output now includes a hex dump of user static.
v The PLIDUMP output now includes the options used to compile each program
in the Language Environment traceback.
v The PLIDUMP output now includes more information on PL/I files.
Improved debug support
v BASED structures using REFER are now supported in DebugTool and in
data-directed I/O statements (with the same restrictions as on all other BASED
variables).
v BASED structures that are BASED on scalar members of other structures
(which, in turn, may be BASED, etc) are now supported in DebugTool and in
data-directed I/O statements (with the same restriction as on all other BASED
variables).
Improved performance
v The compiler now handles even more conversions by generating inline code
which means these conversions will be done much faster than previously. Also,
all conversions done by library call are now flagged by the compiler.
v The compiler-generated code now uses, in various situations, less stack storage.
v The compiler now generates much better code for references to the TRANSLATE
built-in function.
v The compiler-generated code for SUBSCRIPTRANGE checking is now, for arrays
with known bounds, twice as fast as before.
v The ARCH and TUNE options now support 4 as a suboption, thereby allowing
exploitation of instructions new to the zSeries machines.
v ARCH(2), FLOAT(AFP) and TUNE(3) are now the default.
Easier migration
v Compiler defaults have been changed for easier migration and compatibility. The
changed defaults are:
CSECT
CMPAT(V2)
LIMITS(EXTNAME(7))
NORENT
v The compiler now honors the NOMAP, NOMAPIN and NOMAP attributes for
PROCs and ENTRYs with OPTIONS(COBOL).
Introduction xxix
v The compiler now supports PROCs with ENTRY statements that have differing
RETURNS attribute in the same manner as did the old host compiler.
v The compiler will now assume OPTIONS(RETCODE) for PROCs and ENTRYs
with OPTIONS(COBOL).
v The SIZE condition is no longer promoted to ERROR if unhandled.
v Various changes have been made to reduce compile time and storage
requirements.
v The OFFSET option will now produce a statement offset table much like the
ones it produced under the older PL/I compilers.
v The FLAG option now has exactly the same meaning as it had under the old
compilers, while the new MAXMSG option lets you decide if the compiler
should terminate after a specified number of messages of a given severity. For
example, with FLAG(I) MAXMSG(E,10), you can now ask to see all I-level
messages while terminating the compilation after 10 E-level messages.
v The AGGREGATE listing now includes structures with adjustable extents.
v The STMT option is now supported for some sections of the listing.
v The maximum value allowed for LINESIZE has been changed to 32759 for
F-format files and to 32751 for V-format files.
Improved usability
v The defaults for compiler options may now be changed at installation.
v The integrated SQL preprocessor now supports DB2 Unicode.
v The compiler now generates information that allows Debug Tool to support
Auto Monitor, whereby immediately before each statement is executed, all the
values of all the variables used in the statement are displayed.
v The new NOWRITABLE compiler option lets you specify that even under
NORENT and at the expense of optimal performance, the compiler should use
no writable static when generating code to handle FILEs and CONTROLLED.
v The new USAGE compiler option gives you full control over the IBM or ANS
behavior of the ROUND and UNSPEC built-in function without the other effects
of the RULES(IBM|ANS) option.
v The new STDSYS compiler option lets you specify that the compiler should
cause the SYSPRINT file to be equated to the C stdout file.
v The new COMPACT compiler option lets you direct the compiler to favor those
optimizations which tend to limit the growth of the code.
v The LRECL for SYSPRINT has been changed to 137 to match that of the C/C++
compiler.
v POINTERs are now allowed in PUT LIST and PUT EDIT statements: the 8-byte
hex value will be output.
v If specified on a STATIC variable, the ABNORMAL attribute will cause that
variable to be retained even if unused.
v Support for multithreading on z/OS
v Support for IEEE floating-point on z/OS
v Support for the ANSWER statement in the macro prepreprocessor
v SAX-style XML parsing via the PLISAXA and PLISAXB built-in subroutines
v Additional built-in functions:
xxx Enterprise PL/I for z/OS Programming Guide
CS
CDS
ISMAIN
LOWERCASE
UPPERCASE
Enhancements from VisualAge PL/I
This release also provides all of the functional enhancements offered in VisualAge
v Initial UTF-16 support via the WIDECHAR attribute
There is currently no support yet for
WIDECHAR characters in source files
W string constants
use of WIDECHAR expressions in stream I/O
implicit conversion to/from WIDECHAR in record I/O
implicit endianness flags in record I/O
If you create a WIDECHAR file, you should write the endianness flag
('fe_ff'wx) as the first two bytes of the file.
v DESCRIPTORS and VALUE options supported in DEFAULT statements
v PUT DATA enhancements
POINTER, OFFSET and other non-computational variables supported
Type-3 DO specifications allowed
Subscripts allowed
v DEFINE statement enhancements
Unspecified structure definitions
CAST and RESPEC type functions
v Additional built-in functions:
CHARVAL
ISIGNED
IUNSIGNED
ONWCHAR
ONWSOURCE
WCHAR
WCHARVAL
WHIGH
WIDECHAR
WLOW
v Preprocessor enhancements
Support for arrays in preprocessor procedures
WHILE, UNTIL and LOOP keywords supported in %DO statements
%ITERATE statement supported
%LEAVE statement supported
%REPLACE statement supported
%SELECT statement supported
Additional built-in functions:
- COLLATE
- COMMENT
- COMPILEDATE
- COMPILETIME
- COPY
Introduction xxxi
- COUNTER
- DIMENSION
- HBOUND
- INDEX
- LBOUND
- LENGTH
- MACCOL
- MACLMAR
- MACRMAR
- MAX
- MIN
- PARMSET
- QUOTE
- REPEAT
- SUBSTR
- SYSPARM
- SYSTEM
- SYSVERSION
- TRANSLATE
- VERIFY
How to send your comments
Your feedback is important in helping us to provide accurate, high-quality
information. If you have comments about this document or any other PL/I
documentation, contact us in one of these ways:
v Use the Online Readers' Comment Form at
www.ibm.com/software/awdtools/rcf/
or send an e-mail to
[email protected]
Be sure to include the name of the document, the publication number of the
document, the version of PL/I, and, if applicable, the specific location (for
example, page number) of the text that you are commenting on.
v Fill out the Readers' Comment Form at the back of this document, and return it
by mail or give it to an IBM representative. If the form has been removed,
address your comments to:
International Business Machines Corporation
Reader Comments
H150/090
555 Bailey Avenue
San Jose, CA 95141-1003
USA
v Fax your comments to this U.S. number: (800)426-7773.
When you send information to IBM, you grant IBM a nonexclusive right to use or
distribute the information in any way it believes appropriate without incurring any
obligation to you.
xxxii Enterprise PL/I for z/OS Programming Guide
Accessibility
Accessibility features help users who have a disability, such as restricted mobility
or limited vision, to use information technology products successfully. The
accessibility features in z/OS provide accessibility for Enterprise PL/I.
The major accessibility features in z/OS are:
v Interfaces that are commonly used by screen readers and screen-magnifier
software
v Keyboard-only navigation
v Ability to customize display attributes such as color, contrast, and font size
Interface information
Assistive technology products work with the user interfaces that are found in
z/OS. For specific guidance information, see the documentation for the assistive
technology product that you use to access z/OS interfaces.
Keyboard navigation
Users can access z/OS user interfaces by using TSO/E or ISPF. For information
about accessing TSO/E or ISPF interfaces, see the following publications:
v z/OS TSO/E Primer
v z/OS TSO/E Primer
v z/OS TSO/E Primer
These guides describe how to use TSO/E and ISPF, including the use of keyboard
shortcuts or function keys (PF keys). Each guide includes the default settings for
the PF keys and explains how to modify their functions.
Accessibility of this information
The English-language XHTML format of this information that will be provided in
the IBM System z
Enterprise Development Tools & Compilers Information Center

at publib.boulder.ibm.com/infocenter/pdthelp/index.jsp is accessible to visually
impaired individuals who use a screen reader.
To enable your screen reader to accurately read syntax diagrams, source code
examples, and text that contains the period or comma PICTURE symbols, you must
set the screen reader to speak all punctuation.
IBM and accessibility
See the IBM Human Ability and Accessibility Center at www.ibm.com/able for
more information about the commitment that IBM
has to accessibility.
Accessibility
Introduction xxxiii
Accessibility
xxxiv Enterprise PL/I for z/OS Programming Guide
Part 1. Compiling your program
Chapter 1. Using compiler options and facilities . 5
Compile-time option descriptions . . . . . . . 5
AGGREGATE . . . . . . . . . . . . . 9
ARCH . . . . . . . . . . . . . . . 10
ATTRIBUTES . . . . . . . . . . . . . 12
BACKREG. . . . . . . . . . . . . . 13
BIFPREC . . . . . . . . . . . . . . 14
BLANK. . . . . . . . . . . . . . . 15
BLKOFF . . . . . . . . . . . . . . 16
CEESTART . . . . . . . . . . . . . 17
CHECK . . . . . . . . . . . . . . 18
CMPAT. . . . . . . . . . . . . . . 19
CODEPAGE . . . . . . . . . . . . . 21
COMMON . . . . . . . . . . . . . 22
COMPILE . . . . . . . . . . . . . . 23
COPYRIGHT . . . . . . . . . . . . . 24
CSECT . . . . . . . . . . . . . . . 25
CSECTCUT . . . . . . . . . . . . . 26
CURRENCY . . . . . . . . . . . . . 27
DBCS . . . . . . . . . . . . . . . 28
DD . . . . . . . . . . . . . . . . 29
DDSQL. . . . . . . . . . . . . . . 30
DECIMAL. . . . . . . . . . . . . . 31
DEFAULT . . . . . . . . . . . . . . 33
DEPRECATE . . . . . . . . . . . . . 42
DISPLAY . . . . . . . . . . . . . . 44
DLLINIT . . . . . . . . . . . . . . 45
EXIT. . . . . . . . . . . . . . . . 46
EXTRN. . . . . . . . . . . . . . . 47
FLAG . . . . . . . . . . . . . . . 48
FLOAT . . . . . . . . . . . . . . . 49
FLOATINMATH. . . . . . . . . . . . 52
GOFF . . . . . . . . . . . . . . . 53
GONUMBER . . . . . . . . . . . . . 54
GRAPHIC . . . . . . . . . . . . . . 55
HGPR . . . . . . . . . . . . . . . 56
IGNORE . . . . . . . . . . . . . . 57
INCAFTER . . . . . . . . . . . . . 58
INCDIR . . . . . . . . . . . . . . 59
INCPDS . . . . . . . . . . . . . . 60
INITAUTO . . . . . . . . . . . . . 61
INITBASED . . . . . . . . . . . . . 62
INITCTL . . . . . . . . . . . . . . 63
INITSTATIC . . . . . . . . . . . . . 64
INSOURCE . . . . . . . . . . . . . 65
INTERRUPT . . . . . . . . . . . . . 66
LANGLVL. . . . . . . . . . . . . . 67
LIMITS . . . . . . . . . . . . . . . 68
LINECOUNT. . . . . . . . . . . . . 70
LINEDIR . . . . . . . . . . . . . . 71
LIST. . . . . . . . . . . . . . . . 72
LISTVIEW. . . . . . . . . . . . . . 73
MACRO . . . . . . . . . . . . . . 75
MAP . . . . . . . . . . . . . . . 76
MARGINI . . . . . . . . . . . . . . 77
MARGINS. . . . . . . . . . . . . . 78
MAXGEN . . . . . . . . . . . . . . 80
MAXMEM. . . . . . . . . . . . . . 81
MAXMSG . . . . . . . . . . . . . . 82
MAXNEST . . . . . . . . . . . . . 83
MAXSTMT . . . . . . . . . . . . . 84
MAXTEMP . . . . . . . . . . . . . 85
MDECK . . . . . . . . . . . . . . 86
NAME . . . . . . . . . . . . . . . 87
NAMES . . . . . . . . . . . . . . 88
NATLANG . . . . . . . . . . . . . 89
NEST . . . . . . . . . . . . . . . 90
NOT. . . . . . . . . . . . . . . . 91
NUMBER . . . . . . . . . . . . . . 92
OBJECT . . . . . . . . . . . . . . 93
OFFSET . . . . . . . . . . . . . . 94
ONSNAP . . . . . . . . . . . . . . 95
OPTIMIZE. . . . . . . . . . . . . . 96
OPTIONS . . . . . . . . . . . . . . 98
OR . . . . . . . . . . . . . . . . 99
PP . . . . . . . . . . . . . . . . 100
PPCICS . . . . . . . . . . . . . . 101
PPINCLUDE . . . . . . . . . . . . 102
PPLIST . . . . . . . . . . . . . . 103
PPMACRO . . . . . . . . . . . . . 104
PPSQL . . . . . . . . . . . . . . 105
PPTRACE . . . . . . . . . . . . . 106
PRECTYPE . . . . . . . . . . . . . 107
PREFIX . . . . . . . . . . . . . . 108
PROCEED . . . . . . . . . . . . . 109
PROCESS. . . . . . . . . . . . . . 110
QUOTE . . . . . . . . . . . . . . 111
REDUCE . . . . . . . . . . . . . . 112
RENT . . . . . . . . . . . . . . . 113
RESEXP . . . . . . . . . . . . . . 115
RESPECT. . . . . . . . . . . . . . 116
RTCHECK . . . . . . . . . . . . . 117
RULES . . . . . . . . . . . . . . 118
SEMANTIC . . . . . . . . . . . . . 126
SERVICE . . . . . . . . . . . . . . 127
SOURCE . . . . . . . . . . . . . . 128
SPILL . . . . . . . . . . . . . . . 129
STATIC . . . . . . . . . . . . . . 130
STDSYS . . . . . . . . . . . . . . 131
STMT . . . . . . . . . . . . . . . 132
STORAGE . . . . . . . . . . . . . 133
STRINGOFGRAPHIC . . . . . . . . . 134
SYNTAX . . . . . . . . . . . . . . 135
SYSPARM . . . . . . . . . . . . . 136
SYSTEM . . . . . . . . . . . . . . 137
TERMINAL . . . . . . . . . . . . . 138
TEST . . . . . . . . . . . . . . . 139
UNROLL. . . . . . . . . . . . . . 142
USAGE . . . . . . . . . . . . . . 143
WIDECHAR. . . . . . . . . . . . . 144
WINDOW . . . . . . . . . . . . . 145
WRITABLE . . . . . . . . . . . . . 146
Copyright IBM Corp. 1999, 2011 1
XINFO . . . . . . . . . . . . . . 148
XML . . . . . . . . . . . . . . . 151
XREF . . . . . . . . . . . . . . . 152
Blanks, comments and strings in options . . . . 153
Changing the default options . . . . . . . . 153
Specifying options in the %PROCESS or *PROCESS
statements . . . . . . . . . . . . . . 153
Using % statements . . . . . . . . . . . 154
Using the %INCLUDE statement . . . . . . 155
Using the compiler listing . . . . . . . . . 156
Heading information . . . . . . . . . . 156
Options used for compilation . . . . . . . 157
Preprocessor input . . . . . . . . . . 157
SOURCE program. . . . . . . . . . . 157
Statement nesting level . . . . . . . . . 157
ATTRIBUTE and cross-reference table . . . . 158
Attribute table . . . . . . . . . . . 158
Cross-reference table . . . . . . . . . 159
Aggregate length table . . . . . . . . . 159
Statement offset addresses . . . . . . . . 159
Storage offset listing . . . . . . . . . . 161
File reference table . . . . . . . . . . 162
Messages and return codes . . . . . . . . 163
Chapter 2. PL/I preprocessors . . . . . . . 165
Include preprocessor . . . . . . . . . . . 166
Macro preprocessor . . . . . . . . . . . 167
Macro preprocessor options . . . . . . . 167
CASE . . . . . . . . . . . . . . 168
DBCS . . . . . . . . . . . . . . 169
FIXED. . . . . . . . . . . . . . 170
INCONLY . . . . . . . . . . . . 171
NAMEPREFIX . . . . . . . . . . . 172
RESCAN . . . . . . . . . . . . . 173
Macro preprocessor example . . . . . . . 174
SQL preprocessor . . . . . . . . . . . . 175
SQL preprocessor options . . . . . . . . 177
APOSTSQL . . . . . . . . . . . . 178
ATTACH . . . . . . . . . . . . . 179
CCSID0 . . . . . . . . . . . . . 180
CONNECT . . . . . . . . . . . . 181
DATE . . . . . . . . . . . . . . 182
DEC . . . . . . . . . . . . . . 183
FLOAT . . . . . . . . . . . . . 184
GRAPHIC . . . . . . . . . . . . 185
INCONLY . . . . . . . . . . . . 186
LEVEL . . . . . . . . . . . . . 187
NOFOR . . . . . . . . . . . . . 188
SQL . . . . . . . . . . . . . . 189
SQLFLAG . . . . . . . . . . . . 190
STDSQL . . . . . . . . . . . . . 191
TIME . . . . . . . . . . . . . . 192
VERSION . . . . . . . . . . . . 193
XREF . . . . . . . . . . . . . . 194
Coding SQL statements in PL/I applications 195
Defining the SQL communications area. . . 195
Defining SQL descriptor areas . . . . . . 195
Embedding SQL statements . . . . . . 196
Using host variables . . . . . . . . . 197
Using host structures . . . . . . . . . 204
Using indicator variables . . . . . . . 204
Host structure example . . . . . . . . 204
General information on LOBs . . . . . . . 205
Suppressing SQL preprocessor messages . . . 205
CICS Preprocessor. . . . . . . . . . . . 205
CICS preprocessor options . . . . . . . . 206
Coding CICS statements in PL/I applications 206
Embedding CICS statements . . . . . . 206
Writing CICS transactions in PL/I . . . . . 207
Error-handling . . . . . . . . . . . . 207
Chapter 3. Using PL/I cataloged procedures 209
IBM-supplied cataloged procedures . . . . . . 209
Compile only (IBMZC) . . . . . . . . . 210
Compile and bind (IBMZCB) . . . . . . . 211
Compile, bind, and run (IBMZCBG) . . . . . 213
Compile, prelink, and link-edit (IBMZCPL) . . 214
(IBMZCPLG) . . . . . . . . . . . . 216
Compile, prelink, load and run (IBMZCPG) . . 218
Invoking a cataloged procedure . . . . . . . 220
Specifying multiple invocations . . . . . . 220
Modifying the PL/I cataloged procedures . . . . 221
EXEC statement . . . . . . . . . . . 222
DD statement . . . . . . . . . . . . 222
Chapter 4. Compiling your program . . . . . 225
Invoking the compiler under z/OS UNIX . . . . 225
Input files . . . . . . . . . . . . . 225
Specifying compile-time options under z/OS
UNIX . . . . . . . . . . . . . . . 226
-qoption_keyword. . . . . . . . . . . 226
Single and multiletter flags . . . . . . . . 226
Invoking the compiler under z/OS using JCL . . 227
EXEC statement . . . . . . . . . . . 227
DD statements for the standard data sets . . . 228
Input (SYSIN) . . . . . . . . . . . 228
Output (SYSLIN, SYSPUNCH) . . . . . 229
Temporary workfile (SYSUT1) . . . . . . 229
Listing (SYSPRINT) . . . . . . . . . . 229
Source Statement Library (SYSLIB) . . . . . 230
Specifying options. . . . . . . . . . . 230
Specifying options in the EXEC statement . . . 230
Specifying options in the EXEC statement using
an options file . . . . . . . . . . . . 231
Chapter 5. Link-editing and running . . . . . 233
Link-edit considerations . . . . . . . . . . 233
Using the binder . . . . . . . . . . . 233
Using the prelinker . . . . . . . . . . 233
Using the ENTRY card . . . . . . . . . 234
Run-time considerations . . . . . . . . . . 234
Formatting conventions for PRINT files . . . 234
Changing the format on PRINT files. . . . . 234
Automatic prompting . . . . . . . . . 235
Overriding automatic prompting . . . . . 236
Punctuating long input lines . . . . . . . 236
Line continuation character. . . . . . . 236
Punctuating GET LIST and GET DATA
statements . . . . . . . . . . . . . 236
2 Enterprise PL/I for z/OS Programming Guide
Automatic padding for GET EDIT . . . . 237
ENDFILE. . . . . . . . . . . . . . 237
SYSPRINT considerations . . . . . . . . . 237
Using MSGFILE(SYSPRINT) . . . . . . . 239
Using FETCH in your routines . . . . . . . 239
Fetching Enterprise PL/I routines . . . . . 239
Fetching PL/I MAIN Routines . . . . . . 247
Fetching z/OS C routines . . . . . . . . 247
Fetching assembler routines . . . . . . . 248
Invoking MAIN under TSO/E. . . . . . . . 248
Invoking MAIN under z/OS UNIX . . . . . . 249
Part 1. Compiling your program 3
Chapter 1. Using compiler options and facilities
This chapter describes the options that you can use for the compiler, along with
their abbreviations and IBM-supplied defaults. It's important to remember that
PL/I requires access to Language Environment run time when you compile your
applications. You can override most defaults when you compile your PL/I
program. You can also override the defaults when you install the compiler.
Compile-time option descriptions
There are three types of compiler options; however, most compiler options have a
positive and negative form. The negative form is the positive with 'NO' added at
the beginning (as in TEST and NOTEST). Some options have only a positive form
(as in SYSTEM). The three types of compiler options are:
1. Simple pairs of keywords: a positive form that requests a facility, and an
alternative negative form that inhibits that facility (for example, NEST and
NONEST).
2. Keywords that allow you to provide a value list that qualifies the option (for
example, FLAG(W)).
3. A combination of 1 and 2 above (for example, NOCOMPILE(E)).
Table 3 lists all the compiler options with their abbreviations (if any) and their
IBM-supplied default values. If an option has any suboptions which may be
abbreviated, those abbreviations are described in the full description of the option.
For the sake of brevity, some of the options are described loosely in the table (for
example, only one suboption of LANGLVL is mandatory, and similarly, if you
specify one suboption of TEST, you do not have to specify the other). The full and
completely accurate syntax is described in the pages that follow.
The paragraphs following Table 3 describe the options in alphabetical order. For
those options specifying that the compiler is to list information, only a brief
description is included; the generated listing is described under Using the
compiler listing on page 156.
Table 3. Compile-time options, abbreviations, and IBM-supplied defaults
Compile-Time Option Abbreviated Name z/OS Default
AGGREGATE[(DEC|HEX)] | NOAGGREGATE AG | NAG NOAGGREGATE
ARCH( n ) ARCH(5)
ATTRIBUTES[(FULL|SHORT)] |
NOATTRIBUTES
A | NA NA [(FULL)]
1
BACKREG(5 | 11) BACKREG(5)
BIFPREC(15 | 31) BIFPREC(15)
BLANK('c') BLANK('t')
2
BLKOFF | NOBLKOFF BLKOFF
CEESTART(FIRST | LAST) CEESTART(FIRST)
CHECK(STORAGE | NOSTORAGE,
CONFORMANCE | NOCONFORMANCE)
CHECK(NSTG,
NOCONFORMANCE)
CMPAT(LE | V1 | V2 | V3) CMP CMPAT(V2)
CODEPAGE(n) CP CODEPAGE(1140)
COMMON | NOCOMMON NOCOMMON
COMPILE | NOCOMPILE[(W | E | S)] C | NC NOCOMPILE(S)
Table 3. Compile-time options, abbreviations, and IBM-supplied defaults (continued)
COPYRIGHT(string) | NOCOPYRIGHT NOCOPYRIGHT
CSECT | NOCSECT CSE | NOCSE CSECT
CSECTCUT(n) CSECTCUT(4)
CURRENCY('c') CURR CURRENCY($)
DBCS | NODBCS NODBCS
DD(ddname-list) DD(SYSPRINT,SYSIN,
SYSLIB,SYSPUNCH,
SYSLIN,SYSADATA,
SYSXMLSD,SYSDEBUG)
DDSQL(ddname) DDSQL('')
DECIMAL(FOFLONASGN | NOFOFLONASGN,
FOFLONMULT | NOFOFLONMULT,
FORCEDSIGN, NOFORCEDSIGN)
DEC DEC(FOFLONASGN,
NOFOFLONMULT,
NOFORCEDSIGN)
DEFAULT(attribute | option) DFT See page 41
DEPRECATE(BUILTIN(built-in-name)
| ENTRY(entry-name) | INCLUDE(filename)
| VARIABLE(variable-name))
DEPRECATE(BUILTIN() ENTRY()
INCLUDE() VARIABLE())
DISPLAY
(STD | WTO(ROUTCDE(x) DESC(y) REPLY(z)))
DISPLAY(WTO)
DLLINIT | NODLLINIT NODLLINIT
EXIT | NOEXIT NOEXIT
EXTRN(FULL | SHORT) EXTRN(FULL)
FLAG[(I | W | E | S)] F FLAG(W)
FLOAT(AFP(NOVOLATILE | VOLATILE)
| NOAFP, DFP | NODFP)
FLOAT(AFP(NOVOLATILE)
NODFP)
FLOATINMATH(ASIS | LONG | EXTENDED) FLOATINMATH(ASIS)
GOFF | NOGOFF NOGOFF
GONUMBER(SEPARATE | NOSEPARATE) | NOGONUMBER GN | NGN NOGONUMBER
GRAPHIC | NOGRAPHIC GR | NGR NOGRAPHIC
HGPR[(PRESERVE|NOPRESERVE)] | NOHGPR NOHGPR
IGNORE(DISPLAY , PUT) | NOIGNORE NOIGNORE
INCAFTER([PROCESS(filename)]) INCAFTER()
INCDIR('directory name') | NOINCDIR NOINCDIR
INCPDS('PDS name') | NOINCPDS NOINCPDS
INITAUTO([ SHORT | FULL ]) | NOINITAUTO NOINITAUTO
INITBASED | NOINITBASED NOINITBASED
INITCTL | NOINITCTL NOINITCTL
INITSTATIC | NOINITSTATIC NOINITSTATIC
INSOURCE[(FULL|SHORT)] | NOINSOURCE IS | NIS NOINSOURCE
INTERRUPT | NOINTERRUPT INT | NINT NOINTERRUPT
LANGLVL(NOEXT | OS) LANGLVL(OS)
LIMITS(options) See LIMITS on page 68
LINECOUNT(n) LC LINECOUNT(60)
LINEDIR | NOLINEDIR NOLINEDIR
LIST | NOLIST NOLIST
LISTVIEW(SOURCE | AFTERMACRO
| AFTERCICS | AFTERSQL | AFTERALL)
LISTVIEW(SOURCE)
MACRO | NOMACRO M | NM NOMACRO
MAP | NOMAP NOMAP
MARGINI('c') | NOMARGINI MI | NMI NOMARGINI
MARGINS(m,n[,c])| NOMARGINS MAR(m,n) MARGINS
F-format: (2,72)
V-format: (10,100)
MAXGEN(n) MAXGEN(100000)
MAXMEM(n) MAXM MAXMEM(1048576)
MAXMSG(I | W | E | S,n) MAXMSG(W,250)
MAXNEST(BLOCK(x) DO(y) IF(z)) MAXNEST(BLOCK(17)
DO(17) IF(17))
MAXSTMT(n) MAXSTMT(4096)
MAXTEMP(n) MAXTEMP(50000)
MDECK | NOMDECK MD | NMD NOMDECK
NAME[('external name')] | NONAME N NONAME
NAMES('lower'[,upper]) NAMES('#@$','#@$')
NATLANG(ENU | UEN) NATLANG(ENU)
NEST | NONEST NONEST
NOT NOT('')
NUMBER | NONUMBER NUM | NNUM NUMBER
OBJECT | NOOBJECT OBJ | NOBJ OBJECT
OFFSET | NOOFFSET OF | NOF NOOFFSET
ONSNAP(STRINGRANGE, STRINGSIZE) | NOONSNAP NOONSNAP
OPTIMIZE(0 | 2 | 3) | NOOPTIMIZE OPT | NOPT OPT(0)
OPTIONS[(ALL|DOC)] | NOOPTIONS OP | NOP NOOPTIONS
OR('c') OR(' | ')
PP(pp-name) | NOPP NOPP
PPCICS(string ) | NOPPCICS NOPPCICS
PPINCLUDE(string ) | NOPPINCLUDE NOPPINCLUDE
PPLIST(KEEP | ERASE) PPLIST(KEEP)
PPMACRO(string) | NOPPMACRO NOPPMACRO
PPSQL(string) | NOPPSQL NOPPSQL
PPTRACE | NOPPTRACE NOPPTRACE
PREFIX(condition) See page 108
PRECTYPE(ANS | DECDIGIT | DECRESULT) PRECTYPE(ANS)
PROCEED | NOPROCEED[(W | E | S)] PRO | NPRO NOPROCEED(S)
PROCESS[(KEEP | DELETE)] | NOPROCESS PROCESS(DELETE)
QUOTE('"') QUOTE('"')
REDUCE | NOREDUCE REDUCE
RENT | NORENT NORENT
RESEXP | NORESEXP RESEXP
RESPECT([DATE]) RESPECT()
RTCHECK(NULLPTR | NONULLPTR) RTCHECK(NONULLPTR)
RULES(options) See RULES on page 118
SEMANTIC | NOSEMANTIC[(W | E | S)] SEM | NSEM NOSEMANTIC(S)
SERVICE('service string') | NOSERVICE SERV | NOSERV NOSERVICE
SOURCE | NOSOURCE S | NS NOSOURCE
SPILL(n) SP SPILL(512)
STATIC(FULL | SHORT) STATIC(SHORT)
STDSYS | NOSTDSYS NOSTDSYS
STMT | NOSTMT NOSTMT
STORAGE | NOSTORAGE STG | NSTG NOSTORAGE
STRINGOFGRAPHIC(CHAR | GRAPHIC) STRINGOFGRAPHIC
(GRAPHIC)
SYNTAX | NOSYNTAX[(W | E | S)] SYN | NSYN NOSYNTAX(S)
SYSPARM('string') SYSPARM('')
SYSTEM(MVS | CICS | IMS | TSO | OS) SYSTEM(MVS)
TERMINAL | NOTERMINAL TERM | NTERM
Chapter 1. Using compiler options and facilities 7
TEST(options) | NOTEST See TEST on page 139
3
UNROLL(AUTO | NO) UNROLL(AUTO)
USAGE(options) See USAGE on page 143
WIDECHAR(BIGENDIAN | LITTLEENDIAN) WCHAR WIDECHAR(BIGENDIAN)
WINDOW(w) WINDOW(1950)
WRITABLE | NOWRITABLE[(FWS|PRV)] WRITABLE
XINFO(options) XINFO(NODEF,NOMSG,
NOSYMNOSYN,NOXMI,
NOXML)
XML(CASE( UPPER | ASIS)) XML(CASE(UPPER))
XREF[(FULL | SHORT)] | NOXREF X | NX NX [(FULL)]
1
Notes:
1. FULL is the default suboption if the suboption is omitted with ATTRIBUTES or XREF.
2. The default value for the BLANK character is the tab character with value '05'x.
3. (ALL,SYM) is the default suboption if the suboption is omitted with TEST.
AGGREGATE
The AGGREGATE option creates an Aggregate Length Table that gives the lengths
of arrays and major structures in the source program in the compiler listing.
NOAGGREGATE
AGGREGATE
DECIMAL
( HEXADEC )
ABBREVIATIONS: AG, NAG

The suboptions of the AGGREGATE option determine how the offsets of
subelements are displayed in the Aggregate Length Table:
DECIMAL
All offsets are displayed in decimal.
HEXADEC
All offsets are displayed in hexadecimal.
In the Aggregate Length Table, the length of an undimensioned major or minor
structure is always expressed in bytes, but the length might not be accurate if the
major or minor structure contains unaligned bit elements.
The Aggregate Length Table includes structures but not arrays that have
non-constant extents. However, the sizes and offsets of elements within structures
with non-constant extents may be inaccurate or specified as *.
ARCH
The ARCH option specifies the architecture for which the instructions of executable
program are to be generated. It allows the optimizer to take advantage of specific
hardware instruction sets. A subparameter specifies the group to which a model
number belongs.
5
ARCH ( n )
The current values that may be specified for the ARCH level are:
5 Produces code that uses instructions available on model 2064-100 (z900) in
z/Architecture mode.
Specifically, these ARCH(5) machines and their follow-ons include
instructions such as NILL, NILH, OILL, OILH, LLILL and LLILH.
6 Produces code that uses instructions available on the 2084-xxx (z990) and
2086-xxx (z890) models in z/Architecture mode.
Specifically, the compiler on these ARCH(6) machines and their follow-ons
may exploit the long-displacement instruction set. The long-displacement
facility provides a 20-bit signed displacement field in 69 previously existing
instructions (by using a previously unused byte in the instructions) and 44
new instructions. A 20-bit signed displacement allows relative addressing
of up to 524,287 bytes beyond the location designated by a base register or
base and index register pair and up to 524,288 bytes before that location.
The enhanced previously existing instructions generally are ones that
handle 64-bit binary integers. The new instructions generally are new
versions of instructions for 32-bit binary integers. The new instructions also
include:
v A LOAD BYTE instruction that sign-extends a byte from storage to form
a 32-bit or 64-bit result in a general register
v New floating-point LOAD and STORE instructions
The long-displacement facility provides register-constraint relief by
reducing the need for base registers, code size reduction by allowing fewer
instructions to be used, and additional improved performance through
removal of possible address-generation interlocks.
7 Produces code that uses instructions available on the 2094-xxx models in
Specifically, these ARCH(7) machines and their follow-ons add instructions
supported by facilities such as extended-immediate and extended
translation facilities, which may be exploited by the compiler. For further
information about these facilities, see z/Architecture Principles of
Operation.
ARCH(7) machines also support the DFP instructions.
8 Produces code that uses instructions available on the 2097-xxx models
(IBM System z10 EC) in z/Architecture mode.
supported by the general instruction extensions facility, which may be
exploited by the compiler. Also, these machines add instructions supported
by the decimal floating-point facility, which are generated if the DFP
compiler option is specified and there are decimal floating-point data types
in the source code. For further information about these facilities, see
z/Architecture Principles of Operation.
9 Produces code that uses instructions available on the 2817-xxx models in
supported by the following facilities, which may be exploited by the
compiler:
v The high-word facility
v The population count facility
v The distinct-operands facility
v The floating-point extension facility
v The load/store-on-condition facility
For further information about these facilities, see z/Architecture Principles
of Operation.
Notes:
1. If you specify an ARCH value less than 5, the compiler resets it to 5.
2. The "x" in the model numbers above (such as 9672-Rx4 is a "wildcard"
and stands for any alphanumeric machine of that type, such as
9627-RA4).
3. Code that is compiled at ARCH(n) runs on machines in the ARCH(m)
group if and only if m >= n.
ATTRIBUTES
The ATTRIBUTES option specifies that the compiler includes a table of
source-program identifiers and their attributes in the compiler listing.
NOATTRIBUTES
ATTRIBUTES
FULL
( SHORT )
ABBREVIATIONS: A, NA, F, S
FULL
All identifiers and attributes are included in the compiler listing. FULL is the
default.
SHORT
Unreferenced identifiers are omitted, making the listing more manageable.
If you include both ATTRIBUTES and XREF (creates a cross-reference table), the
two tables are combined. However, if the SHORT and FULL suboptions are in
conflict, the last option specified is used. For example, if you specify
ATTRIBUTES(SHORT) XREF(FULL), FULL applies to the combined listing.
BACKREG
The BACKREG option controls the backchain register, which is the register used to
pass the address of the automatic storage of a parent routine when a nested
routine is invoked.
5
BACKREG ( 11 )
For best compatibility with PL/I for MVS & VM, OS PL/I V2R3 and earlier
compilers, BACKREG(5) should be used.
All routines that share an ENTRY VARIABLE must be compiled with the same
BACKREG option, and it is strongly recommended that all code in application be
compiled with the same BACKREG option.
Note that code compiled with VisualAge PL/I for OS/390 effectively used the
BACKREG(11) option. Code compiled with Enterprise PL/I V3R1 or V3R2 also
used the BACKREG(11) option by default.
BIFPREC
The BIFPREC option controls the precision of the FIXED BIN result returned by
various built-in functions.
15
BIFPREC ( 31 )
For best compatibility with PL/I for MVS & VM, OS PL/I V2R3 and earlier
compilers, BIFPREC(15) should be used.
BIFPREC affects the following built-in functions:
v COUNT
v INDEX
v LENGTH
v LINENO
v ONCOUNT
v PAGENO
v SEARCH
v SEARCHR
v SIGN
v VERIFY
v VERIFYR
The effect of the BIFPREC compiler option is most visible when the result of one of
the above built-in functions is passed to an external function that has been
declared without a parameter list. For example, consider the following code
fragment:
dcl parm char(40) var;
dcl funky ext entry( pointer, fixed bin(15) );
dcl beans ext entry;
call beans( addr(parm), verify(parm), ) );
If the function beans actually declares its parameters as POINTER and FIXED
BIN(15), then if the code above were compiled with the option BIFPREC(31) and if
it were run on a big-endian system such as z/OS, the compiler would pass a
four-byte integer as the second argument and the second parameter would appear
to be zero.
Note that the function funky would work on all systems with either option.
The BIFPREC option does not affect the built-in functions DIM, HBOUND and
LBOUND. The CMPAT option determines the precision of the FIXED BIN result
returned these three functions: under CMPAT(V1), these array-handling functions
return a FIXED BIN(15) result, while under CMPAT(V2) and CMPAT(LE), they
return a FIXED BIN(31) result. Under CMPAT(V3), they return a FIXED BIN(63)
result.
BLANK
The BLANK option specifies up to ten alternate symbols for the blank character.
BLANK

( ' char ' )
Note: Do not code any blanks between the quotes.
The IBM-supplied default code point for the BLANK symbol is '05'X.
char
A single SBCS character.
You cannot specify any of the alphabetic characters, digits, and special characters
defined in the PL/I Language Reference.
If you specify the BLANK option, the standard blank symbol is still recognized as
a blank.
BLKOFF
The BLKOFF option controls whether the offsets shown in the pseudo-assembler
listing (produced by the LIST option) and the statement offset listing (produced by
the OFFSET option) are from the start of the current module or from the start of
the current procedure.
BLKOFF
NOBLKOFF
The pseudo-assembler listing also includes the offset of each block from the start of
the current module (so that the offsets shown for each statement can be translated
to either block or module offsets).
CEESTART
The CEESTART option specifies whether the compiler should place the CEESTART
csect before or after all the other generated object code.
FIRST
CEESTART ( LAST )
Under the CEESTART(FIRST) option, the compiler places the CEESTART csect
before all the other generated object code; however, under the CEESTART(LAST)
option, the compiler places it after all the other generated object code.
Using CEESTART(FIRST) will cause the binder to choose CEESTART as the entry
point for a module if no ENTRY card is specified during the bind step.
Those users who want to use linker CHANGE cards must use the
CEESTART(LAST) option.
However, MAIN routines should be linked with an ENTRY CEESTART linkage
editor card. But, if you use the CEESTART(LAST) option, then you must include
an ENTRY CEESTART card when linking your MAIN routine.
CHECK
The CHECK option specifies whether the compiler should generate special code to
detect various programming errors.

,
NOCONFORMANCE
CHECK ( CONFORMANCE )
NOSTORAGE
STORAGE
ABBREVIATIONS: STG, NSTG

Specifying CHECK(CONFORMANCE) causes the compiler to generate, under the
following circumstances, code that checks at run time if the attributes of the
arguments passed to a procedure match those of the declared parameters:
v If a parameter is a string (or an array of strings) declared with a constant length,
then the STRINGSIZE condition will be raised if the argument passed does not
have matching length.
v If a parameter is a string (or an array of strings), then the STRINGSIZE
condition will be raised if the argument does not have the same length type
(VARYING, NONVARYING or VARYINGZ).
v If a parameter is an array (of scalars or structures), then the SUBSCRIPTRANGE
condition will be raised if any constant bounds do not match those of the passed
argument. The SUBSCRIPTRANGE condition will also be raised if all the extents
are constant and the size and spacing of the array elements in the argument do
not match those in the parameter. Arrays inside a structure are not checked.
v If a parameter is a structure or union with constant extents, then the
SUBSCRIPTRANGE condition will be raised if the offset of the last element does
not match that of the passed argument.
v If the procedure has the RETURNS BYADDR attribute and that attribute
specifies a string type, then the STRINGSIZE condition will be raised if the
string passed for the RETURNS value does not have matching length.
This extra code will not be generated if the NODESCRIPTOR option applies to the
procedure or if the block contains ENTRY statements or if the CMPAT(LE) option
is in effect.
When you specify CHECK(STORAGE), the compiler calls slightly different library
routines for ALLOCATE and FREE statements (except when these statements occur
within an AREA). The following built-in functions, described in the PL/I Language
Reference, can be used only when CHECK(STORAGE) has been specified:
v ALLOCSIZE
v CHECKSTG
v UNALLOCATED
AMODE(24) is not recommended for Enterprise PL/I applications. For code
compiled with the CHECK(STORAGE) option, if you have to use AMODE(24),
then you must also specify the HEAP(,,BELOW) runtime option.
CMPAT
The CMPAT option specifies whether object compatibility with OS PL/I Version 1,
OS PL/I Version 2, PL/I for MVS and VM, VisualAge PL/I for OS/390 or
Enterprise PL/I for z/OS is to be maintained for programs sharing strings, AREAs,
arrays and/or structures.
CMPAT (
V2
LE
V1
)
ABBREVIATIONS: CMP
LE Under CMPAT(LE), your program can share strings, AREAs, arrays and/or
structures only with programs compiled with VisualAge PL/I for OS/390 or
Enterprise PL/I for z/OS and only as long as the CMPAT(V1) and CMPAT(V2)
options were not used when they were compiled.
V1 Under CMPAT(V1), you can share strings, AREAs, arrays and/or structures
with programs compiled with the OS PL/I compiler and with programs
compiled with later PL/I compilers as long as the CMPAT(V1) option was
used.
V2 Under CMPAT(V2), you can share strings, AREAs, arrays and/or structures
with programs compiled with the OS PL/I compiler and with programs
compiled with later PL/I compilers as long as the CMPAT(V2) option was
used.
V3 Under CMPAT(V3), you can share strings with programs compiled with the OS
PL/I compiler and with programs compiled with later PL/I compilers as long
as one of the CMPAT(V*) options were used. However, you can not share
AREAs, arrays or structures with any code not compiled with CMPAT(V3).
DB2 stored procedures must not be compiled with CMPAT(LE).
All the modules in an application must be compiled with the same CMPAT option.
Mixing old and new code still has some restrictions. For information about these
restrictions see the Enterprise PL/I for z/OS Compiler and Run-Time Migration Guide.
The DFT(DESCLIST) option conflicts with the CMPAT(V*) options, and if it is
specified with any of them, a message will be issued and the
DFT(DESCLOCATOR) option assumed.
Under CMPAT(V3), arrays may be declared with any value that an 8-byte integer
could assume. However, the total size of an array currently still has the same limit
as an array declared under CMPAT(V2).
Under CMPAT(V3), the following built-in functions will always return a FIXED
BIN(63) result
v CURRENTSIZE/CSTG
v DIMENSION
v HBOUND
v LBOUND
v LOCATION
v SIZE/STG
Because these functions will be returning 8-byte integer values, under CMPAT(V3),
the second option in the FIXEDBIN suboption of the LIMITS option must be 63.
However, even under CMPAT(V3), statement and format label constants must be
specified using 4-byte integers.
CODEPAGE
The CODEPAGE option specifies the code page used for:
v conversions between CHARACTER and WIDECHAR
v the default code page used by the PLISAX built-in subroutines
CODEPAGE ( ccsid )
The supported CCSID's are:
01047
01140
01141
01142
01143
01144
01025
01145
01146
01147
01148
01149
00037
01155
00273
00277
00278
00280
00284
00285
00297
00500
00871
00819
00813
00920
The default CCSID 1140 is an equivalent of CCSID 37 (EBCDIC Latin-1, USA) but
includes the Euro symbol.
COMMON
The COMMON option directs the compiler to generate CM linkage records for
EXTERNAL STATIC variables.
NOCOMMON
COMMON
Under the COMMON option, if the NORENT option applies, then CM linkage
records will be generated for EXTERNAL STATIC variables that are not RESERVED
and which contain no INITIAL values. This matches what the OS PL/I compiler
did.
Under the NOCOMMON option, SD records will be written as was true in earlier
releases of Enterprise PL/I.
The COMMON option must not be used with the RENT option or with
LIMITS(EXTNAME(n)) if n > 7.
COMPILE
The COMPILE option causes the compiler to stop compiling after all semantic
checking of the source program if it produces a message of a specified severity
during preprocessing or semantic checking. Whether the compiler continues or not
depends on the severity of the error detected, as specified by the NOCOMPILE
option in the list below. The NOCOMPILE option specifies that processing stops
unconditionally after semantic checking.
NOCOMPILE
S
( W )
E
COMPILE
ABBREVIATIONS: C, NC
COMPILE
Generates code unless a severe error or unrecoverable error is detected.
Equivalent to NOCOMPILE(S).
NOCOMPILE
Compilation stops after semantic checking.
NOCOMPILE(W)
No code generation if a warning, error, severe error, or unrecoverable error is
detected.
NOCOMPILE(E)
No code generation if an error, severe error, or unrecoverable error is detected.
NOCOMPILE(S)
No code generation if a severe error or unrecoverable error is detected.
If the compilation is terminated by the NOCOMPILE option, the cross-reference
listing and attribute listing can be produced; the other listings that follow the
source program will not be produced.
COPYRIGHT
The COPYRIGHT option places a string in the object module, if generated. This
string is loaded into memory with any load module into which this object is
linked.
NOCOPYRIGHT
COPYRIGHT ( copyright string )
The string is limited to 1000 characters in length. However, if the string is longer
than 100 characters, it will not be shown in the options listing.
To ensure that the string remains readable across locales, only characters from the
invariant character set should be used.
CSECT
The CSECT option ensures that the object module, if generated, contains named
CSECTs. Use this option if you will be using SMP/E to service your product or to
aid in debugging your program.
CSECT
NOCSECT
ABBREVIATIONS: CSE, NOCSE
Under the NOCSECT option, the code and static sections of your object module are
given default names.
Under the CSECT option, the code and static sections of your object module are
given names that depend on the "package name" which is defined as follows:
v if the package statement was used, the "package name" is the leftmost label on
the package statement
v otherwise, the "package name" is the leftmost label on the first procedure
statement.
A "modified package name" of length 7 is then formed as follows:
v when the package name is less than 7 characters long, "*"'s are prefixed to it to
make a modified package name that is 7 characters long
v when the package name is more than 7 characters long, the first n and last 7 - n
characters are used to make the modified package name, where the value n is
set by the CSECTCUT option
v otherwise the package name is copied to the modified package name
The code csect name is built by taking the modified package name and appending
a '1' to it.
The static csect name is built by taking the modified package name and appending
a '2' to it.
So, for a package named "SAMPLE", the code csect name would be "*SAMPLE1",
and the static csect name would be "*SAMPLE2".
CSECTCUT
The CSECTCUT option controls how the compiler, when processing the CSECT
option, handles long names.
4
CSECTCUT ( n )
The CSECTCUT option has no effect unless you specify the CSECT option. It also
has no effect if the "package name" used by the CSECT option has 7 or fewer
characters.
The value n in the CSECTCUT option must be between 0 and 7.
If the "package name" used by the CSECT option has more than 7 characters, the
compiler will collapse the name to 7 characters by taking the first n and last 7 - n
characters.
For example, for a compilation consisting of one procedure with the name
BEISPIEL,
v under CSECTCUT(3), the compiler would collapse the name to BEIPIEL
v under CSECTCUT(4), the compiler would collapse the name to BEISIEL
CURRENCY
The CURRENCY option allows you to specify an alternate character to be used in
picture strings instead of the dollar sign.
$
CURRENCY ( ' x ' )
ABBREVIATIONS: CURR
x Character that you want the compiler and run time to recognize and accept as
the dollar sign in picture strings.
DBCS
The DBCS option ensures that the listing, if generated, is sensitive to the possible
presence of DBCS even though the GRAPHIC option has not been specified.
NODBCS
DBCS
Under z/OS UNIX System Services, the NODBCS option causes the listing, if
generated, to show all DBCS shift-codes as ".".
Under both batch and z/OS UNIX System Services, the NODBCS option ensures
that the header text in the listing page does not contain unmatched shift codes.
The NODBCS option should not be specified if the GRAPHIC option is also
specified.
DD
The DD option allows you to specify alternate DD names for the various datasets
used by the compiler.
DD
( SYSPRINT )
, SYSIN
, SYSLIB
, SYSPUNCH
, SYSLIN
, SYSADATA
, SYSXMLSD
, SYSDEBUG
Up to eight DD names may be specified. In order, they specify alternate DD names

for
v SYSPRINT
v SYSIN
v SYSLIB
v SYSPUNCH
v SYSLIN
v SYSADATA
v SYSXMLSD
v SYSDEBUG
If you wanted to use ALTIN as the DD name for the primary compiler source file,
you would have to specify DD(SYSPRINT,ALTIN). If you specified DD(ALTIN),
SYSIN would be used as the DDNAME for the primary compiler source file and
ALTIN would be used as the DD name for the compiler listing.
You can also use * to indicate that the default DD name should be used. Thus
DD(*,ALTIN) is equivalent to DD(SYSPRINT,ALTIN).
DDSQL
The DDSQL option allows you to specify an alternate DD name for the dataset to
be used by the SQL preprocessor when resolving EXEC SQL INCLUDE statements.
''
DDSQL ( ddname )
Under the DDSQL('') option, the DD name that is used to resolve EXEC SQL
INCLUDE statements is the DD name for SYSLIB from the DD compiler option.
This option may be useful when moving from the SQL precompiler to the
integrated SQL preprocessor.
DECIMAL
The DECIMAL option specifies how the compiler should handle certain FIXED
DECIMAL operations and assignments.

,
FOFLONASGN
DECIMAL ( NOFOFLONASGN )
NOFOFLONMULT
FOFLONMULT
NOFORCEDSIGN
FORCEDSIGN
FOFLONASGN
The FOFLONASGN option requires the compiler to generate code that will
raise the FIXEDOVERFLOW condition, if it is enabled and the SIZE condition
is disabled, whenever a FIXED DECIMAL expression is assigned to a FIXED
DECIMAL target and significant digits are lost.
Conversely, under the NOFOFLONASGN option, the compiler will generate
code that will not raise the FIXEDOVERFLOW condition when significant
digits are lost in such an assignment.
So, for example, given a variable A declared as FIXED DEC(5), the assignment
A = A + 1 might raise FOFL under the FOFLONASGN option, but would
never raise it under the NOFOFLONASGN option.
Note, however, that under the NOFOFLONASGN option, the
FIXEDOVERFLOW condition can still be raised by operations that produce a
result with more digits than allowed by the FIXEDDEC suboption of the
LIMITS option. For example, given a variable B declared as FIXED DEC(15)
with the value 999_999_999_999_999 and given that the FIXEDDEC suboption
of the LIMITS specifies the maximum precision as 15, then the assignment B =
B + 1 will raise the FIXEDOVERFLOW condition (if FOFL is enabled, of
course) since the addition B + 1 will raise the condition.
FOFLONMULT
The FOFLONMULT option requires the compiler to generate code that will
raise the FIXEDOVERFLOW condition for any use of the MULTIPLY built-in
function that would produce a FIXED DEC result that is too large for the
precision specified in the built-in function.
Conversely, under the NOFOFLONMULT option, the compiler will generate
code that will produce a truncated result for any such use of the MULTIPLY
built-in function.
Note that the use of the FOFLONMULT option changes the default language
semantics (which would be to truncate a too large result of the MULTIPLY
built-in function applied to FIXED DEC - unless the SIZE condition were
enabled).
FORCEDSIGN
The FORCEDSIGN option will force the compiler to generate extra code to
insure that whenever a FIXED DECIMAL result with the value zero is
generated, the sign nibble of the result will have the value CX. This option
can cause the compiler to generate code that will perform much, much worse
than the code generated under the NOFORCEDSIGN suboption.
Also, when this option is in effect, more data exceptions may occur when you
run your code. For example, if you assign one FIXED DEC(5) variable to
another FIXED DEC(5) variable, the compiler would normally generate a MVC
instruction to perform the move. However if this option is in effect, in order to
insure that the result has the preferred sign, the compiler will generate a ZAP
instruction to perform the move. If the source contains invalid packed decimal
data, the ZAP instruction, but not the MVC instruction, will raise a decimal
data exception.
Under this option, data exceptions may also be raised when one PICTURE
variable is assigned to another PICTURE variable since that conversion usually
involves an implicit conversion to FIXED DEC which, under this option, will
generate a ZAP instruction that will raise a data exception if the source
contains invalid data.
Under DECIMAL(NOFORCEDSIGN), a "negative zero" may be produced by
certain calculations. However, a programmer should rely on getting a negative
zero only when assigning a negative literal to a FIXED DEC that cannot hold a
value of that small a magnitude (such as assigning -.001 to a FIXED DEC(5,2) )
DEFAULT
The DEFAULT option specifies defaults for attributes and options. These defaults
are applied only when the attributes or options are not specified or implied in the
source code.
DEFAULT (
,
ALIGNED
UNALIGNED
IBM
ANS
EBCDIC
ASCII
ASSIGNABLE
NONASSIGNABLE
BIN1ARG
NOBIN1ARG
BYADDR
BYVALUE
NONCONNECTED
CONNECTED
DESCLOCATOR
DESCLIST
DESCRIPTOR
NODESCRIPTOR
ALIGNED
DUMMY ( )
UNALIGNED
HEXADEC
E ( )
IEEE
EVENDEC
NOEVENDEC
HEXADEC
IEEE
NOINITFILL
INITFILL
( init_value )
NOINLINE
INLINE
OPTLINK
LINKAGE ( )
SYSTEM
LOWERINC
UPPERINC
NATIVE
NONNATIVE
NATIVEADDR
NONNATIVEADDR
NULL370
NULLSYS
NULLSTRADDR
NONULLSTRADDR
NULL
NULLSTRPTR ( )
SYSNULL
REORDER
ORDER
MIN
ORDINAL ( MAX )
NOOVERLAP
OVERLAP
PSEUDODUMMY
NOPSEUDODUMMY
NONRECURSIVE
RECURSIVE
NORETCODE
RETCODE
BYADDR
RETURNS ( )
BYVALUE
HEXADEC
SHORT ( )
IEEE
)
ABBREVIATIONS: DFT, ASGN, NONASGN, NONCONN, CONN, INL, NOINL
ALIGNED | UNALIGNED
This suboption allows you to force byte-alignment on all of your variables.
If you specify ALIGNED, all variables other than character, bit, graphic, and
picture are given the ALIGNED attribute unless the UNALIGNED attribute is
explicitly specified (possibly on a parent structure) or implied by a DEFAULT
statement.
If you specify UNALIGNED, all variables are given the UNALIGNED attribute
unless the ALIGNED attribute is explicitly specified (possibly on a parent
structure) or implied by a DEFAULT statement.
ALIGNED is the default.
IBM | ANS
Use IBM or ANS SYSTEM defaults. The arithmetic defaults for IBM and ANS
are the following:
Attributes DEFAULT(IBM) DEFAULT(ANS)
FIXED DECIMAL (5,0) (10,0)
FIXED BINARY (15,0) (31,0)
FLOAT DECIMAL (6) (6)
FLOAT BINARY (21) (21)
Under the IBM suboption, variables with names beginning from I to N default
to FIXED BINARY and any other variables default to FLOAT DECIMAL. If you
select the ANS suboption, the default for all variables is FIXED BINARY.
IBM is the default.
ASCII | EBCDIC
Use this option to set the default for the character set used for the internal
representation of character problem program data.
Specify ASCII only when compiling programs that depend on the ASCII
character set collating sequence. Such a dependency exists, for example, if your
program relies on the sorting sequence of digits or on lowercase and uppercase
alphabetics. This dependency also exists in programs that create an uppercase
alphabetic character by changing the state of the high-order bit.
Note: The compiler supports A and E as suffixes on character strings. The A
suffix indicates that the string is meant to represent ASCII data, even if
the EBCDIC compiler option is in effect. Alternately, the E suffix
indicates that the string is EBCDIC, even when you select
DEFAULT(ASCII).
123A is the same as 313233X
123E is the same as F1F2F3X
EBCDIC is the default.
ASSIGNABLE | NONASSIGNABLE
This option causes the compiler to apply the specified attribute to all static
variables that are not declared with the ASSIGNABLE or NONASSIGNABLE
attribute. The compiler flags statements in which NONASSIGNABLE variables
are the targets of assignments.
ASSIGNABLE is the default.
BIN1ARG | NOBIN1ARG
This suboption controls how the compiler handles one-byte REAL FIXED BIN
arguments passed to an unprototyped function.
Under BIN1ARG, the compiler will pass a FIXED BIN argument as is to an
unprototyped function.
But under NOBIN1ARG, the compiler will assign any one-byte REAL FIXED
BIN argument passed to an unprototyped function to a two-byte FIXED BIN
temporary and pass that temporary instead.
Consider the following example:
dcl f1 ext entry;
dcl f2 ext entry( fixed bin(15) );
call f1( 1b );
call f2( 1b );
If you specified DEFAULT(BIN1ARG), the compiler would pass the address of
a one-byte FIXED BIN(1) argument to the routine f1 and the address of a
two-byte FIXED BIN(15) argument to the routine f2. However, if you specified
DEFAULT(NOBIN1ARG), the compiler would pass the address of a two-byte
FIXED BIN(15) argument to both routines.
Note that if the routine f1 was a COBOL routine, passing a one-byte integer
argument to it would cause problems since COBOL has no support for
one-byte integers. In this case, using DEFAULT(NOBIN1ARG) might be
helpful; but it would be better to specify the argument attributes in the entry
declare.
BIN1ARG is the default.
BYADDR | BYVALUE
Set the default for whether arguments or parameters are passed by address or
by value. BYVALUE applies only to certain arguments and parameters. See the
PL/I Language Reference for more information.
BYADDR is the default.
CONNECTED | NONCONNECTED
Set the default for whether parameters are connected or nonconnected.
CONNECTED allows the parameter to be used as a target or source in
record-oriented I/O or as a base in string overlay defining.
NONCONNECTED is the default.
DESCLIST | DESCLOCATOR
When you specify DEFAULT(DESCLIST), the compiler passes all descriptors in
a list as a 'hidden' last parameter.
If you specify DEFAULT(DESCLOCATOR), parameters requiring descriptors
are passed using a locator or descriptor in the same way as previous releases
of PL/I. This allows old code to continue to work even if it passed a structure
from one routine to a routine that was expecting to receive a pointer.
The DFT(DESCLIST) option conflicts with the CMPAT(V*) options, and if it is
specified with any of them, a message will be issued and the
DFT(DESCLOCATOR) option assumed.
DESCLOCATOR is the default.
DESCRIPTOR | NODESCRIPTOR
Using DESCRIPTOR with a PROCEDURE indicates that a descriptor list was
passed, while DESCRIPTOR with ENTRY indicates that a descriptor list should
be passed. NODESCRIPTOR results in more efficient code, but has the
following restrictions:
v For PROCEDURE statements, NODESCRIPTOR is invalid if any of the
parameters have:
An asterisk (*) specified for the bound of an array, the length of a string,
or the size of an area except if it is a VARYING or VARYINGZ string with
the NONASSIGNABLE attribute
The NONCONNECTED attribute
The UNALIGNED BIT attribute
v For ENTRY declarations, NODESCRIPTOR is invalid if an asterisk (*) is
specified for the bound of an array, the length of a string, or the size of an
area in the ENTRY description list.
DESCRIPTOR is the default.
DUMMY(ALIGNED | UNALIGNED)
This suboption reduces the number of situations in which dummy arguments
get created.
DUMMY(ALIGNED) indicates that a dummy argument should be created even
if an argument differs from a parameter only in its alignment.
DUMMY(UNALIGNED) indicates that no dummy argument should be created
for a scalar (except a nonvarying bit) or an array of such scalars if it differs
from a parameter only in its alignment.
dcl
1 a1 unaligned,
2 b1 fixed bin(31),
2 b2 fixed bin(15),
2 b3 fixed bin(31),
2 b4 fixed bin(15);
dcl x entry( fixed bin(31) );
call x( b3 );
If you specified DEFAULT(DUMMY(ALIGNED)), a dummy argument would
be created, while if you specified DEFAULT(DUMMY(UNALIGNED)), no
dummy argument would be created.
DUMMY(ALIGNED) is the default.
E (HEXADEC | IEEE)
The E suboption determines how many digits will be used for the exponent in
E-format items.
If you specify E(IEEE), 4 digits will be used for the exponent in E-format items.
If you specify E(HEXADEC), 2 digits will be used for the exponent in E-format
items.
If DFT( E(HEXADEC) ) is specified, an attempt to use an expression whose
exponent has an absolute value greater than 99 will cause the SIZE condition
to be raised.
If the compiler option DFT(IEEE) is in effect, you should normally also use the
option DFT( E(IEEE) ). However, under this option, some E format items that
would be valid under DFT( E(HEXADEC) ) would not be valid. For instance,
under DFT( E(IEEE) ), the statement "put skip edit(x) ( e(15,8));" would be
flagged because the E format item is invalid.
E(HEXADEC) is the default.
EVENDEC | NOEVENDEC
This suboption controls the compiler's tolerance of fixed decimal variables
declared with an even precision.
Under NOEVENDEC, the precision for any fixed decimal variable is rounded
up to the next highest odd number.
If you specify EVENDEC and then assign 123 to a FIXED DEC(2) variable, the
SIZE condition is raised. If you specify NOEVENDEC, the SIZE condition is
not raised.
EVENDEC is the default.
HEXADEC | IEEE
This suboption allows you to specify the default representation used to hold all
FLOAT variables and all floating-point intermediate results. This suboption
also determines whether the compiler evaluates floating-point expressions
using the hexadecimal or IEEE float instructions and math routines.
Programs that communicate with JAVA should probably use the IEEE option,
and programs that pass data to or receive data from platforms where IEEE is
the default representation for floating-point data might also want to use the
IEEE option.
HEXADEC is the default.
INITFILL | NOINITFILL
This suboption controls the default initialization of automatic variables.
If you specify INITFILL with a hex value (nn), that value is used to initialize
the storage used by all automatic variables in a block each time that block is
entered. If you do not enter a hex value, the default is '00'.
Note that the hex value may be specified without or without quotes, but if it is
specified with quotes, the string should not have an X suffix.
Under NOINITFILL, the storage used by an automatic variable may hold
arbitrary bit patterns unless the variable is explicitly initialized.
INITFILL can cause programs to run significantly slower and should not be
specified in production programs. However, the INITFILL option produces
code that runs faster than the LE STORAGE option. Also, during program
development, this option is very useful for detecting uninitialized automatic
variables: a program that runs correctly with DFT(INITFILL('00')) and with
DFT(INITFILL('ff')) probably has no uninitialized automatic variables.
NOINITFILL is the default.
INLINE | NOINLINE
This option sets the default for the inline procedure option.
Specifying INLINE allows your code to run faster but, in some cases, also
creates a larger executable file. For more information on how inlining can
improve the performance of your application, see Chapter 13, Improving
performance, on page 369.
NOINLINE is the default.
LINKAGE
The linkage convention for procedure invocations is:
OPTLINK
The default linkage convention for Enterprise PL/I. This linkage provides
the best performance.
SYSTEM
The standard linking convention for system APIs.
LINKAGE(OPTLINK) should be used for all routines called by or calling to
JAVA, and it should also be used for all routines called by or calling to C
(unless the C code has been compiled with a non-default linkage).
LINKAGE(SYSTEM) should be used for all non-PL/I routines that expect the
high-order bit to be on in the address of the last (and only the last) parameter.
Note that specifying OPTIONS(ASSEMBLER) for a PROCEDURE or an ENTRY
forces LINKAGE(SYSTEM) regardless of the setting of this option.
LINKAGE(OPTLINK) is the default.
LOWERINC | UPPERINC
If you specify LOWERINC, the compiler requires that the actual file names of
INCLUDE files are in lowercase. If you specify UPPERINC, the compiler
requires that the names are in uppercase.
Note: This suboption applies only to compilations under z/OS UNIX.
Under z/OS UNIX, the include name is built using the extension '.inc'. So, for
example, under the DFT(LOWERINC) option, the statement %INCLUDE
STANDARD; will cause the compiler to try to include standard.inc. But, under
the DFT(UPPERINC) option, the statement %INCLUDE STANDARD; will
cause the compiler to try to include STANDARD.INC.
LOWERINC is the default.
NATIVE | NONNATIVE
This option affects only the internal representation of fixed binary, ordinal,
offset, area, and varying string data. When the NONNATIVE suboption is in
effect, the NONNATIVE attribute is applied to all such variables not declared
with the NATIVE attribute.
You should specify NONNATIVE only to compile programs that depend on
the nonnative format for holding these kind of variables.
If your program bases fixed binary variables on pointer or offset variables (or
conversely, pointer or offset variables on fixed binary variables), specify either:
v Both the NATIVE and NATIVEADDR suboptions
v Both the NONNATIVE and NONNATIVEADDR suboptions.
Other combinations produce unpredictable results.
NATIVE is the default.
NATIVEADDR | NONNATIVEADDR
This option affects only the internal representation of pointers. When the
NONNATIVEADDR suboption is in effect, the NONNATIVE attribute is
applied to all pointer variables not declared with the NATIVE attribute.
If your program bases fixed binary variables on pointer or offset variables (or
conversely, pointer or offset variables on fixed binary variables), specify either:
v Both the NATIVE and NATIVEADDR suboptions
v Both the NONNATIVE and NONNATIVEADDR suboptions.
Other combinations produce unpredictable results.
NATIVEADDR is the default.
NULLSYS | NULL370
This suboption determines which value is returned by the NULL built-in
function. If you specify NULLSYS, binvalue(null()) is equal to 0. If you want
binvalue(null()) to equal 'ff_00_00_00'xn as is true with previous releases of
PL/I, specify NULL370.
NULL370 is the default.
NULLSTRADDR | NONULLSTRADDR
This suboption controls how the compiler handles null strings when passed as
arguments.
Under NULLSTRADDR, when a null string is specified as an argument in an
entry invocation, the compiler will pass the address of an initialized piece of
automatic storage. This is compatible with what the OS PL/I and PL/I for
MVS compilers did.
But under NONULLSTRADDR, when a null string is specified as an argument
in an entry invocattion, the compiler will pass a null pointer as the address of
the argument. This is compatible with what early releases of the Enterprise
PL/I compiler did.
NULLSTRADDR is the default.
NULLSTRPTR
This suboption controls how the compiler handles null strings when assigned
to POINTERs.
Under NULLSTRPTR( SYSNULL ), the result of assigning '' to a POINTER is
the same as assigning SYSNULL() to the pointer
Under NULLSTRPTR( NULL ), the result of assigning '' to a POINTER is the
same as assigning NULL() to the pointer
NULLSTRPTR(NULL) is the default.
ORDER | REORDER
This suboption affects the optimization of the object code. Specifying
REORDER allows more optimization of your code. For detailed information,
see Chapter 13, Improving performance, on page 369.
REORDER is the default.
ORDINAL(MIN | MAX)
If you specify ORDINAL(MAX), all ordinals whose definition does not include
a PRECISION attribute is given the attribute PREC(31). Otherwise, they are
given the smallest precision that covers their range of values.
ORDINAL(MIN) is the default.
OVERLAP | NOOVERLAP
If you specify OVERLAP, the compiler presumes the source and target in an
assignment can overlap and generates, as needed, extra code in order to ensure
that the result of the assignment is okay.
NOOVERLAP will produce code that performs better; however, if you use
NOOVERLAP, you must insure that the source and target never overlap.
NOOVERLAP is the default.
PSEUDODUMMY | NOPSEUDODUMMY
This suboption determines if dummy arguments are created when a SUBSTR
reference is specified as an argument to an unprototyped function.
If you specify PSEUDODUMMY, dummy arguments are created when a
SUBSTR reference is specified as an argument to an unprototyped function.
If you specify NOPSEUDODUMMY, dummy arguments are not created when
a SUBSTR reference is specified as an argument to an unprototyped function.
PSEUDODUMMY is the default.
RECURSIVE | NONRECURSIVE
When you specify DEFAULT(RECURSIVE), the compiler applies the
RECURSIVE attribute to all procedures. If you specify
DEFAULT(NONRECURSIVE), all procedures are nonrecursive except
procedures with the RECURSIVE attribute.
NONRECURSIVE is the default.
RETCODE | NORETCODE
If you specify RETCODE, for any external procedure that does not have the
RETURNS attribute, the compiler will generate extra code so that the
procedure returns the integer value obtained by invoking the PLIRETV built-in
function just prior to returning from that procedure.
If you specify NORETCODE, no special code is generated for procedures that
do not have the RETURNS attribute.
NORETCODE is the default.
RETURNS (BYVALUE | BYADDR)
Sets the default for how values are returned by functions. See the PL/I
Language Reference for more information.
You should specify RETURNS(BYADDR) if your application contains ENTRY
statements and the ENTRY statements or the containing procedure statement
have the RETURNS option. You must also specify RETURNS(BYADDR) on the
entry declarations for such entries.
RETURNS(BYADDR) is the default.
SHORT (HEXADEC | IEEE)
This suboption improves compatibility with other non-IBM UNIX compilers.
SHORT (HEXADEC) maps FLOAT BIN (p) to a short (4-byte) floating point
number for p <= 21. SHORT (IEEE) maps FLOAT BIN (p) to a short (4-byte)
floating point number for p <= 24.
SHORT (HEXADEC) is the default.
Default: DEFAULT(ALIGNED IBM EBCDIC ASSIGNABLE BIN1ARG BYADDR
NONCONNECTED DESCLOCATOR DESCRIPTOR DUMMY(ALIGNED)
E(HEXADEC) EVENDEC HEXADEC NOINITFILL NOINLINE
LINKAGE(OPTLINK) LOWERINC NATIVE NATIVEADDR NULL370
NULLSTRPTR(NULL) NULLSTRADDR REORDER ORDINAL(MIN) NOOVERLAP
PSEUDODUMMY NONRECURSIVE NORETCODE RETURNS(BYADDR)
SHORT(HEXADEC))
DEPRECATE
This option flags variable names and included file names that you want to
deprecate.

,
DEPRECATE ( BUILTIN ( ) )
built-in-name
ENTRY ( )
entry-name
INCLUDE ( )
filename
VARIABLE ( )
variable-name
BUILTIN
Flags any declaration of built-in-name with the BUILTIN attribute.
built-in-name
Name of the BUILTIN variable
ENTRY
Flags any declaration of entry-name with the ENTRY attribute.
entry-name
Level-1 name
INCLUDE
Flags any %INCLUDE statement that includes filename.
filename
Name of the file
VARIABLE
Flags any declaration of variable name that does not have the BUILTIN or
ENTRY attributes.
variable-name
Level-1 name
To specify the DEPRECATE option, you must specify at least one of these
suboptions with a possible empty suboption list. For example, both of the
following two specifications are invalid:
v DEPRECATE
v DEPRECATE(BUILTIN)
Specifying one of the suboptions does not change the setting of any of the other
suboptions that are specified previously.
Specifying a suboption a second time replaces the previous specifications.
In all cases, there is no checking of the suboption lists.
Examples:
v The following specifications are equivalent:
DEPRECATE(ENTRY(old)) DEPRECATE(BUILTIN(acos))
DEPRECATE(ENTRY(old) BUILTIN(acos))
v In the following example, x in the first specification is replaced by y:
DEPRECATE(BUILTIN(x)) DEPRECATE(BUILTIN(y))
DEPRECATE(BUILTIN(y))
Default: DEPRECATE(BUILTIN() ENTRY() INCLUDE() VARIABLE())
DISPLAY
The DISPLAY option determines how the DISPLAY statement performs I/O.
DISPLAY ( WTO )
,
( ROUTCDE ( x ) )
,
, DESC ( y )
,
, REPLY ( z )
STD
STD
All DISPLAY statements are completed by writing the text to stdout and
reading any REPLY text from stdin.
WTO
All DISPLAY statements without REPLY are completed via WTOs, and all
DISPLAY statements with REPLY are completed via WTORs. This is the
default.
The following suboptions are supported
ROUTCDE
Specifies one or more values to be used as the ROUTCDE in the WTO. The
default ROUTCDE is 2.
DESC
Specifies one or more values to be used as the DESC in the WTO. The
default DESC is 3.
REPLY
Specifies one or more values to be used as the DESC in the WTOR. If
omitted, the value from the DESC option (or default) is used.
All values specified for the ROUTCDE, DESC and REPLY must between 1 and
16.
DLLINIT
The DLLINIT option applies OPTIONS(FETCHABLE) to all external procedures
that are not MAIN. It should be used only on compilation units containing one
external procedure, and then that procedure should be linked as a DLL.
NODLLINIT
DLLINIT
NODLLINIT has no effect on your programs.
EXIT
The EXIT option enables the compiler user exit to be invoked.
NOEXIT
EXIT
( inparm_string )
inparm_string
A string that is passed to the compiler user exit routine during initialization.
The string can be up to 31 characters long.
See Chapter 22, Using user exits, on page 527 for more information on how to
exploit this option.
EXTRN
The EXTRN option controls when EXTRNs are emitted for external entry constants.
FULL
EXTRN ( SHORT )
FULL
EXTRNs are emitted for all declared external entry constants. This is the
default.
SHORT
EXTRNs are emitted only for those constants that are referenced.
FLAG
The FLAG option specifies the minimum severity of error that requires a message
listed in the compiler listing.
FLAG
W
( I )
E
S
ABBREVIATION: F
I List all messages.
W List all except information messages.
E List all except warning and information messages.
S List only severe error and unrecoverable error messages.
If messages are below the specified severity or are filtered out by a compiler exit
routine, they are not listed.
FLOAT
The FLOAT option controls the use of the additional floating-point registers and
whether Decimal Floating Point is supported.

,
AFP
FLOAT ( )
NOVOLATILE
( VOLATILE )
NOAFP
NODFP
DFP
NOAFP
Compiler-generated code uses the traditional 4 floating-point registers.
AFP
Compiler-generated code may use all 16 floating-point registers.
VOLATILE
The compiler will not expect that the registers FPR8-FPR15 will be
preserved by any called program. Consequently, the compiler will generate
extra code to protect these registers.
Extra code will be generated even if no floating-point registers are used,
and this will hurt performance. If your application makes little or no use of
floating-point, it would be much better to compile with FLOAT(NOAFP)
than with FLOAT(AFP(VOLATILE)).
NOVOLATILE
The compiler will expect that the registers FPR8-FPR15 will be preserved
by any called program (as z/OS says they should be). This option will
produce the most optimal code and is highly recommended if you have no
CICS code. However, it must be used with care in any code included in a
CICS application. For a CICS application,
v either: all code containing EXEC CICS statements must be compiled with
FLOAT(AFP(VOLATILE))
v or: all code using floating-point must be compiled with FLOAT(NOAFP)
or with FLOAT(AFP(VOLATILE))
It is safe and probably simplest not to use the FLOAT(AFP(NOVOLATILE))
option in any code that runs as part of a CICS application.
However, if you are using CICS TS V4.1 or later and are using either z/OS
1.11 or have the PTF for APAR PK71900 applied to z/OS 1.9 or 1.10, then
these restrictions no longer apply to CICS programs.
DFP
The DFP facility is exploited: all DECIMAL FLOAT data will be held in the
DFP format described in the z/OS Principles of Operations manual and
operations using DECIMAL FLOAT will be carried using the DFP hardware
instructions described therein.
The ARCH option must be 7 (or greater) or this option will be rejected.
NODFP
The DFP facility is not exploited.
Under FLOAT(DFP),
v the maximum precision for extended DECIMAL FLOAT will be 34 (not 33 as it
is for hex float)
v the maximum precision for short DECIMAL FLOAT will be 7 (not 6 as it is for
hex float)
v the values for DECIMAL FLOAT for the following built-ins will all have the
appropriate changes
EPSILON
HUGE
MAXEXP
MINEXP
PLACES
RADIX
TINY
v the following built-in functions will all return the appropriate values for
DECIMAL FLOAT (and values that will be much easier for a human to
understand; for example SUCC(1D0) will be 1.000_000_000_000_001, and the
ROUND function will round at a decimal place of course)
EXPONENT
PRED
ROUND
SCALE
SUCC
v decimal floating-point literals, when converted to "right-units-view", i.e. when
the exponent has been adjusted, if needed, so that no non-zero digits follow the
decimal point (for example, as would be done when viewing 3.1415E0 as
31415E-4), must have an exponent within the range of the normal numbers for
the precision given by the literal. These bounds are given by the value of
MINEXP-1 and MAXEXP-1. In particular, the following must hold
For short float, -95 <= exponent <= 90
For long float, -383 <= exponent <= 369
For extended float, -6143 <= exponent <= 6111
v when a DECIMAL FLOAT is converted to CHARACTER, the string will hold 4
digits for the exponent (as opposed to the 2 digits used for hexadecimal float)
v the IEEE and HEXADEC attributes will be accepted only if applied to FLOAT
BIN, and the DEFAULT(IEEE/HEXADEC) option will apply only to FLOAT
BIN.
v the mathematical built-in functions (ACOS, COS, SQRT, etc) will accept
DECIMAL FLOAT arguments and use corresponding Language Environment
functions to evaluate them. DECIMAL FLOAT exponentiation will be handled in
a similar way.
v users of DFP need to be wary of the conversions that will arise in operations
where one operand is FLOAT DECIMAL and the other is binary (i.e. FIXED
BINARY, FLOAT BINARY or BIT). In such operations, the PL/I language rules
dictate that the FLOAT DECIMAL operand be converted to FLOAT BINARY,
and that conversion will require a library call. So, for example, for an
assignment of the form A = A + B; where A is FLOAT DECIMAL and B is FIXED
BINARY, 3 conversions will be done and 2 of those will be library calls:
1. A will be converted via library call from FLOAT DECIMAL to FLOAT
BINARY
2. B will be converted via inline code from FIXED BINARY to FLOAT BINARY
3. the sum A + B will be converted via library call from FLOAT BINARY to
FLOAT DECIMAL
The use of the DECIMAL built-in function would help here: if the statement
were changed to A = A + DEC(B);, the library calls would be eliminated. The
library calls could be eliminated by previously assign B to a FLOAT DECIMAL
temporary variable and then adding that to A.
v the built-in function SQRTF will not be supported for DECIMAL FLOAT
arguments (as there is no hardware instruction to which it can be mapped)
v DFP is not supported by the CAST type function.
FLOATINMATH
The FLOATINMATH option specifies that the precision that the compiler should
use when invoking the mathematical built-in functions.
ASIS
FLOATINMATH ( LONG )
EXTENDED
ASIS
Arguments to the mathematical built-in functions will not be forced to have
long or extended floating-point precision.
LONG
Any argument to a mathematical built-in function with short floating-point
precision will be converted to the maximum long floating-point precision to
yield a result with the same maximum long floating-point precision.
EXTENDED
Any argument to a mathematical built-in function with short or long
floating-point precision will be converted to the maximum extended
floating-point precision to yield a result with the same maximum extended
floating-point precision.
A FLOAT DEC expression with precision p has short floating-point precision if p
<= 6, long floating-point precision if 6 < p <= 16 and extended floating-point
precision if p > 16.
A FLOAT BIN expression with precision p has short floating-point precision if p <=
21, long floating-point precision if 21 <p <= 53 and extended floating-point
precision if p > 53.
The maximum extended floating-point precision depends on the platform.
GOFF
The GOFF option instructs the compiler to produce an object file in the
Generalized Object File Format (GOFF).
NOGOFF
GOFF
When the GOFF and OBJECT options are in effect, the compiler produces an object
file in GOFF format.
When the NOGOFF and OBJECT options are in effect, the compiler produces an
object file in XOBJ format.
The GOFF format supersedes the S/370 Object Module and Extended Object
Module formats. It removes various limitations of the previous format (for
example, 16 MB section size) and provides a number of useful extensions,
including native z/OS support for long names and attributes. GOFF incorporates
some aspects of industry standards such as XCOFF and ELF.
When you specify the GOFF option, you must use the binder to bind the output
object. You cannot use the prelinker to process GOFF objects.
The following options are not supported with the GOFF option:
v COMMON
v NOWRITABLE(PRV)
Note: When using GOFF and source files with duplicate file names, the linker may
emit an error and discard one of the code sections. In this case, turn off the
CSECT option by specifying NOCSECT.
GONUMBER
The GONUMBER option specifies that the compiler produces additional
information that allows line numbers from the source program to be included in
runtime messages.
NOGONUMBER
NOSEPARATE
GONUMBER ( SEPARATE )
ABBREVIATIONS: GN, NGN
SEPARATE
Places the generated statement number table in the separate debug file if the
TEST(SEPARATE) option is specified. When you use GONUMBER(SEPARATE),
the statement numbers are not available for inclusion in runtime messages.
NOSEPARATE
Places the generated statement number table in the object deck.
Alternatively, the line numbers can be derived by using the offset address, which is
always included in runtime messages, and either the table produced by the
OFFSET option or the assembler listing produced by the LIST option.
GONUMBER is forced by the ALL and STMT suboptions of the TEST option.
Note that the GOSTMT option does not exist. The only option that produces
information at run time identifying where an error has occurred is the
GONUMBER option. When the GONUMBER option is used, the term statement in
the runtime error messages refers to the line numbers as used by the NUMBER
compiler option, even if the STMT option is in effect.
If the GONUMBER(SEPARATE) option is specified without TEST(SEPARATE), it is
changed to GONUMBER(NOSEPARATE).
If both TEST and NOGONUMBER are specified, the NOGONUMBER option is
changed to GONUMBER(NOSEPARATE).
The default is NOGONUMBER.
For compatibility, when the GONUMBER option is specified, the default suboption
is SEPARATE.
GRAPHIC
The GRAPHIC option specifies that the source program can contain double-byte
characters. The hexadecimal codes '0E' and '0F' are treated as the shift-out and
shift-in control codes, respectively, wherever they appear in the source program,
including occurrences in comments and string constants.
NOGRAPHIC
GRAPHIC
ABBREVIATIONS: GR, NGR
The GRAPHIC option will also cause the GRAPHIC ENVIRONMENT option to be
applied to any STREAM file used in the compilation.
The GRAPHIC option must be specified if the source program uses any of the
following:
v DBCS identifiers
v Graphic string constants
v Mixed-string constants
v Shift codes anywhere else in the source
HGPR
The HGPR option specifies that the compiler is permitted to exploit 64-bit General
Purpose Registers (GPRs) in 32-bit programs targeting z/Architecture hardware.
NOHGPR
HGPR
NOPRESERVE
( PRESERVE )
PRESERVE
PRESERVE instructs the compiler to preserve the high halves of the 64-bit
GPRs that a function is using, by saving them in the prolog for the function
and restoring them in the epilog. The PRESERVE suboption is necessary only if
the caller is not known to be Enterprise PL/I and/or the z/OS XL C/C++
compiler-generated code.
NOPRESERVE
NOPRESERVE lets the compiler omit preserving the high halves of the 64-bit
GPRs that a function is using. Because of performance considerations, the
default suboption for HGPR is NOPRESERVE.
HGPR means "High half of 64-bit GPR", which refers to the use of native 64-bit
instructions. In particular, if the application use 8-byte integers, it should benefit
from the native 64-bit instructions.
Note: The NOHGPR option must be used in all CICS applications.
IGNORE
The IGNORE option controls whether or not DISPLAY and PUT statements are
ignored.

NOIGNORE
,
IGNORE ( DISPLAY )
PUT
DISPLAY
The compiler will ignore all DISPLAY statements.
PUT
The compiler will ignore all PUT FILE statements.
INCAFTER
The INCAFTER option specifies a file to be included after a particular statement in
your source program.
INCAFTER ( )
PROCESS ( filename )
filename
Name of the file to be included after the last PROCESS statement.
Currently, PROCESS is the only suboption and specifies the name of a file to be
included after the last PROCESS statement.
INCAFTER(PROCESS(DFTS))
This example is equivalent to having the statement %INCLUDE DFTS; after the
last PROCESS statement in your source.
INCDIR
The INCDIR compiler option specifies a directory to be added to the search path
used to locate of include files.
NOINCDIR
INCDIR ( directory name )
directory name
Name of the directory that should be searched for include files. You can
specify the INCDIR option more than once and the directories are searched in
order.
Except under batch, the compiler looks for INCLUDE files in the following order:
1. Current directory
2. Directories specified with the I flag or with the INCDIR compiler option
3. /usr/include directory
4. PDS specified with the INCPDS compiler option
Under batch, this option is probably best used with the DFT(LOWERINC) option,
and it affects only includes of the form "%include x;" For these includes, an hfs file
with the name x.inc will be sought first in the directories specified in this option.
If not found, x must be a member of the pds(e) specified in the syslib dd For
includes of the form "%include dd(x);" no hfs file will ever be included: the
member x must always be a member of the pds named by the specified dd.
INCPDS
The INCPDS option specifies a PDS from which the compiler will include files
when compiling a program under z/OS UNIX.
Note: This option applies only to compilations under z/OS UNIX.
NOINCPDS
INCPDS ( PDS name )
PDS name
Name of a PDS from which files will be included.
For example, if you wanted to compile the program TEST from a PDS named
SOURCE.PLI and wanted to use the INCLUDE files from the PDS SOURCE.INC,
then you could specify the following command
pli -c -qincpds="SOURCE.INC" "//'SOURCE.PLI(TEST)'"
The compiler looks for INCLUDE files in the following order:
2. Directories specified with the I flag or with the INCDIR compiler option
INITAUTO
The INITAUTO option directs the compiler to add an INITIAL attribute to any
AUTOMATIC variable declared without an INITIAL attribute.
NOINITAUTO
INITAUTO
FULL
( SHORT )
Under INITAUTO(FULL), the compiler adds an INITIAL attribute to any

AUTOMATIC variable that does not have an INITIAL attribute according to its
data attributes:
v INIT( (*) 0 ) if it is FIXED or FLOAT
v INIT( (*) '') if it is PICTURE, CHAR, BIT, GRAPHIC or WIDECHAR
v INIT( (*) SYSNULL() ) if it is POINTER or OFFSET
The compiler will not add an INITIAL attribute to variables with other attributes.
INITAUTO will cause more code to be generated in the prologue for each block
containing any AUTOMATIC variables that are not fully initialized (but unlike the
DFT(INITFILL) option, those variables will now have meaningful initial values)
and will have a negative impact on performance.
The INITAUTO option does not apply an INITIAL attribute to any variable
declared with the NOINIT attribute.
Under INITAUTO(SHORT), the compiler adds an INITIAL attribute to an
AUTOMATIC variable that does not have an INITIAL attribute only if it is also a
scalar and has one of the following attributes:
v POINTER
v OFFSET
v FIXED BIN
v FLOAT
v NONVARYING BIT
v NONVARYING CHAR(1)
v NONVARYING WCHAR(1)
As under INITAUTO(FULL), the added INITIAL attribute will be appropriate to
the data type.
If you are using the runtime STORAGE option to zero-out all storage, the
optimizer may still generate undesired code for uninitialized AUTOMATIC
variables and especially for those that are optimized to registers. Scalar variables
with the data types listed above are those most likely to be optimized to registers,
and hence using INITAUTO(SHORT) might have less of a performance impact
than using DFT(INITFILL); however, the latter would leave the optimizer with no
ambiguous code. Moreover, your code is correct only if all variables are explicitly
initialized before use.
INITBASED
The INITBASED option directs the compiler to add an INITIAL attribute to any
BASED variable declared without an INITIAL attribute.
NOINITBASED
INITBASED
This option performs the same function as INITAUTO except for BASED variables.
The INITBASED option will cause more code to be generated for any ALLOCATE
of a BASED variable that is not fully initialized and will have a negative impact on
performance.
The INITBASED option does not apply an INITIAL attribute to any variable
INITCTL
The INITCTL option directs the compiler to add an INITIAL attribute to any
CONTROLLED variable declared without an INITIAL attribute.
NOINITCTL
INITCTL
This option performs the same function as INITAUTO except for CONTROLLED
variables.
The INITCTL option will cause more code to be generated for any ALLOCATE of a
CONTROLLED variable that is not fully initialized and will have a negative
impact on performance.
The INITCTL option does not apply an INITIAL attribute to any variable declared
with the NOINIT attribute.
INITSTATIC
The INITSTATIC option directs the compiler to add an INITIAL attribute to any
STATIC variable declared without an INITIAL attribute.
NOINITSTATIC
INITSTATIC
This option performs the same function as INITAUTO except for STATIC variables.
The INITSTATIC option could make some objects larger and some compilations
longer, but should otherwise have no impact on performance.
The INITSTATIC option does not apply an INITIAL attribute to any variable
INSOURCE
The INSOURCE option specifies that the compiler should include a listing of the
source program before the PL/I macro, CICS, or SQL preprocessors translate it.
NOINSOURCE
INSOURCE
FULL
( SHORT )
ABBREVIATION: IS, NIS

FULL
The INSOURCE listing will ignore %NOPRINT statements and will contain all
the source before the preprocessor translates it.
FULL is the default.
SHORT
The INSOURCE listing will heed %PRINT and %NOPRINT statements.
If the INSOURCE option is in effect, and the compiler invokes more than one of
the macro, CICS or SQL preprocessors, the listing contains the source for each of
these preprocessors.
Under the INSOURCE option, text is included in the listing not according to the
logic of the program, but as each file is read. So, for example, consider the
following simple program which has a %INCLUDE statement between its PROC
and END statements.
insource: proc options(main);
%include member;
end;
The INSOURCE listing will contain all of the main program before any of the
included text from the file "member" (and it would contain all of that file before
any text included by it - and so on).
Under the INSOURCE(SHORT) option, text included by a %INCLUDE statement
inherits the print/noprint status that was in effect when the %INCLUDE statement
was executed, but that print/noprint status is restored at the end of the included
text (however, in the SOURCE listing, the print/noprint status is not restored at
the end of the included text).
INTERRUPT
The INTERRUPT option causes the compiled program to respond to attention
requests (interrupts).
NOINTERRUPT
INTERRUPT
ABBREVIATION: INT, NINT
This option determines the effect of attention interrupts when the compiled PL/I
program runs under an interactive system. This option will have an effect only on
programs running under TSO. If you have written a program that relies on raising
the ATTENTION condition, you must compile it with the INTERRUPT option. This
option allows attention interrupts to become an integral part of programming. This
gives you considerable interactive control of the program.
If you specify the INTERRUPT option, an established ATTENTION ON-unit gets
control when an attention interrupt occurs. When the execution of an ATTENTION
ON-unit is complete, control returns to the point of interrupt unless directed
elsewhere by means of a GOTO statement. If you do not establish an ATTENTION
ON-unit, the attention interrupt is ignored.
If you specify NOINTERRUPT, an attention interrupt during a program run does
not give control to any ATTENTION ON-units.
If you require the attention interrupt capability only for testing purposes, use the
TEST option instead of the INTERRUPT option. For more information see TEST
on page 139.
See Chapter 20, Interrupts and attention processing, on page 521 for more
information about using interrupts in your programs.
LANGLVL
The LANGLVL option specifies the level of PL/I language definition that you want
the compiler to accept.
LANGLVL (
OS
NOEXT )
NOEXT
The only ENVIRONMENT options accepted are:
Bkwd
Consecutive
Ctlasa
Deblock
Genkey
Graphic
Indexed
Keylength
Keyloc
Organization
Recsize
Regional
Relative
Scalarvarying
Vsam
OS All ENVIRONMENT options are allowed. For a complete list of the
ENVIRONMENT options, see Table 14 on page 275.
LIMITS
The LIMITS option specifies various implementation limits.

,
7
LIMITS ( EXTNAME ( n ) )
31
FIXEDBIN ( 63 )
31
, 63
15
FIXEDDEC ( 31 )
15
, 31
100
NAME ( n )
EXTNAME
Specifies the maximum length for EXTERNAL name. The maximum value for
n is 100; the minimum value is 7.
FIXEDDEC
Specifies the maximum precision for FIXED DECIMAL to be either 15 or 31.
The default is 15.
If FIXEDDEC(15,31) is specified, then you may declare FIXED DECIMAL
variables with precision greater than 15, but unless an expression contains an
operand with precision greater than 15, all arithmetic will done using 15 as the
maximum precision.
FIXEDDEC(15,31) will provide much better performance than FIXEDDEC(31).
FIXEDDEC(15) and FIXEDDEC(15,15) are equivalent; similarly, FIXEDDEC(31)
and FIXEDDEC(31,31) are equivalent.
FIXEDDEC(31,15) is not allowed.
FIXEDBIN
Specifies the maximum precision for SIGNED FIXED BINARY to be either 31
or 63. The default is 31.
If FIXEDBIN(31,63) is specified, then you may declare 8-byte integers, but
unless an expression contains an 8-byte integer, all integer arithmetic will done
using 4-byte integers.
Note however, that specifying the FIXEDBIN(31,63) or FIXEDBIN(63) option
may cause the compiler to use 8-byte integer arithmetic for expressions mixing
data types. For example, if a FIXED BIN(31) value is added to a FIXED
DEC(13) value, then the compiler will produce a FIXED BIN result and under
LIMITS(FIXEDBIN(31,63)) that result would have a precision greater than 31
(because the FIXED DEC precision is greater than 9). When this occurs, the
compiler will issue informational message IBM2809.
FIXEDBIN(31,63) will provide much better performance than FIXEDBIN(63).
FIXEDBIN(31) and FIXEDBIN(31,31) are equivalent; similarly, FIXEDBIN(63)
and FIXEDBIN(63,63) are equivalent.
FIXEDBIN(63,31) is not allowed.
The maximum precision for UNSIGNED FIXED BINARY is one greater, that is,
32 and 64.
NAME
Specifies the maximum length of variable names in your program. The
maximum value for n is 100; the minimum value is 31.
LINECOUNT
The LINECOUNT option specifies the number of lines per page for compiler
listings, including blank and heading lines.
LINECOUNT
60
( n )
ABBREVIATION: LC
n The number of lines in a page in the listing. The value can be from 10 to
32,767.
LINEDIR
This option specifies that the compiler should accept %LINE directives.
NOLINEDIR
LINEDIR
If the LINEDIR option is specified, the compiler will reject all %INCLUDE
statements. If the LINEDIR option is specified, the compiler will also reject the use
of the SEPARATE suboption of the TEST option.
LIST
The LIST option specifies that the compiler should produce a pseudo-assembler
listing.
NOLIST
LIST
Specifying the LIST option will increase the time and region required for a
compilation. The OFFSET and MAP options may provide the information you need
at much less cost.
The pseudo-assembler listing will also include at the end of the listing for each
block the offset of the first instruction in that block from the start of the whole
compilation unit.
LISTVIEW
The LISTVIEW option specifies whether the compiler should show the source in
the source listing or whether it should show the source after it has been processed
by one or more of the preprocessors.
SOURCE
LISTVIEW ( AFTERALL )
AFTERCICS
AFTERMACRO
AFTERSQL
SOURCE
Causes the source listing to show the unadulterated source and, more
importantly perhaps, it will cause Debug Tool to bring up this as the source
view.
AFTERALL
Causes the source listing to show the source as if it came from the MDECK
from the last invocation, if any, of the last preprocessor, and, more importantly
perhaps, it will cause Debug Tool to bring up this as the source view if the
SEPARATE suboption of the TEST compiler option is also specified,
AALL may be used as an abbreviation for AFTERALL
AFTERCICS
from the last invocation, if any, of the CICS preprocessor, and, more
view if the SEPARATE suboption of the TEST compiler option is also specified.
ACICS may be used as an abbreviation for AFTERCICS
AFTERMACRO
from the last invocation, if any, of the MACRO preprocessor, and, more
view if the SEPERATE suboption of the TEST compiler option is also specified.
AMACRO may be used as an abbreviation for AFTERMACRO
AFTERSQL
from the last invocation, if any, of the SQL preprocessor, and, more importantly
perhaps, it will cause Debug Tool to bring up this as the source view if the
SEPARATE suboption of the TEST compiler option is also specified.
ASQL may be used as an abbreviation for AFTERSQL
If the TEST option is specified and a suboption other than SOURCE is specified for
LISTVIEW then the SEPARATE suboption must also be specified for the TEST
option.
As an example of the differing effects of the AFTERMACRO, AFTERSQL and
AFTERALL suboptions suppose the PP option was PP( MACRO('INCONLY'), SQL,
MACRO). Then:
v Under LISTVIEW(AFTERMACRO), the "source" in the listing and in the Debug
Tool source window if TEST(SEP) were specified would appear as if it came
from the MDECK that the second invocation of the MACRO preprocessor would
have produced
v Under LISTVIEW(AFTERSQL), the "source" in the listing and in the Debug Tool
source window if TEST(SEP) were specified would appear as if it came from the
MDECK that the invocation of the SQL preprocessor would have produced (and
hence %DCL and other macro statements would still be visible)
v Under LISTVIEW(AFTERALL), the "source" would be as under the
LISTVIEW(AFTERMACRO) option since the MACRO preprocessor is the last in
the PP option
MACRO
The MACRO option invokes the MACRO preprocessor.
NOMACRO
MACRO
ABBREVIATIONS: M, NM
You may also invoke the MACRO preprocessor via the PP(MACRO) option.
However, the use of both the MACRO option and the PP(MACRO) option in the
same compilation is not recommended.
For more discussion of the PP option, see PP on page 100.
For more discussion of the MACRO preprocessor, see Macro preprocessor on
page 167.
MAP
The MAP option specifies that the compiler produces additional information that
can be used to locate static and automatic variables in dumps.
NOMAP
MAP
MARGINI
The MARGINI option specifies a character that the compiler will place in the
column preceding the left-hand margin, and also in the column following the
right-hand margin, of the listings produced by the INSOURCE and SOURCE
options.
NOMARGINI
MARGINI ( ' c ' )
ABBREVIATIONS: MI, NMI
c The character to be printed as the margin indicator.
Note: NOMARGINI is equivalent to MARGINI(' ').
MARGINS
The MARGINS option specifies which part of each compiler input record contains
PL/I statements, and the position of the ANS control character that formats the
listing, if the SOURCE and/or INSOURCE options apply. The compiler does not
process data that is outside these limits, but it does include it in the source listings.
The PL/I source is extracted from the source input records so that the first data
byte of a record immediately follows the last data byte of the previous record. For
variable records, you must ensure that when you need a blank you explicitly insert
it between margins of the records.
2 72
MARGINS ( m , n )
, c
NOMARGINS
ABBREVIATION: MAR
m The column number of the leftmost character (first data byte) that is processed
by the compiler. It must not exceed 100.
n The column number of the rightmost character (last data byte) that is
processed by the compiler. It should be greater than m, but must not exceed
200, except under MVS batch where it must not exceed 100.
Variable-length records are effectively padded with blanks to give them the
maximum record length.
c The column number of the ANS printer control character. It must not exceed
200, except under MVS batch where it must not exceed 100, and it should be
outside the values specified for m and n. A value of 0 for c indicates that no
ANS control character is present. Only the following control characters can be
used:
(blank)
Skip one line before printing
0 Skip two lines before printing
Skip three lines before printing
+ No skip before printing
1 Start new page
Any other character is an error and is replaced by a blank.
Do not use a value of c that is greater than the maximum length of a source
record, because this causes the format of the listing to be unpredictable. To
avoid this problem, put the carriage control characters to the left of the source
margins for variable-length records.
Specifying MARGINS(,,c) is an alternative to using %PAGE and %SKIP
statements (described in PL/I Language Reference).
The IBM-supplied default for fixed-length records is MARGINS(2,72). For
variable-length and undefined-length records, the IBM-supplied default is
MARGINS(10,100). This specifies that there is no printer control character.
Use the MARGINS option to override the default for the primary input in a
program. The secondary input must have the same margins as the primary input.
The NOMARGINS option will suppress any previously encountered instance of the
MARGINS option. The purpose of this option is allow your installation to have a
default set of compile-time options which use a MARGINS option tailored for their
fixed format source preferences while retaining the ability to use variable source
format files.
You would usually specify the NOMARGINS option, if you use it at all, as part of
the parameter-string passed to the compiler. The compiler will ignore
NOMARGINS if it finds the option in a %PROCESS statement.
MAXGEN
MAXGEN (size)
The MAXGEN option specifies the maximum number of intermediate language
statements that should be generated for any one user statement and will cause the
compiler to flag any statement where this maximum is exceeded.
The number of intermediate language statements generated for any user statement
may vary depending on the compiler release, the compiler maintenance level, and
the compiler options in effect. This option is intended to be used only to help find
statements for which excessive amounts of code are generated and which thus may
indicate that they are perhaps are poorly coded.
However, it should be noted that using a preprocessor may cause the number of
intermediate language statements generated for some statements to be very large.
In such a situation, it may be better either to set the MAXGEN threshold to be
larger or to use the LISTVIEW(AFTERALL) option.
The default is MAXGEN(100000).
MAXMEM
When compiling with OPTIMIZE, the MAXMEM option limits the amount of
memory used for local tables of specific, memory intensive optimizations to the
specified number of kilobytes. The minimum number of kilobytes that may be
specified is 1. The maximum number of kilobytes that may be specified is 2097152,
and the default is 1048576.
If you specify the maximum value of 2097152, the compiler will assume that
unlimited memory is available. If you specify any smaller value for MAXMEM, the
compiler, especially when the OPT(2) option is in effect, may issue a message
saying that optimization is inhibited and that you should try using a larger value
for MAXMEM.
Use the MAXMEM option if you know that less (or more) memory is available
than implied by the default value.
If the memory specified by the MAXMEM option is insufficient for a particular
optimization, the compilation is completed in such a way that the quality of the
optimization is reduced, and a warning message is issued.
MAXMEM (size)
ABBREVIATIONS: MAXM
When a large size is specified for MAXMEM, compilation may be aborted because
of insufficient virtual storage, depending on the source file being compiled, the size
of the subprogram in the source, and the virtual storage available for the
compilation.
The advantage of using the MAXMEM option is that, for large and complex
applications, the compiler produces a slightly less-optimized object module and
generates a warning message, instead of terminating the compilation with an error
message of "insufficient virtual storage".
MAXMSG
The MAXMSG option specifies the maximum number of messages with a given
severity (or higher) that the compilation should produce.
MAXMSG
,
W
( I )
E
S
250
n
I Count all messages.

W Count all except information messages.
E Count all except warning and information messages.
S Count only severe error and unrecoverable error messages.
n Terminate the compilation if the number of messages exceeds this value. If
messages are below the specified severity or are filtered out by a compiler exit
routine, they are not counted in the number. The value of n can range from 0
to 32767. If you specify 0, the compilation terminates when the first error of the
specified severity is encountered.
MAXNEST
The MAXNEST option specifies the maximum nesting of various kinds of
statements that should be allowed before the compiler flags your program as too
complex.

,
17
MAXNEST ( BLOCK ( x ) )
17
DO ( x )
17
IF ( z )
BLOCK
Specifies the maximum nesting of BEGIN and PROCEDURE statements.
DO Specifies the maximum nesting of DO statements.
IF Specifies the maximum nesting of IF statements.
The value of any nesting limit must be between 1 and 50 (inclusive).
The default is MAXNEST( BLOCK(17) DO(17) IF(17) ).
MAXSTMT
Under the MAXSTMT option, optimization will be turned off for any block that
has more than the specified number of statements. Use the MAXSTMT option -
with a reasonable limit to the number of statements - if you want the compiler to
optimize the code generated for a program and are willing for the compiler to
optimize only the reasonably sized blocks in that program.
MAXSTMT (size)
When a large size is specified for MAXSTMT and some blocks have a large
number of statements, compilation may be aborted if there is not enough virtual
storage available.
The default for MAXSTMT is 4096.
MAXTEMP
The MAXTEMP option determines when the compiler flags statements using an
excessive amount of storage for compiler-generated temporaries.
MAXTEMP ( max )
max
The limit for the number of bytes that can be used for compiler-generated
temporaries. The compiler flags any statement that uses more bytes than those
specified bymax . The default for max is 50000.
You should examine statements that are flagged under this option - if you code
them differently, you may be able to reduce the amount of stack storage required
by your code.
MDECK
The MDECK option specifies that the preprocessor produces a copy of its output
either on the file defined by the SYSPUNCH DD statement under z/OS, or on the
.dek file under z/OS UNIX.
NOMDECK
MDECK
ABBREVIATIONS: MD, NMD
The MDECK option allows you to retain the output from the preprocessor as a file
of 80-column records. This option is applicable only when the MACRO option is in
effect.
NAME
The NAME option specifies that the TEXT file created by the compiler will contain
a NAME record.
NONAME
NAME
( 'name' )
ABBREVIATIONS: N
If no 'name' is specified as a suboption of the NAME option, then the 'name' used
is determined as follows
v if there is a PACKAGE statement, the leftmost name on it is used
v otherwise, the leftmost name on the first PROCEDURE statement is used
The length of the 'name' must not be greater than 8 characters if the
LIMITS(EXTNAME(n)) option is used with n <= 8.
NAMES
The NAMES option specifies the extralingual characters that are allowed in
identifiers. Extralingual characters are those characters other than the 26 alphabetic,
10 digit, and special characters defined in PL/I Language Reference.
NAMES (

' extraling_char '
' upp_extraling_char '

,
)
extralingual_char
An extralingual character
upp_extraling_char
The extralingual character that you want interpreted as the uppercase version
of the corresponding character in the first suboption.
If you omit the second suboption, PL/I uses the character specified in the first
suboption as both the lowercase and the uppercase values. If you specify the
second suboption, you must specify the same number of characters as you specify
in the first suboption.
The default is NAMES('#@$' '#@$').
NATLANG
The NATLANG option specifies the "language" for compiler messages, headers,
etc.
NATLANG (
ENU
UEN )
ENU
All compiler messages, headers etc will be in mixedcase English.
UEN
All compiler messages, headers etc will be in uppercase English.
NEST
The NEST option specifies that the listing resulting from the SOURCE option
indicates the block level and the do-group level for each statement.
NONEST
NEST
NOT
The NOT option specifies up to seven alternate symbols that can be used as the
logical NOT operator.
NOT

( ' char ' )
char
defined in PL/I Language Reference, except for the standard logical NOT symbol ().
You must specify at least one valid character.
When you specify the NOT option, the standard NOT symbol is no longer
recognized unless you specify it as one of the characters in the character string.
For example, NOT('~') means that the tilde character, 'A1'X, will be recognized as
the logical NOT operator, and the standard NOT symbol, '', '5F'X, will not be
recognized. Similarly, NOT('~') means that either the tilde or the standard NOT
symbol will be recognized as the logical NOT operator.
The IBM-supplied default code point for the NOT symbol is '5F'X. The logical NOT
sign might appear as a logical NOT symbol () or a caret symbol (^) on your
keyboard.
NUMBER
The number option specifies that statements in the source program are to be
identified by the line and file number of the file from which they derived and that
this pair of numbers is used to identify statements in the compiler listings resulting
from the AGGREGATE, ATTRIBUTES, LIST, MAP, OFFSET, SOURCE and XREF
options. The File Reference Table at the end of the listing shows the number
assigned to each of the input files read during the compilation.
NUMBER
NONUMBER
If a preprocessor has been used, more than one line in the source listing may be
identified by the same line and file numbers. For example, almost every EXEC
CICS statement generates several lines of code in the source listing, but these
would all be identified by one line and file number.
In the pseudo-assembler listing produced by the LIST option, the file number is
left blank for the first file.
NUMBER and STMT are mutually exclusive and specifying one will negate the
other.
The default is NUMBER.
OBJECT
The OBJECT option specifies that the compiler creates an object module. Under
batch z/OS, the compiler stores the object in the data set defined by the SYSLIN
DD, while under z/OS UNIX, the compiler creates a .o file.
OBJECT
NOOBJECT
ABBREVIATIONS: OBJ, NOBJ
Under the NOOBJECT option, the compiler does not create an object module.
However, under the NOOBJECT option, the compiler will complete all of its
syntactic and semantic analysis phases as well as all of its detection of uninitialized
variables, and hence it could produce more messages than under the
NOCOMPILE, the NOSEMANTIC or the NOSYNTAX options.
Under the NOOBJECT option, the LIST, MAP, OFFSET and STORAGE options will
be ignored.
OFFSET
The OFFSET option specifies that the compiler is to print a table of line numbers
for each procedure and BEGIN block with their offset addresses relative to the
primary entry point of the procedure. This table can be used to identify a
statement from a run-time error message if the GONUMBER option is not used.
NOOFFSET
OFFSET
ONSNAP

NOONSNAP
,
ONSNAP ( STRINGRANGE )
STRINGSIZE
For a PROCEDURE with either the OPTIONS(MAIN) or the

OPTIONS(FROMALIEN) attribute. the ONSNAP option specifies that the compiler
should insert an ON STRINGRANGE SNAP; or ON STRINGSIZE SNAP; statement
(or both) into the prologue code for that PROCEDURE. This can make it easier to
determine the calling chain if the corresponding conditions are raised in other
routines called from such a PROCEDURE.
The ONSNAP option has no affect on a PROCEDURE without either of these
attributes.
OPTIMIZE
The OPTIMIZE option specifies the type of optimization required:
NOOPTIMIZE
OPTIMIZE
2
( TIME )
0
3
ABBREVIATIONS: OPT, NOPT

OPTIMIZE(0)
Specifies fast compilation speed, but inhibits optimization.
OPTIMIZE(2)
Optimizes the machine instructions generated to produce a more efficient
object program. This type of optimization can also reduce the amount of main
storage required for the object module.
OPTIMIZE(3)
Performs all the optimizations done under OPTIMIZE(2) plus some additional
optimizations. Under OPTIMIZE(3), the compiler will generally, but especially
for programs with large blocks and many variables, generate smaller and more
efficient object code. However, it may also take considerably more time and
region to complete compiles under OPTIMIZE(3) than under OPTIMIZE(2).
It is strongly recommended that the DFT(REORDER) option be used with the
OPTIMIZE option. In fact, the effect of OPTIMIZE is severely limited for any
PROCEDURE or BEGIN-block for which all of the following are true:
v The ORDER option applies to the block.
v The block contains ON-units for hardware-detected conditions (such as
ZERODIVIDE).
v The block has labels that are the (potential) target of branches out of those
ON-units.
The use of OPTIMIZE(2) could result in a substantial increase in compile time over
NOOPTIMIZE and a substantial increase in the space required. For example,
compiling a large program at OPTIMIZE(2) might take several minutes and could
require a region of 75M or more.
The use of OPTIMIZE(3) will increase the time and region needed for a compile
over what is needed under OPTIMIZE(2). For large programs, the time to compile
a program under OPTIMIZE(3) can be more than twice as long as under
OPTIMIZE(2).
During optimization the compiler can move code to increase runtime efficiency. As
a result, statement numbers in the program listing might not correspond to the
statement numbers used in runtime messages.
NOOPTIMIZE is the equivalent of OPTIMIZE(0).
OPTIMIZE(TIME) is the equivalent of OPTIMIZE(2).
Note that the use of OPTIMIZE(2) or OPTIMIZE(3) severely limits the functionality
of the TEST option. For example:
v if the HOOK suboption of TEST is in effect, then only block hooks will be
generated
v if the NOHOOK suboption of TEST is in effect, attempts to list or change a
variable may fail (because the variable may have been optimized into a register)
and attempts to stop at a particular statement may cause the debugger to stop
several times (because the statement may have split up into several parts)
The use of the PREFIX option with one or more of the checkout conditions (SIZE,
STRINGRANGE, STRINGSIZE and SUBSCRIPTRANGE) can significantly increase
the time and space needed for a compile.
For more information on choosing the best options to improve the performance of
your code, see Chapter 13, Improving performance, on page 369.
OPTIONS
The OPTIONS option specifies that the compiler includes a list showing the
compiler options to be used during this compilation in the compiler listing.
NOOPTIONS
OPTIONS
DOC
( ALL )
ABBREVIATIONS: OP, NOP

This list includes all options applied by default, those specified in the PARM
parameter of an EXEC statement or in the invoking command (pli), those specified
in a %PROCESS statement, those specified in the IBM_OPTIONS environment
variable under z/OS, and all those incorporated from any options file.
Under OPTIONS(DOC), the OPTIONS listing will include only those options (and
suboptions) documented in this document at the time of the compilers release.
Under OPTIONS(ALL), the OPTIONS listing will also include any option added by
PTF after the compilers release.
OR
The OR option specifies up to seven alternate symbols as the logical OR operator.
These symbols are also used as the concatenation operator, which is defined as two
consecutive logical OR symbols.
OR

( ' char ' )
The IBM-supplied default code point for the OR symbol (|) is '4F'X.
char
defined in the PL/I Language Reference, except for the standard logical OR symbol
(|). You must specify at least one valid character.
If you specify the OR option, the standard OR symbol is no longer recognized
unless you specify it as one of the characters in the character string.
For example, OR('\') means that the backslash character, 'E0'X, will be recognized
as the logical OR operator, and two consecutive backslashes will be recognized as
the concatenation operator. The standard OR symbol, '|', '4F'X, will not be
recognized as either operator. Similarly, OR('\|') means that either the backslash or
the standard OR symbol will be recognized as the logical OR operator, and either
symbol or both symbols can be used to form the concatenation operator.
PP
The PP option specifies which (and in what order) preprocessors are invoked
before compilation.

NOPP
,
PP ( pp-name )
( pp-string )
pp-name
The name given to a particular preprocessor. CICS, INCLUDE, MACRO and
SQL are the only preprocessors currently supported. Using an undefined name
causes a diagnostic error.
pp-string
A string, delimited by quotation marks, of up to 100 characters representing
the options for the corresponding preprocessor. For example,
PP(MACRO('CASE(ASIS)')) invokes the MACRO preprocessor with the option
CASE(ASIS).
Preprocessor options are processed from left to right, and if two options conflict,
the last (rightmost) option is used. For example, if you invoke the MACRO
preprocessor with the option string 'CASE(ASIS) CASE(UPPER)', then the option
CASE(UPPER) is used.
You can specify a maximum of 31 preprocessor steps, and you can specify the
same preprocessor more than once with the exception of the CICS and SQL
preprocessors. The CICS preprocessor must be invoked at most once, and the SQL
preprocessor must be invoked no more than twice. The SQL preprocessors can be
invoked twice only if the first specification specifies INCONLY as its option.
The MACRO option and the PP(MACRO) option both cause the macro facility to
be invoked before compilation. If both MACRO and PP(MACRO) are specified, the
macro facility is invoked twice. However, the use of both the MACRO option and
the PP(MACRO) option in the same compilation is not recommended.
If you specify the PP option more than once, the compiler effectively concatenates
them. So specifying PP(SQL) PP(CICS) is the same as specifying PP(SQL CICS).
This also means that if you specified PP(MACRO SQL('CCSID0')) and PP(MACRO
SQL('CCSID0 DATE(ISO)')), then the resulting PP option would be PP( MACRO
SQL('CCSID0') MACRO SQL('CCSID0 DATE(ISO)')), both the MACRO and SQL
preprocessor would be invoked twice, and the second invocation of the SQL
preprocessor would be in error. If you were doing this in an attempt to override
the earlier SQL options, it might be better not to specify the preprocessor options
in the PP option, but rather to specify them through the PPSQL option, that is,
specify PP(MACRO SQL) PPSQL('CCSID0 DATE(ISO)').
For more discussion of the preprocessors, see Chapter 2, PL/I preprocessors, on
page 165.
PPCICS
The PPCICS option specifies options to be passed to the CICS preprocessor if it is
invoked.
NOPPCICS
PPCICS ( options string )
So, specifying PPCICS('EDF') PP(CICS) is the same as specifying PP(CICS('EDF')).
This option has no effect unless the PP(CICS) option is specified. However, if you
want to specify a set of CICS preprocessor options that should be used if and
when the CICS preprocessor is invoked, you could specify this option in the
installation options exit. Then whenever you specified PP(CICS), the set of options
specified in the PPCICS option would be used.
Also, any options specified when the preprocessor is invoked overrule those
specified in the PPCICS option. So specifying PPCICS('EDF') PP(CICS('NOEDF')) is
the same as specifying PP(CICS('EDF NOEDF')) or the even simpler
PP(CICS('NOEDF')).
The options string is limited to 1000 characters in length. However, if the string is
longer than 100 characters, it will not be shown in the options listing.
PPINCLUDE
The PPINCLUDE option specifies options to be passed to the INCLUDE
preprocessor if it is invoked.
NOPPINCLUDE
PPINCLUDE ( options string )
So, specifying PPINCLUDE('ID(-inc)') PP(INCLUDE) is the same as specifying
PP(INCLUDE('ID(-inc)')).
This option has no effect unless the PP(INCLUDE) option is specified. However, if
you want to specify a set of INCLUDE preprocessor options that should be used if
and when the INCLUDE preprocessor is invoked, you could specify this option in
the installation options exit. Then whenever you specified PP(INCLUDE), the set of
options specified in the PPINCLUDE option would be used.
specified in the PPINCLUDE option. So specifying PPINCLUDE('ID(-inc)')
PP(INCLUDE('ID(+include)')) is the same as specifying PP(INCLUDE('ID(-inc)
ID(+include)')) or the even simpler PP(INCLUDE('ID(+include)')).
PPLIST
The PPLIST option controls whether the compiler keeps or erases the part of the
listing that is generated by each preprocessor phase.
PPLIST (
KEEP
ERASE )
When you specify PPLIST(KEEP), the compiler keeps the part of the listing that is
generated by each preprocessor phase.
When you specify PPLIST(ERASE), the compiler erases the part of the listing that
is generated by any preprocessor phase that produces no messages.
The compiler does not count messages that are suppressed by the EXIT and FLAG
options. Therefore, specifying both FLAG(W) and PPLIST(ERASE) causes the
compiler to suppress all output from any preprocessor that produces no warning,
error or severe messages.
PPLIST(KEEP) is the default.
PPMACRO
The PPMACRO option specifies options to be passed to the MACRO preprocessor
if it is invoked.
NOPPMACRO
PPMACRO ( options string )
So, specifying PPMACRO('CASE(ASIS)') PP(MACRO) is the same as specifying
PP(MACRO('CASE(ASIS)')).
This option has no effect unless the PP(MACRO) option is specified. However, if
you want to specify a set of MACRO preprocessor options that should be used if
and when the MACRO preprocessor is invoked, you could specify this option in
the installation options exit. Then whenever you specified the MACRO or
PP(MACRO) options, the set of options specified in the PPMACRO option would
be used.
specified in the PPMACRO option. So specifying PPMACRO('CASE(ASIS)')
PP(MACRO('CASE(UPPER)')) is the same as specifying PP(MACRO('CASE(ASIS)
CASE(UPPER)')) or the even simpler PP(MACRO('CASE(UPPER)')).
PPSQL
The PPSQL option specifies options to be passed to the SQL preprocessor.
NOPPSQL
PPSQL ( options string )
Specifying PPSQL('APOSTSQL') PP(SQL) is the same as specifying
PP(SQL('APOSTSQL')).
This option has no effect unless the PP(SQL) option is specified. However, if you
want to specify a set of options that should be used if the SQL preprocessor is
invoked, you can specify this option in the installation options exit. Then,
whenever you specify PP(SQL), the set of options in the PPSQL option are used.
Also, any options that are specified when the preprocessor is invoked overrule
those that are specified in the PPSQL option. Therefore, specifying
PPSQL('APOSTSQL') PP(SQL('QUOTESQL')) is the same as specifying
PP(SQL('APOSTSQL QUOTESQL')) or the even simpler PP(SQL('QUOTESQL')).
PPTRACE
The PPTRACE option specifies that, when a deck file is written for a preprocessor,
every nonblank line in that file is preceded by a line containing a %LINE directive.
The directive indicates the original source file and line to which the nonblank line
should be attributed.
NOPPTRACE
PPTRACE
PRECTYPE
The PRECTYPE option determines how the compiler derives the attributes for the
MULTIPLY, DIVIDE, ADD and SUBTRACT built-in functions when the operands
are FIXED and at least one is FIXED BIN.
ANS
PRECTYPE ( DECDIGIT )
DECRESULT
ANS
under PRECTYPE(ANS), the value p in BIF(x, y, p) and in BIF( x, y, p, 0) is
interpreted as specifying a binary number of digits, the operation is performed
as a binary operation and the result has the attributes FIXED BIN(p,0).
However, for BIF( x, y, p, q ) if q is not-zero, then the operation will be
performed as a decimal operation and the result will have the attributes FIXED
DEC(t,u) where t and u are the decimal equivalents of p and q, namely t = 1 +
ceil(p / 3.32) and u = ceil(q / 3.32). In this case, x, y, p and q are effectively all
converted to decimal (in contrast to the next suboption which converts only x
and y to decimal and does so even if q is zero). The compiler will issue the
Informational message 1BM1053 in this situation.
DECDIGIT
under PRECTYPE(DECDIGIT), the value p in BIF(x, y, p) and BIF(x, y, p, 0) is
interpreted as specifying a decimal number of digits, the operation is
performed as a binary operation, and the result has the attributes FIXED
BIN(s) where s is the corresponding binary equivalent to p (namely s =
ceil(3.32*p)). For an instance of BIF(x, y, p, q) where q is not-zero, the results
under PRECTYPE(DECDIGIT) are the same as described below under
PRECTYPE(DECRESULT).
DECRESULT
under PRECTYPE(DECRESULT), the value p in BIF(x, y, p) and the values p
and q in BIF(x, y, p, q) are interpreted as specifying a decimal number of
digits, the operation is performed as a decimal operation and the result has the
attributes FIXED DEC(p,0) or FIXED DEC(p,q) respectively. The result is the
same as would be produced if the DECIMAL built-in were applied to x and y.
PREFIX
The PREFIX option enables or disables the specified PL/I conditions in the
compilation unit being compiled without your having to change the source
program. The specified condition prefixes are logically prefixed to the beginning of
the first PACKAGE or PROCEDURE statement.
PREFIX (
,
condition
)
condition
Any condition that can be enabled/disabled in a PL/I program, as explained
in PL/I Language Reference.
The use of the PREFIX option with one or more of the checkout conditions (SIZE,
STRINGRANGE, STRINGSIZE and SUBSCRIPTRANGE) can significantly increase
the time and space needed for a compile.
Default PREFIX(CONVERSION FIXEDOVERFLOW INVALIDOP OVERFLOW
NOSIZE NOSTRINGRANGE NOSTRINGSIZE NOSUBSCRIPTRANGE
UNDERFLOW ZERODIVIDE)
PROCEED
The PROCEED option stops the compiler after processing by a preprocessor is
completed depending on the severity of messages issued by previous
preprocessors.
NOPROCEED
S
( W )
E
PROCEED
ABBREVIATIONS: PRO, NPRO
PROCEED
Is equivalent to NOPROCEED(S).
NOPROCEED
Ends the processing after the preprocessor has finished compiling.
NOPROCEED(S)
The invocation of preprocessors and the compiler does not continue if a severe
or unrecoverable error is detected in this stage of preprocessing.
NOPROCEED(E)
The invocation of preprocessors and the compiler does not continue if an error,
severe error, or unrecoverable error is detected in this stage of preprocessing.
NOPROCEED(W)
The invocation of preprocessors and the compiler does not continue if a
warning, error, severe error, or unrecoverable error is detected in this stage of
preprocessing.
PROCESS
The PROCESS option determines if *PROCESS statements are allowed and, if they
are allowed, if they are written to the MDECK file.
PROCESS
DELETE
( KEEP )
NOPROCESS
Under the NOPROCESS option, the compiler will flag any *PROCESS statement
with an E-level message.
Under the PROCESS(KEEP) option, the compiler will not flag *PROCESS
statements, and the compiler will retain any *PROCESS statements in the MDECK
output.
Under the PROCESS(DELETE) option, the compiler will not flag *PROCESS
statements, but the compiler will not retain any *PROCESS statements in the
MDECK output.
QUOTE
The QUOTE option specifies an alternate symbol that can be used as the quote
character.
"
QUOTE ( ' char ' )
The IBM-supplied default code point for the QUOTE symbol is '"'.
char
defined in the PL/I Language Reference, except for the standard QUOTE symbol (").
You must specify a valid character.
The QUOTE option is ignored if the GRAPHIC option is also specified.
REDUCE
The REDUCE option specifies that the compiler is permitted to reduce an
assignment of a null string to a structure into simpler operations - even if that
means padding bytes might be overwritten.
The REDUCE option also allows the compiler to reduce the assignment of
matching structures into a simple aggregate move - even if the structures contain
POINTER fields.
REDUCE
NOREDUCE
The NOREDUCE option specifies that the compiler must decompose an
assignment of a null string to a structure into a series of assignments of the null
string to the base members of the structure.
Under the NOREDUCE option, BY NAME assignments that can be reduced to
aggregate moves are not reduced if the elements that would be moved together
had the AREA or VARYING(Z) attributes.
The REDUCE option causes fewer lines of code to be generated for an assignment
of a null string to a structure, and that usually means your compilation is quicker
and your code runs much faster. However, padding bytes may be zeroed out.
For instance, in the following structure, there is one byte of padding between
field12 and field13.
dcl
1 sample ext,
5 field10 bin fixed(31),
5 field12 bit(8),
5 field13 bin fixed(31);
Now consider the assignment sample = ;
Under the NOREDUCE option, it will cause four assignments to be generated, and
the padding byte will be unchanged.
However, under REDUCE, the assignment would be reduced to one operation, but
the padding byte will be zeroed out.
The NOREDUCE option makes the compiler act more like the OS PL/I and the
PL/I for MVS compilers. These compilers would reduce an assignment of
matching structures into a simple aggregate move unless the structures contain
POINTER fields. The NOREDUCE option will make this compiler act the same
way.
RENT
NORENT
RENT
Your code is "naturally reentrant" if it does not alter any of its static variables.
The RENT option specifies that the compiler is to take code that is not naturally
reentrant and make it reentrant. For a detailed description of reentrancy, see the
z/OS Language Environment Programming Guide. If you use the RENT option,
the Linkage Editor cannot directly process the object module that is produced: you
must use either the prelinker or PDSEs.
The NORENT option specifies that the compiler is not to specifically generate
reentrant code from non-reentrant code. Any naturally reentrant code remains
reentrant.
If you link a module (either MAIN or FETCHABLE) containing one or more
programs compiled with the RENT option, you must specify DYNAM=DLL and
REUS=RENT on the link step.
If you specify the options NORENT and LIMITS(EXTNAME(n)) (with n <= 7), then
the text decks generated by the compiler will have the same format as those
generated by the older PL/I compilers. This means that the prelinker would not be
needed to create a PDS-style load module. If you use any other options, you must
use either the prelinker or PDSE's.
The code generated under the NORENT option may not be reentrant unless the
NOWRITABLE option is also specified.
The use of the NORENT does preclude the use of some features of the compiler. In
particular:
v DLLs cannot be built
v reentrant, writeable static is not supported
v a STATIC ENTRY VARIABLE cannot have an INITIAL value
You may mix RENT and NORENT code subject to the following restrictions:
v code compiled with RENT cannot be mixed with code compiled with NORENT
if they share any EXTERNAL STATIC variables
v code compiled with RENT cannot call an ENTRY VARIABLE set in code
compiled with NORENT
v code compiled with RENT cannot call an ENTRY CONSTANT that was fetched
in code compiled with NORENT
v code compiled with RENT can fetch a module containing code compiled with
NORENT if one of the following is true
all the code in the fetched module was compiled with NORENT
the code containing the entry point to the module was compiled with RENT
v code compiled with NORENT code cannot fetch a module containing any code
compiled with RENT
v code compiled with NORENT WRITABLE cannot be mixed with code compiled
with NORENT NOWRITABLE if they share any external CONTROLLED
variables or any external FILEs
Given the above restrictions, the following is still valid:
v a NORENT routine, called say mnorent, statically links and calls a RENT routine,
called say mrent
v the RENT routine mrent then fetches and CALLs a separately-linked module
with an entry point compiled with RENT
RESEXP
The RESEXP option specifies that the compiler is permitted to evaluate all
restricted expressions at compile time even if this would cause a condition to be
raised and the compilation to end with S-level messages.
RESEXP
NORESEXP
Under the NORESEXP compiler option, the compiler will still evaluate all
restricted expression occurring in declarations, including those in INITIAL value
clauses.
For example, under the NORESEXP option, the compiler would not flag the
following statement (and the ZERODIVIDE exception would be raised at run time)
if preconditions_not_met then
x = 1 / 0;
RESPECT
The RESPECT option causes the compiler to honor any specification of the DATE
attribute and to apply the DATE attribute to the result of the DATE built-in
function.
RESPECT ( )
DATE
Using the default, RESPECT(), causes the compiler to ignore any specification of the
DATE attribute and ensures that the compiler does not apply the DATE attribute to
the result of the DATE built-in function.
RTCHECK
NONULLPTR
RTCHECK ( NULLPTR )
Under the RTCHECK(NULLPTR) option, extra code is generated to force the
ERROR condition to be raised if a null pointer, that is, a pointer equal to
SYSNULL(), is dereferenced (that is, if the pointer is used to change or obtain the
value of a variable).
v Under ARCH with a suboption of 7 or less, RTCHECK(NULLPTR) is not
supported.
v Under ARCH with a suboption of 8 or more, when a null pointer is
dereferenced, a compare-and-trap data exception occurs.
Using the RTCHECK(NONULLPTR) option, no such code will be generated.
RULES
The RULES option allows or disallows certain language capabilities and lets you
choose semantics when alternatives are available. It can help you diagnose
common programming errors.

,
IBM
RULES ( ANS )
BYNAME
NOBYNAME
NODECSIZE
DECSIZE
ELSEIF
NOELSEIF
EVENDEC
NOEVENDEC
GLOBALDO
NOGLOBALDO
GOTO
NOGOTO
STRICT
( LOOSE )
NOLAXBIF
LAXBIF
NOLAXCTL
LAXCTL
LAXDCL
NOLAXDCL
NOLAXDEF
LAXDEF
LAXENTRY
NOLAXENTRY
STRICT
( LOOSE )
LAXIF
NOLAXIF
LAXINOUT
NOLAXINOUT
LAXLINK
NOLAXLINK
LAXMARGINS
NOLAXMARGINS
STRICT
( XNUMERIC )
LAXPUNC
NOLAXPUNC
LAXQUAL
NOLAXQUAL
LOOSE
( STRICT )
LAXRETURN
NOLAXRETURN
LAXSCALE
NOLAXSCALE
LAXSEMI
NOLAXSEMI
LAXSTG
NOLAXSTG
NOLAXSTRZ
LAXSTRZ
MULTICLOSE
NOMULTICLOSE
PADDING
NOPADDING
PROCENDONLY
NOPROCENDONLY
SELFASSIGN
NOSELFASSIGN
STOP
NOSTOP
UNREF
NOUNREF

IBM | ANS
Under the IBM suboption:
v For operations requiring string data, data with the BINARY attribute is
converted to BIT.
v Conversions in arithmetic operations or comparisons occur as described in
the PL/I Language Reference.
v Conversions for the ADD, DIVIDE, MULTIPLY, and SUBTRACT built-in
functions occur as described in the PL/I Language Reference except that
operations specified as scaled fixed binary are evaluated as scaled fixed
decimal.
v Nonzero scale factors are permitted in FIXED BIN declares.
v If the result of any precision-handling built-in function (ADD, BINARY, etc.)
has FIXED BIN attributes, the specified or implied scale factor can be
nonzero.
v Even if all arguments to the MAX or MIN built-in functions are UNSIGNED
FIXED BIN, all the arguments are converted to SIGNED, the function is
evaluated with these converted arguments, and the result is always SIGNED.
v Even when you add, multiply, or divide two UNSIGNED FIXED BIN
operands, all the operands are converted to SIGNED, the operation is
evaluated with these converted arguments, and the result has the SIGNED
attribute.
v Even when you apply the MOD or REM built-in functions to two
UNSIGNED FIXED BIN operands, all the arguments are converted to
SIGNED, the function is evaluated with these converted arguments, and the
result has the SIGNED attribute.
v Declaring a variable with the OPTIONS attribute implies the ENTRY
attribute.
Under the ANS suboption:
v For operations requiring string data, data with the BINARY attribute is
converted to CHARACTER.
v Conversions in arithmetic operations or comparisons occur as described in
the PL/I Language Reference.
v Conversions for the ADD, DIVIDE, MULTIPLY, and SUBTRACT built-in
functions occur as described in the PL/I Language Reference.
v Nonzero scale factors are not permitted in FIXED BIN declares.
v If the result of any precision-handling built-in function (ADD, BINARY, etc.)
has FIXED BIN attributes, the specified or implied scale factor must be zero.
v If all arguments to the MAX or MIN built-in functions are UNSIGNED
FIXED BIN, the result is UNSIGNED.
v When you add, multiply or divide two UNSIGNED FIXED BIN operands,
the result has the UNSIGNED attribute.
v When you apply the MOD or REM built-in functions to two UNSIGNED
FIXED BIN operands, the result has the UNSIGNED attribute.
v Declaring a variable with the OPTIONS attribute does not imply the ENTRY
attribute.
Also, under RULES(ANS), the following errors, which the old compilers
ignored, will produce E-level messages:
v Specifying a string constant as the argument to the STRING built-in
v Giving too many asterisks as subscripts in an array reference
v Qualifying a CONTROLLED variable with a POINTER reference (as if the
CONTROLLED variable were BASED)
BYNAME | NOBYNAME
Specifying NOBYNAME causes the compiler to flag all BYNAME assignments
with an E-level message.
DECSIZE | NODECSIZE
Specifying DECSIZE causes the compiler to flag any assignment of a FIXED
DECIMAL expression to a FIXED DECIMAL variable when the SIZE condition
is disabled if the SIZE condition could be raised by the assignment.
Specifying RULES(DECSIZE) may cause the compiler to produce a large
number of messages since if SIZE is disabled, then any statement of the form X
= X + 1 will be flagged if X is FIXED DECIMAL.
ELSEIF | NOELSEIF
Specifying NOELSEIF causes the compiler to flag any ELSE statement that is
immediately followed by an IF statement and suggest that it be rewritten as a
SELECT statement.
This option can be useful in enforcing that SELECT statements be used rather
than a series of nested IF-THEN-ELSE statements.
EVENDEC | NOEVENDEC
Specifying NOEVENDEC causes the compiler to flag any FIXED DECIMAL
declaration that specifies an even precision.
GLOBALDO | NOGLOBALDO
Specifying NOGLOBALDO instructs the compiler to flag all DO loops with
control variables that are declared in a parent block.
GOTO | NOGOTO
Specifying NOGOTO(LOOSE) causes the compiler to flag any GOTO statement
to a label constant unless the GOTO is exiting an ON-unit or unless the target
label constant is in the same block as the GOTO statement.
Specifying NOGOTO(STRICT) causes the compiler to flag any GOTO statement
to a label constant unless the GOTO is exiting an ON-unit.
LAXBIF | NOLAXBIF
Specifying LAXBIF causes the compiler to build a contextual declaration for
built-in functions, such as NULL, even when used without an empty
parameter list.
LAXCTL | NOLAXCTL
Specifying LAXCTL allows a CONTROLLED variable to be declared with a
constant extent and yet to be allocated with a differing extent. NOLAXCTL
requires that if a CONTROLLED variable is to be allocated with a varying
extent, then that extent must be specified as an asterisk or as a non-constant
expression.
The following code is illegal under NOLAXCTL:
dcl a bit(8) ctl;
alloc a;
alloc a bit(16);
But this code would still be valid under NOLAXCTL:
dcl b bit(n) ctl;
dcl n fixed bin(31) init(8);
alloc b;
alloc b bit(16);
LAXDCL | NOLAXDCL
Specifying LAXDCL allows implicit declarations. NOLAXDCL disallows all
implicit and contextual declarations except for BUILTINs and for files SYSIN
and SYSPRINT.
LAXDEF | NOLAXDEF
Specifying LAXDEF allows so-called illegal defining to be accepted without
any compiler messages (rather than the E-level messages that the compiler
would usually produce).
LAXENTRY | NOLAXENTRY
Specifying LAXENTRY allows unprototyped entry declarations. Specifying
NOLAXENTRY causes the compiler to flag all unprototyped entry declarations,
that is, all ENTRY declares that do not specify a parameter list. Note that this
would mean that if an ENTRY should have no parameters, it should be
declared as ENTRY() rather than simply as ENTRY.
STRICT
Specifying RULES(NOLAXENTRY(STRICT)) causes the compiler to flag
unprototyped entry declarations that have the OPTIONS(ASM) attribute.
LOOSE
Specifying RULES(NOLAXENTRY(LOOSE)) causes the compiler not to flag
unprototyped entry declarations that have the OPTIONS(ASM) attribute.
RULES(LAXENTRY) is the default. When you specify NOLAXENTRY, STRICT
is the default.
LAXIF | NOLAXIF
Specifying RULES(NOLAXIF) will cause the compiler to flag any IF, WHILE,
UNTIL, and WHEN clauses that do not have the attributes BIT(1)
NONVARYING.
The following would all be flagged under NOLAXIF:
dcl i fixed bin;
dcl b bit(8);
.
.
.
if i then ...
if b then ...
LAXINOUT | NOLAXINOUT
Specifying NOLAXINOUT causes the compiler to assume that all
ASSIGNABLE BYADDR parameters are input (and possibly output) parameters
and hence to issue a warning if it thinks such a parameter has not been
initialized.
LAXLINK | NOLAXLINK
Specifying NOLAXLINK causes the compiler to flag any assign or compare of
two ENTRY variables or constants if any of the following do not match:
v the parameter description lists
For instance if A1 is declared as ENTRY(CHAR(8)) and A2 as
ENTRY(POINTER) VARIABLE, then under RULES(NOLAXLINK), the
compiler would flag an attempt to assign A1 to A2.
v the RETURNS attribute
For instance if A3 is declared as ENTRY RETURNS(FIXED BIN(31)) and A4
as an ENTRY VARIABLE without the RETURNS attribute, then under
RULES(NOLAXLINK), the compiler would flag an attempt to assign A3 to
A4.
v the LINKAGE and other OPTIONS suboptions
For instance if A5 is declared as ENTRY OPTIONS(ASM) and A6 as an
ENTRY VARIABLE without the OPTIONS attribute, then under
RULES(NOLAXLINK), the compiler would flag an attempt to assign A5 to
A6 (since the OPTIONS(ASM) in the declare of A5 implies that A5 has
LINKAGE(SYSTEM)) while since A6 has no OPTIONS attribute, it will have
LINKAGE(OPTLINK) by default).
.
LAXMARGINS | NOLAXMARGINS
Specifying NOLAXMARGINS causes the compiler to flag, depending on the
setting of the STRICT and XNUMERIC suboption, lines containing non-blank
characters after the right margin. This can be useful in detecting code, such as
a closing comment, that has accidentally been pushed out into the right
margin.
If the NOLAXMARGINS and STMT options are used together with one of the
preprocessors, then any statements that would be flagged because of the
NOLAXMARGINS option will be reported as statement zero (since statement
numbering occurs only after all the preprocessors are finished, but the
detection of text outside the margins occurs as soon as the source is read).
STRICT
Under the STRICT suboption, the compiler will flag any line containing
non-blank characters after the right margin
XNUMERIC
Under the XNUMERIC suboption, the compiler will flag any line
containing non-blank characters after the right margin except if the right
margin is column 72 and columns 73 through 80 all contain numeric digits
LAXPUNC | NOLAXPUNC
Specifying NOLAXPUNC causes the compiler to flag with an E-level message
any place where it assumes punctuation that is missing.
For instance, given the statement "I = (1 * (2);", the compiler assumes that a
closing right parenthesis was meant before the semicolon. Under
RULES(NOLAXPUNC), this statement would be flagged with an E-level
message; otherwise it would be flagged with a W-level message.
LAXQUAL | NOLAXQUAL
Specifying NOLAXQUAL(LOOSE) causes the compiler to flag any reference to
structure members that are not level 1 and are not dot qualified. Consider the
following example:
dcl
1 a,
2 b,
3 b fixed bin,
3 c fixed bin;
c = 11; /* would be flagged */
b.c = 13; /* would not be flagged */
a.c = 17; /* would not be flagged */
Specifying NOLAXQUAL(STRICT) causes the compiler to flag any reference to
structure members that do not include the level-1 name. Consider the
following example:
dcl
1 a,
2 b,
3 b fixed bin,
3 c fixed bin;
c = 11; /* would be flagged */
b.c = 13; /* would be flagged */
a.c = 17; /* would not be flagged */
LAXRETURN | NOLAXRETURN
Specifying NOLAXRETURN causes the compiler to generate code to raise the
ERROR condition when a RETURN statement is used in either of the following
ways:
v With an expression in a procedure that is coded without the RETURNS
option
v Without an expression in a procedure that is coded with the RETURNS
option
RULES(LAXRETURN) is the default.
LAXSCALE | NOLAXSCALE
Specifying NOLAXSCALE causes the compiler to flag any FIXED BIN(p,q) or
FIXED DEC(p,q) declare where q < 0 or p < q.
LAXSEMI | NOLAXSEMI
Specifying NOLAXSEMI causes the compiler to flag any semicolons appearing
inside comments.
LAXSTG | NOLAXSTG
Specifying NOLAXSTG causes the compiler to flag declares where a variable A
is declared as BASED on ADDR(B) and STG(A) > STG(B) even (and this is the
key part) if B is a parameter.
Note that even with NOLAXSTG specified, if B has subscripts, no IBM2402I
E-level message will be produced.
The compiler would already flag this kind of problem if B were in
AUTOMATIC or STATIC storage, but it does not, by default, flag this when B
is a parameter (since some customers declare B with placeholder attributes that
do not describe the actual argument). For those customers whose parameter
and argument declares match (or should match), specifying
RULES(NOLAXSTG) may help detect more storage overlay problems.
LAXSTRZ | NOLAXSTRZ
Specifying LAXSTRZ causes the compiler not to flag any bit or character
variable that is initialized to or assigned a constant value that is too long if the
excess bits are all zeros (or if the excess characters are all blank).
MULTICLOSE | NOMULTICLOSE
Specifying NOMULTICLOSE causes the compiler to flag all statements that
force the closure of multiple groups of statement with an E-level message.
PADDING | NOPADDING
Specifying NOPADDING causes the compiler to flag all structures that contain
padding.
PROCENDONLY | NOPROCENDONLY
Specifying NOPROCENDONLY causes any END statement closing a
PROCEDURE to be flagged if the END statement does not name the
PROCEDURE, that is, if the END keyword is immediately followed by a
semicolon.
SELFASSIGN | NOSELFASSIGN
Specifying NOSELFASSIGN causes the compiler to flag all assignments where
the source and the target are the same.
RULES(SELFASSIGN) is the default.
STOP | NOSTOP
Specifying NOSTOP causes all STOP and EXIT statements to be flagged.
UNREF | NOUNREF
Specifying NOUNREF causes the compiler to flag any level-1 AUTOMATIC
variable which is not referenced and which, if it is a structure or union,
contains no subelement which is referenced.
Default: RULES (IBM BYNAME NODECSIZE EVENDEC ELSEIF GLOBALDO
GOTO NOLAXBIF NOLAXCTL LAXDCL NOLAXDEF LAXENTRY LAXIF
LAXINOUT LAXLINK LAXPUNC LAXMARGINS(STRICT) LAXQUAL
LAXRETURN LAXSCALE LAXSEMI LAXSTG NOLAXSTRZ MULTICLOSE
PADDING PROCENDONLY SELFASSIGN STOP UNREF)
SEMANTIC
The SEMANTIC option specifies that the execution of the compiler's semantic
checking stage depends on the severity of messages issued prior to this stage of
processing.
NOSEMANTIC
S
( W )
E
SEMANTIC
ABBREVIATIONS: SEM, NSEM
SEMANTIC
Equivalent to NOSEMANTIC(S).
NOSEMANTIC
Processing stops after syntax checking. No semantic checking is performed.
NOSEMANTIC (S)
No semantic checking is performed if a severe error or an unrecoverable error
has been encountered.
NOSEMANTIC (E)
No semantic checking is performed if an error, a severe error, or an
unrecoverable error has been encountered.
NOSEMANTIC (W)
No semantic checking is performed if a warning, an error, a severe error, or an
unrecoverable error has been encountered.
Semantic checking is not performed if certain kinds of severe errors are found. If
the compiler cannot validate that all references resolve correctly (for example, if
built-in function or entry references are found with too few arguments) the
suitability of any arguments in any built-in function or entry reference is not
checked.
SERVICE
The SERVICE option places a string in the object module, if generated. This string
is loaded into memory with any load module into which this object is linked, and
if the LE dump includes a traceback, this string will be included in that traceback.
NOSERVICE
SERVICE ( 'service string' )
ABBREVIATIONS: SERV, NOSERV
The string is limited to 64 characters in length.
To ensure that the string remains readable across locales, only characters from the
invariant character set should be used.
SOURCE
The SOURCE option specifies that the compiler includes a listing of the source
program in the compiler listing. The source program listed is either the original
source input or, if any preprocessors were used, the output from the last
preprocessor.
NOSOURCE
SOURCE
ABBREVIATIONS: S, NS
SPILL
The SPILL option specifies the size of the spill area to be used for the compilation.
When too many registers are in use at once, the compiler dumps some of the
registers into temporary storage that is called the spill area.
SPILL (size)
ABBREVIATIONS: SP
If you have to expand the spill area, you will receive a compiler message telling
you the size to which you should increase it. Once you know the spill area that
your source program requires, you can specify the required size (in bytes) as
shown in the syntax diagram above. The maximum spill area size is 3900.
Typically, you will need to specify this option only when compiling very large
programs with OPTIMIZE.
STATIC
The STATIC option controls whether INTERNAL STATIC variables are retained in
the object module even if unreferenced.
SHORT
STATIC ( FULL )
SHORT
INTERNAL STATIC will be retained in the object module only if used.
FULL
All INTERNAL STATIC with INITIAL will be retained in the object module.
If INTERNAL STATIC variables are used as "eyecatchers", you should specify the
STATIC(FULL) option to insure that they will be in the generated object module.
STDSYS
The STDSYS option specifies that the compiler should cause the SYSPRINT file to
be equated to the C stdout file and the SYSIN file to be equated to the C stdin file.
NOSTDSYS
STDSYS
Using the STDSYS option may make it easier to develop and debug a mixed PL/I
and C application.
When SYSPRINT is equated to stdout, its LINESIZE cannot be greater than 132
(the largest value allowed by C).
STMT
The STMT option specifies that statements in the source program are to be counted
and that this "statement number" is used to identify statements in the compiler
listings resulting from the AGGREGATE, ATTRIBUTES, SOURCE and XREF
options.
NOSTMT
STMT
The default is NOSTMT.
When the STMT option is specified, the source listing will include both the logical
statement numbers and the source file numbers.
Note that the GOSTMT option does not exist. The only option that produces
information at run time identifying where an error has occurred is the
GONUMBER option. When the GONUMBER option is used, the term statement in
the runtime error messages refer to line numbers as used by the NUMBER
compiler option even if the STMT option is in effect.
NUMBER and STMT are mutually exclusive and specifying one will negate the
other.
STORAGE
The STORAGE option directs the compiler to produce as part of the listing a
summary of the storage used by each procedure and begin-block.
NOSTORAGE
STORAGE
ABBREVIATIONS: STG, NSTG
The STORAGE output also includes the amount of storage used for the internal
static for the compilation.
The compiler now generates inline code for STG(x) where x has the BASED
attribute with the REFER option and meets both of the following conditions:
v Elements in x with the NONVARYING and BIT attributes have the ALIGNED
attribute.
v All other elements in x have the UNALIGNED attribute.
STRINGOFGRAPHIC
The STRINGOFGRAPHIC option determines whether the result of the STRING
built-in function when applied to a GRAPHIC aggregate has the attribute
CHARACTER or GRAPHIC.
GRAPHIC
STRINGOFGRAPHIC ( CHARACTER )
ABBREVIATIONS: CHAR, G
CHARACTER
Under STRINGOFGRAPHIC(CHAR), if the STRING built-in is applied to an
array or structure of UNALIGNED NONVARYING GRAPHIC variables, the
result will have the CHARACTER attribute.
GRAPHIC
Under STRINGOFGRAPHIC(GRAPHIC), if the STRING built-in is applied to
an array or structure of GRAPHIC variables, the result will have the GRAPHIC
attribute.
SYNTAX
The SYNTAX option specifies that the compiler continues into syntax checking
after preprocessing when you specify the MACRO option, unless an unrecoverable
error has occurred. Whether the compiler continues with the compilation depends
on the severity of the error, as specified by the NOSYNTAX option.
NOSYNTAX
S
( W )
E
SYNTAX
ABBREVIATIONS: SYN, NSYN
SYNTAX
Continues syntax checking after preprocessing unless a severe error or an
unrecoverable error has occurred. SYNTAX is equivalent to NOSYNTAX(S).
NOSYNTAX
Processing stops unconditionally after preprocessing.
NOSYNTAX(W)
No syntax checking if a warning, error, severe error, or unrecoverable error is
detected.
NOSYNTAX(E)
No syntax checking if the compiler detects an error, severe error, or
unrecoverable error.
NOSYNTAX(S)
No syntax checking if the compiler detects a severe error or unrecoverable
error.
If the NOSYNTAX option terminates the compilation, no cross-reference listing,
attribute listing, or other listings that follow the source program is produced.
You can use this option to prevent wasted runs when debugging a PL/I program
that uses the preprocessor.
If the NOSYNTAX option is in effect, any specification of the CICS preprocessor
via the CICS, XOPT or XOPTS options will be ignored. This allows the MACRO
preprocessor to be invoked before invoking the CICS translator.
SYSPARM
The SYSPARM option allows you to specify the value of the string that is returned
by the macro facility built-in function SYSPARM.
SYSPARM ( 'string' )
string
Can be up to 64 characters long. A null string is the default.
For more information about the macro facility, see PL/I Language Reference.
SYSTEM
The SYSTEM option specifies the format used to pass parameters to the MAIN
PL/I procedure, and generally indicates the host system under which the program
runs.
SYSTEM
MVS
( CICS )
IMS
OS
TSO
Table 4 shows the type of parameter list you can expect, and how the program
runs under the specified host system. It also shows the implied settings of
NOEXECOPS. Your MAIN procedure must receive only those types of parameter
lists that are indicated as valid in this table. Additional runtime information for the
SYSTEM option is provided in z/OS Language Environment Programming Guide. .
Table 4. SYSTEM option table
SYSTEM option Type of parameter
list
Program runs as NOEXECOPS
implied
SYSTEM(MVS) Single CHARACTER
string or no
parameters.
z/OS application
program
NO
Otherwise, arbitrary
parameter list.
YES
SYSTEM(CICS) Pointer(s) CICS
transaction YES
SYSTEM(IMS) Pointer(s) IMS
application
program
YES
SYSTEM(OS) z/OS UNIX
parameter list
z/OS UNIX
application program
YES
SYSTEM(TSO) Pointer to CPPL
1
TSO command
processor
YES
Notes:
1. See Invoking MAIN under TSO/E on page 248 for more details about how to invoke
a MAIN under TSO.
Under SYSTEM(IMS), all pointers are presumed to be passed BYVALUE, but under
SYSTEM(MVS) they are presumed to be passed BYADDR.
MAIN procedures run under CICS must be compiled with SYSTEM(CICS) or
SYSTEM(MVS).
It is highly recommended that NOEXECOPS be specified in the MAIN prodecure
OPTIONS option for code, such as a DB2 stored procedure, compiled with
SYSTEM(MVS) but run where runtime options would not be passed.
The compiler will flag any MAIN program compiled with SYSTEM(MVS) if it has
either more than one parameter or a single parameter that is not CHARACTER
VARYING. It is probably better to compile such MAIN programs with
SYSTEM(OS) because then the library simply passes on to MAIN the parameter list
without any scanning for options or other massaging of the parameter list.
TERMINAL
The TERMINAL option determines whether or not diagnostic and information
messages produced during compilation are displayed on the terminal.
Note: This option applies only to compilations under z/OS UNIX.
TERMINAL
NOTERMINAL
ABBREVIATIONS: TERM, NTERM
TERMINAL
Messages are displayed on the terminal.
NOTERMINAL
No information or diagnostic compiler messages are displayed on the terminal.
TEST
The TEST option specifies the level of testing capability that the compiler generates
as part of the object code. It allows you to control the location of test hooks and to
control whether or not the symbol table will be generated.

NOTEST
,
ALL
TEST ( STMT )
PATH
BLOCK
NONE
SYM
NOSYM
HOOK
NOHOOK
NOSEPARATE
SEPARATE
SEPNAME
NOSEPNAME
ABBREVIATIONS: AALL, ACICS, AMACRO, ASQL

STMT
Causes the compiler to generate a statement table, and if the HOOK suboption
is in effect, inserts hooks at statement boundaries and block boundaries
PATH
is in effect, tells the compiler to insert hooks:
v Before the first statement enclosed by an iterative DO statement
v Before the first statement of the true part of an IF statement
v Before the first statement of the false part of an IF statement
v Before the first statement of a true WHEN or OTHERWISE statement of a
SELECT group
v Before the statement following a user label, excluding labeled FORMAT
statements. If a statement has multiple labels, only one hook is inserted.
v At CALLs or function references - both before and after control is passed to
the routine
v At block boundaries
BLOCK
is in effect, tells the compiler to insert hooks at block boundaries (block entry
and block exit).
ALL
is in effect, inserts hooks at all possible locations and generates a statement
table.
Note: Under opt(2) and opt(3), hooks are set only at block boundaries.
NONE
No hooks are put into the program.
SYM
Creates a symbol table that allows you to examine variables by name.
NOSYM
No symbol table is generated.
NOTEST
Suppresses the generation of all testing information.
HOOK
Causes the compiler to insert hooks into the generated code if any of the TEST
suboptions ALL, STMT, BLOCK or PATH are in effect.
NOHOOK
Causes the compiler not to insert hooks into the generated code.
For Debug Tool to generate overlay hooks, one of the suboptions ALL, PATH,
STMT or BLOCK must be specified, but HOOK need not be specified, and
NOHOOK would in fact be recommended.
If NOHOOK is specified,ENTRY and EXIT breakpoints are the only PATH
breakpoints at which Debug Tool will stop.
SEPARATE
Causes the compiler to place most of the debug information it generates into a
separate debug file. Using this option will substantially reduce the size of the
object deck created by the compiler when the TEST option is in effect.
If your program contains GET or PUT DATA statements, the separate debug
file will contain less debug information since those statements require that
symbol table information be placed into the object deck.
The generated debug information always includes a compressed version of the
source that is passed to the compiler. This means that source may be specified
by using SYSIN DD * or that the source may be a temporary data set that is
created by an earlier job step (for example, the source may be the output of the
old SQL or CICS precompilers). The suboptions that you specify for the
LISTVIEW option control the content of the source.
If SEPARATE is used in a batch compile, then the JCL for that compile must
include a DD card for SYSDEBUG which must name a data set with
RECFM=FB and with 80 <= LRECL <= 1024.
This suboption may not be used with the LINEDIR compiler option.
NOSEPARATE
Causes the compiler to place all of the debug information it generates into the
object deck.
Under this option, the generated debug information will not include a
compressed version of the source passed to the compiler. This means that
source must in a dataset that can be found by DebugTool when you try to
debug the program.
SEPNAME
Causes the compiler to place the name of the separate debug file into the
object deck.
This option is ignored if the SEPARATE option is not in effect.
NOSEPNAME
Causes the compiler not to place the name of the separate debug file into the
object deck.
This option is ignored if the SEPARATE option is not in effect.
Note:
v Under opt(2) or opt(3), hooks are set only at block boundaries. This means
that debugging of optimized code is effectively limited to tracing entry
and exit to PROCEDUREs and BEGIN blocks.
v You must use Debug Tool Version 6 (or later) to debug code compiled
with the SEPARATE compiler option.
v There is no support for an input file that spans concatenated datasets.
Specifying TEST(NONE,NOSYM) causes the compiler to set the option to NOTEST.
Use of TEST(NONE,SYM) is strongly discouraged, and it is unclear what is
intended when you specify these settings. You would probably be much better off
if you specified TEST(ALL,SYM,NOHOOK) or TEST(STMT,SYM,NOHOOK).
Any TEST option other than NOTEST and TEST(NONE,NOSYM) will
automatically provide the attention interrupt capability for program testing.
If the program has an ATTENTION ON-unit that you want invoked, you must
compile the program with either of the following:
v The INTERRUPT option
v A TEST option other than NOTEST or TEST(NONE,NOSYM)
Note: ATTENTION is supported only under TSO.
The TEST option will imply GONUMBER.
Because the TEST option can increase the size of the object code and can affect
performance, you might want to limit the number and placement of hooks.
If the TEST option is specified, no inlining will occur.
Structures with REFER are supported in the symbol table.
If TEST(SYM) is in effect, the compiler will generate tables to enable the
automonitor feature of DebugTool. These tables may substantially increase the size
of the object module unless the TEST(SEPARATE) option is in effect. When the
automonitor feature of DebugTool is activated, these tables will be used to display
the values of the variables used in a statement before the statement executes - as
long as the variable has computational type or has the attribute POINTER, OFFSET
or HANDLE. If the statement is an assignment statement, the value of the target
will also be displayed; however, if the target has not been initialized or assigned
previously, its value will be meaningless.
Any variable declared with an * for its name will not be visible when using
DebugTool. Additionally, if an * is used as the name of a parent structure or
substructure, then all of its children will also be invisible. For these purposes, it
might be better to use a single underscore for the name of any structure elements
that you want to leave "unnamed".
UNROLL
The UNROLL option controls loop unrolling under optimization. Loop unrolling is
an optimization that replicates a loop body multiple times and adjusts the loop
control code accordingly.
AUTO
UNROLL ( NO )
AUTO
Indicates that the compiler is permitted to unroll loops that it determines are
appropriate for unrolling.
NO Indicates that the compiler is not permitted to unroll loops.
The UNROLL option is ignored when the NOOPTIMIZE option is in effect.
Loop unrolling improves the performance of a program by exposing instruction
level parallelism for instruction scheduling and software pipelining. It also creases
code in the new loop body, which might increase pressure on register allocation,
cause register spilling, and thus cause a loss in performance.
Therefore, before you unroll a loop, take the following steps to check if the
UNROLL option improves the performance of a particular application:
1. Compile the program with the usual options.
2. Run the program with a representative workload.
3. Recompile the program with the UNROLL option.
4. Rerun the program under the same conditions.
UNROLL(AUTO) is the default.
USAGE
The USAGE option lets you choose different semantics for selected built-in
functions.
USAGE (

,
SIZE
HEX ( CURRENTSIZE )
IBM
ROUND ( ANS )
STRICT
SUBSTR ( LOOSE )
IBM
UNSPEC ( ANS )
)
HEX( SIZE | CURRENTSIZE )
Under the HEX(SIZE) suboption, when HEX is applied to a VARYING or
VARYINGZ string, it will a return a hex string that represents the maximum
amount of storage used by the string.
Under the HEX(CURRENTSIZE) suboption, when HEX is applied to a
VARYING or VARYINGZ string, it will a return a hex string that represents the
current amount of storage used by the string.
ROUND( IBM | ANS )
Under the ROUND(IBM) suboption, the second argument to the ROUND
built-in function is ignored if the first argument has the FLOAT attribute.
Under the ROUND(ANS) suboption, the ROUND built-in function is
implemented as described in the PL/I Language Reference.
SUBSTR( STRICT | LOOSE )
Under the SUBSTR(STRICT) suboption, if x has CHARACTER type, a
SUBSTR(x,y,z) built-in function reference will return a string whose length is
equal to MIN( z, MAXLENGTH(x) ).
Under the SUBSTR(LOOSE) suboption, the same reference would return a
string whose length is z.
The SUBSTR(LOOSE) suboption may be useful for those who have
SUBSTR(x,y,z) references where x is a CHAR(1) BASED variable.
UNSPEC( IBM | ANS )
Under the UNSPEC(IBM) suboption, UNSPEC cannot be applied to a structure
and, if applied to an array, returns an array of bit strings.
Under the UNSPEC(ANS) suboption, UNSPEC can be applied to structures
and, when applied to a structure or an array, UNSPEC returns a single bit
string.
Default: USAGE( HEX(SIZE) ROUND(IBM) SUBSTR(STRICT) UNSPEC(IBM) )
WIDECHAR
The WIDECHAR option specifies the format in which WIDECHAR data will be
stored.
WIDECHAR (
BIGENDIAN
LITTLEENDIAN )
BIGENDIAN
Indicates that WIDECHAR data will be stored in big-endian format. For
instance, the WIDECHAR value for the UTF-16 character '1' will be stored as
'0031'x.
LITTLEENDIAN
Indicates that WIDECHAR data will be stored in little-endian format. For
instance, the WIDECHAR value for the UTF-16 character '1' will be stored as
'3100'x.
WX constants should always be specified in big-endian format. Thus the value '1'
should always be specified as '0031'wx, even if under the
WIDECHAR(LITTLEENDIAN) option, it is stored as '3100'x.
WINDOW
The WINDOW option sets the value for the w window argument used in various
date-related built-in functions.
1950
WINDOW ( w )
w Either an unsigned integer that represents the start of a fixed window or a
negative integer that specifies a sliding window. For example, WINDOW(-20)
indicates a window that starts 20 years prior to the year when the program
runs.
WRITABLE
The WRITABLE option specifies that the compiler may treat static storage as
writable (and if it does, this would make the resultant code non-reentrant).
WRITABLE
NOWRITABLE
FWS
( PRV )
This option has no effect on programs compiled with the RENT option.
The NORENT WRITABLE options allow the compiler to write on static storage to
implement the following constants or variables:
v CONTROLLED variables
v FETCHABLE ENTRY constants
v FILE constants
Under the NORENT WRITABLE options, a module using CONTROLLED
variables, performing I/O, or using FETCH is not reentrant.
The NORENT NOWRITABLE options require the compiler not to write on static
storage for the following constants or variables:
v FILE constants
Under the NORENT NOWRITABLE options, a module using CONTROLLED
variables, performing I/O, or using FETCH is reentrant.
The FWS and PRV suboptions determine how the compiler handles
CONTROLLED variables:
FWS
Upon entry to an EXTERNAL procedure, the compiler makes a library call to
find storage it can use to address the CONTROLLED variables in that
procedure (and any subprocedures).
PRV
The compiler will use the same pseudoregister variable mechanism used by the
old OS PL/I compiler to address CONTROLLED variables.
Hence, under the NORENT NOWRITABLE(PRV) options, old and new code
may share CONTROLLED variables.
However, this also means that under the NORENT NOWRITABLE(PRV)
options, the use of CONTROLLED variables is subject to all the same
restrictions as under the old compiler.
Under the NORENT NOWRITABLE(FWS) options, the following application might
not perform as well as if they were compiled with the RENT or WRITABLE
options:
v Applications that use CONTROLLED variables
v Applications that assign FILE CONSTANTs to FILE VARIABLEs
The performance of an application under NORENT NOWRITABLE(FWS) may be
especially bad if it uses many CONTROLLED variables in many PROCEDUREs.
Under the NOWRITABLE option, the following may not be declared in a
PACKAGE outside a PROCEDURE:
v FILE constants
Code compiled with NORENT WRITABLE cannot be mixed with code compiled
with NORENT NOWRITABLE if they share any external CONTROLLED variables.
In general, you should avoid mixing code compiled with WRITABLE with code
compiled with NOWRITABLE.
XINFO
The XINFO option specifies that the compiler should generate additional files with
extra information about the current compilation unit.

,
NODEF
XINFO ( DEF )
NOMSG
MSG
NOSYM
SYM
NOSYN
SYN
NOXMI
XMI
NOXML
XML
DEF
A definition side-deck file is created. This file lists, for the compilation unit, all:
v defined EXTERNAL procedures
v defined EXTERNAL variables
v statically referenced EXTERNAL routines and variables
v dynamically called fetched modules
Under batch, this file is written to the file specified by the SYSDEFSD DD
statement. Under z/OS UNIX Systems Services, this file is written to the same
directory as the object deck and has the extension "def".
For instance, given the program:
defs: proc;
dcl (b,c) ext entry;
dcl x ext fixed bin(31) init(1729);
dcl y ext fixed bin(31) reserved;
call b(y);
fetch c;
call c;
end;
The following def file would be produced:
EXPORTS CODE
DEFS
EXPORTS DATA
X
IMPORTS
B
Y
FETCH
C
The def file can be used to be build a dependency graph or cross-reference
analysis of your application.
NODEF
No definition side-deck file is created.
MSG
Message information is generated to the ADATA file. See the appendix for
more details on the format of the ADATA file.
Under batch, the ADATA file is generated to the file specified by the
SYSADATA DD statement. Under z/OS UNIX, the ADATA is generated in the
same directory as the object file and has an extension of adt.
NOMSG
No message information is generated to the ADATA file. If neither MSG nor
SYM is specified, no ADATA file is generated.
SYM
Symbol information is generated to the ADATA file. See the appendix for more
details on the format of the ADATA file.
SYSADATA DD statement. Under z/OS UNIX, the ADATA file is generated in
the same directory as the object file and has en extension of adt.
NOSYM
No symbol information is generated to the ADATA file.
SYN
Syntax information is generated to the ADATA file. See the appendix for more
details on the format of the ADATA file. Specifying the XINFO(SYN) option
can greatly increase the amount of storage, both in memory and for the file
produced, required by the compiler.
SYSADATA DD statement. Under z/OS UNIX, the ADATA file is generated in
the same directory as the object file and has en extension of adt.
NOSYN
No syntax information is generated to the ADATA file.
XMI
An XMI side-file is created. This XMI is not intended to be read or interpreted
except by other tools.
Under batch, this file is written to the file specified by the SYSXMI DD
directory as the object deck and has the extension "xmi".
NOXMI
No XMI side-file is created.
XML
An XML side-file is created. This XML file includes:
v the file reference table for the compilation
v the block structure of the program compiled
v the messages produced during the compilation
Under batch, this file is written to the file specified by the SYSXMLSD DD
directory as the object deck and has the extension "xml".
The DTD file for the XML produced is:
<?xml encoding="UTF-8"?>
<!ELEMENT PACKAGE ((PROCEDURE)*,(MESSAGE)*,FILEREFERNCETABLE)>
<!ELEMENT PROCEDURE (BLOCKFILE,BLOCKLINE,(PROCEDURE)*,(BEGINBLOCK)*)>
<!ELEMENT BEGINBLOCK (BLOCKFILE,BLOCKLINE,(PROCEDURE)*,(BEGINBLOCK)*)>
<!ELEMENT MESSAGE (MSGNUMBER,MSGLINE?,MSGFILE?,MSGTEXT)>
<!ELEMENT FILE (FILENUMBER,INCLUDEDFROMFILE?,INCLUDEDONLINE?,FILENAME)>
<!ELEMENT FILEREFERENCETABLE (FILECOUNT,FILE+)>
<!ELEMENT BLOCKFILE (#PCDATA)>
<!ELEMENT BLOCKLINE (#PCDATA)>
<!ELEMENT MSGNUMBER (#PCDATA)>
<!ELEMENT MSGLINE (#PCDATA)>
<!ELEMENT MSGFILE (#PCDATA)>
<!ELEMENT MSGTEXT (#PCDATA)>
<!ELEMENT FILECOUNT (#PCDATA)>
<!ELEMENT FILENUMBER (#PCDATA)>
<!ELEMENT FILENAME (#PCDATA)>
<!ELEMENT INCLUDEFROMFILE (#PCDATA)>
<!ELEMENT INCLUDEDONLINE (#PCDATA)>
NOXML
No XML side-file is created.
XML
The XML option lets you choose the case of the names in the XML generated by
the XMLCHAR built-in function
UPPER
XML ( CASE ( ASIS ) )
CASE( UPPER | ASIS )
Under the CASE(UPPER) suboption, the names in the XML generated by the
XMLCHAR built-in function will all be in upper case.
Under the CASE(ASIS) suboption, the names in the XML generated by the
XMLCHAR built-in function will be in the case used in their declares. Note
that if you use the MACRO preprocessor without using the macro preprocessor
option CASE(ASIS), then the source seen by the compiler will have all the
names in upper case - and that would make specifying the XML(CASE(ASIS))
option useless.
XREF
The XREF option provides a cross-reference table of names used in the program
together with the numbers of the statements in which they are declared or
referenced in the compiler listing.
NOXREF
XREF
FULL
( SHORT )
ABBREVIATIONS: X, NX
FULL
Includes all identifiers and attributes in the compiler listing.
SHORT
Omits unreferenced identifiers from the compiler listing.
The only names not included in the cross reference listing created when using the
XREF option are label references on END statements. For example, assume that
statement number 20 in the procedure PROC1 is END PROC1;. In this situation,
statement number 20 does not appear in the cross reference listing for PROC1.)
If you specify both the XREF and ATTRIBUTES options, the two listings are
combined. If there is a conflict between SHORT and FULL, the usage is determined
by the last option specified. For example, ATTRIBUTES(SHORT) XREF(FULL)
results in the FULL option for the combined listing.
For a description of the format and content of the cross-reference table, see
Cross-reference table on page 159.
Blanks, comments and strings in options
Wherever you may use a blank when specifying an option, you may also specify
as many blanks or comments as you wish.
However, if a comment is specified in %PROCESS line or in a line in an options
file, then the comment must end on the same line as which it begins.
Similarly, if a comment starts in the command line or the PARM= specification,
then it must also end there.
The same rule applies to strings: if a string is specified in %PROCESS line or in a
line in an options file, then the string must end on the same line as which it
begins. Similarly, if a string starts in the command line or the PARM= specification,
then it must also end there.
Changing the default options
If you want to change the supplied default compiler options, during installation of
the compiler you should edit and submit sample job IBMZWIOP.
This job will let you specify options that will be applied before any other options,
thus effectively changing the default options. This job will also let you specify
options that will be applied after all other options, thus effectively changing the
default options and preventing them from being overridden.
If you want to change the defaults for the macro preprocessor options, you can
also do this at installation time by specifying the appropriate PPMACRO option as
part of this job. The PPCICS and PPSQL options let you make the corresponding
changes for the CICS and SQL preprocessors respectively.
Consult the instructions in the sample job for more information.
Specifying options in the %PROCESS or *PROCESS statements
You can use either %PROCESS or *PROCESS in your program; they are equally
acceptable. For consistency and readability in this book, we will always refer to
%PROCESS.
The %PROCESS statement identifies the start of each external procedure and
allows compiler options to be specified for each compilation. The options you
specify in adjacent %PROCESS statements apply to the compilation of the source
statements to the end of input, or the next %PROCESS statement.
To specify options in the %PROCESS statement, code as follows:
%PROCESS options;
where options is a list of compiler options. You must end the list of options with a
semicolon, and the options list should not extend beyond the default right-hand
source margin. The percent sign (%) or asterisk (*) must appear in the first column
of the record. The keyword PROCESS can follow in the next byte (column) or after
any number of blanks. You must separate option keywords by a comma or at least
one blank.
The number of characters is limited only by the length of the record. If you do not
wish to specify any options, code:
%PROCESS;
If you find it necessary to continue the %PROCESS statement onto the next record,
terminate the first part of the list after any delimiter, and continue on the next
record. You cannot split keywords or keyword arguments across records. You can
continue a %PROCESS statement on several lines, or start a new %PROCESS
statement. An example of multiple adjacent %PROCESS statements is as follows:
%PROCESS INT F(I) AG A(F) OP STG NEST X(F) SOURCE ;
%PROCESS LIST TEST ;
Compile-time options, their abbreviated syntax, and their IBM-supplied defaults
are shown in Table 3 on page 5.
How the compiler determines if there are any %PROCESS statements depends on
the format of the initial source file:
F or FB format
if the first character in the record is a "*" or a "%", then the compiler will
check if the next non-blank characters are "PROCESS".
V or VB format
if the first character in the record is a numeric, then the compiler will
assume that the first 8 characters are sequence numbers, and if the ninth
character is a "*" or a "%", then it will check if the next non-blank
characters are "PROCESS". However, if the first character is not a numeric,
then if the first character is a "*" or "%", the compiler will check if the next
non-blank characters are "PROCESS".
U format
if the first character in the record is a "*" or a "%", then the compiler will
check if the next non-blank characters are "PROCESS".
Using % statements
Statements that direct the operation of the compiler begin with a percent (%)
symbol. % statements allow you to control the source program listing and to
include external strings in the source program. % statements must not have label
or condition prefixes and cannot be a unit of a compound statement. You should
place each % statement on a line by itself.
The usage of each % control statement is listed below. For a complete description
of these statements, see PL/I Language Reference.
%INCLUDE
Directs the compiler to incorporate external strings of characters and/or
graphics into the source program.
%PRINT
Directs the compiler to resume printing the source and insource listings.
%NOPRINT
Directs the compiler to suspend printing the source and insource listings
until a %PRINT statement is encountered.
%PAGE
Directs the compiler to print the statement immediately after a %PAGE
statement in the program listing on the first line of the next page.
%POP Directs the compiler to restore the status of the %PRINT and %NOPRINT
saved by the most recent %PUSH.
%PUSH
Saves the current status of the %PRINT and %NOPRINT in a push down
stack on a last-in, first-out basis.
%SKIP Specifies the number of lines to be skipped.
Using the %INCLUDE statement
%INCLUDE statements are used to include additional PL/I files at specified points
in a compilation unit. The PL/I Language Reference describes how to use the
%INCLUDE statement to incorporate source text from a library into a PL/I
program.
For a batch compile
A library is an z/OS partitioned data set that can be used to store other
data sets called members. Source text that you might want to insert into a
PL/I program using a %INCLUDE statement must exist as a member
within a library. Source Statement Library (SYSLIB) on page 230 further
describes the process of defining a source statement library to the compiler.
The statement:
%INCLUDE DD1 (INVERT);
specifies that the source statements in member INVERT of the library
defined by the DD statement with the name DD1 are to be inserted
consecutively into the source program. The compilation job step must
include appropriate DD statements.
If you omit the ddname, the ddname SYSLIB is assumed. In such a case,
you must include a DD statement with the name SYSLIB. (The
IBM-supplied cataloged procedures do not include a DD statement with
this name in the compilation procedure step.)
For a z/OS UNIX compile
The name of the actual include file must be lowercase, unless you specify
UPPERINC. For example, if you used the include statement %include
sample, the compiler would find the file sample.inc, but would not find
the file SAMPLE.inc. Even if you used the include statement %include
SAMPLE, the compiler would still look for sample.inc.
The compiler looks for INCLUDE files in the following order:
2. Directories specified with the I flag or with the INCDIR compiler
option
The first file found by the compiler is included into your source.
A %PROCESS statement in source text included by a %INCLUDE statement results
in an error in the compilation.
Figure 1 on page 156 shows the use of a %INCLUDE statement to include the
source statements for FUN in the procedure TEST. The library HPU8.NEWLIB is
defined in the DD statement with the qualified name PLI.SYSLIB, which is added
to the statements of the cataloged procedure for this job. Since the source statement
library is defined by a DD statement with the name SYSLIB, the %INCLUDE
statement need not include a ddname.
It is not necessary to invoke the preprocessor if your source program, and any text
to be included, does not contain any macro statements.
Using the compiler listing
During compilation, the compiler generates a listing, most of which is optional,
that contains information about the source program, the compilation, and the
object module. The following description of the listing refers to its appearance on a
printed page.
Note:
Although the compiler listing is for your use, it is not a programming
interface and is subject to change.
Of course, if compilation terminates before reaching a particular stage of
processing, the corresponding listings do not appear.
Heading information
The first page of the listing is identified by the product number, the compiler
version number, a string specifying when the compiler was built, and the date and
the time compilation commenced. This page and subsequent pages are numbered.
The listing will then show any options that have been specified for this
compilation. These options will be shown even if the NOOPTIONS option has
been specified and will include, in order, the following:
v the initial install options (those are the options set at install time and which are
applied before any other options)
v under z/OS UNIX, any options specified in the IBM_OPTIONS environment
variable
v any options specified in the parameter string passed to the compiler (i.e. in the
command line under z/OS UNIX or in the PARM= under batch)
//OPT4#9 JOB
//STEP3 EXEC IBMZCBG,PARM.PLI='INC,S,A,X,NEST'
//PLI.SYSLIB DD DSN=HPU8.NEWLIB,DISP=OLD
//PLI.SYSIN DD *
TEST: PROC OPTIONS(MAIN) REORDER;
DCL ZIP PIC '99999'; /* ZIP CODE */
DCL EOF BIT INIT('0'B);
ON ENDFILE(SYSIN) EOF = '1'B;
GET EDIT(ZIP) (COL(1), P'99999');
DO WHILE(EOF);
PUT SKIP EDIT(ZIP, CITYFUN(ZIP)) (P'99999', A(16));
GET EDIT(ZIP) (COL(1), P'99999');
END;
%PAGE;
%INCLUDE FUN;
END; /* TEST */
//GO.SYSIN DD *
95141
95030
94101
//
Figure 1. Including source statements from a library
v any options specified in options files named in the compiler parameter string
This will include the name of each options file and its contents in the same form
as read by the compiler.
v any options specified in *PROCESS or %PROCESS lines in the source
v the final install options (those are the options set at install time and which are
applied after any other options)
Near the end of the listing you will find either a statement that no errors or
warning conditions were detected during the compilation, or a message that one or
more errors were detected. The format of the messages is described under
Messages and return codes on page 163. The second to the last line of the listing
shows the time taken for the compilation. The last line of the listing is END OF
COMPILATION OF xxxx. where xxxx is the external procedure name. If you specify
the NOSYNTAX compiler option, or the compiler aborts early in the compilation,
the external procedure name xxxx is not included and the line truncates to END
OF COMPILATION.
The following sections describe the optional parts of the listing in the order in
which they appear.
Options used for compilation
If you specify the OPTIONS option, a complete list of the options specified for the
compilation, including the default options, appears on the following pages. It will
show the settings of all the options finally in effect during the compilation. If the
setting of an option differs from the default setting after the initial install options
were applied, then that line will be marked with a +
Preprocessor input
If you specify both the MACRO and INSOURCE options, the compiler lists input
to the preprocessor, one record per line, each line numbered sequentially at the left.
If the preprocessor detects an error, or the possibility of an error, it prints a
message on the page or pages following the input listing. The format of these
messages is the same as the format for the compiler messages described under
Messages and return codes on page 163.
SOURCE program
If you specify the SOURCE option, the compiler lists one record per line. These
records will always include the source line and file numbers. However, if a file
contains 999,999 or more lines, then the compiler will flag the file as too large and
will list only the last 6 digits in the source line numbers for that file.
If the input records contain printer control characters, or %SKIP or %PAGE
statements, the lines are spaced accordingly. Use %NOPRINT and %PRINT
statements to stop and restart the printing of the listing.
If you specify the MACRO option, the source listing shows the included text in
place of the %INCLUDE statements in the primary input data set.
Statement nesting level
If you specify the NEST option, the block level and the DO-level are printed to the
right of the statement or line number under the headings LEV and NT respectively,
as in the following example:
Line.File LV NT
1.0 A: PROC OPTIONS(MAIN);
2.0 1 B: PROC;
3.0 2 DCL K(10,10) FIXED BIN (15);
4.0 2 DCL Y FIXED BIN (15) INIT (6);
5.0 2 DO I=1 TO 10;
6.0 2 1 DO J=1 TO 10;
7.0 2 2 K(I,J) = N;
8.0 2 2 END;
9.0 2 1 BEGIN;
10.0 3 1 K(1,1)=Y;
11.0 3 1 END;
12.0 2 1 END B;
13.0 1 END A;
ATTRIBUTE and cross-reference table
If you specify the ATTRIBUTES option, the compiler prints an attribute table
containing a list of the identifiers in the source program together with their
declared and default attributes.
If you specify the XREF option, the compiler prints a cross-reference table
containing a list of the identifiers in the source program together with the file and
line numbers of the statements in which they appear.
If you specify both ATTRIBUTES and XREF, the two tables are combined. In these
tables, if you explicitly declare an identifier, the compiler will list file number and
line number of its DECLARE. Contextually declared variables are marked by
+++++, and other implicitly declared variables are marked by *****.
Attribute table
The compiler never includes the attributes INTERNAL and REAL. You can assume
them unless the respective conflicting attributes, EXTERNAL and COMPLEX,
appear.
For a file identifier, the attribute FILE always appears, and the attribute
EXTERNAL appears if it applies; otherwise, the compiler lists only explicitly
declared attributes.
The OPTIONS attribute will not appear unless the ENTRY attribute applies, and
then only the following options would appear (as appropriate)
v ASSEMBLER
v COBOL
v FETCHABLE
v FORTRAN
v NODESCRIPTOR
v RETCODE
The compiler prints the dimension attribute for an array first. It prints the bounds
as in the array declaration, but expressions are replaced by asterisks unless they
have been reduced by the compiler to a constant, in which case the value of the
constant is shown.
For a character string, a bit string, a graphic string, or an area variable, the
compiler prints the length, as in the declaration, but expressions are replaced by
asterisks unless they have been reduced by the compiler to a constant, in which
case the value of the constant is shown.
Cross-reference table
If you combine the cross-reference table with the attribute table, the list of
attributes for a name is identified by file number and line number. An identifier
appears in the Sets: part of the cross-reference table if it is:
v The target of an assignment statement
v Used as a loop control variable in DO loops
v Used in the SET option of an ALLOCATE or LOCATE statement
v Used in the REPLY option of a DISPLAY statement
If you specify ATTRIBUTES and XREF, the two tables are combined.
If there are unreferenced identifiers, they are displayed in a separate table.
Aggregate length table
An aggregate length table is obtained by using the AGGREGATE option. The table
includes structures but not arrays that have non-constant extents, but the sizes and
offsets of elements within structures with non-constant extents may be inaccurate
or specified as *. For the aggregates listed, the table contains the following
information:
v Where the aggregate is declared.
v The name of the aggregate and each element within the aggregate.
v The byte offset of each element from the beginning of the aggregate.
v The length of each element.
v The total length of each aggregate, structure, and substructure.
v The total number of dimensions for each element.
Be careful when interpreting the data offsets indicated in the data length table. An
odd offset does not necessarily represent a data element without halfword,
fullword, or even double word alignment. If you specify or infer the aligned
attribute for a structure or its elements, the proper alignment requirements are
consistent with respect to other elements in the structure, even though the table
does not indicate the proper alignment relative to the beginning of the table.
If there is padding between two structure elements, a /*PADDING*/ comment
appears, with appropriate diagnostic information.
Statement offset addresses
If the LIST compile option is used, the compiler includes a pseudo-assembler
listing in the compiler listing. This listing includes, for each instruction, an offset
whose meaning depends on the setting of the BLKOFF compiler option:
v Under the BLKOFF option, this offset is the offset of the instruction from the
primary entry point for the function or subroutine to which it belongs. Thus
under this option, the offsets are reset with each new block.
v Under the NOBLKOFF option, this offset is the offset of the instruction from the
start of the compilation unit. Thus under this option, the offsets are cumulative.
The pseudo-assembler listing also includes, at the end of the code for each block,
the offset of the block from the start of the current module (so that the offsets
shown for each statement can be translated to either block or module offsets).
These offsets can be used with the offset given in a runtime error message to
determine the statement to which the message applies.
The OFFSET option produces a table that gives for each statement, the offset of the
first instruction belonging to that statement.
In the example shown in Figure 2, the message indicates that the condition was
raised at offset +98 from the SUB1 entry. The compiler listing excerpt shows this
offset associated with line number 8. The runtime output from this erroneous
statement is shown if Figure 3 on page 161.
Compiler Source
Line.File
1.0
2.0 TheMain: proc options( main );
3.0 call sub1();
4.0 Sub1: proc;
5.0 dcl (i, j) fixed bin(31);
6.0
7.0 i = 0; j = 0;
8.0 j = j / i;
9.0 put skip data( j );
10.0 end Sub1;
11.0 end TheMain;
. . .
OFFSET OBJECT CODE LINE# FILE# P S E U D O A S S E M B L Y L I S T I N G
000000 000002 | THEMAIN DS 0D
000000 47F0 F024 000002 | B 36(,r15)
000004 01C3C5C5 CEE eyecatcher
000008 000000B0 DSA size
00000C 000001F8 =A(PPA1-THEMAIN)
000010 47F0 F001 000002 | B 1(,r15)
. . .
000000 000004 | SUB1 DS 0D
000000 47F0 F024 000004 | B 36(,r15)
000004 01C3C5C5 CEE eyecatcher
000008 00000140 DSA size
00000C 00000190 =A(PPA1-SUB1)
000010 47F0 F001 000004 | B 1(,r15)
...
000086 5020 D0B8 000007 | ST r2,I(,r13,184)
00008A 1842 000007 | LR r4,r2
00008C 5040 D0BC 000007 | ST r4,J(,r13,188)
000090 5800 D0B8 000008 | L r0,I(,r13,184)
000094 8E40 0020 000008 | SRDA r4,32
000098 1D40 000008 | DR r4,r0
00009A 1805 000008 | LR r0,r5
00009C 5000 D0BC 000008 | ST r0,J(,r13,188)
0000A0 4100 D0C0 000009 | LA r0,_temp1(,r13,192)
0000A4 5000 D130 000009 | ST r0,_temp2(,r13,304)
0000A8 A708 5A88 000009 | LHI r0,H23176
0000AC 4000 D0EC 000009 | STH r0,_temp1(,r13,236)
0000B0 5800 6004 000009 | L r0,SYSPRINT(,r6,4)
0000B4 5000 D12C 000009 | ST r0,_temp2(,r13,300)
0000B8 4100 0001 000009 | LA r0,1
0000BC 5000 D0C0 000009 | ST r0,_temp1(,r13,192)
0000C0 4100 D128 000009 | LA r0,_temp2(,r13,296)
...
Figure 2. Finding statement number (compiler listing example)
Entry offsets given in dump and ON-unit SNAP error messages can be compared
with this table and the erroneous statement discovered. The statement is identified
by finding the section of the table that relates to the block named in the message
and then finding the largest offset less than or equal to the offset in the message.
The statement number associated with this offset is the one needed.
Storage offset listing
If the MAP compile option is used, the compiler includes a storage offset listing in
the compiler listing. This listing gives the location in storage of the following
level-1 variables if they are used in the program:
v AUTOMATIC
v CONTROLLED except for PARAMETERs
v STATIC except for ENTRY CONSTANTs that are not FETCHABLE
The listing may also include some compiler generated temporaries.
For an AUTOMATIC variable with adjustable extents, there will be two entries in
this table:
v an entry with '_addr' prefixing the variable name - this entry gives the location
of the address of the variable
v an entry with '_desc' prefixing the variable name - this entry gives the location
of the address of the variable's descriptor
For STATIC and CONTROLLED variables, the storage location will depend on the
RENT/NORENT compiler option, and if the NORENT option is in effect, the
location of CONTROLLED variables will also depend on the WRITABLE/
NOWRITABLE compiler option.
The first column in the Storage Offset Listing is labeled IDENTIFIER and holds the
name of the variable whose location appears in the fourth column.
The second column in the Storage Offset Listing is labeled DEFINITION and holds
a string in the format "B-F:N" where
v B is the number of the block where the variable is declared
You can find the name of the block corresponding to this block number in the
Block Name List which will proceed the Storage Offset Listing (and the Pseudo
Assembly Listing, if any)
v F is the number of the source file where the variable is declared
You can find the name of the file corresponding to this file number in the File
Reference Table which will appear very near the end of the entire compilation
listing.
v N is the number of the source line where the variable is declared in that source
file
Message :
IBM0301S ONCODE=320 The ZERODIVIDE condition was raised.
From entry point SUB1 at compile unit offset +00000098
at entry offset +00000098 at address 0EB00938.
Figure 3. Finding statement number (runtime message example)
The third column in the Storage Offset Listing is labeled ATTRIBUTES and
indicates the storage class of the variable.
The fourth column in the Storage Offset Listing is unlabeled and tells how to find
the location of the variable.
This storage offset listing is sorted by block and by variable name, and it also
includes only user variables. However, specifying the MAP option also causes the
compiler to produce
v a "static map" which lists all STATIC variables but sorted by hex offset
v an "automatic map" which lists, for each block, all AUTOMATIC variables but
sorted by hex offset
The PL/L language's mapping rules may require that a structure be offset by up to
8 bytes from where it would seem to start. For example. consider the AUTOMATIC
structure A declared as
dcl
1 A,
2 B char(2),
2 C fixed bin(31);
Since C must be aligned on a 4-byte boundary, 2 bytes of padding will be needed
for this structure. However, PL/I places those 2 bytes not after B, but before B.
These 2 bytes of "padding" before a structure starts are referred to as the "hang
bytes" for the structure.
These hang bytes will also be reflected in the "automatic map" generated by the
compiler. While the "storage offset listing" would show the offset and length for A
without including its hang bytes
A Class = automatic, Location = 186 : 0xBA(r13), Length = 6
the "automatic map" would show the offset and length for A with its hang bytes
included
OFFSET (HEX) LENGTH (HEX) NAME
98 8 #MX_TEMP1
A0 18 _Sfi
B8 8 A
File reference table
The file reference table consists of three columns which list the following
information about the files read during the compile:
v the number assigned by the compiler to the file
v the included-from data for the file
v the name of the file
The first entry in the included-from column is blank because the first file listed is
the source file. Subsequent entries in this column show the line number of the
include statement followed by a period and the file number of the source file
containing the include.
If the file is a member of a PDS or PDSE, the file name lists the fully qualified
dataset name and the member name.
If the file is included via a subsystem (such as Librarian), then the file name will
have the form DD:ddname(member), where
v ddname is the ddname specified on the %INCLUDE statement (or SYSLIB if no
ddname was specified)
v member is the member name specified on the %INCLUDE statement.
Messages and return codes
If the preprocessor or the compiler detects an error, or the possibility of an error,
messages are generated. Messages generated by the preprocessor appear in the
listing immediately after the listing of the statements processed by the
preprocessor. You can generate your own messages in the preprocessing stage by
use of the %NOTE statement. Such messages might be used to show how many
times a particular replacement had been made. Messages generated by the
compiler appear at the end of the listing.
For compilations that produce no messages, the compiler will include a line saying
"no compiler messages" where the compiler messages would have been listed.
Messages are displayed in the following format:
PPPnnnnI X
where PPP is the prefix identifying the origin of the message (for example, IBM
indicates the PL/I compiler), nnnn is the 4-digit message number, and X identifies
the severity code. All messages are graded according to their severity, and the
severity codes are I, W, E, S, and U.
For every compilation job or job step, the compiler generates a return code that
indicates to the operating system the degree of success or failure it achieved. For
z/OS, this code appears in the end-of-step message that follows the listing of the job
control statements and job scheduler messages for each step.
Table 5 provides an explanation of the severity codes and the comparable return
code for each:
Table 5. Description of PL/I error codes and return codes
Severity
Code
Return
Code
Message
Type
Description
I 0000 Informational The compiled program should run correctly. The
compiler might inform you of a possible inefficiency
in your code or some other condition of interest.
W 0004 Warning A statement might be in error (warning) even
though it is syntactically valid. The compiled
program should run correctly, but it might produce
different results than expected or be significantly
inefficient.
E 0008 Error A simple error fixed by the compiler. The compiled
program should run correctly, but it might produce
different results than expected.
S 0012 Severe An error not fixed by the compiler. If the program
is compiled and an object module is produced, it
should not be used.
U 0016 Unrecoverable An error that forces termination of the compilation.
An object module is not successfully created.
Table 5. Description of PL/I error codes and return codes (continued)
Severity
Code
Return
Code
Message
Type
Description
Note: Compiler messages are printed in groups according to these severity levels.
The compiler lists only messages that have a severity equal to or greater than that
specified by the FLAG option, as shown in Table 6.
Table 6. Using the FLAG option to select the lowest message severity listed
Type of Message Option
Information FLAG(I)
Warning FLAG(W)
Error FLAG(E)
Severe Error FLAG(S)
Unrecoverable Error Always listed
The text of each message, an explanation, and any recommended programmer
response, are given in Enterprise PL/I Messages and Codes.
Chapter 2. PL/I preprocessors
The PL/I compiler allows you to select one or more of the integrated preprocessors
as required for use in your program. You can select the include preprocessor, the
macro preprocessor, the SQL preprocessor, or the CICS preprocessor and the order
in which you would like them to be called.
v The include preprocessor processes special include directives and incorporates
external source files.
v The macro preprocessor, based on %statements and macros, modifies your
source program.
v The SQL preprocessor modifies your source program and translates EXEC SQL
statements into PL/I statements.
v The CICS preprocessor modifies your source program and translates EXEC CICS
statements into PL/I statements.
Each preprocessor supports a number of options to allow you to tailor the
processing to your needs.
The three compile-time options MDECK, INSOURCE, and SYNTAX are meaningful
only when you also specify the PP option. For more information about these
options, see MDECK on page 86, INSOURCE on page 65, and SYNTAX on
page 135.
Include preprocessor
The include preprocessor allows you to incorporate external source files into your
programs by using include directives other than the PL/I directive %INCLUDE.
The following syntax diagram illustrates the options supported by the INCLUDE
preprocessor:
PP ( INCLUDE ( ' ID(<directive>) ' ) )
ID Specifies the name of the include directive. Any line that starts with this
directive as the first set of nonblank characters is treated as an include
directive.
The specified directive must be followed by one or more blanks, an include
member name, and finally an optional semicolon. Syntax for
ddname(membername) is not supported.
In the following example, the first include directive is valid and the second one
is not:
++include payroll
++include syslib(payroll)
This first example causes all lines that start with -INC (and possibly preceding
blanks) to be treated as include directives:
pp( include( id(-inc)))
This second example causes all lines that start with ++INCLUDE (and possibly
preceding blanks) to be treated as include directives:
pp( include( id(++include)))
Include preprocessor
Macro preprocessor
Macros allow you to write commonly used PL/I code in a way that hides
implementation details and the data that is manipulated, and exposes only the
operations. In contrast with a generalized subroutine, macros allow generation of
only the code that is needed for each individual use.
The macro preprocessing facilities of the compiler are described in the PL/I
Language Reference.
You can invoke the macro preprocessor by specifying either the MACRO option or
the PP(MACRO) option. You can specify PP(MACRO) without any options or with
options from the list below.
The defaults for all these options cause the macro preprocessor to behave the same
as the OS PL/I V2R3 macro preprocessor.
If options are specified, the list must be enclosed in quotes (single or double, as
long as they match); for example, to specify the FIXED(BINARY) option, you must
specify PP(MACRO(FIXED(BINARY))).
If you want to specify more than one option, you must separate them with a
comma and/or one or more blanks. For example, to specify the CASE(ASIS) and
RESCAN(UPPER) options, you can specify PP(MACRO(CASE(ASIS)
RESCAN(UPPER))) or PP(MACRO("CASE(ASIS),RESCAN(UPPER)")). You may
specify the options in any order.
Macro preprocessor options
This section describes the options that the macro preprocessor supports.
Macro preprocessor
Chapter 2. PL/I preprocessors 167
CASE
This option specifies if the preprocessor should convert the input text to uppercase.
UPPER
CASE ( ASIS )
ASIS
the input text is left "as is".
UPPER
the input text is to be converted to upper case.
Macro preprocessor
DBCS
This option specifies if the preprocessor should normalize DBCS during text
replacement.
INEXACT
DBCS ( EXACT )
EXACT
the input text is left "as is", and the preprocessor will treat <kk.B> and <kk>B
as different names.
INEXACT
the input text is "normalized", and the preprocessor will treat <kk.B> and
<kk>B as two versions of the same name.
Macro preprocessor
FIXED
This option specifies the default base for FIXED variables.
DECIMAL
FIXED ( BINARY )
DECIMAL
FIXED variables will have the attributes REAL FIXED DEC(5).
BINARY
FIXED variables will have the attributes REAL SIGNED FIXED BIN(31).
Macro preprocessor
INCONLY
The INCONLY option specifies that the preprocessor should process only
%INCLUDE and %XINCLUDE statements.
The NOINCONLY option specifies that the preprocessor should process all
preprocessor statements and not only %INCLUDE and %XINCLUDE statements
NOINCONLY
INCONLY
When the INCONLY option is in effect, you may use neither INCLUDE nor
XINCLUDE as a macro
v procedure name
v statement label
v variable name
The INCONLY option and the NOINCONLY option are mutually exclusive.
For compatibility, the default is NOINCONLY.
Macro preprocessor
NAMEPREFIX
The NAMEPREFIX option specifies that the names of preprocessor procedures and
variables must start with the specified character.
The NONAMEPREFIX option specifies that the names of preprocessor procedures
and variables are not required to start with one particular character.
NONAMEPREFIX
NAMEPREFIX (character)
The character should be specified "as is" and should not be enclosed in quotes.
The default is NONAMEPREFIX.
Macro preprocessor
RESCAN
This option specifies how the preprocessor should handle the case of identifiers
when rescanning text.
ASIS
RESCAN ( UPPER )
UPPER
rescans will not be case-sensitive.
ASIS
rescans will be case-sensitive.
To see the effect of this option, consider the following code fragment
%dcl eins char ext;
%dcl text char ext;
%eins = zwei;
%text = EINS;
display( text );
%text = eins;
display( text );
When compiled with PP(MACRO('RESCAN(ASIS)')), in the second display
statement, the value of text is replaced by eins, but no further replacement occurs
since under RESCAN(ASIS), eins does not match the macro variable eins since the
former is left asis while the latter is uppercased. Hence the following text would be
generated
DISPLAY( zwei );
DISPLAY( eins );
But when compiled with PP(MACRO('RESCAN(UPPER)')), in the second display
statement, the value of text is replaced by eins, but further replacement does
occur since under RESCAN(UPPER), eins does match the macro variable eins
since both are uppercased. Hence the following text would be generated
DISPLAY( zwei );
DISPLAY( zwei );
In short: RESCAN(UPPER) ignores case while RESCAN(ASIS) does not.
Macro preprocessor
Macro preprocessor example
A simple example of the use of the preprocessor to produce a source deck is
shown in Figure 4. According to the value assigned to the preprocessor variable
USE, the source statements will represent either a subroutine (CITYSUB) or a
function (CITYFUN).
The DSNAME used for SYSPUNCH specifies a source program library on which
the preprocessor output will be placed. Normally compilation would continue and
the preprocessor output would be compiled.
//OPT4#8 JOB
//STEP2 EXEC IBMZC,PARM.PLI='MACRO,MDECK,NOCOMPILE,NOSYNTAX'
//PLI.SYSPUNCH DD DSNAME=HPU8.NEWLIB(FUN),DISP=(NEW,CATLG),UNIT=SYSDA,
// SPACE=(TRK,(1,1,1)),DCB=(RECFM=FB,LRECL=80,BLKSIZE=400)
//PLI.SYSIN DD *
/* GIVEN ZIP CODE, FINDS CITY */
%DCL USE CHAR;
%USE = 'FUN' /* FOR SUBROUTINE, %USE = 'SUB' */ ;
%IF USE = 'FUN' %THEN %DO;
CITYFUN: PROC(ZIPIN) RETURNS(CHAR(16)) REORDER; /* FUNCTION */
%END;
%ELSE %DO;
CITYSUB: PROC(ZIPIN, CITYOUT) REORDER; /* SUBROUTINE */
DCL CITYOUT CHAR(16); /* CITY NAME */
%END;
DCL (LBOUND, HBOUND) BUILTIN;
DCL ZIPIN PIC '99999'; /* ZIP CODE */
DCL 1 ZIP_CITY(7) STATIC, /* ZIP CODE - CITY NAME TABLE */
2 ZIP PIC '99999' INIT(
95141, 95014, 95030,
95051, 95070, 95008,
0), /* WILL NOT LOOK AT LAST ONE */
2 CITY CHAR(16) INIT(
'SAN JOSE', 'CUPERTINO', 'LOS GATOS',
'SANTA CLARA', 'SARATOGA', 'CAMPBELL',
'UNKNOWN CITY'); /* WILL NOT LOOK AT LAST ONE */
DCL I FIXED BIN(31);
DO I = LBOUND(ZIP,1) TO /* SEARCH FOR ZIP IN TABLE */
HBOUND(ZIP,1)-1 /* DON'T LOOK AT LAST ELEMENT */
WHILE(ZIPIN = ZIP(I));
END;
%IF USE = 'FUN' %THEN %DO;
RETURN(CITY(I)); /* RETURN CITY NAME */
%END;
%ELSE %DO;
CITYOUT=CITY(I); /* RETURN CITY NAME */
%END;
END;
Figure 4. Using the macro preprocessor to produce a source deck
Macro preprocessor
SQL preprocessor
In general, the coding for your PL/I program will be the same whether or not you
want it to access a DB2 database. However, to retrieve, update, insert, and delete
DB2 data and use other DB2 services, you must use SQL statements. You can use
dynamic and static EXEC SQL statements in PL/I applications.
To communicate with DB2, you need to do the following:
v Code any SQL statements you need, delimiting them with EXEC SQL.
v Use the DB2 precompiler or, if using DB2 for z/OS Version 7 Release 1 or later.
compile with the PL/I PP(SQL()) compiler option.
Before you can take advantage of EXEC SQL support, you must have authority to
access a DB2 system. Contact your local DB2 Database Administrator for your
authorization.
Programming and compilation considerations
When you use the PL/I SQL Preprocessor, the PL/I compiler handles your source
program containing embedded SQL statements at compile time, without your
having to use a separate precompile step. Although the use of a separate
precompile step continues to be supported, use of the PL/I SQL Preprocessor is
recommended.
Interactive debugging with Debug Tool is enhanced when you use the
preprocessor because you see only the SQL statements while debugging (and not
the generated PL/I source). However, you must have DB2 for z/OS Version 7
Release 1 or later to use the SQL preprocessor.
Using the preprocessor lifts some of the DB2 precompiler's restrictions on SQL
programs. When you process SQL statements with the preprocessor, you can:
v Use nested SQL INCLUDE statements.
v Use fully-qualified names for structured host variables.
v Include SQL statements at any level of a nested PL/I program, instead of in only
the top-level source file.
v Use the SQL TYPE attribute anywhere you can specify a PL/I data type. All
such attributes can be factored and can be used in structure elements, in arrays,
and with any storage class like BASED.
The PL/I SQL Preprocessor supports DBCS in the same manner as the PL/I
compiler does. When the GRAPHIC PL/I compiler option is in effect, some source
language elements can be written using DBCS and SBCS characters. In particular,
you can use DBCS characters in the source program in following places:
v Inside comments
v As part of statement labels and identifiers
v In G or M literals
The following restrictions apply to the use of PL/I built-in functions, compiler
options, and statements when you program and compile SQL statements:
v When EXEC SQL statements are translated to PL/I, the following built-in
functions might be included in the generated code. If you use any of the
following built-in functions as the names of elements in structures, you must
also explicitly declare them as BUILTIN:
ADDR
SQL preprocessor
LENGTH
MAXLENGTH
PTRVALUE
SYSNULL
v You must not use PL/I type functions, such as BIND(: t, p :), in EXEC SQL
statements.
v You must not use the DFT(ASCII) compiler option when compiling with the
preprocessor.
v Do not use DECLARE STATEMENT statements in SQL queries, because the PL/I
preprocessor always ignores these these statements.
Compiling with the PL/I SQL Preprocessor option generates a DB2 database
request module (DBRM) along with the usual PL/I compiler outputs such as object
module and listing. As input to the DB2 bind process, the DBRM data set contains
information about the SQL statements and host variables in the program. Not all of
the information in the DBRM is important in terms of the bind or runtime
processing, however. For example, if the HOST value in the DBRM specifies a
language other than PL/I, there is no reason to be concerned. All this means is that
the other language is selected as the installation default for the HOST value, which
does not effect the bind or runtime processing of your program.
The PL/I compiler listing includes the error diagnostics (such as syntax errors in
the SQL statements) that the preprocessor generates.
To use the preprocessor, you need to do the following things:
v Specify the following option when you compile your program:
PP(SQL(options))
This compiler option indicates that you want the compiler to invoke the
integrated PL/I SQL preprocessor. Specify a list of SQL processing options in the
parenthesis after the SQL keyword. The options can be separated by a comma or
by a space and the list of options must be enclosed in quotation marks (single or
double, as long as they match).
For example, PP(SQL('DATE(USA),TIME(USA)') tells the preprocessor to use the
USA format for both DATE and TIME data types.
In addition, for LOB support you must specify the following option:
LIMITS( FIXEDBIN(31,63) FIXEDDEC(31) )
An alternative way to specify SQL preprocessor options is to use the PPSQL
compiler option. For information about how to use it, see PPSQL on page 105.
v Include DD statements for the following data sets in the JCL for your compile
step:
DB2 load library (prefix.SDSNLOAD)
The PL/I SQL preprocessor calls DB2 modules to do the SQL statement
processing. You therefore need to include the name of the DB2 load library
data set in the STEPLIB concatenation for the compile step.
Library for SQL INCLUDE statements
If your program contains SQL INCLUDE member-name statements that specify
secondary input to the source program, you need to include the name of the
data set that contains member-name in the SYSLIB concatenation for the
compile step.
DBRM library
The compilation of the PL/I program generates a DB2 database request
module (DBRM), and the DBRMLIB DD statement is required to designate
the data set to which the DBRM is written.
The DBRMLIB DD statement must name a data set that can be opened and
closed more than once during the compilation.
For example, you might have the following lines in your JCL:
//STEPLIB DD DSN=DSN710.SDSNLOAD,DISP=SHR
//SYSLIB DD DSN=PAYROLL.MONTHLY.INCLUDE,DISP=SHR
//DBRMLIB DD DSN=PAYROLL.MONTHLY.DBRMLIB.DATA(MASTER),DISP=SHR
SQL preprocessor options
When you specify SQL preprocessor options, the list of options must be enclosed
in quotation marks (single or double, as long as they match). For example, to
specify the DATE(ISO) option, you must specify PP(SQL(DATE(ISO))).
Table 7 lists all the SQL preprocessor options with their abbreviations (if any) and
their IBM-supplied default values. This table uses a vertical bar(|) to separate
mutually exclusive options, and brackets ([ ]) to indicate that you can sometimes
omit the enclosed option.
The paragraphs following Table 7 describe the options in the alphabetical order.
In addition to these preprocessor options, you can pass DB2 Coprocessor options
in on the PP(SQL('option-list')) compiler option. For more information about DB2
Coprocessor options, see DB2 UDB for z/OS Application Programming and SQL Guide.
Table 7. SQL preprocessor options and IBM-supplied defaults
SQL Preprocessor Option Abbreviated Name z/OS Default
APOSTSQL | QUOTESQL APOSTSQL
ATTACH(TSO | CAF | RRSAF) ATTACH(TSO)
CCSID0 | NOCCSID0 CCSID0
CONNECT(2 | 1) CT CONNECT(2)
DATE(ISO | USA | EUR |
JIS | LOCAL)
In the field DATE FORMAT on the Application
Programming Defaults Panel 2 when DB2 is installed
DEC(15 | 31) In the field DECIMAL ARITHMETIC on the Application
FLOAT(S390 | IEEE) FLOAT(S390)
GRAPHIC | NOGRAPHIC In the field MIXED DATA on the Application
INCONLY | NOINCONLY NOINCONLY
LEVEL[(aaaa)] L
NOFOR
SQL(ALL | DB2) SQL(DB2)
SQLFLAG(IBM |
STD[(ssname[,qualifier])])

STDSQL(NO | YES) In the field STD SQL LANGUAGE on the Application
TIME(ISO | USA | EUR |
JIS | LOCAL)
In the field TIME FORMAT on the Application
VERSION(aaaa | AUTO)
XREF | NOXREF NOXREF
APOSTSQL
The APOSTSQL option specifies that the apostrophe (') is recognized as the string
delimiter and the quotation mark (") as the SQL escape character within SQL
statements.
The QUOTESQL option specifies that the quotation mark (") is recognized as the
string delimiter and the apostrophe () as the SQL escape character within SQL
statements.
APOSTSQL
QUOTESQL
For compatibility with older PL/I programs which used the DB2 precompiler,
APOSTSQL should be chosen.
APOSTSQL and QUOTESQL are mutually exclusive options.
The default is APOSTSQL.
ATTACH
The ATTACH option specifies the attachment facility that the application uses to
access DB2, TSO, CAF, and RRSAF.
TSO
ATTACH ( )
CAF
RRSAF
Applications that load the attachment facility can use this option to specify the
correct attachment facility, instead of coding a dummy DSNHLI entry point.
The default is ATTACH(TSO).
CCSID0
The CCSID0 option specifies that no host variables be assigned a CCSID value by
the PL/I SQL preprocessor.
The NOCCSID0 option allows host variables to be assigned a CCSID value by the
PL/I SQL preprocessor.
CCSID0
NOCCSID0
If your program updates FOR BIT DATA columns with a data type that is not BIT
data, you will want to choose CCSID0. CCSID0 tells DB2 that the host variable is
not associated with a CCSID, allowing the assignment to be made. Otherwise, if a
host variable that is associated with a CCSID that is not BIT data is assigned to a
FOR BIT DATA column, a DB2 error occurs.
For compatibility with older PL/I programs which used the DB2 precompiler,
CCSID0 should be chosen.
CCSID0 and NOCCSID0 are mutually exclusive options.
The default is CCSID0.
CONNECT
The CONNECT option determines whether to apply type 1 or type 2 CONNECT
statement rules.
2
CONNECT( )
1
ABBREVIATION: CT
CONNECT(2)
applies rules for the CONNECT (Type 2) statement.
CONNECT(1)
applies rules for the CONNECT (Type 1) statement.
The default is CONNECT(2).
For more information about this option, see DB2 SQL Reference.
DATE
The DATE option specifies that date output should always be returned in a
particular format, regardless of the format specified as the location default. For a
description of these formats, see DB2 SQL Reference.
DATE ( )
ISO
USA
EUR
JIS
LOCAL
You cannot use the LOCAL option unless you have a date exit routine.
The default is in the field DATE FORMAT on the Application Programming
Defaults Panel 2 when DB2 is installed.
DEC
The DEC option specifies the maximum precision for decimal arithmetic
operations.
15
DEC ( )
31
The default is in the field DECIMAL ARITHMETIC on the Application

Programming Defaults Panel 1 when DB2 is installed.
FLOAT
The FLOAT option determines whether the contents of floating point host variables
are in System/390 hexadecimal format or in IEEE format.
S390
FLOAT ( )
IEEE
An error message is issued if this FLOAT option is different than the PL/I
compiler's DEFAULT(HEXADEC|IEEE) option.
The default is FLOAT(S390).
GRAPHIC
The GRAPHIC option indicates that the source code might use mixed data, and
that X'0E' and X'0F' are special control characters (shift-out and shift-in) for
EBCDIC data.
The NOGRAPHIC option permits the use of X'0E' and X'0F' in a string, but not as
control characters.
GRAPHIC
NOGRAPHIC
GRAPHIC and NOGRAPHIC are mutually exclusive options.

The default is in the field MIXED DATA on the Application Programming Defaults
Panel 1 when DB2 is installed.
INCONLY
The INCONLY option specifies that the SQL preprocessor should process only
EXEC SQL INCLUDE statements. No code is generated by the SQL preprocessor
when this option is in effect.
The NOINCONLY option specifies that the SQL preprocessor should process all
statements and not only EXEC SQL INCLUDE statements.
NOINCONLY
INCONLY
When you specify the INCONLY option, the compiler does not produce the SQL
options listing, because all other options are ignored under INCONLY.
The INCONLY option and the NOINCONLY option are mutually exclusive.
For compatibility, the default is NOINCONLY.
LEVEL
The LEVEL option defines the level of a module.
LEVEL
(aaaa)
ABBREVIATION: L
aaaa
is any alphanumeric value of up to seven characters.
You can omit the suboption (aaaa). The resulting consistency token is blank.
This option is not recommended for general use, and the DSNH CLIST and the
DB2I panels do not support it.
NOFOR
In static SQL, NOFOR eliminates the need for the FOR UPDATE of FOR UPDATE
OF clause in DECLARE CURSOR statements.
NOFOR
When you use NOFOR, your program can make positioned updates to any
columns that the program has DB2 authority to update.
When you do not use NOFOR, if you want to make positioned updates to any
columns that the program has DB2 authority to update, you need to specify FOR
UPDATE with no column list in your DECLARE CURSOR statements. The FOR
UPDATE clause with no column list applies to static or dynamic SQL statements.
Whether you use or do not use NOFOR, you can specify FOR UPDATE OF with a
column list to restrict updates to only the columns named in the clause and specify
the acquisition of update locks.
You imply NOFOR when you use the option STDSQL(YES).
If the resulting DBRM is very large, you might need extra storage when you
specify NOFOR or use the FOR UPDATE clause with no column list.
SQL
The SQL option indicates whether the source contains SQL statements other than
those recognized by DB2 for z/OS.
DB2
SQL ( )
ALL
DB2
means to interpret SQL statements and check syntax for use by DB2 for z/OS.
SQL(DB2) is recommended when the database server is DB2 for z/OS.
ALL
indicates that the SQL statements in the program are not necessarily for DB2
for z/OS.
SQL(ALL) is recommended for application programs whose SQL statements
must execute on a server other than DB2 for z/OS using DRDA access.
When SQL(ALL) is specified, the SQL statement processor takes the following
actions:
v The SQL statement processor accepts statements that do not conform to the
DB2 syntax rules.
v The SQL statement processor interprets and processes SQL statements
according to distributed relational database architecture (DRDA) rules.
v The SQL statement processor also issues an informational message if the
program attempts to use IBM SQL reserved words as ordinary identifiers.
v SQL(ALL) does not affect the limits of the SQL statement processor.
The default is SQL(DB2).
SQLFLAG
The SQLFLAG option specifies the standard used to check the syntax of SQL
statements.
SQLFLAG ( )
STD
(ssname )
,qualifier
IBM
IBM
checks SQL statements against the syntax of IBM SQL Version 1.
STD
checks SQL statements against the syntax of the entry level of the ANSI/ISO
SQL standard of 1992. You can also use 86 for option, as in releases before
Version 7.
ssname
requests semantics checking, using the specified DB2 subsystem name for
catalog access. If you do not specify ssname, the SQL statement processor
checks only the syntax.
qualifier
specifies the qualifier used for flagging. If you specify a qualifier, you must
always specify the ssname first. If qualifier is not specified, the default is the
authorization ID of the process that started the SQL statement processor.
When statements deviate from the standard, the SQL statement processor writes
informational messages (flags) to the output listing. The SQLFLAG option is
independent of other SQL statement processor options, including SQL and
STDSQL.
STDSQL
The STDSQL option indicates to which rules the output statements should
conform.
NO
STDSQL ( )
YES
NO indicates conformance to DB2 rules.

YES
indicates that the precompiled SQL statements in the source program conform
to certain rules of the SQL standard.
STDSQL(YES) automatically implies the NOFOR option.
The default is in the field STD SQL LANGUAGE on the Application Programming
TIME
The TIME option specifies that time output should always be returned in a
particular format, regardless of the format specified as the location default. For a
description of these formats, see DB2 SQL Reference.
TIME ( )
ISO
USA
EUR
JIS
LOCAL
You cannot use the LOCAL option unless you have a date exit routine.
The default is in the field TIME FORMAT on the Application Programming
VERSION
The VERSION option defines the version identifier of a package, program, and the
resulting DBRM.
VERSION ( )
aaaa
AUTO
When you specify VERSION, the SQL statement processor creates a version
identifier in the program and DBRM. This affects the size of the load module and
DBRM. DB2 uses the version identifier when you bind the DBRM to a plan or
package.
If you do not specify a version at precompile time, then an empty string is the
default version identifier.
If you specify AUTO, the SQL statement processor uses the consistency token to
generate the version identifier. If the consistency token is a timestamp, the
timestamp is converted into ISO character format and used as the version
identifier. The timestamp used is based on the System/370 Store Clock value.
When you compile your PL/I program against a DB2 V9 (or later) database the
options provided in the listing are divided into the following two categories:
SQL Preprocessor Options Used
A list of the PL/I SQL preprocessor options that were in effect at the time of
the compile.
DB2 for z/OS Coprocessor Options used
A list of the DB2 for z/OS Coprocessor options that were in effect at the time
of the compile. See DB2 UDB for z/OS Application Programming and SQL Guide
for information about how they are determined.
XREF
The (NO)XREF option determines whether a cross-reference table of names that are
used in SQL statements is included in the compiler listing.
NOXREF
XREF
When the XREF option is specified, the compiler listing includes a cross-reference
table that includes the following items:
v The names that are used in SQL statements
v The numbers of the statements in which the names are declared or referenced
When the NOXREF option is specified, the compiler listing does not include the
cross-reference table of names that are used in SQL statement.
The use of the XREF SQL preprocessor option requires DB2 for z/OS V10 or later.
The default is NOXREF.
Coding SQL statements in PL/I applications
You can code SQL statements in your PL/I applications using the language defined
in DB2 UDB for z/OS SQL Reference. Specific requirements for your SQL code are
described in the sections that follow.
Defining the SQL communications area
A PL/I program that contains SQL statements must include either an SQLCODE
variable (if the STDSQL(86) preprocessor option is used) or an SQL
communications area (SQLCA). As shown in Figure 5, part of an SQLCA consists
of an SQLCODE variable and an SQLSTATE variable.
v The SQLCODE value is set by the Database Services after each SQL statement is
executed. An application can check the SQLCODE value to determine whether
the last SQL statement was successful.
v The SQLSTATE variable can be used as an alternative to the SQLCODE variable
when analyzing the result of an SQL statement. Like the SQLCODE variable, the
SQLSTATE variable is set by the Database Services after each SQL statement is
executed.
The SQLCA declaration should be included by using the EXEC SQL INCLUDE
statement:
exec sql include sqlca;
The SQLCA structure must not be defined within an SQL declare section. The
scope of the SQLCODE and SQLSTATE declarations must include the scope of all
SQL statements in the program.
Defining SQL descriptor areas
The following statements require an SQLDA:
PREPARE statement-name INTO descriptor-name FROM host-variable
EXECUTE...USING DESCRIPTOR descriptor-name
FETCH...USING DESCRIPTOR descriptor-name
OPEN...USING DESCRIPTOR descriptor-name
Dcl
1 Sqlca,
2 sqlcaid char(8), /* Eyecatcher = SQLCA */
2 sqlcabc fixed binary(31), /* SQLCA size in bytes = 136 */
2 sqlcode fixed binary(31), /* SQL return code */
2 sqlerrmc char(70) var, /* Error message tokens */
2 sqlerrp char(8), /* Diagnostic information */
2 sqlerrd(0:5) fixed binary(31), /* Diagnostic information */
2 sqlwarn, /* Warning flags */
3 sqlwarn0 char(1),
3 sqlwarn1 char(1),
3 sqlwarn2 char(1),
3 sqlwarn3 char(1),
3 sqlwarn4 char(1),
3 sqlwarn5 char(1),
3 sqlwarn6 char(1),
3 sqlwarn7 char(1),
2 sqlext,
3 sqlwarn8 char(1),
3 sqlwarn9 char(1),
3 sqlwarna char(1),
3 sqlstate char(5); /* State corresponding to SQLCODE */
Figure 5. The PL/I declaration of SQLCA
DESCRIBE statement-name INTO descriptor-name
Unlike the SQLCA, there can be more than one SQLDA in a program, and an
SQLDA can have any valid name. An SQLDA should be included by using the
EXEC SQL INCLUDE statement:
exec sql include sqlda;
The SQLDA must not be defined within an SQL declare section.
Embedding SQL statements
The first statement of your program must be a PROCEDURE or a PACKAGE
statement. You can add SQL statements to your program wherever executable
statements can appear. Each SQL statement must begin with EXEC (or EXECUTE)
SQL and end with a semicolon (;).
For example, an UPDATE statement might be coded as follows:
exec sql update DSN8710.DEPT
set Mgrno = :Mgr_Num
where Deptno = :Int_Dept;
Comments: In addition to SQL statements, comments can be included in
embedded SQL statements wherever a blank is allowed.
SQL style comments ('--') are supported when embedded in SQL statements.
Dcl
1 Sqlda based(Sqldaptr),
2 sqldaid char(8), /* Eye catcher = SQLDA */
2 sqldabc fixed binary(31), /* SQLDA size in bytes=16+44*SQLN*/
2 sqln fixed binary(15), /* Number of SQLVAR elements*/
2 sqld fixed binary(15), /* # of used SQLVAR elements*/
2 sqlvar(Sqlsize refer(sqln)), /* Variable Description */
3 sqltype fixed binary(15), /* Variable data type */
3 sqllen fixed binary(15), /* Variable data length */
3 sqldata pointer, /* Pointer to variable data value*/
3 sqlind pointer, /* Pointer to Null indicator*/
3 sqlname char(30) var ; /* Variable Name */
Dcl
1 Sqlda2 based(Sqldaptr),
2 sqldaid2 char(8), /* Eye catcher = SQLDA */
2 sqldabc2 fixed binary(31), /* SQLDA size in bytes=16+44*SQLN*/
2 sqln2 fixed binary(15), /* Number of SQLVAR elements*/
2 sqld2 fixed binary(15), /* # of used SQLVAR elements*/
2 sqlvar2(Sqlsize refer(sqln2)), /* Variable Description */
3 sqlbiglen,
4 sqllongl fixed binary(31),
4 sqlrsvdl fixed binary(31),
3 sqldatal pointer,
3 sqltname char(30) var;
dcl Sqlsize fixed binary(15); /* number of sqlvars (sqln) */
dcl Sqldaptr pointer;
dcl Sqltripled char(1) value(3);
dcl Sqldoubled char(1) value(2);
dcl Sqlsingled char(1) value( );
Figure 6. The PL/I declaration of an SQL descriptor area
Continuation for SQL statements: The line continuation rules for SQL statements
are the same as those for other PL/I statements.
Including code: SQL statements or PL/I host variable declaration statements can
be included by placing the following SQL statement in the source code. Place it at
the point where the statements are to be embedded.
exec sql include member;
Margins: SQL statements must be coded in columns m through n where m and n
are specified in the MARGINS(m,n) compiler option.
Names: Any valid PL/I variable name can be used for a host variable. The length
of a host variable name must not exceed the value n specified in the
LIMITS(NAME(n)) compiler option.
Statement labels: With the exception of the END DECLARE SECTION statement,
and the INCLUDE text-file-name statement, executable SQL statements, like PL/I
statements, can have a label prefix.
WHENEVER statement: The target for the GOTO clause in an SQL WHENEVER
statement must be a label in the PL/I source code and must be within the scope of
any SQL statements affected by the WHENEVER statement.
Using host variables
All host variables used in SQL statements must be explicitly declared, and all host
variables within an SQL statement must be preceded by a colon (:).
Subscripts must not be used in host variable references.
The following topics describe the details of using host variables:
v Using arrays as host variables
v Declaring host variables
v Determining equivalent SQL and PL/I data types on page 200
v Determining compatibility of SQL and PL/I data types on page 203
Using arrays as host variables: You can use an array as a host variable only in
the following two ways:
v As an array of indicator variables for a host structure
v As an array of host variables when used in any of the following statements:
A FETCH statement for a multiple row fetch
An INSERT statement with a multiple row insert
A multiple row MERGE statement
All such arrays must be one-dimensional, have the CONNECTED attribute, and
have constant bounds.
All other use of arrays as host variables is invalid.
Declaring host variables: Host variable declarations can be made at the same
place as regular PL/I variable declarations.
Only a subset of valid PL/I declarations are recognized as valid host variable
declarations. The preprocessor does not use the data attribute defaults specified in
the PL/I DEFAULT statement. If the declaration for a variable is not recognized,
any statement that references the variable might result in the message :
'The host variable token ID is not valid'
Only the names and data attributes of the variables are used by the preprocessor;
the alignment, scope, and storage attributes are ignored.
Declaring scalar host variables: You must declare a scalar host variable with one of
the following data attributes:
CHARACTER, GRAPHIC, or WIDECHAR
Host variables that are declared with the CHARACTER, GRAPHIC, or
WIDECHAR attribute are called string host variables. The following
restrictions apply to them:
v A string host variable must be either NONVARYING or VARYING
attribute.
v If it has the VARYING attribute, it must have the NATIVE attribute.
FIXED BIN, FIXED DEC, or FLOAT
Host variables that are declared with the FIXED BIN, FIXED DEC, or
FLOAT attribute are called numeric host variables. The following
restrictions apply to them:
v A numeric host variable must have the REAL attribute.
v If it has FIXED BIN(p,s), it must have the SIGNED NATIVE attribute, a
zero scale factor, and a precision greater than seven.
v If it has FIXED BIN(p,s), it must have a non-negative scale factor that is
smaller than its precision.
v If it has FLOAT DEC, it must have a precision that is less than 17 unless
the FLOAT(DFP) option is in effect.
v If it has FLOAT BIN, it must have a precision that is less than 54.
SQL TYPE
Host variables that are declared with the SQL TYPE attribute are called
SQL TYPE host variables. The attribute specification must conform to one
of the following syntax diagrams:
BINARY
SQL TYPE IS BINARY ( length )
VARBINARY
SQL TYPE IS VARBINARY ( length )
Result set locator
SQL TYPE IS RESULT_SET_LOCATOR
ROWID
SQL TYPE IS ROWID
Table locator
SQL TYPE IS TABLE LIKE table-name AS LOCATOR
LOB file reference
SQL TYPE IS BLOB_FILE
CLOB_FILE
DBCLOB_FILE
LOB locator
SQL TYPE IS BLOB_LOCATOR
CLOB_LOCATOR
DBCLOB_LOCATOR
LOB variable
SQL TYPE IS BLOB ( length )
CLOB K
DBCLOB M
G
BLOB
You can also use BINARY LARGE OBJECT as an alternative
for BLOB.
CLOB
You can also use either CHARACTER LARGE OBJECT or
CHAR LARGE OBJECT as an alternative for CLOB.
XML LOB variable
SQL TYPE IS XML AS BLOB ( length )
CLOB K
DBCLOB M
G
BLOB
You can also use BINARY LARGE OBJECT as an alternative
for BLOB.
CLOB
You can also use either CHARACTER LARGE OBJECT or
CHAR LARGE OBJECT as an alternative for CLOB.
XML file reference
SQL TYPE IS XML AS BLOB_FILE
CLOB_FILE
DBCLOB_FILE
The following constant declarations are generated by the SQL preprocessor.

You can use them to set the file option variable when you use the file
reference host variables:
DCL SQL_FILE_READ FIXED BIN(31) VALUE(2);
DCL SQL_FILE_CREATE FIXED BIN(31) VALUE(8);
DCL SQL_FILE_OVERWRITE FIXED BIN(31) VALUE(16);
DCL SQL_FILE_APPEND FIXED BIN(31) VALUE(32);
Determining equivalent SQL and PL/I data types: The base SQLTYPE and
SQLLEN of host variables are determined according to Table 8 and Table 9 on page
201. If a host variable appears with an indicator variable, the SQLTYPE is the base
SQLTYPE plus one.
To determine the PL/I data type that is equivalent to a given SQL data type, you
can use Table 10 on page 202 and Table 11 on page 203.
Table 8. SQL data types generated from PL/I declarations
PL/I Data Type SQLTYPE of
Host
Variable
SQLLEN of Host
Variable
SQL Data Type
BIN FIXED(p), 7 < p <= 15 500 2 SMALLINT
BIN FIXED(p), 15 < p <= 31 496 4 INTEGER
BIN FIXED(p), 31 < p <= 63 492 8 BIGINT
DEC FIXED(p,s) 0<=p<=15
and 0<=s<=p
484 p (byte 1)
s (byte 2)
DECIMAL(p,s)
BIN FLOAT(p), 1 p 21 480 4 REAL or FLOAT(n) 1<=n<=21
BIN FLOAT(p), 22 p 53 480 8 DOUBLE PRECISION or
FLOAT(n) 22<=n<=53
Under FLOAT(NODFP):
DEC FLOAT(p), 1 p 6 480 4 FLOAT (single precision)
DEC FLOAT(p), 7 p 16 480 8 FLOAT (double precision)
Under FLOAT(DFP):
DEC FLOAT(p), 1 p 7 996 4 DECFLOAT (single precision)
DEC FLOAT(p), >7 p 16 996 8 DECFLOAT (double precision)
DEC FLOAT(p), >16 p 34 996 16 DECFLOAT (extended decimal)
CHAR(n), 452 n CHAR(n)
CHAR(n) VARYING, 1 n 255 448 n VARCHAR(n)
CHAR(n) VARYING, n > 255 456 n VARCHAR(n)
GRAPHIC(n), 1 n 127 468 n GRAPHIC(n)
GRAPHIC(n) VARYING, 1 n 2000 464 n VARGRAPHIC(n)
GRAPHIC(n) VARYING, n > 2000 472 n LONG VARGRAPHIC
Table 9. SQL data types generated from SQL TYPE declarations
PL/I Data Type SQLTYPE of Host
Variable
SQLLEN of Host
Variable
SQL Data Type
SQL TYPE IS BLOB(n)
1<n<2147483647
404 n BLOB(n)
SQL TYPE IS CLOB(n)
1<n<2147483647
408 n CLOB(n)
SQL TYPE IS DBCLOB(n)
1<n<1073741823 (2)
412 n DBCLOB(n) (2)
SQL TYPE IS ROWID 904 40 ROWID
SQL TYPE IS VARBINARY(n)
1<n<32704
908 n VARBINARY(n)
SQL TYPE IS BINARY(n) 1<n<255 912 n BINARY(n)
SQL TYPE IS BLOB_FILE 916 267 BLOB File Reference (1)
SQL TYPE IS CLOB_FILE 920 267 CLOB File Reference (1)
SQL TYPE IS DBCLOB_FILE 924 267 DBCLOB File Reference (1)
SQL TYPE IS BLOB_LOCATOR 960 4 BLOB Locator (1)
SQL TYPE IS CLOB_LOCATOR 964 4 CLOB Locator (1)
SQL TYPE IS DBCLOB_LOCATOR 968 4 DBCLOB Locator (1)
SQL TYPE IS
RESULT_SET_LOCATOR
972 4 Result Set Locator
SQL TYPE IS TABLE LIKE
table-name AS LOCATOR
976 4 Table Locator (1)
Note:
1. Do not use this data type as a column type.
2. n is the number of double-byte characters.
Table 10. SQL data types mapped to PL/I declarations
SQL Data Type PL/I Equivalent Notes
SMALLINT BIN FIXED(15)
INTEGER BIN FIXED(31)
BIGINT BIN FIXED(63)
DECIMAL(p,s) DEC FIXED(p) or
DEC FIXED(p,s)
p = precision and s = scale;
1 p 31 and 0 s p
Under FLOAT(NODFP):
REAL or FLOAT(n) BIN FLOAT(p) or
DEC FLOAT(m)
1 p 21
and 1 m 6
DOUBLE PRECISION,
DOUBLE, or FLOAT(n)
BIN FLOAT(p) or
DEC FLOAT(m)
22 p 53
and 7 m 16
Under FLOAT(DFP):
DECFLOAT DEC FLOAT(m) Short Decimal Float
1 m 7
DECFLOAT DEC FLOAT(m) Long Decimal Float
7 m 16
DECFLOAT DEC FLOAT(m) Extended Decimal Float
16 m 34
CHAR(n) CHAR(n) 1 n 32767
VARCHAR(n) CHAR(n) VAR
GRAPHIC(n) GRAPHIC(n) n is a positive integer from 1 to 127 that refers
to the number of double-byte characters, not to
the number of bytes
VARGRAPHIC(n) GRAPHIC(n) VAR n is a positive integer that refers to the number
of double-byte characters, not to the number of
bytes; 1 n 2000
LONG VARGRAPHIC GRAPHIC(n) VAR n > 2000
DATE CHAR(n) n must be at least 10
TIME CHAR(n) n must be at least 8
TIMESTAMP CHAR(n) n must be at least 26
Table 11. SQL data types mapped to SQL TYPE declarations
SQL Data Type PL/I Equivalent Notes
Result set locator SQL TYPE IS
RESULT_SET_LOCATOR
Use this data type only for receiving result sets.
Do not use this data type as a column type.
Table locator SQL TYPE IS TABLE LIKE
table-name AS LOCATOR
Use this data type only in a user-defined
function or stored procedure to receive rows of a
transition table. Do not use this data type as a
column type.
BLOB locator SQL TYPE IS BLOB_LOCATOR Use this data type only to manipulate data in
BLOB columns. Do not use this data type as a
column type.
CLOB locator SQL TYPE IS CLOB_LOCATOR Use this data type only to manipulate data in
CLOB columns. Do not use this data type as a
column type.
DBCLOB locator SQL TYPE IS
DBCLOB_LOCATOR
Use this data type only to manipulate data in
DBCLOB columns. Do not use this data type as
a column type.
BLOB file reference SQL TYPE IS BLOB_FILE Use this data type only as a reference to a BLOB
file. Do not use this data type as a column type.
CLOB file reference SQL TYPE IS CLOB_FILE Use this data type only as a reference to a CLOB
file. Do not use this data type as a column type.
DBCLOB file reference SQL TYPE IS DBCLOB_FILE Use this data type only as a reference to a
DBCLOB file. Do not use this data type as a
column type.
BLOB(n) SQL TYPE IS BLOB(n) 1<n<2147483647
CLOB(n) SQL TYPE IS CLOB(n) 1<n<2147483647
DBCLOB(n) SQL TYPE IS DBCLOB(n) n is the number of double-byte characters.
1<n<1073741823
ROWID SQL TYPE IS ROWID
XML AS SQL TYPE IS XML AS ... Used to describe an XML version of a BLOB,
CLOB, DBCLOB, BLOB_FILE, CLOB_FILE, or
DBCLOB_FILE
Determining compatibility of SQL and PL/I data types: PL/I host variables in
SQL statements must be type compatible with the columns which use them:
v Numeric data types are compatible with each other. A SMALLINT, INTEGER,
DECIMAL, or FLOAT column is compatible with a PL/I host variable of BIN
FIXED(15), BIN FIXED(31), DECIMAL(p,s), BIN FLOAT(n) where n is from 22 to
53, or DEC FLOAT(m) where m is from 7 to 16.
v Character data types are compatible with each other. A CHAR or VARCHAR
column is compatible with a fixed-length or varying-length PL/I character host
variable.
v Datetime data types are compatible with character host variables. A DATE,
TIME, or TIMESTAMP column is compatible with a fixed-length or
varying-length PL/I character host variable.
When necessary, the Database Manager automatically converts a fixed-length
character string to a varying-length string or a varying-length string to a
fixed-length character string.
Using host structures
A PL/I host structure name can be a structure name with members that are not
structures or unions. For example:
dcl 1 A,
2 B,
3 C1 char(...),
3 C2 char(...);
In this example, B is the name of a host structure consisting of the scalars C1 and
C2.
Host structures are limited to two levels. A host structure can be thought of as a
named collection of host variables.
Each leaf element of a host structure must have one of the following valid host
data attributes as discussed in Declaring host variables on page 197:
v CHARACTER, GRAPHIC, or WIDECHAR
v FIXED BIN, FIXED DEC, or FLOAT
v SQL TYPE
Using indicator variables
An indicator variable is a two-byte integer (BIN FIXED(15)). On retrieval, an
indicator variable is used to show whether its associated host variable has been
assigned a null value. On assignment to a column, a negative indicator variable is
used to indicate that a null value should be assigned.
Indicator variables are declared in the same way as host variables and the
declarations of the two can be mixed in any way that seems appropriate to the
programmer.
The SQL Preprocessor does not require that indicator arrays have a lower bound of
one.
An indicator variable must have the attribute REAL NATIVE SIGNED FIXED
BIN(15).
Given the statement:
exec sql fetch Cls_Cursor into :Cls_Cd,
:Day :Day_Ind,
:Bgn :Bgn_Ind,
:End :End_Ind;
Variables can be declared as follows:
exec sql begin declare section;
dcl Cls_Cd char(7);
dcl Day bin fixed(15);
dcl Bgn char(8);
dcl End char(8);
dcl (Day_Ind, Bgn_Ind, End_Ind) bin fixed(15);
exec sql end declare section;
Host structure example
The following example shows the declaration of a host structure and an indicator
array followed by an SQL statement that could be used to retrieve the data into the
host structure:
Using host structures
dcl 1 games,
5 sunday,
10 opponents char(30),
10 gtime char(10),
10 tv char(6),
10 comments char(120) var;
dcl indicator(4) fixed bin (15);
exec sql
fetch cursor_a
into :games.sunday:indicator;
General information on LOBs
LOBS, CLOBS, and BLOBS can be as large as 2,147,483,647 bytes long (2
Gigabytes). Double Byte CLOBS can be 1,073,741,823 characters long (1 Gigabyte).
Note: Under CMPAT(V1), the size of a BLOB, CLOB, or DBCLOB must not exceed
32767 KB.
Suppressing SQL preprocessor messages
You can use the IBM supplied compiler user exit (IBMUEXIT) to suppress a
message or to change the severity of a message. See Example of suppressing SQL
messages on page 533 for an example of suppressing the preprocessor messages
by modifying the user exit.
CICS Preprocessor
You can use EXEC CICS statements in PL/I applications that run as transactions
under CICS.
If you do not specify the PP(CICS) option, EXEC CICS statements are parsed and
variable references in them are validated. If they are correct, no messages are
issued as long as the NOCOMPILE option is in effect. If you do not invoke the
CICS translator and the COMPILE option is in effect, the compiler will issue
S-level messages.
The compiler will invoke the CICS preprocessor if you specify the CICS suboption
of the PP option. For compatibility, it will also invoke the CICS preprocessor if the
compiler sees any of these options: CICS, XOPT or XOPTS. However, you should
not specify one of these options and also the PP(CICS) options.
When you are developing programs for execution under CICS, all the EXEC CICS
commands must be translated in one of two ways:
v by the command language translator provided by CICS in a job step prior to the
PL/I compilation
v by the PL/I CICS preprocessor as part of the PL/I compilation (this requires
CICS TS 2.2 or later)
To use the CICS preprocessor, you must also specify the PP(CICS) compile-time
option.
Unless you specify CICS as one of the suboptions of the PP(CICS) option, the
compiler will flag any EXEC CICS statements in the source. Similarly, it will flag
Host structure example
any EXEC CPSM or EXEC DLI statements if you do not specify CPSM or DLI
respectively as a suboption of the PP(CICS) option.
If your CICS program is a MAIN procedure, you must also compile it with the
SYSTEM(CICS) or the SYSTEM(MVS) option. If compiled with SYSTEM(MVS), the
PTFs for runtime APAR PQ91318 must be applied. NOEXECOPS is implied with
this option and all parameters passed to the MAIN procedure must be POINTERs.
For a description of the SYSTEM compile-time option, see SYSTEM on page 137.
If your want your CICS program to be reentrant and if your program uses FILEs
or CONTROLLED variables, then you must compile it with the NOWRITABLE as
well.
If your CICS program includes any files or uses any macros that contain EXEC
CICS statements, you must also run the MACRO preprocessor before your code is
translated (in either of the ways described above). If you are using the CICS
preprocessor, you can specify this with one PP option as illustrated in the
following example:
pp (macro(...) cics(...) )
The CICS preprocessor will add a set of declares for CICS variables and APIs to all
non-nested procedures. Consequently, it keeps tracks of all statements that require
a matching END statement, and if some of these statements are missing or
incorrect, the preprocessor may be misled and not insert these declares. The
preprocessor will also terminate with a severe message if such statements are
nested more than 150 deep.
Finally, in order to use the CICS preprocessor, you must have the CICS
SDFHLOAD dataset as part of the STEPLIB DD for the PL/I compiler.
CICS preprocessor options
There are many options supported by CICS translator. For a description of these
options, see the CICS Application Programming Guide.
Note that these options should be enclosed in quotes (single or double, as long as
they match). For instance, to invoke the CICS preprocessor with the EDF option,
you should specify the option PP(CICS(EDF)).
Coding CICS statements in PL/I applications
You can code CICS statements in your PL/I applications using the language
defined in CICS on Open Systems Application Programming Guide. Specific
requirements for your CICS code are described in the sections that follow.
Embedding CICS statements
If you use the CICS translator, rather than the integrated preprocessor, then the
first statement of your PL/I program must be a PROCEDURE statement. You can
add CICS statements to your program wherever executable statements can appear.
Each CICS statement must begin with EXEC (or EXECUTE) CICS and end with a
semicolon (;).
For example, the GETMAIN statement might be coded as follows:
EXEC CICS GETMAIN SET(BLK_PTR) LENGTH(STG(BLK));
Comments: In addition to the CICS statements, PL/I comments can be included
in embedded CICS statements wherever a blank is allowed.
Continuation for CICS statements: Line continuation rules for CICS statements
are the same as those for other PL/I statements.
Including code: If included code contains EXEC CICS statements or your
program uses PL/I macros that generate EXEC CICS statements, you must use one
of the following:
v The MACRO compile-time option
v The MACRO option of the PP option (before the CICS option of the PP option)
Margins: CICS statements must be coded within the columns specified in the
MARGINS compile-time option.
Statement labels: EXEC CICS statements, like PL/I statements, can have a label
prefix.
Writing CICS transactions in PL/I
You can use PL/I with CICS facilities to write application programs (transactions)
for CICS subsystems. If you do this, CICS provides facilities to the PL/I program
that would normally be provided directly by the operating system. These facilities
include most data management facilities and all job and task management
facilities.
You must observe the following restrictions on PL/I CICS programs:
v Macro-level CICS is not supported.
v PL/I input or output cannot be used except for:
PUT FILE(SYSPRINT)
DISPLAY
CALL PLIDUMP
v The PLISRTx built-in subroutines cannot be used.
v Routines written in a language other than PL/I cannot be called from a PL/I
CICS program if those routines contain their own EXEC CICS statements. If you
want to communicate with a non-PL/I program that contains EXEC CICS
statements, you must use EXEC CICS LINK or EXEC CICS XCTL to do so.
Although PUT FILE(SYSPRINT) is permitted under CICS, you should generally not
use it in production programs as it will degrade performance.
Since the CICS EIB address is only generated by either the CICS translator or the
PL/I CICS preprocessor for an OPTIONS(MAIN) program, it is the users
responsibility to establish the addressability to the EIB for the
OPTIONS(FETCHABLE) routine.
Addressability can be achieved either by using this command:
EXEC CICS ADDRESS EIB(DFHEIPTR)
or by passing the EIB address as an argument to the CALL statement that invokes
the external procedure.
Error-handling
Language Environment prohibits the use of the following EXEC CICS commands
in any PL/I ON-unit or in any code called from a PL/I ON-unit.
v EXEC CICS ABEND
v EXEC CICS HANDLE AID
Coding CICS statements in PL/I applications
v EXEC CICS HANDLE ABEND
v EXEC CICS HANDLE CONDITION
v EXEC CICS IGNORE CONDITION
v EXEC CICS POP HANDLE
v EXEC CICS PUSH HANDLE
All other EXEC CICS commands are allowed within an ON-unit. However, they
must be coded using the NOHANDLE option, the RESP option, or the RESP2
option.
Error-handling
Chapter 3. Using PL/I cataloged procedures
This chapter describes the standard cataloged procedures supplied by IBM for use
with the IBM Enterprise PL/I for z/OS compiler. It explains how to invoke them,
and how to temporarily or permanently modify them. The Language Environment
SCEERUN data set must be located in STEPLIB and accessible to the compiler
when you use any of the cataloged procedures described in this chapter.
A cataloged procedure is a set of job control statements, stored in a library, that
includes one or more EXEC statements, each of which can be followed by one or
more DD statements. You can retrieve the statements by naming the cataloged
procedure in the PROC parameter of an EXEC statement in the input stream.
You can use cataloged procedures to save time and reduce Job Control Language
(JCL) errors. If the statements in a cataloged procedure do not match your
requirements exactly, you can easily modify them or add new statements for the
duration of a job. You should review these procedures and modify them to obtain
the most efficient use of the facilities available and to allow for your own
conventions.
IBM-supplied cataloged procedures
The PL/I cataloged procedures supplied for use with Enterprise PL/I for z/OS are:
IBMZC
Compile only
IBMZCB
Compile and bind
IBMZCPL
Compile, prelink, and link-edit
IBMZCBG
Compile, bind, and run
IBMZCPLG
IBMZCPG
Compile, prelink, load, and run
Cataloged procedures IBMZCB and IBMZCBG use features of the program
management binder introduced in DFSMS/MVS
1.4 in place of the prelinker

supplied with Language Environment. These procedures produce a program object
in a PDSE.
Cataloged procedures IBMZCPL, IBMZCPLG and IBMZCPG use the prelinker
supplied with Language Environment and produce a load module in PDS. Use
these procedures if you do not want to use a PDSE. The information in this section
describes the procedure steps of the different cataloged procedures. For a
description of the individual statements for compiling and link editing,
seeInvoking the compiler under z/OS using JCL on page 227 and z/OS Language
Environment Programming Guide. These cataloged procedures do not include a DD
statement for the input data set; you must always provide one. The example
shown in Figure 7 on page 210 illustrates the JCL statements you might use to
invoke the cataloged procedure IBMZCBG to compile, bind, and run a PL/I
program.
Enterprise PL/I requires a minimum REGION size of 32M. Large programs require
more storage. If you do not specify REGION on the EXEC statement that invokes
the cataloged procedure you are running, the compiler uses the default REGION
size for your site. The default size might or might not be adequate, depending on
the size of your PL/I program.
If you compile your programs with optimization turned on, the REGION size (and
time) required may be much, much larger.
For an example of specifying REGION on the EXEC statement, see Figure 7.
Compile only (IBMZC)
The IBMZC cataloged procedure, shown in Figure 8 on page 211, includes only one
procedure step, in which the options specified for the compilation are OBJECT and
OPTIONS. (IBMZPLI is the symbolic name of the compiler.) In common with the
other cataloged procedures that include a compilation procedure step, IBMZC does
not include a DD statement for the input data set; you must always supply an
appropriate statement with the qualified ddname PLI.SYSIN.
The OBJECT compile-time option causes the compiler to place the object module,
in a syntax suitable for input to the linkage editor, in the standard data set defined
by the DD statement with the name SYSLIN. This statement defines a temporary
data set named &&LOADSET on a sequential device; if you want to retain the
object module after the end of your job, you must substitute a permanent name for
&&LOADSET (that is, a name that does not start with &&) and specify KEEP in
the appropriate DISP parameter for the last procedure step that used the data set.
You can do this by providing your own SYSLIN DD statement, as shown below.
The data set name and disposition parameters on this statement will override those
on the IBMZC procedure SYSLIN DD statement. In this example, the compile step
is the only step in the job.
//PLICOMP EXEC IBMZC
//PLI.SYSLIN DD DSN=MYPROG,DISP=SHR
//PLI.SYSIN DD ...
The term MOD in the DISP parameter in Figure 8 on page 211 allows the compiler
to place more than one object module in the data set, and PASS ensures that the
data set is available to a later procedure step providing a corresponding DD
statement is included there.
//COLEGO JOB
//STEP1 EXEC IBMZCBG, REGION.PLI=32M
//PLI.SYSIN DD *
.
.
.
(insert PL/I program to be compiled here)
.
.
.
/*
Figure 7. Invoking a cataloged procedure
The SYSLIN SPACE parameter allows an initial allocation of 1 cylinder and, if
necessary, 15 further allocations of 1 cylinder (a total of 16 cylinders).
Compile and bind (IBMZCB)
The IBMZCB cataloged procedure, shown in Figure 9 on page 212, includes two
procedure steps: PLI, which is identical to cataloged procedure IBMZC, and BIND,
which invokes the Program Management binder (symbolic name IEWBLINK) to
bind the object module produced in the first procedure step.
Input data for the compilation procedure step requires the qualified ddname
PLI.SYSIN. The COND parameter in the EXEC statement BIND specifies that this
procedure step should be bypassed if the return code produced by the compiler is
greater than 8 (that is, if a severe or unrecoverable error occurs during
compilation).
//IBMZC PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200
//*
//********************************************************************
//* *
//* LICENSED MATERIALS - PROPERTY OF IBM *
//* *
//* 5655-H31 (C) COPYRIGHT IBM CORP. 1999, 2009 *
//* ALL RIGHTS RESERVED. *
//* *
//* US GOVERNMENT USERS RESTRICTED RIGHTS - USE, *
//* DUPLICATION OR DISCLOSURE RESTRICTED BY GSA *
//* ADP SCHEDULE CONTRACT WITH IBM CORP. *
//* *
//********************************************************************
//*
//* IBM ENTERPRISE PL/I FOR Z/OS
//* VERSION 3 RELEASE 9 MODIFICATION 0
//*
//* COMPILE A PL/I PROGRAM
//*
//* PARAMETER DEFAULT VALUE USAGE
//* LNGPRFX IBMZ.V3R9M0 PREFIX FOR LANGUAGE DATA SET NAMES
//* LIBPRFX CEE PREFIX FOR LIBRARY DATA SET NAMES
//* SYSLBLK 3200 BLKSIZE FOR OBJECT DATA SET
//*
//* USER MUST SUPPLY //PLI.SYSIN DD STATEMENT THAT IDENTIFIES
//* LOCATION OF COMPILER INPUT
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
//PLI EXEC PGM=IBMZPLI,PARM=OBJECT,OPTIONS
//STEPLIB DD DSN=&LNGPRFX;.SIBMZCMP,DISP=SHR
// DD DSN=&LIBPRFX;.SCEERUN,DISP=SHR
//SYSPRINT DD SYSOUT=*
//SYSOUT DD SYSOUT=*
//SYSLIN DD DSN=&&LOADSET,DISP=(MOD,PASS),UNIT=SYSALLDA,
// SPACE=(CYL,(1,1)),DCB=(LRECL=80,BLKSIZE=&SYSLBLK)
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSALLDA,
// SPACE=(1024,(200,50),,CONTIG,ROUND),DCB=BLKSIZE=1024
Figure 8. Cataloged Procedure IBMZC
The Program Management binder always places the program objects it creates in
the standard data set defined by the DD statement with the name SYSLMOD. This
statement in the cataloged procedure specifies a new temporary library &&GOSET,
in which the program object will be placed and given the member name GO. In
specifying a temporary library, the cataloged procedure assumes that you will run
the program object in the same job; if you want to retain the program object, you
must substitute your own statement for the DD statement with the name
SYSLMOD.
//IBMZCB PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200,GOPGM=GO
//*
//********************************************************************
//* *
//* *
//* *
//* *
//********************************************************************
//*
//*
//* COMPILE AND BIND A PL/I PROGRAM
//*
//* GOPGM GO MEMBER NAME FOR PROGRAM OBJECT
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
//*********************************************************************
//* BIND STEP
//*********************************************************************
//BIND EXEC PGM=IEWBLINK,COND=(8,LT,PLI),
// PARM=XREF,COMPAT=PM3
//SYSLIB DD DSN=&LIBPRFX..SCEELKED,DISP=SHR
//SYSLIN DD DSN=*.PLI.SYSLIN,DISP=(OLD,DELETE)
// DD DDNAME=SYSIN
//SYSLMOD DD DSN=&&GOSET(&GOPGM),DISP=(MOD,PASS),UNIT=SYSALLDA,
// SPACE=(1024,(50,20,1)),DSNTYPE=LIBRARY
//SYSDEFSD DD DUMMY
//SYSIN DD DUMMY
Figure 9. Cataloged Procedure IBMZCB
Compile, bind, and run (IBMZCBG)
The IBMZCBG cataloged procedure, shown in Figure 10, includes three procedure
steps: PLI, BIND, and GO. PLI and BIND are identical to the two procedure steps
of IBMZCB, and GO runs the program object created in the step BIND. The GO
step is executed only if no severe or unrecoverable errors occurred in the preceding
procedure steps.
Input data for the compilation procedure step should be specified in a DD
statement with the name PLI.SYSIN, and for the GO step in a DD statement with
the name GO.SYSIN.
//IBMZCBG PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200,GOPGM=GO
//*
//********************************************************************
//* *
//* *
//* *
//* *
//********************************************************************
//*
//*
//* COMPILE, BIND, AND RUN A PL/I PROGRAM
//*
//* GOPGM GO MEMBER NAME FOR PROGRAM OBJECT
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
Figure 10. Cataloged Procedure IBMZCBG (Part 1 of 2)
Compile, prelink, and link-edit (IBMZCPL)
The IBMZCPL cataloged procedure, shown in Figure 11 on page 215, includes three
procedure steps: PLI, which is identical to cataloged procedure IBMZC; PLKED,
which invokes the Language Environment prelinker; and LKED, which invokes the
linkage editor (symbolic name IEWL) to link-edit the object module produced in
the first procedure step.
Input data for the compilation procedure step requires the qualified ddname
PLI.SYSIN. The COND parameter in the EXEC statement LKED specifies that this
procedure step should be bypassed if the return code produced by the compiler is
greater than 8 (that is, if a severe or unrecoverable error occurs during
compilation).
//*********************************************************************
//* BIND STEP
//*********************************************************************
//BIND EXEC PGM=IEWBLINK,COND=(8,LT,PLI),
// PARM=XREF,COMPAT=PM3
//SYSLIN DD DSN=*.PLI.SYSLIN,DISP=(OLD,DELETE)
// DD DDNAME=SYSIN
// SPACE=(1024,(50,20,1)),DSNTYPE=LIBRARY
//SYSDEFSD DD DUMMY
//SYSIN DD DUMMY
//*********************************************************************
//* RUN STEP
//*********************************************************************
//GO EXEC PGM=*.BIND.SYSLMOD,COND=((8,LT,PLI),(8,LE,BIND))
//STEPLIB DD DSN=&LIBPRFX;.SCEERUN,DISP=SHR
//CEEDUMP DD SYSOUT=*
//SYSUDUMP DD SYSOUT=*
Figure 10. Cataloged Procedure IBMZCBG (Part 2 of 2)
//IBMZCPL PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200,PLANG=EDCPMSGE,GOPGM=GO
//*
//********************************************************************
//* *
//* *
//* *
//* *
//********************************************************************
//*
//*
//* COMPILE, PRELINK, LINK-EDIT A PL/I PROGRAM
//*
//* PLANG EDCPMSGE PRELINKER MESSAGES MEMBER NAME
//* GOPGM GO MEMBER NAME FOR LOAD MODULE
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
Figure 11. Cataloged Procedure IBMZCPL (Part 1 of 2)
The linkage editor always places the load modules it creates in the standard data
set defined by the DD statement with the name SYSLMOD. This statement in the
cataloged procedure specifies a new temporary library &&GOSET, in which the
load module will be placed and given the member name GO. In specifying a
temporary library, the cataloged procedure assumes that you will run the load
module in the same job; if you want to retain the module, you must substitute
your own statement for the DD statement with the name SYSLMOD.
The SYSLIN DD statement in Figure 11 on page 215 shows how to concatenate a
data set defined by a DD statement named SYSIN with the primary input
(SYSLIN) to the linkage editor. You could place linkage editor control statements in
the input stream by this means, as described in the z/OS Language Environment
Programming Guide.
Compile, prelink, link-edit, and run (IBMZCPLG)
The IBMZCPLG cataloged procedure, shown in Figure 12 on page 217, includes
four procedure steps: PLI, PLKED, LKED, and GO. PLI, PLKED, and LKED are
identical to the three procedure steps of IBMZCPL, and GO runs the load module
created in the step LKED. The GO step is executed only if no severe or
unrecoverable errors occurred in the preceding procedure steps.
Input data for the compilation procedure step should be specified in a DD
statement with the name PLI.SYSIN, and for the GO step in a DD statement with
the name GO.SYSIN.
//*********************************************************************
//* PRE-LINK-EDIT STEP
//*********************************************************************
//PLKED EXEC PGM=EDCPRLK,COND=(8,LT,PLI)
//SYSMSGS DD DSN=&LIBPRFX..SCEEMSGP(&PLANG),DISP=SHR
//SYSLIB DD DUMMY
//SYSMOD DD DSN=&&PLNK,DISP=(,PASS),
// UNIT=SYSALLDA,SPACE=(CYL,(1,1)),
// DCB=(RECFM=FB,LRECL=80,BLKSIZE=&SYSLBLK)
//SYSIN DD DSN=&&LOADSET,DISP=(OLD,DELETE)
//*********************************************************************
//* LINK-EDIT STEP
//*********************************************************************
//LKED EXEC PGM=IEWL,PARM=XREF,COND=((8,LT,PLI),(8,LE,PLKED))
//SYSLIN DD DSN=*.PLKED.SYSMOD,DISP=(OLD,DELETE)
// DD DDNAME=SYSIN
// SPACE=(1024,(50,20,1))
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSALLDA,SPACE=(1024,(200,20)),
// DCB=BLKSIZE=1024
//SYSIN DD DUMMY
Figure 11. Cataloged Procedure IBMZCPL (Part 2 of 2)
//IBMZCPLG PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200,PLANG=EDCPMSGE,GOPGM=GO
//*
//********************************************************************
//* *
//* *
//* *
//* *
//********************************************************************
//*
//*
//* COMPILE, PRELINK, LINK-EDIT AND RUN A PL/I PROGRAM
//*
//* GOPGM GO MEMBER NAME FOR LOAD MODULE
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
//*********************************************************************
//*********************************************************************
//SYSLIB DD DUMMY
//SYSMOD DD DSN=&&PLNK,DISP=(,PASS),UNIT=SYSALLDA,SPACE=(CYL,(1,1)),
//SYSIN DD DSN=*.PLI.SYSLIN,DISP=(OLD,DELETE)
Figure 12. Cataloged Procedure IBMZCPLG (Part 1 of 2)
Compile, prelink, load and run (IBMZCPG)
The IBMZCPG cataloged procedure, shown in Figure 13 on page 219, achieves the
same results as IBMZCPLG but uses the loader instead of the linkage editor.
Instead of using four procedure steps (compile, prelink, link-edit, and run), it has
only three (compile, prelink, and load-and-run). The third procedure step runs the
loader program. The loader program processes the object module produced by the
compiler and runs the resultant executable program immediately. You must
provide input data for the compilation step by supplying a qualified ddname
PLI.SYSIN.
The use of the loader imposes certain restrictions on your PL/I program; before
using this cataloged procedure, see z/OS Language Environment Programming Guide,
which explains how to use the loader.
//*********************************************************************
//* LINK-EDIT STEP
//*********************************************************************
//LKED EXEC PGM=IEWL,PARM=XREF,COND=((8,LT,PLI),(8,LE,PLKED))
// DD DDNAME=SYSIN
// SPACE=(1024,(50,20,1))
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSALLDA,SPACE=(1024,(200,20)),
// DCB=BLKSIZE=1024
//SYSIN DD DUMMY
//*********************************************************************
//* RUN STEP
//*********************************************************************
//GO EXEC PGM=*.LKED.SYSLMOD,
// COND=((8,LT,PLI),(8,LE,PLKED),(8,LE,LKED))
Figure 12. Cataloged Procedure IBMZCPLG (Part 2 of 2)
//IBMZCPG PROC LNGPRFX=IBMZ.V3R9M0,LIBPRFX=CEE,
// SYSLBLK=3200,PLANG=EDCPMSGE
//*
//********************************************************************
//* *
//* *
//* *
//* *
//********************************************************************
//*
//*
//* COMPILE, PRELINK, LOAD AND RUN A PL/I PROGRAM
//*
//*
//*********************************************************************
//* COMPILE STEP
//*********************************************************************
//*********************************************************************
//*********************************************************************
//SYSLIB DD DUMMY
//SYSMOD DD DSN=&&PLNK,DISP=(,PASS),
// UNIT=SYSALLDA,SPACE=(CYL,(1,1)),
//SYSIN DD DSN=*.PLI.SYSLIN,DISP=(OLD,DELETE)
Figure 13. Cataloged Procedure IBMZCPG (Part 1 of 2)
For more information on other cataloged procedures, see z/OS Language
Environment Programming Guide.
Invoking a cataloged procedure
To invoke a cataloged procedure, specify its name in the PROC parameter of an
EXEC statement. For example, to use the cataloged procedure IBMZC, you could
include the following statement in the appropriate position among your other job
control statements in the input stream:
//stepname EXEC PROC=IBMZC
You do not need to code the keyword PROC. If the first operand in the EXEC
statement does not begin PGM= or PROC=, the job scheduler interprets it as the
name of a cataloged procedure. The following statement is equivalent to that given
above:
//stepname EXEC IBMZC
If you include the parameter MSGLEVEL=1 in your JOB statement, the operating
system will include the original EXEC statement in its listing, and will add the
statements from the cataloged procedure. In the listing, cataloged procedure
statements are identified by XX or X/ as the first two characters; X/ signifies a
statement that was modified for the current invocation of the cataloged procedure.
You might be required to modify the statements of a cataloged procedure for the
duration of the job step in which it is invoked, either by adding DD statements or
by overriding one or more parameters in the EXEC or DD statements. For example,
cataloged procedures that invoke the compiler require the addition of a DD
statement with the name SYSIN to define the data set containing the source
statements. Also, whenever you use more than one standard link-edit procedure
step in a job, you must modify all but the first cataloged procedure that you
invoke if you want to run more than one of the load modules.
Specifying multiple invocations
You can invoke different cataloged procedures, or invoke the same cataloged
procedure several times, in the same job. No special problems are likely to arise
unless more than one of these cataloged procedures involves a link-edit procedure
step, in which case you must take the following precautions to ensure that all your
load modules can be run.
//*********************************************************************
//* LOAD AND RUN STEP
//*********************************************************************
//GO EXEC PGM=LOADER,PARM=MAP,PRINT,
// COND=((8,LT,PLI),(8,LE,PLKED))
//SYSLOUT DD SYSOUT=*
Figure 13. Cataloged Procedure IBMZCPG (Part 2 of 2)
When the linkage editor creates a load module, it places the load module in the
standard data set defined by the DD statement with the name SYSLMOD. When
the binder creates a program object, it places the program object in the PDSE
defined by the DD statement with the name SYSLMOD. In the absence of a linkage
editor NAME statement, the linkage editor or the binder uses the member name
specified in the DSNAME parameter as the name of the module. In the standard
cataloged procedures, the DD statement with the name SYSLMOD always specifies
a temporary library &&GOSET with the member name GO.
If you use the cataloged procedure IBMZCBG twice within the same job to
compile, bind, and run two PL/I programs, and do not name each of the two
program objects that the binder creates, the first program object runs twice, and the
second one not at all.
To prevent this, use one of the following methods:
v Delete the library &&GOSET at the end of the GO step. In the first invocation of
the cataloged procedure at the end of the GO step, add a DD statement with the
syntax:
//GO.SYSLMOD DD DSN=&&GOSET,
// DISP=(OLD,DELETE)
v Modify the DD statement with the name SYSLMOD in the second and
subsequent invocations of the cataloged procedure so as to vary the names of
the load modules. For example:
//BIND.SYSLMOD DD DSN=&&GOSET(GO1)
and so on.
v Use the NAME linkage editor option to give a different name to each program
object and change each job step EXEC statement to specify the running of the
program object with the name for that job step.
To assign a membername to the program object, you can use the linkage editor
NAME option with the DSNAME parameter on the SYSLMOD DD statement.
When you use this procedure, the membername must be identical to the name on
the NAME option if the EXEC statement that runs the program refers to the
SYSLMOD DD statement for the name of the module to be run.
Another option is to give each program a different name by using GOPGM on the
EXEC procedure statement. For example:
// EXEC IBMZCBG,GOPGM=GO2
Modifying the PL/I cataloged procedures
You can modify a cataloged procedure temporarily by including parameters in the
EXEC statement that invokes the cataloged procedure, or by placing additional DD
statements after the EXEC statement. Temporary modifications apply only for the
duration of the job step in which the procedure is invoked. They do not affect the
master copy of the cataloged procedure in the procedure library.
Temporary modifications can apply to EXEC or DD statements in a cataloged
procedure. To change a parameter of an EXEC statement, you must include a
corresponding parameter in the EXEC statement that invokes the cataloged
procedure. To change one or more parameters of a DD statement, you must
include a corresponding DD statement after the EXEC statement that invokes the
cataloged procedure. Although you cannot add a new EXEC statement to a
cataloged procedure, you can always include additional DD statements.
EXEC statement
If a parameter of an EXEC statement that invokes a cataloged procedure has an
unqualified name, the parameter applies to all the EXEC statements in the
cataloged procedure. The effect on the cataloged procedure depends on the
parameters, as follows:
v PARM applies to the first procedure step and nullifies any other PARM
parameters.
v COND and ACCT apply to all the procedure steps.
v TIME and REGION apply to all the procedure steps and override existing
values.
For example, the statement:
//stepname EXEC IBMZCBG,PARM='OFFSET',REGION=32M
v Invokes the cataloged procedure IBMZCBG.
v Substitutes the option OFFSET for OBJECT and OPTIONS in the EXEC
statement for procedure step PLI.
v Nullifies the PARM parameter in the EXEC statement for procedure step BIND.
v Specifies a region size of 32M for all three procedure steps.
To change the value of a parameter in only one EXEC statement of a cataloged
procedure, or to add a new parameter to one EXEC statement, you must identify
the EXEC statement by qualifying the name of the parameter with the name of the
procedure step. For example, to alter the region size for procedure step PLI only in
the preceding example, code:
//stepname EXEC PROC=IBMZCBG,PARM='OFFSET',REGION.PLI=90M
A new parameter specified in the invoking EXEC statement overrides completely
the corresponding parameter in the procedure EXEC statement.
You can nullify all the options specified by a parameter by coding the keyword
and equal sign without a value. For example, to suppress the bulk of the linkage
editor listing when invoking the cataloged procedure IBMZCBG, code:
//stepname EXEC IBMZCBG,PARM.BIND=
DD statement
To add a DD statement to a cataloged procedure, or to modify one or more
parameters of an existing DD statement, you must include a DD statement with
the form procstepname.ddname in the appropriate position in the input stream. If
ddname is the name of a DD statement already present in the procedure step
identified by procstepname, the parameters in the new DD statement override the
corresponding parameters in the existing DD statement; otherwise, the new DD
statement is added to the procedure step. For example, the statement:
//PLI.SYSIN DD *
adds a DD statement to the procedure step PLI of cataloged procedure IBMZC and
the effect of the statement:
//PLI.SYSPRINT DD SYSOUT=C
is to modify the existing DD statement SYSPRINT (causing the compiler listing to
be transmitted to the system output device of class C).
Overriding DD statements must appear after the procedure invocation and in the
same order as they appear in the cataloged procedure. Additional DD statements
can appear after the overriding DD statements are specified for that step.
To override a parameter of a DD statement, code either a revised form of the
parameter or a replacement parameter that performs a similar function (for
example, SPLIT for SPACE). To nullify a parameter, code the keyword and equal
sign without a value. You can override DCB subparameters by coding only those
you wish to modify; that is, the DCB parameter in an overriding DD statement
does not necessarily override the entire DCB parameter of the corresponding
statement in the cataloged procedures.
Chapter 4. Compiling your program
This chapter describes how to invoke the compiler under z/OS UNIX System
Services (z/OS UNIX) and the job control statements used for compiling under
z/OS. The Language Environment SCEERUN data set must be accessible to the
compiler when you compile your program.
Invoking the compiler under z/OS UNIX
To compile your program under the z/OS UNIX environment, use the pli
command.

pli
command_line_option input_file
command_line_option
You can specify a command_line_option in the following ways:
v -qoption
v Option flag (usually a single letter preceded by -)
If you choose to specify compile-time options on the command line, the format
differs from either setting them in your source file using %PROCESS
statements. See Specifying compile-time options under z/OS UNIX on page
226.
input_file
The z/OS UNIX file specification for your program files. If you omit the
extension from your file specification, the compiler assumes an extension of
.pli. If you omit the complete path, the current directory is assumed.
Input files
The pli command compiles PL/I source files, links the resulting object files with
any object files and libraries specified on the command line in the order indicated,
and produces a single executable file.
The pli command accepts the following types of files:
Source files.pli
All .pli files are source files for compilation. The pli command sends source
files to the compiler in the order they are listed. If the compiler cannot find a
specified source file, it produces an error message and the pli command
proceeds to the next file if one exists.
All HFS source files must be line-delimited and encoded in EBCDIC.
Object files.o
All .o files are object files. The pli command sends all object files along with
library files to the linkage editor at link-edit time unless you specify the -c
option. After it compiles all the source files, the compiler invokes the linkage
editor to link-edit the resulting object files with any object files specified in the
input file list, and produces a single executable output file.
Library files.a
The pli command sends all of the library files (.a files) to the linkage editor at
link-edit time.
Specifying compile-time options under z/OS UNIX
Enterprise PL/I provides compile-time options to change any of the compiler's
default settings. You can specify options on the command line, and they remain in
effect for all compilation units in the file, unless %PROCESS statements in your
source program override them.
See Compile-time option descriptions on page 5 for a description of these
options.
When you specify options on the command line, they override the default settings
of the option. They are overridden by options set in the source file.
You can specify compile-time options on the command line in three ways:
v -qoption_keyword (compiler-specific)
v Single and multiletter flags
v -q+/u/myopts.txt
-qoption_keyword
You can specify options on the command line using the -qoption format.
-q option_keyword
:
= suboption
suboption=argument
You can have multiple -qoptions on the same command line, but they must be
separated by blanks. Option keywords can appear in either uppercase or
lowercase, but you must specify the -q in lowercase.
Some compile-time options allow you to specify suboptions. These suboptions are
indicated on the command line with an equal sign following the -qoption_keyword.
Multiple suboptions must be separated with a colon(:) and no intervening blanks.
An option, for example, that contains multiple suboptions is RULES (RULES on
page 118). To specify RULES(LAXDCL) on the command line, you would enter:
-qrules=ibm:laxdcl
The LIMITS option (LIMITS on page 68) is slightly more complex since each of
its suboptions also has an argument. You would specify
LIMITS(EXTNAME(31),FIXEDDEC(15)) on the command line as shown in the
following example:
-qlimits=extname=31:fixeddec=15
Single and multiletter flags
The z/OS UNIX family of compilers uses a number of common conventional flags.
Each language has its own set of additional flags.
Some flag options have arguments that form part of the flag, for example:
pli samp.pli -I/home/test3/include
In this case, /home/test3/include is an include directory to be searched for
INCLUDE files.
Each flag option should be specified as a separate argument.
Table 12. Compile-time option flags supported by Enterprise PL/I under z/OS UNIX
Option Description
-c Compile only.
-e Create names and entries for a fetchable load module.
-I<dir>
*
Add path <dir> to the directories to be searched for INCLUDE files. -I
must be followed by a path and only a single path is allowed per -I
option. To add multiple paths, use multiple -I options. There shouldn't
be any spaces between -I and the path name.
-O, -O2 Optimize generated code. This option is equivalent to -qOPT=2.
-q<option>
*
Pass it to the compiler. <option> is a compile-time option. Each option
should be delimited by a comma and each suboption should be
delimited by an equal sign or colon. There shouldn't be any spaces
between -q and <option>.
-v Display compile and link steps and execute them.
-# Display compile and link steps, but do not execute them.
Note:
*
You must specify an argument where indicated; otherwise, the results are
unpredictable.
Invoking the compiler under z/OS using JCL
Although you will probably use cataloged procedures rather than supply all the
JCL required for a job step that invokes the compiler, you should be familiar with
these statements so that you can make the best use of the compiler and, if
necessary, override the statements of the cataloged procedures.
So-called "batch compilation", whereby one compilation produced more than one
object deck, is not supported.
Invoking the compiler via BPXBATCH is also not supported.
The following section describes the JCL needed for compilation. The IBM-supplied
cataloged procedures described in IBM-supplied cataloged procedures on page
209 contain these statements. You need to code them yourself only if you are not
using the cataloged procedures.
EXEC statement
The basic EXEC statement is:
//stepname EXEC PGM
512K is required for the REGION parameter of this statement.
If you compile your programs with optimization turned on, the REGION size (and
time) required may be much, much larger.
Specifying compile-time options
The PARM parameter of the EXEC statement can be used to specify one or more of
the optional facilities provided by the compiler. These facilities are described under
Specifying options in the EXEC statement on page 230. See Chapter 1, Using
compiler options and facilities, on page 5 for a description of the options.
DD statements for the standard data sets
The compiler requires several standard data sets, the number of data sets depends
on the optional facilities specified. You must define these data sets in DD
statements with the standard ddnames shown, together with other characteristics
of the data sets, in Table 13. The DD statements SYSIN, SYSUT1, and SYSPRINT
are always required.
You can store any of the standard data sets on a direct access device, but you must
include the SPACE parameter in the DD statement. This parameter defines the data
set to specify the amount of auxiliary storage required. The amount of auxiliary
storage allocated in the IBM-supplied cataloged procedures should suffice for most
applications.
Table 13. Compiler standard data sets
Standard
DDNAME
Contents of data set Possible
device
classes
1
Record
format
(RECFM)
Record
size
(LRECL)
SYSDEBUG TEST(SEPARATE) output SYSDA F,FB >=80 and
<=1024
SYSDEFSD XINFO(DEF) output SYSDA F,FB 128
SYSIN Input to the compiler SYSSQ F,FB,U
VB,V
<101(100)
<105(104)
SYSLIB Source statements for
INCLUDE files
SYSDA F,FB,U
V,VB
<101
<105
SYSLIN Object module SYSSQ FB 80
SYSPRINT Listing, including
messages
SYSSQ VBA 137
SYSPUNCH Preprocessor output,
compiler output
SYSSQ
SYSCP
FB 80 or
MARGINS() value
SYSUT1 Temporary workfile SYSDA F 4051
SYSXMI XINFO(XMI) output SYSDA VB 16383
SYSXMLSD XINFO(XML) output SYSDA VB 16383
SYSADATA XINFO(MSG) output SYSDA U 1024
Notes:
The only value for compile-time SYSPRINT that can be overridden is BLKSIZE.
1. The possible device classes are:
SYSSQ Sequential device
SYSDA direct access device
Block size can be specified except for SYSUT1. The block size and logical record length
for SYSUT1 is chosen by the compiler.
Input (SYSIN)
Input to the compiler must be a data set defined by a DD statement with the name
SYSIN. This data set must have CONSECUTIVE organization. The input must be
one or more external PL/I procedures. If you want to compile more than one
external procedure in a single job or job step, precede each procedure, except
possibly the first, with a %PROCESS statement.
80-byte records are commonly used as the input medium for PL/I source
programs. The input data set can be on a direct access device or some other
sequential media. The input data set can contain either fixed-length records
(blocked or unblocked), variable-length records (coded or uncoded), or
undefined-length records. The maximum record size is 100 bytes.
The maximum number of lines in the input file is 999,999.
When data sets are concatenated for input to the compiler, the concatenated data
sets must have similar characteristics (for example, block size and record format).
Output (SYSLIN, SYSPUNCH)
Output in the form of one or more object modules from the compiler will be stored
in the data set SYSLIN if you specify the OBJECT compile-time option. This data
set is defined by the DD statement.
The object module is always in the form of 80-byte fixed-length records, blocked or
unblocked. If the BLKSIZE is specified for SYSLIN and is something other than 80,
then the LRECL must be specified as 80.
The SYSLIN DD must name either a temporary dataset or a permanent dataset: it
cannot specify a concatenation of datasets of any type.
The SYSLIN DD must specify a sequential dataset, not a PDS or PDSE.
The data set defined by the DD statement with the name SYSPUNCH is also used
to store the output from the preprocessor if you specify the MDECK compile-time
option.
Temporary workfile (SYSUT1)
The compiler requires a data set for use as a temporary workfile. It is defined by a
DD statement with the name SYSUT1, and is known as the spill file. It must be on
a direct access device, and must not be allocated as a multi-volume data set.
The spill file is used as a logical extension to main storage and is used by the
compiler and by the preprocessor to contain text and dictionary information. The
LRECL and BLKSIZE for SYSUT1 is chosen by the compiler based on the amount
of storage available for spill file pages.
The DD statements given in this publication and in the cataloged procedures for
SYSUT1 request a space allocation in blocks of 1024 bytes. This is to insure that
adequate secondary allocations of direct access storage space are acquired.
Listing (SYSPRINT)
The compiler generates a listing that includes all the source statements that it
processed, information relating to the object module, and, when necessary,
messages. Most of the information included in the listing is optional, and you can
specify those parts that you require by including the appropriate compile-time
options. The information that can appear, and the associated compile-time options,
are described under Using the compiler listing on page 156.
You must define the data set, in which you wish the compiler to store its listing, in
a DD statement with the name SYSPRINT. This data set must have
CONSECUTIVE organization. Although the listing is usually printed, it can be
stored on any sequential or direct access device. For printed output, the following
statement will suffice if your installation follows the convention that output class A
refers to a printer:
//SYSPRINT DD SYSOUT=A
Source Statement Library (SYSLIB)
If you use the %INCLUDE statement to introduce source statements into the PL/I
program from a library, you can either define the library in a DD statement with
the name SYSLIB, or you can choose your own ddname (or ddnames) and specify
a ddname in each %INCLUDE statement.
The DD should specify a PDS or PDSE, but not the actual member. For example to
include the file HEADER from the library SYSLIB using the dataset INCLUDE.PLI,
the %INCLUDE statement would like
%INCLUDE HEADER;
or
%INCLUDE SYSLIB( HEADER );
The DD statement should be
SYSLIB DD DISP=SHR,DSN=INCLUDE.PLI
but it should not be
SYSLIB DD DISP=SHR,DSN=INCLUDE.PLI(HEADER)
All %INCLUDE files must have the same record format (fixed, variable,
undefined), the same logical record length, and the same left and right margins as
the SYSIN source file.
The BLOCKSIZE of the library must be less than or equal to 32,760 bytes.
The maximum number of lines in any one include file is 999,999.
Specifying options
For each compilation, the IBM-supplied or installation default for a compile-time
option applies unless it is overridden by specifying the option in a %PROCESS
statement or in the PARM parameter of an EXEC statement.
An option specified in the PARM parameter overrides the default value, and an
option specified in a %PROCESS statement overrides both that specified in the
PARM parameter and the default value.
Note: When conflicting attributes are specified either explicitly or implicitly by the
specification of other options, the latest implied or explicit option is
accepted. No diagnostic message is issued to indicate that any options are
overridden in this way.
Specifying options in the EXEC statement
To specify options in the EXEC statement, code PARM= followed by the list of
options, in any order separating the options with commas and enclosing the list
within single quotation marks, for example:
//STEP1 EXEC PGM=IBMZPLI,PARM='OBJECT,LIST'
Any option that has quotation marks, for example MARGINI('c'), must have the
quotation marks duplicated. The length of the option list must not exceed 100
characters, including the separating commas. However, many of the options have
an abbreviated syntax that you can use to save space. If you need to continue the
statement onto another line, you must enclose the list of options in parentheses
(instead of in quotation marks) enclose the options list on each line in quotation
marks, and ensure that the last comma on each line except the last line is outside
of the quotation marks. An example covering all the above points is as follows:
//STEP1 EXEC PGM=IBMZPLI,PARM=('AG,A',
// 'C,F(I)',
// 'M,MI(''X''),NEST,STG,X')
If you are using a cataloged procedure, and want to specify options explicitly, you
must include the PARM parameter in the EXEC statement that invokes it,
qualifying the keyword PARM with the name of the procedure step that invokes
the compiler. For example:
//STEP1 EXEC nnnnnnn,PARM.PLI='A,LIST'
Specifying options in the EXEC statement using an options
file
Another way to specify options in the EXEC statement is by declaring all your
options in an options file and coding the following:
//STEP1 EXEC PGM=IBMZPLI,PARM=+DD:OPTIONS
This method allows you to provide a consistent set of options that you frequently
use. This is especially effective if you want other programmers to use a common
set of options. It also gets you past the 100-character limit.
The MARGINS option does not apply to options files: the data in column 1 will be
read as part of the options. Also, if the file is F-format, any data after column 72
will be ignored.
The parm string can contain "normal" options and can point to more than one
options file. For instance, to specify the option LIST as well as options from both
the file in the GROUP DD and in the PROJECT DD, you could specify
PARM=LIST +DD:GROUP +DD:PROJECT
The options in the PROJECT file would have precedence over options in the
GROUP file.
Also, in this case, the LIST option might be turned off by a NOLIST option
specified in either of the options files. To insure that the LIST option is on, you
could specify
PARM=+DD:GROUP +DD:PROJECT LIST
Options files may also be used under z/OS UNIX. For example, in z/OS UNIX, to
compile sample.pli with options from the file /u/pli/group.opt, you would specify
pli -q+/u/pli/group.opt sample.pli
Earlier releases of the compiler used the character '@' as the trigger character that
preceded the options file specification. This character is not part of the invariant set
of EBCDIC code points, and for that reason the character '+', which is invariant, is
preferred. However, the '@' character may be still be used as long as it is specified
with the hex value '7C'x.
Chapter 5. Link-editing and running
After compilation, your program consists of one or more object modules that
contain unresolved references to each other, as well as references to the Language
Environment run-time library. These references are resolved during link-editing
(statically) or during execution (dynamically). There are two ways to link-edit
statically:
1. Use the prelinker prior to the traditional link step
2. Link without the prelinker, which is similar to linking with PL/I for MVS &
VM except that depending on which compile-time options you use, you may
now need to use a PDSE to hold the resultant load module.
After you compile your PL/I program, the next step is to link and run your
program with test data to verify that it produces the results you expect. When
using Enterprise PL/I we recommend you select the method of linking without the
prelinker (as described in Item 2 above).
Language Environment provides the run-time environment and services you need
to execute your program. For instructions on linking and running PL/I and all
other Language Environment-conforming language programs, see z/OS Language
Environment Programming Guide. For information about migrating your existing
PL/I programs to Language Environment, see Enterprise PL/I for z/OS Compiler and
Run-Time Migration Guide.
Link-edit considerations
If you compile with the option RENT or the option LIMITS(EXTNAME(n)) with n
> 8, then you must use the prelinker or use a PDSE for your linker output.
Using the binder
You must place the binder output into a PDSE.
When linking a DLL, you must specify any needed definition side-decks during
the bind step.
You can use the binder in place of the prelinker and linkage-editor, with the
following exceptions:
v Releases of CICS prior to CICS Transaction Server 1.3 do not support PDSEs.
From CICS Transaction Server 1.3 onwards, there is support in CICS for PDSEs.
Refer to the CICS Transaction Server for z/OS Release Guide, GC34-5701 for a
list of prerequisite APAR fixes.
v MTF does not support PDSEs.
Using the prelinker
If you use the prelinker, you must prelink together in one job step all object decks
defining external references in any of the input object decks.
For instance, if A and B are separately compiled programs and A statically calls B,
then you cannot prelink A and B separately and then later link them together.
Instead. you must link A and B together in one prelink job.
Using the ENTRY card
If you are building a module that will be fetched and which has an Enterprise
PL/I routine as its entry point, then the ENTRY card should specify the name of
that PL/I entry point. If the module is to be fetched from Enterprise PL/I, you
may specify CEESTART on the ENTRY card, although this is strongly not
recommended. However, if the module is to be fetched from COBOL or assembler,
then the ENTRY card absolutely must specify the name of the PL/I entry point
into the module and not CEESTART.
Run-time considerations
You can specify run-time options as parameters passed to the program
initialization routine. You can also specify run-time options in the PLIXOPT
variable. It might also prove beneficial, from a performance standpoint, if you alter
your existing programs by using the PLIXOPT variable to specify your run-time
options and recompiling your programs. For a description of using PLIXOPT, see
z/OS Language Environment Programming Guide.
To simplify input/output at the terminal, various conventions have been adopted
for stream files that are assigned to the terminal. Three areas are affected:
1. Formatting of PRINT files
2. The automatic prompting feature
3. Spacing and punctuation rules for input.
Note: No prompting or other facilities are provided for record I/O at the terminal,
so you are strongly advised to use stream I/O for any transmission to or
from a terminal.
Formatting conventions for PRINT files
When a PRINT file is assigned to the terminal, it is assumed that it will be read as
it is being printed. Spacing is therefore reduced to a minimum to reduce printing
time. The following rules apply to the PAGE, SKIP, and ENDPAGE keywords:
v PAGE options or format items result in three lines being skipped.
v SKIP options or format items larger than SKIP (2) result in three lines being
skipped. SKIP (2) or less is treated in the usual manner.
v The ENDPAGE condition is never raised.
Changing the format on PRINT files
If you want normal spacing to apply to output from a PRINT file at the terminal,
you must supply your own tab table for PL/I. This is done by declaring an
external structure called PLITABS in the main program or in a program linked
with the main program and initializing the element PAGELENGTH to the number
of lines that can fit on your page. This value differs from PAGESIZE, which defines
the number of lines you want to print on the page before ENDPAGE is raised (see
Figure 15 on page 235). If you require a PAGELENGTH of 64 lines, declare
PLITABS as shown in Figure 14 on page 235. For information on overriding the tab
table, see Overriding the tab control table on page 302.
If your code contains a declare for PLITABS, not only must the values be valid, but
the first field in the PLITABS structure must also be valid. This field is supposed to
hold the offset to the field specifying the number of tabs set by the structure, and
the Enterprise PL/I library code will not work correctly if this is not true.
Automatic prompting
When the program requires input from a file that is associated with a terminal, it
issues a prompt. This takes the form of printing a colon on the next line and then
skipping to column 1 on the line following the colon. This gives you a full line to
enter your input, as follows:
:
(space for entry of your data)
DCL 1 PLITABS STATIC EXTERNAL,
( 2 OFFSET INIT (14),
2 PAGESIZE INIT (60),
2 LINESIZE INIT (120),
2 PAGELENGTH INIT (64),
2 FILL1 INIT (0),
2 FILL2 INIT (0),
2 FILL3 INIT (0),
2 NUMBER_OF_TABS INIT (5),
2 TAB1 INIT (25),
2 TAB2 INIT (49),
2 TAB3 INIT (73),
2 TAB4 INIT (97),
2 TAB5 INIT (121)) FIXED BIN (15,0);
Figure 14. Declaration of PLITABS. This declaration gives the standard page size, line size
and tabulating positions
Figure 15. PAGELENGTH and PAGESIZE. PAGELENGTH defines the size of your paper,
PAGESIZE the number of lines in the main printing area.
This type of prompt is referred to as a primary prompt.
Overriding automatic prompting
You can override the primary prompt by making a colon the last item in the
request for the data. You cannot override the secondary prompt. For example, the
two PL/I statements:
PUT SKIP EDIT ('ENTER TIME OF PERIHELION') (A);
GET EDIT (PERITIME) (A(10));
result in the terminal displaying:
ENTER TIME OF PERIHELION
: (automatic prompt)
(space for entry of data)
However, if the first statement has a colon at the end of the output, as follows:
PUT EDIT ('ENTER TIME OF PERIHELION:') (A);
the sequence is:
ENTER TIME OF PERIHELION: (space for entry of data)
Note: The override remains in force for only one prompt. You will be
automatically prompted for the next item unless the automatic prompt is
again overridden.
Punctuating long input lines
Line continuation character
To transmit data that requires 2 or more lines of space at the terminal as one
data-item, type an SBCS hyphen as the last character in each line except the last
line. For example, to transmit the sentence this data must be transmitted as one
unit. you enter:
:'this data must be transmitted -
+:as one unit.'
Transmission does not occur until you press ENTER after unit.'. The hyphen is
removed. The item transmitted is called a logical line.
Note: To transmit a line whose last data character is a hyphen or a PL/I minus
sign, enter two hyphens at the end of the line, followed by a null line as the
next line. For example:
xyz--
(press ENTER only, on this line)
Punctuating GET LIST and GET DATA statements
For GET LIST and GET DATA statements, a comma is added to the end of each
logical line transmitted from the terminal, if the programmer omits it. Thus there is
no need to enter blanks or commas to delimit items if they are entered on separate
logical lines. For the PL/I statement GET LIST(A,B,C); you can enter at the
terminal:
:1
+:2
+:3
This rule also applies when entering character-string data. Therefore, a character
string must transmit as one logical line. Otherwise, commas are placed at the break
points. For example, if you enter:
:'COMMAS SHOULD NOT BREAK
+:UP A CLAUSE.'
the resulting string is: COMMAS SHOULD NOT BREAK, UP A CLAUSE. The
comma is not added if a hyphen was used as a line continuation character.
Automatic padding for GET EDIT
For a GET EDIT statement, there is no need to enter blanks at the end of the line.
The data will be padded to the specified length. Thus, for the PL/I statement:
GET EDIT (NAME) (A(15));
you can enter the 5 characters SMITH. The data will be padded with ten blanks so
that the program receives the fifteen characters:
'SMITH '
Note: A single data item must transmit as a logical line. Otherwise, the first line
transmitted will be padded with the necessary blanks and taken as the
complete data item.
Use of SKIP for terminal input: All uses of SKIP for input are interpreted as
SKIP(1) when the file is allocated to the terminal. SKIP(1) is treated as an
instruction to ignore all unused data on the currently available logical line.
ENDFILE
The end-of-file can be entered at the terminal by keying in a logical line that
consists of the two characters /*. Any further attempts to use the file without
closing it result in the ENDFILE condition being raised.
SYSPRINT considerations
The PL/I standard SYSPRINT file is shared by multiple enclaves within an
application. You can issue I/O requests, for example STREAM PUT, from the same
or different enclaves. These requests are handled using the standard PL/I
SYSPRINT file as a file which is common to the entire application. The SYSPRINT
file is implicitly closed only when the application terminates, not at the termination
of the enclave.
The standard PL/I SYSPRINT file contains user-initiated output only, such as
STREAM PUTs. Run-time library messages and other similar diagnostic output are
directed to the Language Environment MSGFILE. See the z/OS Language
Environment Programming Guide for details on redirecting SYSPRINT file output to
the Language Environment MSGFILE.
To be shared by multiple enclaves within an application, the PL/I SYSPRINT file
must be declared as an EXTERNAL FILE constant with a file name of SYSPRINT
and also have the attributes STREAM and OUTPUT as well as the (implied)
attribute of PRINT, when OPENed. This is the standard SYSPRINT file as
defaulted by the compiler.
There exists only one standard PL/I SYSPRINT FILE within an application and
this file is shared by all enclaves within the application. For example, the
SYSPRINT file can be shared by multiple nested enclaves within an application or
by a series of enclaves that are created and terminated within an application by the
Language Environment preinitialization function. To be shared by an enclave
within an application, the PL/I SYSPRINT file must be declared in that enclave.
The standard SYSPRINT file cannot be shared by passing it as a file argument
between enclaves. The declared attributes of the standard SYSPRINT file should be
the same throughout the application, as with any EXTERNALly declared constant.
PL/I does not enforce this rule. Both the TITLE option and the
MSGFILE(SYSPRINT) option attempt to route SYSPRINT to another data set. As
such, if the two options were used together, there will be a conflict and the TITLE
option will be ignored.
Having a common SYSPRINT file within an application can be an advantage to
applications that utilize enclaves that are closely tied together. However, since all
enclaves in an application write to the same shared data set, this might require
some coordination among the enclaves.
The SYSPRINT file is opened (implicitly or explicitly) when first referenced within
an enclave of the application. When the SYSPRINT file is CLOSEd, the file
resources are released (as though the file had never been opened) and all enclaves
are updated to reflect the closed status.
If SYSPRINT is utilized in a multiple enclave application, the LINENO built-in
function only returns the current line number until after the first PUT or OPEN in
an enclave has been issued. This is required in order to maintain full compatibility
with old programs.
The COUNT built-in function is maintained at an enclave level. It always returns a
value of zero until the first PUT in the enclave is issued. If a nested child enclave
is invoked from a parent enclave, the value of the COUNT built-in function is
undefined when the parent enclave regains control from the child enclave.
The TITLE option can be used to associate the standard SYSPRINT file with
different operating system data sets, keeping in mind that a particular open
association has to be closed before another one is opened. This association is
retained across enclaves for the duration of the open.
PL/I condition handling associated with the standard PL/I SYSPRINT file retains
its current semantics and scope. For example, an ENDPAGE condition raised
within a child enclave will only invoke an established ON-unit within that child
enclave. It does not cause invocation of an ON-unit within the parent enclave.
The tabs for the standard PL/I SYSPRINT file can vary when PUTs are done from
different enclaves, if the enclaves contain a user PLITABS table.
If the PL/I SYSPRINT file is utilized as a RECORD file or as a STREAM INPUT
file, PL/I supports it at an individual enclave or task level, but not as a shareable
file among enclaves. If the PL/I SYSPRINT file is open at the same time with
different file attributes (e.g. RECORD and STREAM) in different enclaves of the
same application, results are unpredictable.
SYSPRINT may also be shared between code compiled by Enterprise PL/I and by
older PL/I compilers, but the following must all apply:
v SYSPRINT must be declared as STREAM OUTPUT
v the application must not be running under TSO
v if the runtime option MSGFILE(SYSPRINT) is in effect, then there must be no
preinitialized programs and no stored procedures in the application
Using MSGFILE(SYSPRINT)
Any file attributes that are specified in the ENVIRONMENT option of the file
declaration for SYSPRINT STREAM PRINT are ignored.
Any attributes that are specified on the OPEN statement for SYSPRINT are
ignored.
When you use the OPEN statement to open the PL/I SYSPRINT STREAM PRINT file,
the file is marked as opened in the PL/I control blocks, but it is actually opened by
the Language Environment.
When you use the CLOSE statements to close the PL/I SYSPRINT STREAM PRINT file,
the file is marked as closed in the PL/I control blocks, but Language Environment
still keeps it open.
The synchronization between the Language Environment messages and PL/I
user-specified output is not provided, so the order of the output is unpredictable.
The use of MSGFILE(SYSPRINT) restricts the line size specified by the LINESIZE
option to a maximum of 225 characters.
Using FETCH in your routines
In Enterprise PL/I, you can fetch routines compiled by PL/I, C, COBOL or
assembler.
Fetching Enterprise PL/I routines
Almost all the restrictions imposed by the older PL/I compilers on fetched
modules have been removed. So a FETCHed module can now:
v Fetch other modules
v Perform any I/O operations on any PL/I file. The file can be opened either by
the fetched module, by the main module, or by some other fetched module.
v ALLOCATE and FREE its own CONTROLLED variables
There are, however, a few restrictions on a Enterprise PL/I module that is to be
fetched. These restrictions are:
1. OPTIONS(FETCHABLE) should be specified on the PROCEDURE statement of
the fetched routine if there is no ENTRY card provided during the link-edit
step.
2. The ENTRY card should specify the name of that PL/I entry point
v If the module is to be fetched from Enterprise PL/I, you may specify
CEESTART on the ENTRY card, although this is strongly not recommended.
v However, if the module is to be fetched from COBOL or assembler, then the
ENTRY card absolutely must specify the name of the PL/I entry point into
the module and not CEESTART.
3. If the RENT compiler option was used to compile any of the fetched code, then
the module must be linked as a DLL.
4. If the NORENT compiler option was used to compile the fetching code, then
the fetched module must satisfy at least one of the following:
v It is a MAIN module
v It consists only of NORENT code
v It has as its entry point code compiled with the C or Enterprise PL/I
compiler with the NORENT option in effect. In this case, the module may
also contain code compiled with the RENT option, but calling that code is
not supported
5. If the RENT compiler option was used to compile the fetching code, then the
ENTRY that is fetched must not be declared in the fetching module as
OPTIONS(COBOL) or OPTIONS(ASM). If you want to avoid passing
descriptors in this situation, you should specify the
OPTIONS(NODESCRIPTOR) attribute on the ENTRY declare.
NORENT WRITABLE code is serially usable, and for that reason, the pointer that
is used to represent a FFETCHABLE constant is zeroed out in the prologue code of
any NORENT WRITABLE routine. While this insures that the code is serially
reusable while also providing the correct PL/I semantics, it does impose a
restriction on the use of FETCH with TITLE in NORENT WRITABLE code: if a
routine that did a FETCH A TITLE('B') is exited and reentered, then it must
re-execute the FETCH A TITLE('B'), before executing any CALL A statements
(otherwise it would do an implicit FETCH of A (but without any TITLE) before
making the CALL).
As an illustration of these restrictions, consider the compiler user exit. If you
specify the EXIT compile-time option, the compiler will fetch and call a Enterprise
PL/I module named IBMUEXIT.
First note that the compiler user exit must be compiled with the RENT option
since the compiler expects it to be a DLL.
In accordance with Item 1 above, the PROCEDURE statement for this routine looks
like:
ibmuexit:
proc ( addr_Userexit_Interface_Block,
addr_Request_Area )
options( fetchable );
dcl addr_Userexit_Interface_Block pointer byvalue;
dcl addr_Request_Area pointer byvalue;
In accordance with Item 3 above, the linker option DYNAM=DLL must be
specified when linking the user exit into a DLL. The DLL must be linked either
into a PDSE or into a temporary dataset (in which case DSNTYPE=LIBRARY must
be specified on the SYSLMOD DD statement).
All the JCL to compile, link, and invoke the user exit is given in the JCL below in
Figure 16 on page 241. The one significant difference between the sample below
and the code excerpts above is that, in the code below, the fetched user exit does
not receive two BYVALUE pointers to structures, but instead it receives the two
structures BYADDR. In order to make this change work, the code specifies
OPTIONS(NODESCRIPTOR) on each of its PROCEDURE statements.
//*
//*********************************************************************
//* compile the user exit
//*********************************************************************
//PLIEXIT EXEC PGM=IBMZPLI,
//STEPLIB DD DSN=IBMZ.V3R9M0.SIBMZCMP,DISP=SHR
//SYSLIN DD DSN=&&LOADSET,DISP=(MOD,PASS),UNIT=SYSSQ,
// SPACE=(CYL,(3,1))
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSDA,
//SYSIN DD *
*Process or(|) not(!);
*Process limits(extname(31));
*Process RENT;
/********************************************************************/
/* */
/* NAME - IBMUEXIT.PLI */
/* */
/* DESCRIPTION */
/* User-exit sample program. */
/* */
/* Licensed Materials - Property of IBM */
/* 5639-A83, 5639-A24 (C) Copyright IBM Corp. 1992,2009. */
/* All Rights Reserved. */
/* US Government Users Restricted Rights-- Use, duplication or */
/* disclosure restricted by GSA ADP Schedule Contract with */
/* IBM Corp. */
/* */
/* DISCLAIMER OF WARRANTIES */
/* The following "enclosed" code is sample code created by IBM */
/* Corporation. This sample code is not part of any standard */
/* IBM product and is provided to you solely for the purpose of */
/* assisting you in the development of your applications. The */
/* code is provided "AS IS", without warranty of any kind. */
/* IBM shall not be liable for any damages arising out of your */
/* use of the sample code, even if IBM has been advised of the */
/* possibility of such damages. */
/* */
/********************************************************************/
/********************************************************************/
/* */
/* During initialization, IBMUEXIT is called. It reads */
/* information about the messages being screened from a text */
/* file and stores the information in a hash table. IBMUEXIT */
/* also sets up the entry points for the message filter service */
/* and termination service. */
/* */
/* For each message generated by the compiler, the compiler */
/* calls the message filter registered by IBMUEXIT. The filter */
/* looks the message up in the hash table previously created. */
/* */
/* The termination service is called at the end of the compile */
/* but does nothing. It could be enhanced to generates reports */
/* or do other cleanup work. */
/* */
/********************************************************************/
Figure 16. Sample JCL to compile, link, and invoke the user exit (Part 1 of 7)
pack: package exports(*);
Dcl
1 Uex_UIB native Based( null() ),
2 Uex_UIB_Length fixed bin(31),
2 Uex_UIB_Exit_token pointer, /* for user exits use*/
2 Uex_UIB_User_char_str pointer, /* to exit option str */
2 Uex_UIB_User_char_len fixed bin(31),
2 Uex_UIB_Filename_str pointer, /* to source filename */
2 Uex_UIB_Filename_len fixed bin(31),
2 Uex_UIB_return_code fixed bin(31), /* set by exit procs */
2 Uex_UIB_reason_code fixed bin(31), /* set by exit procs */
2 Uex_UIB_Exit_Routs, /* exit entries setat
initialization */
3 ( Uex_UIB_Termination,
Uex_UIB_Message_Filter, /* call for each msg */
*, *, *, * )
limited entry (
*, /* to Uex_UIB */
* /* to a request area */
);
/*******************************************************************/
/* */
/* Request Area for Initialization exit */
/* */
/*******************************************************************/
Dcl 1 Uex_ISA native based( null() ),
2 Uex_ISA_Length fixed bin(31);
/*******************************************************************/
/* */
/* Request Area for Message_Filter exit */
/* */
/*******************************************************************/
Dcl 1 Uex_MFA native based( null() ),
2 Uex_MFA_Length fixed bin(31),
2 Uex_MFA_Facility_Id char(3),
2 * char(1),
2 Uex_MFA_Message_no fixed bin(31),
2 Uex_MFA_Severity fixed bin(15),
2 Uex_MFA_New_Severity fixed bin(15); /* set by exit proc */
/*******************************************************************/
/* */
/* Request Area for Terminate exit */
/* */
/*******************************************************************/
Dcl 1 Uex_TSA native based( null() ),
2 Uex_TSA_Length fixed bin(31);
/*******************************************************************/
/* */
/* Severity Codes */
/* */
/*******************************************************************/
dcl uex_Severity_Normal fixed bin(15) value(0);
dcl uex_Severity_Warning fixed bin(15) value(4);
dcl uex_Severity_Error fixed bin(15) value(8);
dcl uex_Severity_Severe fixed bin(15) value(12);
dcl uex_Severity_Unrecoverable fixed bin(15) value(16);
/*******************************************************************/
/* */
/* Return Codes */
/* */
/*******************************************************************/
dcl uex_Return_Normal fixed bin(15) value(0);
dcl uex_Return_Warning fixed bin(15) value(4);
dcl uex_Return_Error fixed bin(15) value(8);
dcl uex_Return_Severe fixed bin(15) value(12);
dcl uex_Return_Unrecoverable fixed bin(15) value(16);
/*******************************************************************/
/* */
/* Reason Codes */
/* */
/*******************************************************************/
dcl uex_Reason_Output fixed bin(15) value(0);
dcl uex_Reason_Suppress fixed bin(15) value(1);
dcl hashsize fixed bin(15) value(97);
dcl hashtable(0:hashsize-1) ptr init((hashsize) null());
dcl 1 message_item native based,
2 message_Info,
3 facid char(3),
3 msgno fixed bin(31),
3 newsev fixed bin(15),
3 reason fixed bin(31),
2 link pointer;
ibmuexit: proc ( ue, ia )
options( fetchable nodescriptor );
dcl 1 ue like uex_Uib byaddr;
dcl 1 ia like uex_Isa byaddr;
dcl sysuexit file stream input env(recsize(80));
dcl p pointer;
dcl bucket fixed bin(31);
dcl based_Chars char(8) based;
dcl title_Str char(8) var;
ue.uex_Uib_Message_Filter = message_Filter;
ue.uex_Uib_Termination = exitterm;
on undefinedfile(sysuexit)
begin;
put edit (** User exit unable to open exit file )
(A) skip;
put skip;
signal error;
end;
if ue.uex_Uib_User_Char_Len = 0 then
do;
open file(sysuexit);
end;
else
do;
title_Str
= substr( ue.uex_Uib_User_Char_Str->based_Chars,
1, ue.uex_Uib_User_Char_Len );
open file(sysuexit) title(title_Str);
end;
on error, endfile(sysuexit)
goto done;
allocate message_item set(p);
/***************************************************************/
/* */
/* Skip header lines and read first data line */
/* */
/***************************************************************/
get file(sysuexit) list(p->message_info) skip(3);
do loop;
/*************************************************************/
/* */
/* Put message information in hash table */
/* */
/*************************************************************/
bucket = mod(p->msgno, hashsize);
p->link = hashtable(bucket);
hashtable(bucket) = p;
/*************************************************************/
/* */
/* Read next data line */
/* */
/*************************************************************/
allocate message_item set(p);
get file(sysuexit) skip;
get file(sysuexit) list(p->message_info);
end;
/***************************************************************/
/* */
/* Clean up */
/* */
/***************************************************************/
done:
free p->message_Item;
close file(sysuexit);
end;
message_Filter:
proc ( ue, mf )
options( nodescriptor );
dcl 1 mf like uex_Mfa byaddr;
dcl p pointer;
dcl bucket fixed bin(15);
on error snap system;
ue.uex_Uib_Reason_Code = uex_Reason_Output;
ue.uex_Uib_Return_Code = 0;
mf.uex_Mfa_New_Severity = mf.uex_Mfa_Severity;
/***************************************************************/
/* */
/* Calculate bucket for error message */
/* */
/***************************************************************/
bucket = mod(mf.uex_Mfa_Message_No, hashsize);
/***************************************************************/
/* */
/* Search bucket for error message */
/* */
/***************************************************************/
do p = hashtable(bucket) repeat (p->link) while(p!=null())
until (p->msgno = mf.uex_Mfa_Message_No &
p->facid = mf.Uex_Mfa_Facility_Id);
end;
if p = null() then;
else
do;
/***********************************************************/
/* */
/* Filter error based on information in has table */
/* */
/***********************************************************/
ue.uex_Uib_Reason_Code = p->reason;
if p->newsev < 0 then;
else
mf.uex_Mfa_New_Severity = p->newsev;
end;
end;
exitterm:
proc ( ue, ta )
dcl 1 ta like uex_Tsa byaddr;
ue.uex_Uib_return_Code = 0;
ue.uex_Uib_reason_Code = 0;
end;
end pack;
//*********************************************************************
//* link the user exit
//*********************************************************************
//LKEDEXIT EXEC PGM=IEWL,PARM=XREF,LIST,LET,DYNAM=DLL,
// COND=(9,LT,PLIEXIT),REGION=5000K
//SYSLIB DD DSN=CEE.SCEELKED,DISP=SHR
//SYSLMOD DD DSN=&&EXITLIB(IBMUEXIT),DISP=(NEW,PASS),UNIT=SYSDA,
// SPACE=(TRK,(7,1,1)),DSNTYPE=LIBRARY
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSDA,SPACE=(CYL,(3,1)),
// DCB=BLKSIZE=1024
//SYSPRINT DD SYSOUT=X
//SYSDEFSD DD DUMMY
//SYSLIN DD DSN=&&LOADSET,DISP=SHR
// DD DDNAME=SYSIN
//LKED.SYSIN DD *
ENTRY IBMUEXIT
Fetching PL/I MAIN Routines
When you fetch a PL/I MAIN program, a child enclave is created in Language
Environment.
You need to follow these rules to fetch a PL/I MAIN program:
v You cannot pass any runtime options to a fetched MAIN program in the
parameter string. If you want the fetched MAIN routine to have its own runtime
option, use either of the following ways:
Declare a PLIXOPT variable in the fetched MAIN routine.
Link the fetched MAIN routine with a CEEUOPT module.
v You can pass only a VARYING CHARACTER string to a fetched MAIN routine
that is compiled with the SYSTEM(MVS) compiler option.
v You must specify NOEXECOPS in the option on the PROCEDURE statement of
a fetched MAIN routine.
v You must not specify OPTION(ASM) or OPTION(NODESCRIPTOR) in the
ENTRY declaration for the fetched MAIN routine in the fetching program.
Fetching z/OS C routines
Unless the NORENT option has been specified, the ENTRY declaration in the
routine that fetches an z/OS C routine must not specify OPTIONS(COBOL) or
OPTIONS(ASM)these should be specified only for COBOL or ASM routines not
linked as DLLs
The z/OS C documentation provides instructions on how to compile and link an
z/OS C DLL.
//*********************************************************************
//* compile main
//*********************************************************************
//PLI EXEC PGM=IBMZPLI,PARM=F(I),EXIT,
// REGION=256K
//STEPLIB DD DSN=&&EXITLIB,DISP=SHR
// DD DSN=IBMZ.V3R9M0.SIBMZCMP,DISP=SHR
//SYSLIN DD DSN=&&LOADSET2,DISP=(MOD,PASS),UNIT=SYSSQ,
// SPACE=(CYL,(3,1))
//SYSUT1 DD DSN=&&SYSUT1,UNIT=SYSDA,
//SYSIN DD *
*process;
MainFet: Proc Options(Main);
/* the exit will suppress the message for the next dcl */
dcl one_byte_integer fixed bin(7);
End ;
//*
//SYSUEXIT DD DISP=SHR,DSN=hlq.some.user.exit.input.file *
Fac Id Msg No Severity Suppress Comment
+--------+--------+----------+----------+-------------------------------
IBM 1042 -1 1 String spans multiple lines
IBM 1044 -1 1 FIXED BIN 7 mapped to 1 byte
Fetching assembler routines
Unless the NORENT option has been specified, the ENTRY declaration in the
routine that fetches an assembler routine must specify OPTIONS(ASM).
Invoking MAIN under TSO/E
If you compile your MAIN program with the SYSTEM(MVS) option, you can
invoke the program using the TSO CALL command or as a TSO command
processor. Both the run-time options and parameters can be passed the same way
as under MVS batch.
For example, if the MAIN program TSOARG1 was link-edited as a member of a
data set named userid.TEST.load, it can be invoked by:
CALL TEST(TSOARG1) RPTSTG(ON),TRAP(ON)/THIS IS MY ARGUMENT
or
CALL TEST(TSOARG1) /THIS IS MY ARGMENT
or
TSOARG1 TRAP(ON)/THIS IS MY ARGUMENT
or
TSOARG1 /THIS IS MY ARGUMENT
NOTE: The data set containing TSOARG1 (userid.TEST.load) must be in the
standard TSO program search list to run the program without using the CALL
statement. This can be accomplished by issuing the TSO command
TSOLIB ACTIVATE DSN(userid.TEST.load)
However, if you compile your MAIN program with the SYSTEM(TSO) option, the
program will be passed a pointer to the Command Processor Parameter List
(CPPL). In this case, NOEXECOPS is in effect. Your program can be invoked as a
TSO command, but it cannot be invoked by a TSO CALL statement. For example:
TSOARG2 This is my argument
The program in Figure 17 on page 249 uses the SYSTEM(TSO) interface to address
and display the program arguments from the CPPL.
Whether your MAIN program is invoked as a command or via CALL, you can
always specify run-time options via the PLIXOPT string.
Invoking MAIN under z/OS UNIX
Under z/OS UNIX, if you compile a MAIN program with the SYSTEM(MVS)
option, the program will be passed, as usual, one CHARACTER VARYING string
containing the parameters specified when it was invoked.
However, under z/OS UNIX, if you compile a MAIN program with the
SYSTEM(OS) option, the program will be passed 7 parameters as specified in the
z/OS UNIX manuals. These 7 parameters include:
v the argument count (which includes the name of the executable as the first
"argument"
v the address of an array of addresses of the lengths of the arguments
v the address of an array of addresses of the arguments as null-terminated
character strings
v the count of environment variables set
v the address of an array of addresses of the lengths of the environment variables
v the address of an array of addresses of the environment variables as
null-terminated character strings
The program in Figure 18 on page 250 uses the SYSTEM(OS) interface to address
and display the individual arguments and environment variables.
*process system(tso);
tsoarg2: proc (cppl_ptr) options(main);
dcl cppl_ptr pointer;
dcl 1 cppl based(cppl_ptr),
2 cpplcbuf pointer,
2 cpplupt pointer,
2 cpplpscb pointer,
2 cpplect pointer;
dcl 1 cpplbuf based(cpplcbuf),
2 len fixed bin(15),
2 offset fixed bin(15),
2 argstr char(1000);
dcl my_argument char(1000) varying;
dcl my_argument_len fixed bin(31);
dcl length builtin;
my_argument_len = len - offset - 4;
if my_argument_len = 0 then
my_argument = ;
else
my_argument = substr(argstr,offset + 1, my_argument_len);
display(Program args: || my_argument);
end tsoarg2;
Figure 17. Sample program to display program arguments from the CPPL under TSO when
using SYSTEM(STD) option
*process display(std) system(os);
sayargs:
proc(argc, pArgLen, pArgStr, envc, pEnvLen, pEnvStr, pParmSelf)
options( main, noexecops );
dcl argc fixed bin(31) nonasgn byaddr;
dcl pArgLen pointer nonasgn byvalue;
dcl pArgStr pointer nonasgn byvalue;
dcl envc fixed bin(31) nonasgn byaddr;
dcl pEnvLen pointer nonasgn byvalue;
dcl pEnvStr pointer nonasgn byvalue;
dcl pParmSelf pointer nonasgn byvalue;
dcl q(4095) pointer based;
dcl bxb fixed bin(31) based;
dcl bcz char(31) varz based;
display( argc = || argc );
do jx = 1 to argc;
display( pargStr(jx) = || pArgStr->q(jx)->bcz );
end;
display( envc = || envc );
do jx = 1 to envc;
display( pEnvStr(jx) = || pEnvStr->q(jx)->bcz );
end;
end;
Figure 18. Sample program to display z/OS UNIX args and environment variables
Part 2. Using I/O facilities
Chapter 6. Using data sets and files . . . . . 253
Allocating files . . . . . . . . . . . . . 253
Associating data sets with files under z/OS . . . 254
Associating several files with one data set . . . 256
Associating several data sets with one file . . . 256
Concatenating several data sets . . . . . . 257
Accessing HFS files under z/OS . . . . . . 257
Associating data sets with files under z/OS UNIX 258
Using environment variables . . . . . . . 259
Attempting to use files not associated with data
sets. . . . . . . . . . . . . . . . 261
How PL/I finds data sets . . . . . . . . 261
Specifying characteristics using DD_DDNAME
environment variables . . . . . . . . . 261
APPEND. . . . . . . . . . . . . 262
BUFSIZE . . . . . . . . . . . . . 263
CHARSET for record I/O . . . . . . . 263
CHARSET for stream I/O . . . . . . . 263
DELAY . . . . . . . . . . . . . 263
DELIMIT . . . . . . . . . . . . . 264
LRECL . . . . . . . . . . . . . 264
LRMSKIP. . . . . . . . . . . . . 264
PROMPT . . . . . . . . . . . . . 264
PUTPAGE . . . . . . . . . . . . 264
RECCOUNT. . . . . . . . . . . . 265
RECSIZE . . . . . . . . . . . . . 265
SAMELINE . . . . . . . . . . . . 265
SKIP0 . . . . . . . . . . . . . . 266
TYPE . . . . . . . . . . . . . . 266
Establishing data set characteristics . . . . . . 267
Blocks and records . . . . . . . . . . 267
Information interchange codes. . . . . . 268
Record formats . . . . . . . . . . . . 268
Fixed-length records . . . . . . . . . 268
Variable-length records . . . . . . . . 269
Undefined-length records . . . . . . . 270
Data set organization. . . . . . . . . . 270
Labels . . . . . . . . . . . . . . . 271
Data Definition (DD) statement . . . . . . 271
Use of the conditional subparameters . . . 271
Data set characteristics . . . . . . . . 272
Associating PL/I files with data sets . . . . 273
Opening a file . . . . . . . . . . . 273
Closing a file . . . . . . . . . . . 274
Specifying characteristics in the
ENVIRONMENT attribute . . . . . . . . 275
The ENVIRONMENT attribute . . . . . 275
Record formats for record-oriented data
transmission. . . . . . . . . . . . 277
Record formats for stream-oriented data
transmission. . . . . . . . . . . . 277
RECSIZE option . . . . . . . . . . 277
BLKSIZE option . . . . . . . . . . 278
Record format, BLKSIZE, and RECSIZE
defaults . . . . . . . . . . . . . 279
GENKEY option key classification . . . 280
SCALARVARYING option varying-length
strings. . . . . . . . . . . . . . 281
KEYLENGTH option . . . . . . . . . 282
ORGANIZATION option . . . . . . . 282
Data set types used by PL/I record I/O . . 282
Setting environment variables . . . . . . 283
PL/I standard files (SYSPRINT and SYSIN) 284
Redirecting standard input, output, and error
devices . . . . . . . . . . . . . 284
Chapter 7. Using libraries . . . . . . . . 285
Types of libraries . . . . . . . . . . . . 285
How to use a library . . . . . . . . . . 285
Creating a library . . . . . . . . . . . . 286
SPACE parameter . . . . . . . . . . . 286
Creating and updating a library member . . . . 287
Examples. . . . . . . . . . . . . . 287
Extracting information from a library directory . . 289
Chapter 8. Defining and using consecutive data
sets . . . . . . . . . . . . . . . . 291
Using stream-oriented data transmission . . . . 291
Defining files using stream I/O . . . . . . 291
Defining stream files using PL/I dynamic
allocation. . . . . . . . . . . . . . 292
CONSECUTIVE . . . . . . . . . . 292
Record format options . . . . . . . . 293
RECSIZE . . . . . . . . . . . . . 293
Defaults for record format, BLKSIZE, and
RECSIZE . . . . . . . . . . . . . 293
GRAPHIC option . . . . . . . . . . 294
Creating a data set with stream I/O. . . . . 294
Essential information . . . . . . . . . 295
Examples. . . . . . . . . . . . . 295
Accessing a data set with stream I/O . . . . 297
Essential information . . . . . . . . . 298
Record format . . . . . . . . . . . 298
Example . . . . . . . . . . . . . 298
Using PRINT files with stream I/O . . . . . 299
Controlling printed line length . . . . . 300
Overriding the tab control table . . . . . 302
Using SYSIN and SYSPRINT files . . . . . 303
Controlling input from the terminal . . . . . . 304
Format of data . . . . . . . . . . . . . 305
Defining QSAM files using PL/I dynamic
allocation. . . . . . . . . . . . . . . 306
Capital and lowercase letters . . . . . . . . 307
End-of-file . . . . . . . . . . . . . . 307
COPY option of GET statement . . . . . . . 307
Chapter 9. Controlling output to the terminal 309
Format of PRINT files . . . . . . . . . . 309
Output from the PUT EDIT command . . . . . 309
Chapter 10. Using record-oriented data
transmission . . . . . . . . . . . . . 311
Specifying record format . . . . . . . . . 312
Defining files using record I/O . . . . . . . 312
Specifying ENVIRONMENT options . . . . . 312
CONSECUTIVE . . . . . . . . . . . 312
ORGANIZATION(CONSECUTIVE) . . . . . 313
CTLASA|CTL360 . . . . . . . . . . . 313
LEAVE|REREAD . . . . . . . . . . . 314
Creating a data set with record I/O . . . . . . 315
Accessing and updating a data set with record I/O 316
Example of consecutive data sets . . . . . . 317
Chapter 11. Defining and using regional data
sets . . . . . . . . . . . . . . . . 321
Defining REGIONAL(1) data sets using PL/I
dynamic allocation . . . . . . . . . . . 323
Defining files for a regional data set . . . . . . 323
REGIONAL option . . . . . . . . . 324
Using keys with REGIONAL data sets . . . . 324
Using REGIONAL(1) data sets . . . . . . . 324
Dummy Records . . . . . . . . . . . 325
Creating a REGIONAL(1) data set . . . . . 325
Example . . . . . . . . . . . . . 325
Accessing and updating a REGIONAL(1) data
set . . . . . . . . . . . . . . . . 326
Sequential access . . . . . . . . . . 327
Direct access. . . . . . . . . . . . 327
Example . . . . . . . . . . . . . 327
Essential information for creating and accessing
regional data sets . . . . . . . . . . . . 329
Chapter 12. Defining and using VSAM data sets 333
Defining VSAM file using PL/I dynamic allocation 333
Using VSAM data sets . . . . . . . . . . 333
How to run a program with VSAM data sets 333
Pairing an Alternate Index Path with a File . . 334
VSAM organization . . . . . . . . . . . 334
Keys for VSAM data sets . . . . . . . . 336
Keys for indexed VSAM data sets . . . . 336
Relative byte addresses (RBA) . . . . . . 337
Relative record numbers. . . . . . . . 337
Choosing a data set type . . . . . . . . 337
Defining files for VSAM data sets . . . . . . 339
BKWD option . . . . . . . . . . . 340
BUFND option . . . . . . . . . . . 340
BUFNI option . . . . . . . . . . . 341
BUFSP option . . . . . . . . . . . 341
GENKEY option . . . . . . . . . . 341
PASSWORD option . . . . . . . . . 341
REUSE option . . . . . . . . . . . 341
SKIP option . . . . . . . . . . . . 342
VSAM option . . . . . . . . . . . 342
Performance options . . . . . . . . . . 342
Defining Files for Alternate Index Paths . . . . 342
Defining VSAM data sets . . . . . . . . . 343
Entry-sequenced data sets . . . . . . . . . 344
Loading an ESDS . . . . . . . . . . . 344
Using a SEQUENTIAL file to access an ESDS 345
Defining and loading an ESDS . . . . . 345
Updating an ESDS . . . . . . . . . 346
Key-sequenced and indexed entry-sequenced
data sets . . . . . . . . . . . . . 347
Alternate Indexes for KSDSs or Indexed
ESDSs . . . . . . . . . . . . . . 353
Relative-record data sets. . . . . . . . 360
Using Files Defined for non-VSAM Data Sets. . . 366
Using Shared Data Sets . . . . . . . . . 366
Chapter 6. Using data sets and files
Your PL/I programs process and transmit units of information called records. A
collection of records is called a data set. Data sets are physical collections of
information external to PL/I programs; they can be created, accessed, or modified
by programs written in PL/I or other languages or by the utility programs of the
operating system.
Your PL/I program recognizes and processes information in a data set by using a
symbolic or logical representation of the data set called a file. This chapter
describes how to allocate files and associate data sets with the files known within
your program. It introduces the five major types of data sets, how they are
organized and accessed, and some of the file and data set characteristics you need
to know how to specify.
Note: INDEXED implies VSAM and is supported only under batch.
Allocating files
For any type of file, including sequential, VSAM, REGIONAL(1) and HFS, you can
define the external name by using the following methods:
v In the MVS or TSO environment:
A ddname in the JCL
An environment variable name
The TITLE option of the OPEN statement
v In the USS environment:
An environment variable name
The TITLE option of the OPEN statement
The following rules of precedence apply in determining when the dynamic
allocation takes place:
1. If one of the following DD statement exists for the file, it is used. This rule is
not valid in the USS environment.
v JCL DD
v TSO ALLOCATE
v User-initiated dynamic allocation
2. If a DD statement does not exist for the file and the TITLE option is specified
in the OPEN statement, the TITLE option is used by associating it with the
external name for the file.
3. If a DD statement does not exist for the file and the TITLE option is not
specified, but an environment variable exists for the file, the environment
variable is used by associating it with the external name for the file.
The file is dynamically allocated by using the attributes specified by the
environment variable or TITLE option. At a minimum, the PATH() or DSN() option
must be specified, with the DSN() option accompanied by a disposition value. All
options and attributes must be in uppercase, except for the pathname suboption of
the PATH option, which is case sensitive. Temporary data set names in the DSN()
option are not allowed.
For MVS data sets, the Enterprise PL/I run time checks the contents of the
environment variable or TITLE option at each OPEN statement:
v If a file with the same external name was dynamically allocated by a previous
OPEN statement, and the contents of the environment variable or TITLE option
have changed since that OPEN statement, the run time dynamically deallocates
the previous allocation and reallocates the file by using the options that is
currently set in the environment variable or specified in the TITLE option.
v If the contents of the environment variable or TITLE option have not changed,
the run time uses the current allocation without taking the action of deallocation
or reallocation.
For HFS files, the DD statement is deallocated and reallocated at each subsequent
OPEN statement.
To use PL/I dynamic allocation, you must specify the file names by using the
DSN() format for MVS data sets, or using the PATH() format for HFS files as
shown in the following examples:
OPEN FILE(FILEIN) TITLE(DSN(USER.FILE.EXT),SHR);
OPEN FILE(FILEIN) TITLE(PATH(/usr/FILE.EXT));
EXPORT DD_FILE="DSN(USER.FILE.EXT),SHR"
EXPORT DD_FILE="PATH(/usr/FILE.EXT)"
Notes:
1. When you use the DSN() or the PATH() format specification for PL/I dynamic
allocation, if a DD statement and the TITLE option of the OPEN statement are
both specified for a file, the DD statement is used and the TITLE option is
ignored.
2. Under the USS environment, user-initiated dynamic allocation is not supported.
If it is attempted, the UNDEFINEDFILE condition is raised, even if the TITLE
or environment variable option is used, because the external name is in use.
Associating data sets with files under z/OS
A file used within a PL/I program has a PL/I file name. The physical data set
external to the program has a name by which it is known to the operating system:
a data set name or dsname. In some cases the data set has no name; it is known to
the system by the device on which it exists.
The operating system needs a way to recognize which physical data set is referred
to by your program, so you must write a data definition or DD statement, external
to your program, that associates the PL/I file name with a dsname. For example, if
you have the following file declaration in your program:
DCL STOCK FILE STREAM INPUT;
you should create a DD statement with a data definition name (ddname) that matches
the name of the PL/I file. The DD statement specifies a physical data set name
(dsname) and gives its characteristics:
//GO.STOCK DD DSN=PARTS.INSTOCK, . . .
You'll find some guidance in writing DD statements in this manual, but for more
detail refer to the job control language (JCL) manuals for your system.
There is more than one way to associate a data set with a PL/I file. You associate a
data set with a PL/I file by ensuring that the ddname of the DD statement that
defines the data set is the same as one of the following:
v The declared PL/I file name
v The character-string value of the expression specified in the TITLE option of the
associated OPEN statement.
You must choose your PL/I file names so that the corresponding ddnames
conform to the following restrictions:
v If a file is opened implicitly, or if no TITLE option is included in the OPEN
statement that explicitly opens the file, the ddname defaults to the file name. If
the file name is longer than 8 characters, the default ddname is composed of the
first 8 characters of the file name.
v The character set of the JCL does not contain the break character (_).
Consequently, this character cannot appear in ddnames. Do not use break
characters among the first 8 characters of file names, unless the file is to be
opened with a TITLE option with a valid ddname as its expression. The
alphabetic extender characters $, @, and #, however, are valid for ddnames, but
the first character must be one of the letters A through Z.
Since external names are limited to 7 characters, an external file name of more than
7 characters is shortened into a concatenation of the first 4 and the last 3 characters
of the file name. Such a shortened name is not, however, the name used as the
ddname in the associated DD statement.
Consider the following statements:
1. OPEN FILE(MASTER);
2. OPEN FILE(OLDMASTER);
3. READ FILE(DETAIL) ...;
When statement number 1 is run, the file name MASTER is taken to be the same
as the ddname of a DD statement in the current job step. When statement number
2 is run, the name OLDMASTE is taken to be the same as the ddname of a DD
statement in the current job step. (The first 8 characters of a file name form the
ddname. If OLDMASTER is an external name, it will be shortened by the compiler
to OLDMTER for use within the program.) If statement number 3 causes implicit
opening of the file DETAIL, the name DETAIL is taken to be the same as the
ddname of a DD statement in the current job step.
In each of the above cases, a corresponding DD statement must appear in the job
stream; otherwise, the UNDEFINEDFILE condition is raised. The three DD
statements could start as follows:
1. //MASTER DD ...
2. //OLDMASTE DD ...
3. //DETAIL DD ...
If the file reference in the statement which explicitly or implicitly opens the file is
not a file constant, the DD statement name must be the same as the value of the
file reference. The following example illustrates how a DD statement should be
associated with the value of a file variable:
DCL PRICES FILE VARIABLE,
RPRICE FILE;
PRICES = RPRICE;
OPEN FILE(PRICES);
The DD statement should associate the data set with the file constant RPRICE,
which is the value of the file variable PRICES, thus:
//RPRICE DD DSNAME=...
Use of a file variable also allows you to manipulate a number of files at various
times by a single statement. For example:
DECLARE F FILE VARIABLE,
A FILE,
B FILE,
C FILE;
.
.
.
DO F=A,B,C;
READ FILE (F) ...;
.
.
.
END;
The READ statement reads the three files A, B, and C, each of which can be
associated with a different data set. The files A, B, and C remain open after the
READ statement is executed in each instance.
The following OPEN statement illustrates use of the TITLE option:
OPEN FILE(DETAIL) TITLE('DETAIL1');
For this statement to be executed successfully, you must have a DD statement in
the current job step with DETAIL1 as its ddname. It could start as follows:
//DETAIL1 DD DSNAME=DETAILA,...
Thus, you associate the data set DETAILA with the file DETAIL through the
ddname DETAIL1.
Associating several files with one data set
You can use the TITLE option to associate two or more PL/I files with the same
external data set, provided the first file association is closed before opening a
second file association against the same TITLE name. The following example,
where INVNTRY is the name of a DD statement defining a data set to be
associated with two files:
OPEN FILE (FILE1) TITLE('INVNTRY');
.....
CLOSE FILE (FILE1);
.....
OPEN FILE (FILE2) TITLE('INVNTRY');
If the file is not closed first before the second open is done, the UNDEFINEDFILE
condition will be raised with a subcode1 value of 59, stating that an open was
attempted against a file that was already open.
Associating several data sets with one file
The file name can, at different times, represent entirely different data sets. In the
above example of the OPEN statement, the file DETAIL1 is associated with the
data set named in the DSNAME parameter of the DD statement DETAIL1. If you
closed and reopened the file, you could specify a different ddname in the TITLE
option to associate the file with a different data set.
Use of the TITLE option allows you to choose dynamically, at open time, one
among several data sets to be associated with a particular file name. Consider the
following example:
DO IDENT='A','B','C';
OPEN FILE(MASTER)
TITLE('MASTER1'||IDENT);
.
.
.
CLOSE FILE(MASTER);
END;
In this example, when MASTER is opened during the first iteration of the
do-group, the associated ddname is taken to be MASTER1A. After processing, the
file is closed, dissociating the file name and the ddname. During the second
iteration of the do-group, MASTER is opened again. This time, MASTER is
associated with the ddname MASTER1B. Similarly, during the final iteration of the
do-group, MASTER is associated with the ddname MASTER1C.
Concatenating several data sets
For input only, you can concatenate two or more sequential or regional data sets
(that is, link them so that they are processed as one continuous data set) by
omitting the ddname from all but the first of the DD statements that describe
them. For example, the following DD statements cause the data sets LIST1, LIST2,
and LIST3 to be treated as a single data set for the duration of the job step in
which the statements appear:
//GO.LIST DD DSNAME=LIST1,DISP=OLD
// DD DSNAME=LIST2,DISP=OLD
// DD DSNAME=LIST3,DISP=OLD
When read from a PL/I program, the concatenated data sets need not be on the
same volume. You cannot process concatenated data sets backward.
Accessing HFS files under z/OS
You can access HFS files from a batch program by specifying the HFS file name in
the DD statement or in the TITLE option of the OPEN statement.
For example, to access the HFS file /u/USER/sample.txt by using the DD HFS, you
would code the DD statement as follows:
//HFS DD PATH=/u/USER/sample.txt,PATHOPTS=ORDONLY,DSNTYPE=HFS
To access the same file by using the TITLE option of the OPEN statement, you
would code:
OPEN FILE(HFS) TITLE(//u/USER/sample.txt);
Note the two forward slashes in the TITLE option: the first indicates that what
follows is a file name (rather than a DD name), and the second is the start of the
fully qualified HFS file name (and fully qualified names have to be used when
HFS files are referenced under batch, because there is no current directory that
could be used to complete a file specification).
You can access HFS files from a batch program using PL/I dynamic allocation by
specifying the HFS file name using one of the following methods:
v The DD statement
v The TITLE option of the OPEN statement
v The PUTENV built-in function
v The PLIXOPT string using the ENVAR option
The following example shows how to access HFS files by using these methods:
//HFS DD PATH=/u/USER/sample.txt,PATHOPTS=ORDONLY...
OPEN FILE(HFS) TITLE(PATH(/u/USER/sample.txt));
xx = putenv(DD_HFS=/u/USER/sample.txt);
Dcl plixopt char(100) var ext static
init(ENVAR("DD_HFS=PATH(/u/USER/sample.txt)"));
Note: To use PL/I dynamic allocation, specify the file names by using the DSN()
format for MVS data sets, or the PATH() format for HFS files.
PL/I decides how to treat HFS files under batch in the following order:
1. If ENV(F) is specified in the file declaration, the file is assumed to consist of
fixed length records.
2. If ENV(V) is specified in the file declaration, the file is assumed to consist of
lf-delimited records.
3. If FILEDATA=BINARY is specified on the file's DD statement, the file is
assumed to consist of fixed length records.
4. Otherwise the file is assumed to consist of lf-delimited records.
To access a fixed length z/OS file from UNIX, the record size of the file must be
specified either in the ENVIRONMENT attribute of the file or in the TITLE option
on the OPEN statement. So,
v If the file declaration does not contain ENV( F RECSIZE(...) ), you must specify
the data set name and these attributes in the TITLE option. For example, for a
file of fixed length 80-byte records, you could specify a TITLE option as follows:
/dataset.name,type(fixed),recsize(80)
v If the ENVIRONMENT attribute specifies only one of F or RECSIZE, you must
specify the data set name and the omitted attribute in the TITLE option.
v If the ENVIRONMENT attribute specifies both F and RECSIZE, you need to
specify only the data set name in the TITLE option.
Associating data sets with files under z/OS UNIX
A file used within a PL/I program has a PL/I file name. A data set also has a
name by which it is known to the operating system.
PL/I needs a way to recognize the data sets to which the PL/I files in your
program refer, so you must provide an identification of the data set to be used, or
allow PL/I to use a default identification.
You can identify the data set explicitly using either an environment variable or the
TITLE option of the OPEN statement.
To use PL/I dynamic allocation, you must specify the file names by using the
DSN() format for MVS data sets, or the PATH() format for HFS files.
To access HFS files from z/OS UNIX program using PL/I dynamic allocation, you
can specify the HFS filename by using one of the following methods:
v The TITLE option of the OPEN statement
v The PUTENV built-in function
v The PLIXOPT string using the ENVAR option
v The EXPORT statement
Using environment variables
Use the export command to establish an environment variable that identifies the
data set to be associated with a PL/I file, and, optionally, to specify the
characteristics of that data set. The information provided by the environment
variable is called data definition (or DD) information.
These environment variable names have the form DD_DDNAME where the
DDNAME is the name of a PL/I file constant (or an alternate DDNAME, as defined
below). If the filename refers to an HFS file, the filename has to be properly
qualified. Otherwise, the PL/I library assumes the filename refers to an MVS data
set.
Examples:
declare MyFile stream output;
export DD_MYFILE=/datapath/mydata.dat
/datapath/mydata.dat refers to an HFS file. The filename is fully-qualified.
export DD_MYFILE=./mydata.dat
./mydata.dat refers to an HFS file residing in the current directory.
export DD_MYFILE=mydata.dat
mydata.dat refers to an MVS data set.
The following examples show that the HFS files are accessed when using PL/I
dynamic allocation:
export DD_HFS="PATH(/u/USER/sample.txt)"
export DD_FILE="DSN(USER.FILE.EXT),SHR"
If you are familiar with the IBM mainframe environment, you can think of the
environment variable much like:
DD statement in z/OS
ALLOCATE statement in TSO
For more information about the syntax and options you can use with the
DD_DDNAME environment variable, see Specifying characteristics using
DD_DDNAME environment variables on page 261.
Under z/OS UNIX, where more types of varying length HFS files are supported
than under batch, PL/I treats an HFS file as follows:
1. If ENV(F) is specified in the file declaration, the file is assumed to consist of
fixed length records.
2. If a TYPE is specified in the file's EXPORT statement, the file is assumed to
consist of records of that type.
3. Otherwise the file is assumed to consist of lf-delimited records.
Using the TITLE option of the OPEN statement
You can use the TITLE option of the OPEN statement to identify the data set to be
associated with a PL/I file, and, optionally, to provide additional characteristics of
that data set.
TITLE ('expression')
The expression must yield a character string with the following syntax:
alternate_ddname
/filespec
, dd_option
DSN ( dsname ) , dsn_option
PATH ( pathname )
alternate_ddname
The name of an alternate DD_DDNAME environment variable. An alternate
DD_DDNAME environment variable is not named after a file constant. For
example, if you have a file named INVENTRY in your program, and you
establish two DD_DDNAME environment variablesthe first named
INVENTRY and the second named PARTSyou can associate the file with the
second one using this statement:
open file(Inventry) title(PARTS);
filespec
Any valid file specification on the system you are using. The maximum length
of filespec is 1023 characters.
dd_option
One or more options allowed in a DD_DDNAME environment variable
For more information about the options of the DD_DDNAME environment
variable, see Specifying characteristics using DD_DDNAME environment
variables on page 261.
dsname
The fully qualified MVS data set name
dsn_option
One or more DSN options
For more information about the DSN options, see Defining QSAM files using
PL/I dynamic allocation on page 306, Defining REGIONAL(1) data sets
using PL/I dynamic allocation on page 323, and Defining VSAM file using
PL/I dynamic allocation on page 333.
pathname
The fully qualified HFS path name
Here is an example of using the OPEN statement in this manner with a z/OS
DSN:
open file(Payroll) title('/June.Dat,append(n),recsize(52)');
Note the required leading forward slash in the TITLE option. This leading forward
slash indicates that what follows is a file name (rather than a DD name). In this
case, June.Dat refers to an MVS dataset.
If June.Dat is an HFS file, the example looks like this:
open file(Payroll) title('//u/USER/June.Dat,append(n),recsize(52)');
Note the two forward slashes in the TITLE option: the first indicates that what
follows is a file name (rather than a DD name), and the second is the start of the
fully qualified HFS file name.
Relative HFS file names can also be specified in place of fully qualified names, for
example:
open file(Payroll) title('./June.Dat,append(n),recsize(52)');
The dataset name June.Dat will be prefixed with the pathname of the current z/OS
UNIX directory.
With this form, PL/I obtains all DD information either from the TITLE expression
or from the ENVIRONMENT attribute of a file declaration - a DD_DDNAME
environment variable is not referenced.
The following examples show how to use the OPEN statement with PL/I dynamic
allocation support:
v For a z/OS data set:
OPEN FILE(QSAM01) title(DSN(USER.FILE.EXT),SHR);
v For an HFS file:
OPEN FILE(QSAM01) title(PATH(/u/USER/sample.txt));
Attempting to use files not associated with data sets
If you attempt to use a file that has not been associated with a data set, (either
through the use of the TITLE option of the OPEN statement or by establishing a
DD_DDNAME environment variable), the UNDEFINEDFILE condition is raised.
The only exceptions are the files SYSIN and SYSPRINT; these default to stdin and
stdout, respectively.
How PL/I finds data sets
PL/I establishes the path for creating new data sets or accessing existing data sets
in one of the following ways:
v The current directory.
v The paths as defined by the export DD_DDNAME environment variable.
Specifying characteristics using DD_DDNAME environment
variables
You use the export command to establish an environment variable that identifies
the data set to be associated with a PL/I file, and, optionally, provide additional
characteristics of that data set. This information provided by the environment
variable is called data definition (or DD) information.
The syntax of the DD_DDNAME environment variable is:
DD_DDNAME = " filespec "

, option
DSN ( dsname ) , dsn_option
PATH ( pathname )
Blanks are acceptable within the syntax. In addition, the syntax of the statement is
not checked at the time the command is entered. It is verified when the data set is
opened. If the syntax is wrong, UNDEFINEDFILE is raised with the oncode 96.
DD_DDNAME
The name of the environment variable. The DDNAME must be in uppercase
and can be either the name of a file constant or an alternate DDNAME that
you specify in the TITLE option of your OPEN statement. The TITLE option is
described in Using the TITLE option of the OPEN statement on page 260.
If you use an alternate DDNAME, and it is longer than 31 characters, only the
first 31 characters are used in forming the environment variable name.
filespec
The specifications of a file or the name of a device to be associated with the
PL/I file
option
The options that you can specify as DD information
dsname
The fully-qualified MVS data set name
pathname
The fully-qualified HFS path name
The options that you can specify as DD information are described in the following
topics, beginning with APPEND and ending with TYPE on page 266. Note
that these options do no apply to the DSN() or PATH() formats.
APPEND
The APPEND option specifies whether an existing data set is to be extended or
re-created.
APPEND
Y
( N )
Y Specifies that new records are to be added to the end of a sequential data set,
or inserted in a relative or indexed data set.
N Specifies that, if the file exists, it is to be recreated.
The APPEND option applies only to OUTPUT files. APPEND is ignored if:
v The file does not exist
v The file does not have the OUTPUT attribute
v The organization is REGIONAL(1)
BUFSIZE
The BUFSIZE option specifies the number of bytes for a buffer.
BUFSIZE (n)
RECORD output is buffered by default and has a default value for BUFSIZE of
64k. STREAM output is buffered, but not by default, and has a default value for
BUFSIZE of zero.
If the value of zero is given to BUFSIZE, the number of bytes for buffering is equal
to the value specified in the RECSIZE or LRECL option.
The BUFSIZE option is valid only for a consecutive binary file. If the file is used
for terminal input, you should assign the value of zero to BUFSIZE for increased
efficiency.
CHARSET for record I/O
This version of the CHARSET option applies only to consecutive files using record
I/O. It gives the user the capability of using ASCII data files as input files, and
specifying the character set of output files.
CHARSET
ASIS
( EBCDIC )
ASCII
Choose a suboption of CHARSET based on what form the file has (input) or what
form you want the file have (output).
CHARSET for stream I/O
This version of the CHARSET option applies for stream input and output files. It
gives the user the capability of using ASCII data files as input files, and specifying
the character set of output files. If you attempt to specify ASIS when using stream
I/O, no error is issued and character sets are treated as EBCDIC.
CHARSET
EBCDIC
( ASCII )
Choose a suboption of CHARSET based on what form the file has (input) or what
form you want the file to have (output).
DELAY
The DELAY option specifies the number of milliseconds to delay before retrying an
operation that fails when a file or record lock cannot be obtained by the system.
DELAY
0
( n )
This option is applicable only to VSAM files.
DELIMIT
The DELIMIT option specifies whether the input file contains field delimiters or
not. A field delimiter is a blank or a user-defined character that separates the fields
in a record. This is applicable for sort input files only.
DELIMIT
N
( Y )
The sort utility distinguishes text files from binary files with the presence of field
delimiters. Input files that contain field delimiters are processed as text files;
otherwise, they are considered to be binary files. The library needs this information
in order to pass the correct parameters to the sort utility.
LRECL
The LRECL option is the same as the RECSIZE option.
LRECL (n)
If LRECL is not specified and not implied by a LINESIZE value (except for
TYPE(FIXED) files, the default is 1024.
LRMSKIP
The LRMSKIP option allows output to commence on the nth (n refers to the value
specified with the SKIP option of the PUT or GET statement) line of the first page
for the first SKIP format item to be executed after a file is opened.
LRMSKIP
N
( Y )
If n is zero or 1, output commences on the first line of the first page.
PROMPT
The PROMPT option specifies whether or not colons should be visible as prompts
for stream input from the terminal.
PROMPT
N
( Y )
PUTPAGE
The PUTPAGE option specifies whether or not the form feed character should be
followed by a carriage return character. This option applies only to printer-destined
files. Printer-destined files are stream output files declared with the PRINT
attribute, or record output files declared with the CTLASA environment option.
PUTPAGE
NOCR
( CR )
NOCR
Indicates that the form feed character ('0C'x) is not followed by a carriage
return character ('0D'x).
CR Indicates that the carriage return character is appended to the form feed
character. This option should be specified if output is sent to non-IBM printers.
RECCOUNT
The RECCOUNT option specifies the maximum number of records that can be
loaded into a relative or regional data set that is created during the PL/I file
opening process.
RECCOUNT (n)
The RECCOUNT option is ignored if PL/I does not create, or recreate, the data set.
The default for the RECCOUNT option is 50.
Note: Under z/OS, it is recommended to omit the TITLE option with both the
/filespec parameter and RECCOUNT parameter for improved functionality and
performance of REGIONAL(1) data sets. In such a case, the number of records that
will be loaded into the file depends on the space allocated to the first extent of the
data set. See Chapter 11, Defining and using regional data sets, on page 321 for
additional information.
RECSIZE
The RECSIZE option specifies the length, n, of records in the data set.
RECSIZE
512
( n )
For regional and fixed-length data sets, RECSIZE specifies the length of each
record in the data set; for all other data set types, RECSIZE specifies the maximum
length records may have.
SAMELINE
The SAMELINE option specifies whether the system prompt occurs on the same
line as the statement that prompts for input.
SAMELINE
N
( Y )
The following examples show the results of certain combinations of the PROMPT
and SAMELINE options:
Example 1
Given the statement PUT SKIP LIST(ENTER:);, output is as follows:
prompt(y) sameline(y) ENTER: (cursor)
prompt(y) sameline(n) ENTER:
(cursor)
prompt(n) sameline(y) ENTER: (cursor)
prompt(n) sameline(n) ENTER:
(cursor)
Example 2
Given the statement PUT SKIP LIST(ENTER);, output is as follows:
prompt(y) sameline(y) ENTER: (cursor)
prompt(y) sameline(n) ENTER
:
(cursor)
prompt(n) sameline(y) ENTER (cursor)
prompt(n) sameline(n) ENTER
(cursor)
SKIP0
The SKIP0 option specifies where the line cursor moves when SKIP(0) statement is
coded in the source program. SKIP0 applies to terminal files that are not linked as
PM applications.
SKIP0
N
( Y )
SKIP0(N)
Specifies that the cursor is to be moved to the beginning of the next line.
SKIP0(Y)
Specifies that the cursor to be moved to the beginning of the current line.
The following example shows how you could make the output to the terminal skip
zero lines so that the cursor moves to the beginning of the current output line:
export DD_SYSPRINT=stdout:,SKIP0(Y)
TYPE
The TYPE option specifies the format of records in a native file.
TYPE
LF
( CRLF )
TEXT
FIXED
CRLFEOF
U
CRLF
Specifies that records are delimited by the CR - LF character combination. ('CR'
and 'LF' represent the ASCII values of carriage return and line feed, '0D'x and
'0A'x, respectively. For an output file, PL/I places the characters at the end of
each record; for an input file, PL/I discards the characters. For both input and
output, the characters are not counted in consideration for RECSIZE.
The data set must not contain any record that is longer than the value
determined for the record length of the data set.
LF Specifies that records are delimited by the LF character combination. ('LF'
represents the ASCII values of feed or '0A'x.) For an output file, PL/I places
the characters at the end of each record; for an input file, PL/I discards the
characters. For both input and output, the characters are not counted in
consideration for RECSIZE.
The data set must not contain any record that is longer than the value
determined for the record length of the data set.
TEXT
Equivalent to LF.
FIXED
Specifies that each record in the data set has the same length. The length
determined for records in the data set is used to recognize record boundaries.
All characters in a TYPE(FIXED) file are considered as data, including control
characters if they exist. Make sure the record length you specify reflects the
presence of these characters or make sure the record length you specify
accounts for all characters in the record.
CRLFEOF
Except for output files, this suboption specifies the same information as CRLF.
When one of these files is closed for output, an end-of-file marker is appended
to the last record.
U Indicates that records are unformatted. These unformatted files cannot be used
by any record or stream I/O statements except OPEN and CLOSE. You can
read from a TYPE(U) file only by using the FILEREAD built-in function. You
can write to a TYPE(U) file only by using the FILEWRITE built-in function.
The TYPE option applies only to CONSECUTIVE files, except that it is ignored for
printer-destined files with ASA(N) applied.
If your program attempts to access an existing data set with TYPE(FIXED) in effect
and the length of the data set is not a multiple of the logical record length you
specify, PL/I raises the UNDEFINEDFILE condition.
When using nonprint files with the TYPE(FIXED) attribute, SKIP is replaced by
trailing blanks to the end of the line. If TYPE(LF) is being used, SKIP is replaced
by LF with no trailing blanks.
Establishing data set characteristics
A data set consists of records stored in a particular format which the operating
system data management routines understand. When you declare or open a file in
your program, you are describing to PL/I and to the operating system the
characteristics of the records that file will contain. You can also use JCL or an
expression in the TITLE option of the OPEN statement to describe to the operating
system the characteristics of the data in data sets or in the PL/I files associated
with them.
You do not always need to describe your data both within the program and
outside it; often one description will serve for both data sets and their associated
PL/I files. There are, in fact, advantages to describing your data's characteristics in
only one place. These are described later in this chapter and in following chapters.
To effectively describe your program data and the data sets you will be using, you
need to understand something of how the operating system moves and stores data.
Blocks and records
The items of data in a data set are arranged in blocks separated by interblock gaps
(IBG). (Some manuals refer to these as interrecord gaps.)
A block is the unit of data transmitted to and from a data set. Each block contains
one record, part of a record, or several records. You can specify the block size in
the BLKSIZE parameter of the DD statement or in the BLKSIZE option of the
ENVIRONMENT attribute.
A record is the unit of data transmitted to and from a program. You can specify the
record length in the LRECL parameter of the DD statement, in the TITLE option of
the OPEN statement, or in the RECSIZE option of the ENVIRONMENT attribute.
When writing a PL/I program, you need consider only the records that you are
reading or writing; but when you describe the data sets that your program will
create or access, you must be aware of the relationship between blocks and records.
Blocking conserves storage space in a magnetic storage volume because it reduces
the number of interblock gaps, and it can increase efficiency by reducing the
number of input/output operations required to process a data set. Records are
blocked and deblocked by the data management routines.
Information interchange codes
The normal code in which data is recorded is the Extended Binary Coded Decimal
Interchange Code (EBCDIC).
Each character in the ASCII code is represented by a 7-bit pattern and there are 128
such patterns. The ASCII set includes a substitute character (the SUB control
character) that is used to represent EBCDIC characters having no valid ASCII code.
The ASCII substitute character is translated to the EBCDIC SUB character, which
has the bit pattern 00111111.
Record formats
The records in a data set have one of the following formats:
Fixed-length
Variable-length
Undefined-length.
Records can be blocked if required. The operating system will deblock fixed-length
and variable-length records, but you must provide code in your program to
deblock undefined-length records.
You specify the record format in the RECFM parameter of the DD statement, in the
TITLE option of the OPEN statement, or as an option of the ENVIRONMENT
attribute.
Fixed-length records
You can specify the following formats for fixed-length records:
F Fixed-length, unblocked
FB Fixed-length, blocked
FS Fixed-length, unblocked, standard
FBS Fixed-length, blocked, standard.
In a data set with fixed-length records, as shown in Figure 19 on page 269, all
records have the same length. If the records are blocked, each block usually
contains an equal number of fixed-length records (although a block can be
truncated). If the records are unblocked, each record constitutes a block.
Because it bases blocking and deblocking on a constant record length, the
operating system processes fixed-length records faster than variable-length records.
Variable-length records
You can specify the following formats for variable-length records:
V Variable-length, unblocked
VB Variable-length, blocked
VS Variable-length, unblocked, spanned
VBS Variable-length, blocked, spanned
V-format allows both variable-length records and variable-length blocks. A 4-byte
prefix of each record and the first 4 bytes of each block contain control information
for use by the operating system (including the length in bytes of the record or
block). Because of these control fields, variable-length records cannot be read
backward.
V-format signifies unblocked variable-length records. Each record is treated as a
block containing only one record. The first 4 bytes of the block contain block
control information, and the next 4 contain record control information.
VB-format signifies blocked variable-length records. Each block contains as many
complete records as it can accommodate. The first 4 bytes of the block contain
block control information, and a 4-byte prefix of each record contains record
control information.
Spanned Records: A spanned record is a variable-length record in which the length
of the record can exceed the size of a block. If this occurs, the record is divided
into segments and accommodated in two or more consecutive blocks by specifying
the record format as either VS or VBS. Segmentation and assembly are handled by
the operating system. The use of spanned records allows you to select a block size,
independently of record length, that will combine optimum use of auxiliary
storage with maximum efficiency of transmission.
VS-format is similar to V-format. Each block contains only one record or segment
of a record. The first 4 bytes of the block contain block control information, and the
next 4 contain record or segment control information (including an indication of
whether the record is complete or is a first, intermediate, or last segment).
Figure 19. Fixed-length records
VBS-format differs from VS-format in that each block contains as many complete
records or segments as it can accommodate; each block is, therefore, approximately
the same size (although there can be a variation of up to 4 bytes, since each
segment must contain at least 1 byte of data).
Undefined-length records
U-format allows the processing of records that do not conform to F- and V-formats.
The operating system and the compiler treat each block as a record; your program
must perform any required blocking or deblocking.
Data set organization
The data management routines of the operating system can handle a number of
types of data sets, which differ in the way data is stored within them and in the
allowed means of access to the data. The three main types of non-VSAM data sets
and the corresponding keywords describing their PL/I organization
1
are as follows:
Type of data set PL/I organization
Sequential CONSECUTIVE or ORGANIZATION(consecutive)
Indexed INDEXED or ORGANIZATION(indexed)
Direct REGIONAL or ORGANIZATION(relative)
A fourth type, partitioned, has no corresponding PL/I organization.
PL/I also provides support for three types of VSAM data organization: ESDS,
KSDS, and RRDS. For more information about VSAM data sets, see Chapter 12,
Defining and using VSAM data sets, on page 333.
In a sequential (or CONSECUTIVE) data set, records are placed in physical
sequence. Given one record, the location of the next record is determined by its
physical position in the data set. Sequential organization can be selected for direct
access devices.
An indexed sequential (or INDEXED) data set must reside on a direct access volume.
An index or set of indexes maintained by the operating system gives the location
of certain principal records. This allows direct retrieval, replacement, addition, and
deletion of records, as well as sequential processing.
A direct (or REGIONAL) data set must reside on a direct access volume. The data
set is divided into regions, each of which contains one or more records. A key that
specifies the region number allows direct access to any record; sequential
processing is also possible.
In a partitioned data set, independent groups of sequentially organized data, each
called a member, reside in a direct access data set. The data set includes a directory
that lists the location of each member. Partitioned data sets are often called libraries.
The compiler includes no special facilities for creating and accessing partitioned
data sets. Each member can be processed as a CONSECUTIVE data set by a PL/I
program. The use of partitioned data sets as libraries is described under Chapter 7,
Using libraries, on page 285.
1. Do not confuse the terms sequential and direct with the PL/I file attributes SEQUENTIAL and DIRECT. The attributes refer
to how the file is to be processed, and not to the way the corresponding data set is organized.
Labels
The operating system uses internal labels to identify direct access volumes and to
store data set attributes (for example, record length and block size). The attribute
information must originally come from a DD statement or from your program.
IBM standard labels have two parts: the initial volume label and header labels. The
initial volume label identifies a volume and its owner; the header labels precede
and follow each data set on the volume. Header labels contain system information,
device-dependent information (for example, recording technique), and dataset
characteristics.
direct access volumes have IBM standard labels. Each volume is identified by a
volume label, which is stored on the volume. This label contains a volume serial
number and the address of a volume table of contents (VTOC). The table of
contents, in turn, contains a label, termed a data set control block (DSCB), for each
data set stored on the volume.
Data Definition (DD) statement
A data definition (DD) statement is a job control statement that defines a data set
to the operating system, and is a request to the operating system for the allocation
of input/output resources. If the data sets are not dynamically allocated, each job
step must include a DD statement for each data set that is processed by the step.
Your z/OS JCL User's Guide describes the syntax of job control statements. The
operand field of the DD statement can contain keyword parameters that describe
the location of the data set (for example, volume serial number and identification
of the unit on which the volume will be mounted) and the attributes of the data
itself (for example, record format).
The DD statement enables you to write PL/I source programs that are independent
of the data sets and input/output devices they will use. You can modify the
parameters of a data set or process different data sets without recompiling your
program.
The following paragraphs describe the relationship of some operands of the DD
statement to your PL/I program.
The LEAVE and REREAD options of the ENVIRONMENT attribute allow you to
use the DISP parameter to control the action taken when the end of a
magnetic-tape volume is reached or when a magnetic-tape data set is closed. The
LEAVE and REREAD options are described under LEAVE|REREAD on page
314.
Write validity checking, which was standard in PL/I Version 1, is no longer
performed. Write validity checking can be requested through the OPTCD
subparameter of the DCB parameter of the JCL DD statement. See OS/VS2 Job
Control Language manual.
Use of the conditional subparameters
If you use the conditional subparameters of the DISP parameter for data sets
processed by PL/I programs, the step abend facility must be used. The step abend
facility is obtained as follows:
1. The ERROR condition should be raised or signaled whenever the program is to
terminate execution after a failure that requires the application of the
conditional subparameters.
2. The PL/I user exit must be changed to request an ABEND.
Data set characteristics
The DCB (data control block) parameter of the DD statement allows you to
describe the characteristics of the data in a data set, and the way it will be
processed, at run time. Whereas the other parameters of the DD statement deal
chiefly with the identity, location, and disposal of the data set, the DCB parameter
specifies information required for the processing of the records themselves. The
subparameters of the DCB parameter are described in your z/OS JCL User's Guide.
The DCB parameter contains subparameters that describe:
v The organization of the data set and how it will be accessed (CYLOFL, DSORG,
LIMCT, NTM, and OPTCD subparameters)
v Device-dependent information such as the line spacing for a printer (CODE,
FUNC, MODE, OPTCD=J, PRTSP, and STACK subparameters)
v The record format (BLKSIZE, KEYLEN, LRECL, and RECFM subparameters)
v The ASA control characters (if any) that will be inserted in the first byte of each
record (RECFM subparameter).
You can specify BLKSIZE, LRECL, KEYLEN, and RECFM (or their equivalents) in
the ENVIRONMENT attribute of a file declaration in your PL/I program instead of
in the DCB parameter.
You cannot use the DCB parameter to override information already established for
the data set in your PL/I program (by the file attributes declared and the other
attributes that are implied by them). DCB subparameters that attempt to change
information already supplied are ignored.
For a new dataset, the attributes of the file defined in the program will be used if
there is a conflict with the DD statement.
You may see message IEC225I with RC=4 issued when closing PDS files. This
message can be safely ignored.
An example of the DCB parameter is:
DCB=(RECFM=FB,BLKSIZE=400,LRECL=40)
which specifies that fixed-length records, 40 bytes in length, are to be grouped
together in a block 400 bytes long.
Using the TITLE option of the OPEN statement
You can use the TITLE option of the OPEN statement to identify the data set to be
associated with a PL/I file and, optionally, to provide additional characteristics of
the data set.
For more information about using the TITLE option of the OPEN statement, see
Using the TITLE option of the OPEN statement on page 260.
Associating PL/I files with data sets
Opening a file
The execution of a PL/I OPEN statement associates a file with a data set. This
requires merging of the information describing the file and the data set. If any
conflict is detected between file attributes and data set characteristics, the
UNDEFINEDFILE condition is raised.
Subroutines of the PL/I library create a skeleton data control block for the data set.
They use the file attributes from the DECLARE and OPEN statements and any
attributes implied by the declared attributes, to complete the data control block as
far as possible. (See Figure 20 on page 274.) They then issue an OPEN macro
instruction, which calls the data management routines to check that the correct
volume is mounted and to complete the data control block.
The data management routines examine the data control block to see what
information is still needed and then look for this information, first in the DD
statement, and finally, if the data set exists and has standard labels, in the data set
labels. For new data sets, the data management routines begin to create the labels
(if they are required) and to fill them with information from the data control block.
For INPUT data sets, the PL/I program can override the DCB attributes as long as
there is no conflict in attributes. For OUTPUT data sets, the PL/I program cannot
override the DCB attributes. However, if any DCB attributes are missing from the
data set when it is opened, they will be obtained from the PL/I program, if
provided.
When the DCB fields are filled in from these sources, control returns to the PL/I
library subroutines. If any fields still are not filled in, the PL/I OPEN subroutine
provides default information for some of them. For example, if LRECL is not
specified, it is provided from the value given for BLKSIZE.
Using system-determined block size: If you use the system-determined block
size function of the Data Facility Product on z/OS, DFP determines the optimal
block size for the device type that is assigned. When you create a new DASD data
set, the system derives the optimum block size and saves it in the data set label
when all of the following conditions are true:
v Block size is not available or specified from any source. BLKSIZE=0 can be
specified.
v You specify LRECL or it is in the data class. The data set does not have to be
SMS managed.
v You specify RECFM or it is in the data class. It must be fixed or variable.
v You specify DSORG as PS or PO or you omit DSORG and it is PS or PO in the
data class.
When system-determined block size is active, DFP determines the block size and
places it in the data set label before PL/I opens the file. Therefore, if the PL/I
program specifies a block size with its ENVIRONMENT option, it must not be in
conflict with the value from system-determined block size.
For detailed information about system-determined block size, see related topics in
z/OS DFSMS Using Data Sets.
Closing a file
The execution of a PL/I CLOSE statement dissociates a file from the data set with
which it was associated. The PL/I library subroutines first issue a CLOSE macro
instruction and, when control returns from the data management routines, release
the data control block that was created when the file was opened. The data
management routines complete the writing of labels for new data sets and update
the labels of existing data sets.
DCL MASTER FILE ENV(FB BLKSIZE(400),
RECSIZE(40));
OPEN FILE(MASTER);
//MASTER DD UNIT=2400
VOLUME=SER= 1791,
DSNAME=LIST,
DCB=( BUFNO=3,
RECFM=F,
BLKSIZE=400,
LRECL=100)
Record format=F
Record length=100
Blocking factor=4
Recording density=1600
Record format
Block size
Record length
Device type
Number of buffers
Recording density
DATA CONTROL BLOCK
FB
400
40
2400
3
1600
PL/I PROGRAM
DD STATEMENT
DATA SET LABEL
Note: Information from the PL/I program overrides that from the DD statement and the data set label.
Information from the DD statement overrides that from the data set label.
Figure 20. How the operating system completes the DCB
Specifying characteristics in the ENVIRONMENT attribute
You can use various options in the ENVIRONMENT attribute. Each type of file has
different attributes and environment options, which are listed below.
The ENVIRONMENT attribute
You use the ENVIRONMENT attribute of a PL/I file declaration file to specify
information about the physical organization of the data set associated with a file,
and other related information. The format of this information must be a
parenthesized option list.
ENVIRONMENT ( option-list )
Abbreviation: ENV
You can specify the options in any order, separated by blanks or commas.
The following example illustrates the syntax of the ENVIRONMENT attribute in
the context of a complete file declaration (the options specified are for VSAM and
are discussed in Chapter 12, Defining and using VSAM data sets, on page 333).
DCL FILENAME FILE RECORD SEQUENTIAL
INPUT ENV(VSAM GENKEY);
Table 14 summarizes the ENVIRONMENT options and file attributes. Certain
qualifications on their use are presented in the notes and comments for the figure.
Those options that apply to more than one data set organization are described in
the remainder of this chapter. In addition, in the following chapters, each option is
described with each data set organization to which it applies.
Table 14. Attributes of PL/I file declarations
Data set type
S
t
r
e
a
m
Record
Legend:
C Checked for VSAM
D Default
I Must be specified or implied
N Ignored for VSAM
O Optional
S Must be specified
- Invalid
File
Type
C
o
n
s
e
c
u
t
i
v
e
Sequential Direct
Consecutive
T
e
l
e
p
r
o
c
e
s
s
i
n
g
I
n
d
e
x
e
d
V
S
A
M
R
e
g
i
o
n
a
l
I
n
d
e
x
e
d
V
S
A
M
B
u
f
f
e
r
e
d
U
n
b
u
f
f
e
r
e
d
R
e
g
i
o
n
a
l
File attributes
1
Attributes implied
File I I I I I I I I I I
Input
1
D D D D D D D D D D File
Output O O O O O O O O O O File
Environment I I I S S S S S S S File
Stream D - - - - - - - - - File
Print
1
O - - - - - - - - - File stream output
Record - I I I I I I I I I File
Update - O O O - O O O O O File record
Sequential - D D D - D D - - D File record
Buffered - D - - I D D - - S File record
Keyed
2
- - - O I O O I I O File record
Direct - - - - - - S S S S File record keyed
ENVIRONMENT options Comments
F|FB|FS|FBS|V|
VB|VS|VBS||U
I S S - - - N - - N VS and VBS are invalid
with Stream
F|FB|U S S - - - - N - - N ASCII data sets only
F|V|U - - - S - - N S - N Only F for REGIONAL(1)
F|FB|V|VB - - - - - S N - S N
RECSIZE(n) I I I I S I C I I C
RECSIZE and/or BLKSIZE must
be specified for consecutive
BLKSIZE(n) I I I I - I N I I N indexed, and regional files
SCALARVARYING - O O O - O O O O O Invalid for ASCII data sets
CONSECUTIVE D D D - - - O - - O Allowed for VSAM ESDS
LEAVE|REREAD O O O - - - - - - -
CTLASA|CTL360 - O O - - - - - - - Invalid for ASCII data sets
GRAPHIC O - - - - - - - - -
INDEXED - - - - - S O - S O Allowed for VSAM ESDS
KEYLOC(n) - - - - - O - - O -
ORGANIZATION D - - - - - - - - -
GENKEY - - - - - O O - O O
INPUT or UPDATE files only;
KEYED is required
REGIONAL(1) - - - S - - - S - -
VSAM - - - - - - S - - S
BKWD - - - - - - O - - O
REUSE - - - - - - O - - O OUTPUT file only
Notes:
1. A file with the INPUT attribute cannot have the PRINT attribute.
2. Keyed is required for INDEXED and REGIONAL output.
Data set organization options: The options that specify data set organization are:
CONSECUTIVE
INDEXED
REGIONAL ( 1 )
VSAM
Each option is described in the discussion of the data set organization to which it
applies.
Other ENVIRONMENT options: You can use a constant or variable with those
ENVIRONMENT options that require integer arguments, such as block sizes and
record lengths. The variable must not be subscripted or qualified, and must have
attributes FIXED BINARY(31,0) and STATIC.
The list of equivalents for ENVIRONMENT options and DCB parameters are:
ENVIRONMENT option DCB subparameter
Record format RECFM
1
RECSIZE LRECL
BLKSIZE BLKSIZE
CTLASA|CTL360 RECFM
KEYLENGTH KEYLEN
Note:
1
VS must be specified as an ENVIRONMENT option, not in the DCB.
Record formats for record-oriented data transmission
Record formats supported depend on the data set organization.
F
FS
FB
FBS
V
VS
VB
VBS
U
Records can have one of the following formats:

Fixed-length F
FB
FS
FBS
unblocked
blocked
unblocked, standard
blocked, standard
Variable-length V
VB
VS
VBS
unblocked
blocked
spanned
blocked, spanned
Undefined-length U (cannot be blocked)
When U-format records are read into a varying-length string, PL/I sets the length
of the string to the block length of the retrieved data.
These record format options do not apply to VSAM data sets. If you specify a
record format option for a file associated with a VSAM data set, the option is
ignored.
You can only specify VS-format records for data sets with consecutive organization.
Record formats for stream-oriented data transmission
The record format options for stream-oriented data transmission are discussed in
Using stream-oriented data transmission on page 291.
RECSIZE option
The RECSIZE option specifies the record length.
RECSIZE ( record-length )
For files associated with VSAM data sets, record-length is the sum of:
1. The length required for data. For variable-length and undefined-length records,
this is the maximum length.
2. Any control bytes required. Variable-length records require 4 (for the
record-length prefix); fixed-length and undefined-length records do not require
any.
For VSAM data sets, the maximum and average lengths of the records are
specified to the Access Method Services utility when the data set is defined. If you
include the RECSIZE option in the file declaration for checking purposes, you
should specify the maximum record size. If you specify RECSIZE and it conflicts
with the values defined for the data set, the UNDEFINEDFILE condition is raised.
You can specify record-length as an integer or as a variable with attributes FIXED
BINARY(31,0) STATIC.
The value is subject to the following conventions:
Maximum:
Fixed-length, and undefined (except ASCII data sets): 32760
V-format, and VS- and VBS-format with UPDATE files: 32756
VS- and VBS-format with INPUT and OUTPUT files: 16777215
ASCII data sets: 9999
VSAM data sets: 32761
Note: For VS- and VBS-format records longer than 32,756 bytes, you must
specify the length in the RECSIZE option of ENVIRONMENT, and for the
DCB subparameter of the DD statement you must specify LRECL=X. If
RECSIZE exceeds the allowed maximum for INPUT or OUTPUT, either a
record condition occurs or the record is truncated. UPDATE files are not
supported with LRECL=X.
Zero value:
A search for a valid value is made first:
v In the DD statement for the data set associated with the file, and second
v In the data set label.
If neither of these provides a value, default action is taken (see Record
format, BLKSIZE, and RECSIZE defaults on page 279).
Negative Value:
The UNDEFINEDFILE condition is raised.
BLKSIZE option
The BLKSIZE option specifies the maximum block size on the data set.
BLKSIZE ( block-size )
block-size is the sum of:
1. The total length(s) of one of the following:
v A single record
v A single record and either one or two record segments
v Several records
v Several records and either one or two record segments
v Two record segments
v A single record segment.
For variable-length records, the length of each record or record segment
includes the 4 control bytes for the record or segment length.
The preceding list summarizes all the possible combinations of records and
record segments options: fixed- or variable-length blocked or unblocked,
spanned or unspanned. When specifying a blocksize for spanned records, note
that each record and each record segment requires 4 control bytes for the record
length and that these quantities are in addition to the 4 control bytes required
for each block.
2. Any further control bytes required.
v Variable-length blocked records require 4 (for the block size).
v Fixed-length and undefined-length records do not require any further control
bytes.
3. Any block prefix bytes required (ASCII data sets only).
block-size can be specified as an integer, or as a variable with attributes FIXED
BINARY(31,0) STATIC.
The value is subject to the following conventions:
Maximum:
32760
Zero value:
If you set BLKSIZE to 0, under z/OS the Data Facility Product sets the
block size. For an elaboration of this topic, see Record format, BLKSIZE,
and RECSIZE defaults.
Negative value:
The UNDEFINEDFILE condition is raised.
The relationship of block size to record length depends on the record format:
FB-format or FBS-format
The block size must be a multiple of the record length.
VB-format:
The block size must be equal to or greater than the sum of:
1. The maximum length of any record
2. Four control bytes.
VS-format or VBS-format:
The block size can be less than, equal to, or greater than the record length.
Notes:
v Use the BLKSIZE option with unblocked (F- or V-format) records in either of the
following ways:
Specify the BLKSIZE option, but not the RECSIZE option. Set the record
length equal to the block size (minus any control or prefix bytes), and leave
the record format unchanged.
Specify both BLKSIZE and RECSIZE and ensure that the relationship of the
two values is compatible with blocking for the record format you use. Set the
record format to FB or VB, whichever is appropriate.
v If for FB-format or FBS-format records the block size equals the record length,
the record format is set to F.
v The BLKSIZE option does not apply to VSAM data sets, and is ignored if you
specify it for one.
Record format, BLKSIZE, and RECSIZE defaults
If you do not specify either the record format, block size, or record length for a
non-VSAM data set, the following default action is taken:
Record format:
A search is made in the associated DD statement or data set label. If the
search does not provide a value, the UNDEFINEDFILE condition is raised,
except for files associated with dummy data sets or the foreground
terminal, in which case the record format is set to U.
Block size or record length:
If one of these is specified, a search is made for the other in the associated
DD statement or data set label. If the search provides a value, and if this
value is incompatible with the value in the specified option, the
UNDEFINEDFILE condition is raised. If the search is unsuccessful, a value
is derived from the specified option (with the addition or subtraction of
any control or prefix bytes).
If neither is specified, the UNDEFINEDFILE condition is raised, except for
files associated with dummy data sets, in which case BLKSIZE is set to 121
for F-format or U-format records and to 129 for V-format records. For files
associated with the foreground terminal, RECSIZE is set to 120.
If you are using z/OS with the Data Facility Product system-determined
block size, DFP determines the optimum block size for the device type
assigned. If you specify BLKSIZE(0) in either the DD assignment or the
ENVIRONMENT statement, DFP calculates BLKSIZE using the record
length, record format, and device type.
GENKEY option key classification
The GENKEY (generic key) option applies only to INDEXED and VSAM
key-sequenced data sets. It enables you to classify keys recorded in a data set and
to use a SEQUENTIAL KEYED INPUT or SEQUENTIAL KEYED UPDATE file to
access records according to their key classes.
GENKEY
A generic key is a character string that identifies a class of keys; all keys that begin
with the string are members of that class. For example, the recorded keys ABCD,
ABCE, and ABDF are all members of the classes identified by the generic keys
A and AB, and the first two are also members of the class ABC; and the
three recorded keys can be considered to be unique members of the classes
ABCD, ABCE, and ABDF, respectively.
The GENKEY option allows you to start sequential reading or updating of a
VSAM data set from the first record that has a key in a particular class, and for an
INDEXED data set from the first nondummy record that has a key in a particular
class. You identify the class by including its generic key in the KEY option of a
READ statement. Subsequent records can be read by READ statements without the
KEY option. No indication is given when the end of a key class is reached.
Although you can retrieve the first record having a key in a particular class by
using a READ with the KEY option, you cannot obtain the actual key unless the
records have embedded keys, since the KEYTO option cannot be used in the same
statement as the KEY option.
In the following example, a key length of more than 3 bytes is assumed:
DCL IND FILE RECORD SEQUENTIAL KEYED
UPDATE ENV (GENKEY);
.
.
.
READ FILE(IND) INTO(INFIELD)
KEY (ABC);
.
.
.
NEXT: READ FILE (IND) INTO (INFIELD);
.
.
.
GO TO NEXT;
The first READ statement causes the first nondummy record in the data set whose
key begins with ABC to be read into INFIELD; each time the second READ
statement is executed, the nondummy record with the next higher key is retrieved.
Repeated execution of the second READ statement could result in reading records
from higher key classes, since no indication is given when the end of a key class is
reached. It is your responsibility to check each key if you do not wish to read
beyond the key class. Any subsequent execution of the first READ statement
would reposition the file to the first record of the key class ABC.
If the data set contains no records with keys in the specified class, or if all the
records with keys in the specified class are dummy records, the KEY condition is
raised. The data set is then positioned either at the next record that has a higher
key or at the end of the file.
The presence or absence of the GENKEY option affects the execution of a READ
statement which supplies a source key that is shorter than the key length specified
in the KEYLEN subparameter. This KEYLEN subparameter is found in the DD
statement that defines the indexed data set. If you specify the GENKEY option, it
causes the source key to be interpreted as a generic key, and the data set is
positioned to the first nondummy record in the data set whose key begins with the
source key. If you do not specify the GENKEY option, a READ statement's short
source key is padded on the right with blanks to the specified key length, and the
data set is positioned to the record that has this padded key (if such a record
exists). For a WRITE statement, a short source key is always padded with blanks.
Use of the GENKEY option does not affect the result of supplying a source key
whose length is greater than or equal to the specified key length. The source key,
truncated on the right if necessary, identifies a specific record (whose key can be
considered to be the only member of its class).
SCALARVARYING option varying-length strings
You use the SCALARVARYING option in the input/output of varying-length
strings; you can use it with records of any format.
SCALARVARYING
When storage is allocated for a varying-length string, the compiler includes a
2-byte prefix that specifies the current length of the string. For an element
varying-length string, this prefix is included on output, or recognized on input,
only if SCALARVARYING is specified for the file.
When you use locate mode statements (LOCATE and READ SET) to create and
read a data set with element varying-length strings, you must specify
SCALARVARYING to indicate that a length prefix is present, since the pointer that
locates the buffer is always assumed to point to the start of the length prefix.
When you specify SCALARVARYING and element varying-length strings are
transmitted, you must allow two bytes in the record length to include the length
prefix.
A data set created using SCALARVARYING should be accessed only by a file that
also specifies SCALARVARYING.
You must not specify SCALARVARYING and CTLASA/CTL360 for the same file,
as this causes the first data byte to be ambiguous.
KEYLENGTH option
Use the KEYLENGTH option to specify the length of the recorded key for KEYED
files where n is the length. You can specify KEYLENGTH for INDEXED files.
KEYLENGTH ( n )
If you include the KEYLENGTH option in a VSAM file declaration for checking
purposes, and the key length you specify in the option conflicts with the value
defined for the data set, the UNDEFINEDFILE condition is raised.
ORGANIZATION option
The ORGANIZATION option specifies the organization of the data set associated
with the PL/I file.
ORGANIZATION
CONSECUTIVE
( INDEXED )
RELATIVE
CONSECUTIVE
Specifies that the files is associated with a consecutive data set. A consecutive
file can be either a native data set or a VSAM, ESDS, RRDS, or KSDS data set.
RELATIVE
Specifies that the file is associated with a relative data set. RELATIVE specifies
that the data set contains records that do not have recorded keys. A relative file
is a VSAM direct data set. Relative keys range from 1 to nnnn.
Data set types used by PL/I record I/O
Data sets with the RECORD attribute are processed by record-oriented data
transmission in which data is transmitted to and from auxiliary storage exactly as
it appears in the program variables; no data conversion takes place. A record in a
data set corresponds to a variable in the program.
Table 15 shows the facilities that are available with the various types of data sets
that can be used with PL/I Record I/O.
Table 15. A comparison of data set types available to PL/I record I/O
VSAM
KSDS
VSAM
ESDS
VSAM
RRDS INDEXED CONSECUTIVE
REGIONAL
(1)
SEQUENCE Key
order
Entry
order
Num-
bered
Key
order
Entry
order
By
region
DEVICES DASD DASD DASD DASD DASD,
card, etc.
DASD
ACCESS
1 By key
2 Sequential
3 Backward
123 123 123 12 2 12
Table 15. A comparison of data set types available to PL/I record I/O (continued)
VSAM
KSDS
VSAM
ESDS
VSAM
RRDS INDEXED CONSECUTIVE
REGIONAL
(1)
Alternate
index
access
as above
123 123 No No No No
How
extended
With
new
keys
At
end
In
empty
slots
With
new
keys
At
end
In
empty
slots
DELETION
1 Space
reusable
2 Space not
reusable
Yes, 1 No Yes, 1 Yes, 2 No Yes, 1
The following chapters describe how to use Record I/O data sets for different
types of data sets:
v Chapter 8, Defining and using consecutive data sets, on page 291
v Chapter 11, Defining and using regional data sets, on page 321
v Chapter 12, Defining and using VSAM data sets, on page 333
Setting environment variables
z/OS UNIX System Services Only
There are a number of environment variables that can be set and exported for use
with z/OS UNIX.
To set the environment variables system wide so all users have access to them, add
the lines suggested in the subsections to the file /etc/profile. To set them for a
specific user only, add them to the file .profile in the user's home directory. The
variables are set the next time the user logs on.
The following example illustrates how to set environment variables:
LANG=ja_JP
NLSPATH=/usr/lib/nls/msg/%L/%N:/usr/lib/nls/msg/prime/%N
LIBPATH=/home/joe/usr/lib:/home/joe/mylib:/usr/lib
export LANG NLSPATH LIBPATH
Rather than using the last statement in the previous example, you could have
added export to each of the preceding lines (export LANG=ja_JP...).
You can use the ECHO command to determine the current setting of an
environment variable. To define the value of BYPASS, you can use either of the
following two examples:
echo $LANG
echo $LIBPATH
End of z/OS UNIX System Services Only
PL/I standard files (SYSPRINT and SYSIN)
SYSIN is read from stdin and SYSPRINT is directed to stdout by default. If you
want either to be associated differently, you must use the TITLE option of the
OPEN statement, or establish a DD_DDNAME environment variable naming a
data set or another device. Environment variables are discussed above in Setting
environment variables on page 283.
Redirecting standard input, output, and error devices
You can also redirect standard input, standard output, and standard error devices
to a file. You can use redirection in the following program:
Hello2: proc options(main);
put list('Hello!');
end;
After compiling and linking the program, you can invoke it from the command
line by entering:
hello2 > hello2.out
If you want to combine stdout and stderr in a single file, enter the following
command:
hello2 > hello2.out 2>&1
As is true with display statements, the greater than sign redirects the output to the
file that is specified after it, in this case hello2.out. This means that the word
'Hello' is written in the file hello2.out. Note also that the output includes printer
control characters since the PRINT attribute is applied to SYSPRINT by default.
READ statements can access data from stdin; however, the record into which the
data is to be put must have an LRECL equal to 1.
Chapter 7. Using libraries
Within the z/OS operating system, the terms partitioned data set, partitioned
data set/extension, and library are synonymous and refer to a type of data set
that can be used for the storage of other data sets (usually programs in the form of
source, object or load modules). A library must be stored on direct access storage
and be wholly contained in one volume. It contains independent, consecutively
organized data sets, called members. Each member has a unique name, not more
than 8 characters long, which is stored in a directory that is part of the library. All
the members of one library must have the same data characteristics because only
one data set label is maintained.
You can create members individually until there is insufficient space left for a new
entry in the directory, or until there is insufficient space for the member itself. You
can access members individually by specifying the member name.
Use DD statements or their conversational mode equivalent to create and access
members.
You can delete members by means of the IBM utility program IEHPROGM. This
deletes the member name from the directory so that the member can no longer be
accessed, but you cannot use the space occupied by the member itself again unless
you recreate the library or compress the unused space using, for example, the IBM
utility program IEBCOPY. If you attempt to delete a member by using the DISP
parameter of a DD statement, it causes the whole data set to be deleted.
Types of libraries
You can use the following types of libraries with a PL/I program:
v The system program library SYS1.LINKLIB or its equivalent. This can contain all
system processing programs such as compilers and the linkage editor.
v Private program libraries. These usually contain user-written programs. It is
often convenient to create a temporary private library to store the load module
output from the linkage editor until it is executed by a later job step in the same
job. The temporary library will be deleted at the end of the job. Private libraries
are also used for automatic library call by the linkage editor and the loader.
v The system procedure library SYS1.PROCLIB or its equivalent. This contains the
job control procedures that have been cataloged for your installation.
How to use a library
A PL/I program can use a library directly. If you are adding a new member to a
library, its directory entry will be made by the operating system when the
associated file is closed, using the member name specified as part of the data set
name.
If you are accessing a member of a library, its directory entry can be found by the
operating system from the member name that you specify as part of the data set
name.
More than one member of the same library can be processed by the same PL/I
program, but only one such file can be open as output at any one time. You access
different members by giving the member name in a DD statement.
Creating a library
To create a library include in your job step a DD statement containing the
information given in Table 16. The information required is similar to that for a
consecutively organized data set (see Defining files using record I/O on page
312) except for the SPACE parameter.
Table 16. Information required when creating a library
Information Required Parameter of DD statement
Type of device that will be used UNIT=
Serial number of the volume that will contain the
library
VOLUME=SER
Name of the library DSNAME=
Amount of space required for the library SPACE=
Disposition of the library DISP=
SPACE parameter
The SPACE parameter in a DD statement that defines a library must always be of
the form:
SPACE=(units,(quantity,increment,directory))
Although you can omit the third term (increment), indicating its absence by a
comma, the last term, specifying the number of directory blocks to be allocated,
must always be present.
The amount of auxiliary storage required for a library depends on the number and
sizes of the members to be stored in it and on how often members will be added
or replaced. (Space occupied by deleted members is not released.) The number of
directory blocks required depends on the number of members and the number of
aliases. You can specify an incremental quantity in the SPACE parameter that
allows the operating system to obtain more space for the data set, if such is
necessary at the time of creation or at the time a new member is added; the
number of directory blocks, however, is fixed at the time of creation and cannot be
increased.
For example, the DD statement:
// PDS DD UNIT=SYSDA,VOL=SER=3412,
// DSNAME=ALIB,
// SPACE=(CYL,(5,,10)),
// DISP=(,CATLG)
requests the job scheduler to allocate 5 cylinders of the DASD with a volume serial
number 3412 for a new library name ALIB, and to enter this name in the system
catalog. The last term of the SPACE parameter requests that part of the space
allocated to the data set be reserved for ten directory blocks.
Creating and updating a library member
The members of a library must have identical characteristics. Otherwise, you might
later have difficulty retrieving them. Identical characteristics are necessary because
the volume table of contents (VTOC) will contain only one data set control block
(DSCB) for the library and not one for each member. When using a PL/I program
to create a member, the operating system creates the directory entry; you cannot
place information in the user data field.
When creating a library and a member at the same time, your DD statement must
include all the parameters listed under Creating a library on page 286 (although
you can omit the DISP parameter if the data set is to be temporary). The DSNAME
parameter must include the member name in parentheses. For example,
DSNAME=ALIB(MEM1) names the member MEM1 in the data set ALIB. If the
member is placed in the library by the linkage editor, you can use the linkage
editor NAME statement or the NAME compile-time option instead of including the
member name in the DSNAME parameter. You must also describe the
characteristics of the member (record format, etc.) either in the DCB parameter or
in your PL/I program. These characteristics will also apply to other members
added to the data set.
When creating a member to be added to an existing library, you do not need the
SPACE parameter. The original space allocation applies to the whole of the library
and not to an individual member. Furthermore, you do not need to describe the
characteristics of the member, since these are already recorded in the DSCB for the
library.
To add two more members to a library in one job step, you must include a DD
statement for each member, and you must close one file that refers to the library
before you open another.
Examples
The use of the cataloged procedure IBMZC to compile a simple PL/I program and
place the object module in a new library named EXLIB is shown in Figure 21 on
page 288. The DD statement that defines the new library and names the object
module overrides the DD statement SYSLIN in the cataloged procedure. (The PL/I
program is a function procedure that, given two values in the form of the character
string produced by the TIME built-in function, returns the difference in
milliseconds.)
The use of the cataloged procedure IBMZCL to compile and link-edit a PL/I
program and place the load module in the existing library HPU8.CCLM is shown
in Figure 22 on page 288.
Chapter 7. Using libraries 287
To use a PL/I program to add or delete one or more records within a member of a
library, you must rewrite the entire member in another part of the library. This is
rarely an economic proposition, since the space originally occupied by the member
cannot be used again. You must use two files in your PL/I program, but both can
be associated with the same DD statement. The program shown in Figure 24 on
page 289 updates the member created by the program in Figure 23 on page 289. It
copies all the records of the original member except those that contain only blanks.
//OPT10#1 JOB
//TR EXEC IBMZC
//PLI.SYSLIN DD UNIT=SYSDA,DSNAME=HPU8.EXLIB(ELAPSE),
// SPACE=(TRK,(1,,1)),DISP=(NEW,CATLG)
//PLI.SYSIN DD *
ELAPSE: PROC(TIME1,TIME2);
DCL (TIME1,TIME2) CHAR(9),
H1 PIC '99' DEF TIME1,
M1 PIC '99' DEF TIME1 POS(3),
MS1 PIC '99999' DEF TIME1 POS(5),
H2 PIC '99' DEF TIME2,
M2 PIC '99' DEF TIME2 POS(3),
MS2 PIC '99999' DEF TIME2 POS(5),
ETIME FIXED DEC(7);
IF H2<H1 THEN H2=H2+24;
ETIME=((H2*60+M2)*60000+MS2)-((H1*60+M1)*60000+MS1);
RETURN(ETIME);
END ELAPSE;
/*
Figure 21. Creating new libraries for compiled object modules
//OPT10#2 JOB
//TRLE EXEC IBMZCL
//PLI.SYSIN DD *
MNAME: PROC OPTIONS(MAIN);
.
.
.
program
.
.
.
END MNAME;
/*
//LKED.SYSLMOD DD DSNAME=HPU8.CCLM(DIRLIST),DISP=OLD
Figure 22. Placing a load module in an existing library
Extracting information from a library directory
The directory of a library is a series of records (entries) at the beginning of the data
set. There is at least one directory entry for each member. Each entry contains a
member name, the relative address of the member within the library, and a
variable amount of user data.
//OPT10#3 JOB
//TREX EXEC IBMZCBG
//PLI.SYSIN DD *
NMEM: PROC OPTIONS(MAIN);
DCL IN FILE RECORD SEQUENTIAL INPUT,
OUT FILE RECORD SEQUENTIAL OUTPUT,
P POINTER,
IOFIELD CHAR(80) BASED(P),
EOF BIT(1) INIT('0'B);
OPEN FILE(IN),FILE (OUT);
ON ENDFILE(IN) EOF='1'B;
READ FILE(IN) SET(P);
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (IOFIELD) (A);
WRITE FILE(OUT) FROM(IOFIELD);
READ FILE(IN) SET(P);
END;
CLOSE FILE(IN),FILE(OUT);
END NMEM;
/*
//GO.OUT DD UNIT=SYSDA,DSNAME=HPU8.ALIB(NMEM),
// DISP=(NEW,CATLG),SPACE=(TRK,(1,1,1)),
// DCB=(RECFM=FB,BLKSIZE=3600,LRECL=80)
//GO.IN DD *
MEM: PROC OPTIONS(MAIN);
/* this is an incomplete dummy library member */
Figure 23. Creating a library member in a PL/I program
//OPT10#4 JOB
//TREX EXEC IBMZCBG
//PLI.SYSIN DD *
UPDTM: PROC OPTIONS(MAIN);
DCL (OLD,NEW) FILE RECORD SEQUENTIAL,
EOF BIT(1) INIT('0'B),
DATA CHAR(80);
ON ENDFILE(OLD) EOF = '1'B;
OPEN FILE(OLD) INPUT,FILE(NEW) OUTPUT TITLE('OLD');
READ FILE(OLD) INTO(DATA);
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (DATA) (A);
IF DATA=' ' THEN ;
ELSE WRITE FILE(NEW) FROM(DATA);
READ FILE(OLD) INTO(DATA);
END;
CLOSE FILE(OLD),FILE(NEW);
END UPDTM;
/*
//GO.OLD DD DSNAME=HPU8.ALIB(NMEM),DISP=(OLD,KEEP)
Figure 24. Updating a library member
Chapter 7. Using libraries 289
User data is information inserted by the program that created the member. An
entry that refers to a member (load module) written by the linkage editor includes
user data in a standard format, described in the systems manuals.
If you use a PL/I program to create a member, the operating system creates the
directory entry for you and you cannot write any user data. However, you can use
assembler language macro instructions to create a member and write your own
user data. The method for using macro instructions to do this is described in the
data management manuals.
Chapter 8. Defining and using consecutive data sets
This chapter covers consecutive data set organization and the ENVIRONMENT
options that define consecutive data sets for stream and record-oriented data
transmission. It then covers how to create, access, and update consecutive data sets
for each type of transmission.
In a data set with consecutive organization, records are organized solely on the
basis of their successive physical positions; when the data set is created, records
are written consecutively in the order in which they are presented. You can retrieve
the records only in the order in which they were written. See Table 14 on page 275
for valid file attributes and ENVIRONMENT options for consecutive data sets.
Using stream-oriented data transmission
This section covers how to define data sets for use with PL/I files that have the
STREAM attribute. It covers the ENVIRONMENT options you can use and how to
create and access data sets. The essential parameters of the DD statements you use
in creating and accessing these data sets are summarized in tables, and several
examples of PL/I programs are included to illustrate the text.
Data sets with the STREAM attribute are processed by stream-oriented data
transmission, which allows your PL/I program to ignore block and record
boundaries and treat a data set as a continuous stream of data values in character
or graphic form.
You create and access data sets for stream-oriented data transmission using the
list-, data-, and edit-directed input and output statements described in the PL/I
Language Reference.
For output, PL/I converts the data items from program variables into character
form if necessary, and builds the stream of characters or graphics into records for
transmission to the data set.
For input, PL/I takes records from the data set and separates them into the data
items requested by your program, converting them into the appropriate form for
assignment to program variables.
You can use stream-oriented data transmission to read or write graphic data. There
are terminals, printers, and data-entry devices that, with the appropriate
programming support, can display, print, and enter graphics. You must be sure
that your data is in a format acceptable for the intended device, or for a print
utility program.
Defining files using stream I/O
You define files for stream-oriented data transmission by a file declaration with the
following attributes:
DCL filename FILE STREAM
INPUT | {OUTPUT [PRINT]}
ENVIRONMENT(options);
Default file attributes are shown in Table 14 on page 275; the FILE attribute is
described in the PL/I Language Reference. The PRINT attribute is described further
in Using PRINT files with stream I/O on page 299. Options of the
ENVIRONMENT attribute are discussed below.
Defining stream files using PL/I dynamic allocation
To define the stream files, you can use a DD statement, an environment variable,
or the TITLE option of the OPEN statement.
When an environment variable or the TITLE option is used, the name must be in
uppercase. Specify the MVS data set in one of the following ways:
*DSN(data-set-name)
*DSN(data-set-name(member-name))
data-set-name must be fully qualified and cannot be a temporary data set; for
example, it must not start with &.
Specify the HFS file as follows:
PATH(absolute-path-name)
You can specify the following attributes in any order after the DSN keyword:
NEW, OLD, SHR, or MOD
TRACKS or CYL
SPACE(n,m)
VOL(volser)
UNIT(type)
KEEP, DELETE, CATALOG, or UNCATALOG
STORCLAS(storageclass)
MGMTCLAS(managementclass)
DATACLAS(dataclass)
Note: You cannot create a PDS or PDSE using an environment variable or the
TITLE option of the OPEN statement, but you can create a new member in
an existing PDS or PDSE.
Specifying ENVIRONMENT options
Table 14 on page 275 summarizes the ENVIRONMENT options. The options
applicable to stream-oriented data transmission are:
CONSECUTIVE or ORGANIZATION(CONSECUTIVE)
F|FB|FS|FBS|V|VB|VS|VBS|U
RECSIZE(record-length)
BLKSIZE(block-size)
GRAPHIC
LEAVE|REREAD
BLKSIZE is described in Chapter 6, Using data sets and files, beginning on page
278. LEAVE and REREAD are described later in this chapter, beginning at
LEAVE|REREAD on page 314. Descriptions of the rest of these options follow
immediately below.
CONSECUTIVE
STREAM files must have CONSECUTIVE data set organization; however, it is not
necessary to specify this in the ENVIRONMENT options since CONSECUTIVE is
the default data set organization. The CONSECUTIVE option for STREAM files is
the same as that described in Data set organization on page 270.
CONSECUTIVE
Record format options
Although record boundaries are ignored in stream-oriented data transmission,
record format is important when creating a data set. This is not only because
record format affects the amount of storage space occupied by the data set and the
efficiency of the program that processes the data, but also because the data set can
later be processed by record-oriented data transmission.
Having specified the record format, you need not concern yourself with records
and blocks as long as you use stream-oriented data transmission. You can consider
your data set a series of characters or graphics arranged in lines, and you can use
the SKIP option or format item (and, for a PRINT file, the PAGE and LINE options
and format items) to select a new line.
F
FS
FB
FBS
V
VS
VB
VBS
U
Records can have one of the following formats, which are described in Record
formats on page 268.
Fixed-length F
FB
FS
FBS
unblocked
blocked
unblocked, standard
blocked, standard
Variable-length V
VB
VS
VBS
unblocked
blocked
Undefined-length U (cannot be blocked)
Blocking and deblocking of records are performed automatically.
RECSIZE
RECSIZE for stream-oriented data transmission is the same as that described in
Specifying characteristics in the ENVIRONMENT attribute on page 275.
Additionally, a value specified by the LINESIZE option of the OPEN statement
overrides a value specified in the RECSIZE option. LINESIZE is discussed in the
PL/I Language Reference.
Additional record-size considerations for list- and data-directed transmission of
graphics are given in the PL/I Language Reference.
Defaults for record format, BLKSIZE, and RECSIZE
If you do not specify the record format, BLKSIZE, or RECSIZE option in the
ENVIRONMENT attribute, or in the associated DD statement or data set label, the
following action is taken:
Chapter 8. Defining and using consecutive data sets 293
Input files:
Defaults are applied as for record-oriented data transmission, described in
Record format, BLKSIZE, and RECSIZE defaults on page 279.
Output files:
Record format
Set to VB-format
Record length
The specified or default LINESIZE value is used:
v PRINT files:
F, FB, FBS, or U: line size + 1
V or VB: line size + 5
v Non-PRINT files:
F, FB, FBS, or U: linesize
V or VB: linesize + 4
Block Size
Files associated with SYSOUT:
v F, FB, or FBS: record length
v V or VB: record length + 4
New or Temporary data set:
v Optimum block size determined by DFP
GRAPHIC option
Specify the GRAPHIC option for edit-directed I/O.
GRAPHIC
The ERROR condition is raised for list- and data-directed I/O if you have graphics
in input or output data and do not specify the GRAPHIC option.
For edit-directed I/O, the GRAPHIC option specifies that left and right delimiters
are added to DBCS variables and constants on output, and that input graphics will
have left and right delimiters. If you do not specify the GRAPHIC option, left and
right delimiters are not added to output data, and input graphics do not require
left and right delimiters. When you do specify the GRAPHIC option, the ERROR
condition is raised if left and right delimiters are missing from the input data.
For information on the graphic data type, and on the G-format item for
edit-directed I/O, see the PL/I Language Reference.
Creating a data set with stream I/O
To create a data set, you must give the operating system certain information either
in your PL/I program or in the DD statement that defines the data set. For z/OS
UNIX, use one of the following to provide the additional information:
v TITLE option of the OPEN statement
v DD_DDNAME environment variable
v ENVIRONMENT attribute
The following paragraphs indicate the essential information, and discuss some of
the optional information you can supply.
Essential information
When your application creates a STREAM file, PL/I will derive a line-size value
for that file from one of the following sources in order of declining precedence.
v LINESIZE option of the OPEN statement
v RECSIZE option of the ENVIRONMENT attribute
v RECSIZE option of the TITLE option of the OPEN statement
v RECSIZE option of the DD_DDNAME environment variable
v PL/I-supplied default value
If a LINESIZE value is supplied, but a RECSIZE valus is not, then PL/I derives the
record-length value as follows:
v For a V-format PRINT file, the value is LINESIZE + 5
v For a V-format non-PRINT file, the value is LINESIZE + 4
v For a F-format PRINT file, the value is LINESIZE + 1
v In all other cases, the value is LINESIZE
If a LINESIZE value is not supplied, but a RECSIZE value is, then PL/I derives the
line-size value from RECSIZE as follows:
v For a V-format PRINT file, the value is RECSIZE - 5
v For a V-format non-PRINT file, the value is RECSIZE - 4
v For a F-format PRINT file, the value is RECSIZE - 1
v In all other cases, the value is RECSIZE
If neither LINESIZE nor RECSIZE is supplied, then PL/I determines a default
line-size value based on the attributes of the file and the type of associated data
set. In cases where PL/I cannot supply an appropriate default line size, the
UNDEFINEDFILE condition is raised.
A default line-size value is supplied for an OUTPUT file when:
v The file has the PRINT attribute. In this case, the value is obtained from the tab
control table.
v The associated data set is the terminal (stdout: or stderr:). In this case the value
is 120.
Note that if the LINESIZE option is specified (on the OPEN statement) and
RECSIZE is also specified (in the ENVIRONMENT attribute, the TITLE option or
the DD statement), if the record size value is too small to hold the LINESIZE
(taking into account the record format and appropriate control byte overhead),
then
v For users who have LE for z/OS 1.9 or earlier releases
For DD SYSOUT= files, the LINESIZE option will be used to determine a new
record size that matches the given LINESIZE. For DD DSN= files and all other
files, the UNDEFINEDFILE condition will be raised.
v For users with LE for z/OS releases subsequent to 1.9
The UNDEFINEDFILE condition will be raised for all files
Examples
The use of edit-directed stream-oriented data transmission to create a data set on a
direct access storage device is shown in Figure 25 on page 296. The data read from
the input stream by the file SYSIN includes a field VREC that contains five
unnamed 7-character subfields; the field NUM defines the number of these
subfields that contain information. The output file WORK transmits to the data set
the whole of the field FREC and only those subfields of VREC that contain
information.
Figure 26 on page 297 shows an example of a program using list-directed output to
write graphics to a stream file. It assumes that you have an output device that can
print graphic data. The program reads employee records and selects persons living
in a certain area. It then edits the address field, inserting one graphic blank
between each address item, and prints the employee number, name, and address.
//EX7#2 JOB
//STEP1 EXEC IBMZCBG
//PLI.SYSIN DD *
PEOPLE: PROC OPTIONS(MAIN);
DCL WORK FILE STREAM OUTPUT,
1 REC,
2 FREC,
3 NAME CHAR(19),
3 NUM CHAR(1),
3 PAD CHAR(25),
2 VREC CHAR(35),
EOF BIT(1) INIT('0'B),
IN CHAR(80) DEF REC;
ON ENDFILE(SYSIN) EOF='1'B;
OPEN FILE(WORK) LINESIZE(400);
GET FILE(SYSIN) EDIT(IN)(A(80));
DO WHILE (EOF);
PUT FILE(WORK) EDIT(IN)(A(45+7*NUM));
GET FILE(SYSIN) EDIT(IN)(A(80));
END;
CLOSE FILE(WORK);
END PEOPLE;
/*
//GO.WORK DD DSN=HPU8.PEOPLE,DISP=(NEW,CATLG),UNIT=SYSDA,
// SPACE=(TRK,(1,1))
//GO.SYSIN DD *
R.C.ANDERSON 0 202848 DOCTOR
B.F.BENNETT 2 771239 PLUMBER VICTOR HAZEL
R.E.COLE 5 698635 COOK ELLEN VICTOR JOAN ANN OTTO
J.F.COOPER 5 418915 LAWYER FRANK CAROL DONALD NORMAN BRENDA
A.J.CORNELL 3 237837 BARBER ALBERT ERIC JANET
E.F.FERRIS 4 158636 CARPENTER GERALD ANNA MARY HAROLD
/*
Figure 25. Creating a data set with stream-oriented data transmission
Accessing a data set with stream I/O
A data set accessed using stream-oriented data transmission need not have been
created by stream-oriented data transmission, but it must have CONSECUTIVE
organization, and all the data in it must be in character or graphic form. You can
open the associated file for input, and read the records the data set contains; or
you can open the file for output, and extend the data set by adding records at the
end.
//EX7#3 JOB
//PLI.SYSIN DD *
% PROCESS GRAPHIC;
XAMPLE1: PROC OPTIONS(MAIN);
DCL INFILE FILE INPUT RECORD,
OUTFILE FILE OUTPUT STREAM ENV(GRAPHIC);
/* GRAPHIC OPTION MEANS DELIMITERS WILL BE INSERTED ON OUTPUT FILES. */
DCL
1 IN,
3 EMPNO CHAR(6),
3 SHIFT1 CHAR(1),
3 NAME,
5 LAST G(7),
5 FIRST G(7),
3 SHIFT2 CHAR(1),
3 ADDRESS,
5 ZIP CHAR(6),
5 SHIFT3 CHAR(1),
5 DISTRICT G(5),
5 CITY G(5),
5 OTHER G(8),
5 SHIFT4 CHAR(1);
DCL EOF BIT(1) INIT('0'B);
DCL ADDRWK G(20);
ON ENDFILE (INFILE) EOF = '1'B;
READ FILE(INFILE) INTO(IN);
DO WHILE(EOF);
DO;
IF SUBSTR(ZIP,1,3)='300'
THEN LEAVE;
L=0;
ADDRWK=DISTRICT;
DO I=1 TO 5;
IF SUBSTR(DISTRICT,I,1)= < >
THEN LEAVE; /* SUBSTR BIF PICKS 3P */
END; /* THE ITH GRAPHIC CHAR */
L=L+I+1; /* IN DISTRICT */
SUBSTR(ADDRWK,L,5)=CITY;
DO I=1 TO 5;
IF SUBSTR(CITY,I,1)= < >
THEN LEAVE;
END;
L=L+I;
SUBSTR(ADDRWK,L,8)=OTHER;
PUT FILE(OUTFILE) SKIP /* THIS DATA SET */
EDIT(EMPNO,IN.LAST,FIRST,ADDRWK) /* REQUIRES UTILITY */
(A(8),G(7),G(7),X(4),G(20)); /* TO PRINT GRAPHIC */
/* DATA */
END; /* END OF NON-ITERATIVE DO */
READ FILE(INFILE) INTO (IN);
END; /* END OF DO WHILE(EOF) */
END XAMPLE1;
/*
//GO.OUTFILE DD SYSOUT=A,DCB=(RECFM=VB,LRECL=121,BLKSIZE=129)
//GO.INFILE DD *
ABCDEF<
>300099< 3 3 3 3 3 3 3 >
ABCD <
>300011< 3 3 3 3 >
/*
Figure 26. Writing graphic data to a stream file
To access a data set, you must use one of the following to identify it:
v ENVIRONMENT attribute
v DD_DDNAME environment variable
v TITLE option of the OPEN statement
The following paragraphs describe the essential information you must include in
the DD statement, and discuss some of the optional information you can supply.
The discussions do not apply to data sets in the input stream.
When your application accesses an existing STREAM file, PL/I must obtain a
record-length value for that file. If the data set does not have a record-length, the
value can come from one of the following sources:
v The LINESIZE option of the OPEN statement
v The RECSIZE option of the ENVIRONMENT attribute
v The RECSIZE option of the DD_DDNAME environment variable
v The RECSIZE option of the TITLE option of the OPEN statement
v PL/I-supplied default value
If you are using an existing OUTPUT file, or if you supply a RECSIZE value, PL/I
determines the record-length value as described in Creating a data set with
stream I/O on page 294.
PL/I uses a default record-length value for an INPUT file when:
v The file is SYSIN, value = 80
v The file is associated with the terminal (stdout: or stderr:), value = 120
Record format
When using stream-oriented data transmission to access a data set, you do not
need to know the record format of the data set (except when you must specify a
block size); each GET statement transfers a discrete number of characters or
graphics to your program from the data stream.
If you do give record-format information, it must be compatible with the actual
structure of the data set. For example, if a data set is created with F-format records,
a record size of 600 bytes, and a block size of 3600 bytes, you can access the
records as if they are U-format with a maximum block size of 3600 bytes; but if
you specify a block size of 3500 bytes, your data will be truncated.
Example
The program in Figure 27 on page 299 reads the data set created by the program in
Figure 25 on page 296 and uses the file SYSPRINT to list the data it contains. (For
details on SYSPRINT, see Using SYSIN and SYSPRINT files on page 303.) Each
set of data is read, by the GET statement, into two variables: FREC, which always
contains 45 characters; and VREC, which always contains 35 characters. At each
execution of the GET statement, VREC consists of the number of characters
generated by the expression 7*NUM, together with sufficient blanks to bring the
total number of characters to 35. The DISP parameter of the DD statement could
read simply DISP=OLD; if DELETE is omitted, an existing data set will not be
deleted.
Using PRINT files with stream I/O
Both the operating system and the PL/I language include features that facilitate
the formatting of printed output. The operating system allows you to use the first
byte of each record for a print control character. The control characters, which are
not printed, cause the printer to skip to a new line or page. (Tables of print control
characters are given in Figure 30 on page 313 and Table 18 on page 314.)
In a PL/I program, the use of a PRINT file provides a convenient means of
controlling the layout of printed output from stream-oriented data transmission.
The compiler automatically inserts print control characters in response to the
PAGE, SKIP, and LINE options and format items.
You can apply the PRINT attribute to any STREAM OUTPUT file, even if you do
not intend to print the associated data set directly. When a PRINT file is associated
with a direct access data set, the print control characters have no effect on the
layout of the data set, but appear as part of the data in the records.
PRINT files opened with FB or VB will cause the UNDEFINEDFILE condition to
be raised. PRINT files should be opened with the "A" option; i.e., FBA or VBA.
The compiler reserves the first byte of each record transmitted by a PRINT file for
an American National Standard print control character, and inserts the appropriate
characters automatically.
A PRINT file uses only the following five print control characters:
Character
Action
Space 1 line before printing (blank character)
//EX7#5 JOB
//PLI.SYSIN DD *
PEOPLE: PROC OPTIONS(MAIN);
DCL WORK FILE STREAM INPUT,
1 REC,
2 FREC,
3 NAME CHAR(19),
3 NUM CHAR(1),
3 SERNO CHAR(7),
3 PROF CHAR(18),
2 VREC CHAR(35),
IN CHAR(80) DEF REC,
ON ENDFILE(WORK) EOF='1'B;
OPEN FILE(WORK);
GET FILE(WORK) EDIT(IN,VREC)(A(45),A(7*NUM));
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT(IN)(A);
GET FILE(WORK) EDIT(IN,VREC)(A(45),A(7*NUM));
END;
CLOSE FILE(WORK);
END PEOPLE;
/*
//GO.WORK DD DSN=HPU8.PEOPLE,DISP=(OLD,DELETE)
Figure 27. Accessing a data set with stream-oriented data transmission
0 Space 2 lines before printing
Space 3 lines before printing
+ No space before printing
1 Start new page
The compiler handles the PAGE, SKIP, and LINE options or format items by
padding the remainder of the current record with blanks and inserting the
appropriate control character in the next record. If SKIP or LINE specifies more
than a 3-line space, the compiler inserts sufficient blank records with appropriate
control characters to accomplish the required spacing. In the absence of a print
control option or format item, when a record is full the compiler inserts a blank
character (single line space) in the first byte of the next record.
If a PRINT file is being transmitted to a terminal, the PAGE, SKIP, and LINE
options will never cause more than 3 lines to be skipped, unless formatted output
is specified.
Controlling printed line length
You can limit the length of the printed line produced by a PRINT file either by
specifying a record length in your PL/I program (ENVIRONMENT attribute) or in
a DD statement, or by giving a line size in an OPEN statement (LINESIZE option).
The record length must include the extra byte for the print control character, that
is, it must be 1 byte larger than the length of the printed line (5 bytes larger for
V-format records). The value you specify in the LINESIZE option refers to the
number of characters in the printed line; the compiler adds the print control
character.
The blocking of records has no effect on the appearance of the output produced by
a PRINT file, but it does result in more efficient use of auxiliary storage when the
file is associated with a data set on a direct access device. If you use the LINESIZE
option, ensure that your line size is compatible with your block size. For F-format
records, block size must be an exact multiple of (line size+1); for V-format records,
block size must be at least 9 bytes greater than line size.
Although you can vary the line size for a PRINT file during execution by closing
the file and opening it again with a new line size, you must do so with caution if
you are using the PRINT file to create a data set on a direct access device. You
cannot change the record format that is established for the data set when the file is
first opened. If the line size you specify in an OPEN statement conflicts with the
record format already established, the UNDEFINEDFILE condition is raised. To
prevent this, either specify V-format records with a block size at least 9 bytes
greater than the maximum line size you intend to use, or ensure that the first
OPEN statement specifies the maximum line size. (Output destined for the printer
can be stored temporarily on a direct access device, unless you specify a printer by
using UNIT=, even if you intend it to be fed directly to the printer.)
Since PRINT files have a default line size of 120 characters, you need not give any
record format information for them. In the absence of other information, the
compiler assumes V-format records. The complete default information is:
BLKSIZE=129
LRECL=125
RECFM=VBA.
Example: Figure 28 on page 302 illustrates the use of a PRINT file and the
printing options of stream-oriented data transmission statements to format a table
and write it onto a direct access device for printing on a later occasion. The table
comprises the natural sines of the angles from 0 to 359 54' in steps of 6'.
The statements in the ENDPAGE ON-unit insert a page number at the bottom of
each page, and set up the headings for the following page.
The DD statement defining the data set created by this program includes no
record-format information. The compiler infers the following from the file
declaration and the line size specified in the statement that opens the file TABLE:
Record format =
V (the default for a PRINT file).
Record size =
98 (line size + 1 byte for print control character + 4 bytes for record control
field).
Block size =
102 (record length + 4 bytes for block control field).
The program in Figure 32 on page 320 uses record-oriented data transmission to
print the table created by the program in Figure 28 on page 302.
Overriding the tab control table
Data-directed and list-directed output to a PRINT file are aligned on preset
tabulator positions. See Figure 14 on page 235 and Figure 29 on page 303 for
examples of declaring a tab table. The definitions of the fields in the table are as
follows:
OFFSET OF TAB COUNT:
Halfword binary integer that gives the offset of Tab count, the field that
indicates the number of tabs to be used.
PAGESIZE:
Halfword binary integer that defines the default page size. This page size
is used for dump output to the PLIDUMP data set as well as for stream
output.
%PROCESS LIST;
SINE: PROC OPTIONS(MAIN);
DCL TABLE FILE STREAM OUTPUT PRINT;
DCL DEG FIXED DEC(5,1) INIT(0); /* INIT(0) FOR ENDPAGE */
DCL MIN FIXED DEC(3,1);
DCL PGNO FIXED DEC(2) INIT(0);
DCL ONCODE BUILTIN;
ON ERROR
BEGIN;
ON ERROR SYSTEM;
DISPLAY ('ONCODE = '|| ONCODE);
END;
ON ENDPAGE(TABLE)
BEGIN;
DCL I;
IF PGNO = 0 THEN
PUT FILE(TABLE) EDIT ('PAGE',PGNO)
(LINE(55),COL(80),A,F(3));
IF DEG = 360 THEN
DO;
PUT FILE(TABLE) PAGE EDIT ('NATURAL SINES') (A);
IF PGNO = 0 THEN
PUT FILE(TABLE) EDIT ((I DO I = 0 TO 54 BY 6))
(SKIP(3),10 F(9));
PGNO = PGNO + 1;
END;
ELSE
PUT FILE(TABLE) PAGE;
END;
OPEN FILE(TABLE) PAGESIZE(52) LINESIZE(93);
SIGNAL ENDPAGE(TABLE);
PUT FILE(TABLE) EDIT
((DEG,(SIND(DEG+MIN) DO MIN = 0 TO .9 BY .1) DO DEG = 0 TO 359))
(SKIP(2), 5 (COL(1), F(3), 10 F(9,4) ));
PUT FILE(TABLE) SKIP(52);
END SINE;
Figure 28. Creating a print file via stream data transmission. The example in Figure 32 on
page 320 will print the resultant file.
LINESIZE:
Halfword binary integer that defines the default line size.
PAGELENGTH:
Halfword binary integer that defines the default page length for printing at
a terminal.
FILLERS:
Three halfword binary integers; reserved for future use.
TAB COUNT:
Halfword binary integer that defines the number of tab position entries in
the table (maximum 255). If tab count = 0, any specified tab positions are
ignored.
Tab1Tabn:
n halfword binary integers that define the tab positions within the print
line. The first position is numbered 1, and the highest position is numbered
255. The value of each tab should be greater than that of the tab preceding
it in the table; otherwise, it is ignored. The first data field in the printed
output begins at the next available tab position.
You can override the default PL/I tab settings for your program by causing the
linkage editor to resolve an external reference to PLITABS. To cause the reference
to be resolved, supply a table with the name PLITABS, in the format described
above.
To supply this tab table, include a PL/I structure in your source program with the
name PLITABS, which you must declare to be STATIC EXTERNAL in your MAIN
procedure or in a program linked with your MAIN procedure. An example of the
PL/I structure is shown in Figure 29. This example creates three tab settings, in
positions 30, 60, and 90, and uses the defaults for page size and line size. Note that
TAB1 identifies the position of the second item printed on a line; the first item on a
line always starts at the left margin. The first item in the structure is the offset to
the NO_OF_TABS field. The FILL fields must not be omitted.
Using SYSIN and SYSPRINT files
If you code a GET statement without the FILE option in your program, the
compiler inserts the file name SYSIN. If you code a PUT statement without the
FILE option, the compiler inserts the name SYSPRINT.
If you do not declare SYSPRINT, the compiler gives the file the attribute PRINT in
addition to the normal default attributes. The complete set of attributes will be:
DCL 1 PLITABS STATIC EXT,
2 (OFFSET INIT(14),
PAGESIZE INIT(60),
LINESIZE INIT(120),
PAGELENGTH INIT(0),
FILL1 INIT(0),
FILL2 INIT(0),
FILL3 INIT(0),
NO_OF_TABS INIT(3),
TAB1 INIT(30),
TAB2 INIT(60),
TAB3 INIT(90)) FIXED BIN(15,0);
Figure 29. PL/I structure PLITABS for modifying the preset tab settings
FILE STREAM OUTPUT PRINT EXTERNAL
Since SYSPRINT is a PRINT file, the compiler also supplies a default line size of
120 characters and a V-format record. You need give only a minimum of
information in the corresponding DD statement; if your installation uses the usual
convention that the system output device of class A is a printer, the following is
sufficient:
Note: SYSIN and SYSPRINT are established in the User Exit during initialization.
IBM-supplied defaults for SYSIN and SYSPRINT are directed to the
terminal.
You can override the attributes given to SYSPRINT by the compiler by explicitly
declaring or opening the file. For more information about the interaction between
SYSPRINT and the z/OS Language Environment message file option, see the z/OS
Language Environment Programming Guide.
The compiler does not supply any special attributes for the input file SYSIN; if you
do not declare it, it receives only the default attributes. The data set associated
with SYSIN is usually in the input stream; if it is not in the input stream, you must
supply full DD information.
For more information about SYSPRINT, see SYSPRINT considerations on page
237.
Controlling input from the terminal
You can enter data at the terminal for an input file in your PL/I program if you do
the following:
1. Declare the input file explicitly or implicitly with the CONSECUTIVE
environment option (all stream files meet this condition)
2. Allocate the input file to the terminal.
You can usually use the standard default input file SYSIN because it is a stream
file and can be allocated to the terminal.
You are prompted for input to stream files by a colon (:). You will see the colon
each time a GET statement is executed in the program. The GET statement causes
the system to go to the next line. You can then enter the required data. If you enter
a line that does not contain enough data to complete execution of the GET
statement, a further prompt, which is a plus sign followed by a colon (+:), is
displayed.
By adding a hyphen to the end of any line that is to continue, you can delay
transmission of the data to your program until you enter two or more lines.
If you include output statements that prompt you for input in your program, you
can inhibit the initial system prompt by ending your own prompt with a colon. For
example, the GET statement could be preceded by a PUT statement such as:
PUT SKIP LIST('ENTER NEXT ITEM:');
To inhibit the system prompt for the next GET statement, your own prompt must
meet the following conditions:
1. It must be either list-directed or edit-directed, and if list-directed, must be to a
PRINT file.
2. The file transmitting the prompt must be allocated to the terminal. If you are
merely copying the file at the terminal, the system prompt is not inhibited.
Under TSO, there is support of an environment variable called TSO_INPUT_OPT
to help control the input from a terminal in a more flexible way than the default
way described above.
The syntax for the TSO_INPUT_OPT environment variable, which has to be in
uppercase, is:
TSO_INPUT_OPT =option
When specifying the options, which could be in uppercase or lowercase, blanks are
not allowed. If more than one option is specified, they have to be separated by a
comma. In addition, the syntax of the statement is not checked at the time the
command is entered. It is verified when the data set is opened. If the syntax is
wrong, the UNDEFINEDFILE condition is raised with the oncode 96.
The options you can specify are:
PROMPT
The PROMPT option specifies whether or not a colon should be visible as
prompts for stream input from the terminal.
PROMPT
N
( Y )
SAMELINE
The SAMELINE option specifies whether the system prompt occurs on the
same line as the statement that prompts for input.
SAMELINE
N
( Y )
For more information on how these options will affect your input and output to
the terminal, see Example 1 and Example 2 under the SAMELINE option in
Chapter 6.
One way to specify this environment variable under TSO is via the PLIXOPT
string. For example:
DCL PLIXOPT char(50) var ext static
init(ENVAR("TSO_INPUT_OPT=PROMPT(Y),SAMELINE(Y)"));
Format of data
The data you enter at the terminal should have exactly the same format as stream
input data in batch mode, except for the following variations:
v Simplified punctuation for input: If you enter separate items of input on
separate lines, there is no need to enter intervening blanks or commas; the
compiler will insert a comma at the end of each line.
For instance, in response to the statement:
GET LIST(I,J,K);
your terminal interaction could be as follows:
:
1
+:2
+:3
with a carriage return following each item. It would be equivalent to:
:
1,2,3
If you wish to continue an item onto another line, you must end the first line
with a continuation character. Otherwise, for a GET LIST or GET DATA
statement, a comma will be inserted, and for a GET EDIT statement, the item
will be padded (see next paragraph).
v Automatic padding for GET EDIT: There is no need to enter blanks at the end of
a line of input for a GET EDIT statement. The item you enter will be padded to
the correct length.
For instance, for the PL/I statement:
GET EDIT(NAME)(A(15));
you could enter the five characters:
SMITH
followed immediately by a carriage return. The item will be padded with 10
blanks, so that the program receives a string 15 characters long. If you wish to
continue an item on a second or subsequent line, you must add a continuation
character to the end of every line except the last; the first line transmitted would
otherwise be padded and treated as the complete data item.
v SKIP option or format item: A SKIP in a GET statement asks the program to
ignore data not yet entered. All uses of SKIP(n) where n is greater than one are
taken to mean SKIP(1). SKIP(1) is taken to mean that all unused data on the
current line is ignored.
Stream and record files
You can allocate both stream and record files to the terminal. However, no
prompting is provided for record files. If you allocate more than one file to the
terminal, and one or more of them is a record file, the output of the files will not
necessarily be synchronized. The order in which data is transmitted to and from
the terminal is not guaranteed to be the same order in which the corresponding
PL/I I/O statements are executed.
Also, record file input from the terminal is received in upper case letters because of
a TCAM restriction. To avoid problems you should use stream files wherever
possible.
Defining QSAM files using PL/I dynamic allocation
QSAM or HFS files can be defined using a DD statement, an environment variable,
or the TITLE option of the OPEN statement.
When an environment variable or TITLE option is used, the name must be in
uppercase. Specify the MVS data set in one of the following ways:
v DSN(data-set-name)
v DSN(data-set-name (member-name))
Specify the HFS file as follows:
PATH (absolute-path-name)
The following attributes can be specified in any order after the DSN keyword:
NEW, OLD, SHR, or MOD
TRACKS or CYL
SPACE(n,m)
VOL(volser)
UNIT(type)
KEEP, DELETE, CATALOG, or UNCATALOG
STORCLAS(storageclass)
MGMTCLAS(managementclass)
DATACLAS(dataclass)
Note: You cannot create a PDS or PDSE by using an environment variable or the
TITLE option of the OPEN statement, but you can create a new member in
an existing PDS or PDSE.
Capital and lowercase letters
For stream files, character strings are transmitted to the program as entered in
lowercase or uppercase. For record files, all characters become uppercase.
End-of-file
The characters /* in positions one and two of a line that contains no other
characters are treated as an end-of-file mark, that is, they raise the ENDFILE
condition.
COPY option of GET statement
The GET statement can specify the COPY option; but if the COPY file, as well as
the input file, is allocated to the terminal, no copy of the data will be printed.
Chapter 9. Controlling output to the terminal
At your terminal you can obtain data from a PL/I file that has been both:
1. Declared explicitly or implicitly with the CONSECUTIVE environment option.
All stream files meet this condition.
2. Allocated to the terminal.
The standard print file SYSPRINT generally meets both these conditions.
Format of PRINT files
Data from SYSPRINT or other PRINT files is not normally formatted into pages at
the terminal. Three lines are always skipped for PAGE and LINE options and
format items. The ENDPAGE condition is normally never raised. SKIP(n), where n
is greater than three, causes only three lines to be skipped. SKIP(0) is implemented
by backspacing, and should therefore not be used with terminals that do not have
a backspace feature.
You can cause a PRINT file to be formatted into pages by inserting a tab control
table in your program. The table must be called PLITABS, and its contents are
explained in Overriding the tab control table on page 302. You must initialize the
element PAGELENGTH to the length of page you requirethat is, the length of
the sheet of paper on which each page is to be printed, expressed as the maximum
number of lines that could be printed on it. You must initialize the element
PAGESIZE to the actual number of lines to be printed on each page. After the
number of lines in PAGESIZE has been printed on a page, ENDPAGE is raised, for
which standard system action is to skip the number of lines equal to
PAGELENGTH minus PAGESIZE, and then start printing the next page. For other
than standard layout, you must initialize the other elements in PLITABS to the
values shown in Figure 14 on page 235. You can also use PLITABS to alter the
tabulating positions of list-directed and data-directed output. You can use PLITABS
for SYSPRINT when you need to format page breaks in ILC applications. Set
PAGESIZE to 32767 and use the PUT PAGE statement to control page breaks.
Although some types of terminals have a tabulating facility, tabulating of
list-directed and data-directed output is always achieved by transmission of blank
characters.
Stream and record files
You can allocate both stream and record files to the terminal. However, if you
allocate more than one file to the terminal, and one of the files is SYSPRINT or a
record file, the output of the files is not necessarily synchronized. There is no
guarantee that the order in which data is transmitted between the program and the
terminal is the same as the order in which the corresponding PL/I output
statements are executed.
Output from the PUT EDIT command
The format of the output from a PUT EDIT command to a terminal is line mode
TPUTs with Start of field and end of field characters appearing as blanks on
the screen.
Chapter 10. Using record-oriented data transmission
PL/I supports various types of data sets with the RECORD attribute (see Table 20
on page 315). This section covers how to use consecutive data sets.
Table 17 lists the statements and options that you can use to create and access a
consecutive data set using record-oriented data transmission.
Table 17. Statements and options allowed for creating and accessing consecutive data sets
File declaration
1
Valid statements,
2
with
Options you must specify
Other options you
can specify
SEQUENTIAL OUTPUT
BUFFERED
WRITE FILE(file-reference)
FROM(reference);
LOCATE based-variable
FILE(file-reference);
SET(pointer-reference)
SEQUENTIAL OUTPUT WRITE FILE(file-reference)
FROM(reference);
SEQUENTIAL INPUT
BUFFERED
READ FILE(file-reference)
INTO(reference);
SET(pointer-reference);
IGNORE(expression);
SEQUENTIAL INPUT READ FILE(file-reference)
INTO(reference);
IGNORE(expression);
SEQUENTIAL UPDATE
BUFFERED
INTO(reference);
IGNORE(expression);
REWRITE FILE(file-reference); FROM(reference)
SEQUENTIAL UPDATE READ FILE(file-reference)
INTO(reference);
IGNORE(expression);
REWRITE FILE(file-reference)
FROM(reference);
Table 17. Statements and options allowed for creating and accessing consecutive data
sets (continued)
File declaration
1
Valid statements,
2
with
Options you must specify
Other options you
can specify
Notes:
1. The complete file declaration would include the attributes FILE, RECORD and
ENVIRONMENT.
2. The statement READ FILE (file-reference); is a valid statement and is equivalent to
READ FILE(file-reference) IGNORE (1);
Specifying record format
If you give record-format information, it must be compatible with the actual
structure of the data set. For example, if you create a data set with FB-format
records, with a record size of 600 bytes and a block size of 3600 bytes, you can
access the records as if they are U-format with a maximum block size of 3600
bytes. If you specify a block size of 3500 bytes, your data is truncated.
Defining files using record I/O
You define files for record-oriented data transmission by using a file declaration
with the following attributes:
DCL filename FILE RECORD
INPUT | OUTPUT | UPDATE
SEQUENTIAL
BUFFERED
Default file attributes are shown in Table 14 on page 275. The file attributes are
described in the PL/I Language Reference. Options of the ENVIRONMENT attribute
are discussed below.
The ENVIRONMENT options applicable to consecutive data sets are:
F|FB|FS|FBS|V|VB|U
BLKSIZE(block-size)
SCALARVARYING
CONSECUTIVE or ORGANIZATION(CONSECUTIVE)
CTLASA|CTL360
LEAVE|REREAD
The options through SCALARVARYING are described in Specifying characteristics
in the ENVIRONMENT attribute on page 275, and those after SCALARVARYING
are described below.
See Table 14 on page 275 to find which options you must specify, which are
optional, and which are defaults.
CONSECUTIVE
The CONSECUTIVE option defines a file with consecutive data set organization,
which is described in this chapter and in Data set organization on page 270.
CONSECUTIVE
CONSECUTIVE is the default.
ORGANIZATION(CONSECUTIVE)
Specifies that the file is associated with a consecutive data set. The
ORGANIZATION option is described in ORGANIZATION option on page 282.
The file can be either a native data set or a VSAM data set.
CTLASA|CTL360
The printer control options CTLASA and CTL360 apply only to OUTPUT files
associated with consecutive data sets. They specify that the first character of a
record is to be interpreted as a control character.
CTLASA
CTL360
The CTLASA option specifies American National Standard Vertical Carriage

Positioning Characters or American National Standard Pocket Select Characters
(Level 1). The CTL360 option specifies IBM machine-code control characters.
The American National Standard control characters, listed in Figure 30, cause the
specified action to occur before the associated record is printed or punched.
The machine code control characters differ according to the type of device. The
IBM machine code control characters for printers are listed in Table 18 on page 314.
Code Action
Space 1 line before printing (blank code)
0 Space 2 lines before printing
Space 3 lines before printing
+ Suppress space before printing
1 Skip to channel 1
2 Skip to channel 2
3 Skip to channel 3
4 Skip to channel 4
5 Skip to channel 5
6 Skip to channel 6
7 Skip to channel 7
8 Skip to channel 8
9 Skip to channel 9
A Skip to channel 10
B Skip to channel 11
C Skip to channel 12
V Select stacker 1
W Select stacker 2
Figure 30. American National Standard print and card punch control characters (CTLASA)
Chapter 10. Using record-oriented data transmission 313
Table 18. IBM machine code print control characters (CTL360)
Print and
Then Act
Code byte
Action Act immediately
(no printing)
Code byte
00000001 Print only (no space)
00001001 Space 1 line 00001011
00010001 Space 2 lines 00010011
00011001 Space 3 lines 00011011
10001001 Skip to channel 1 10001011
10010001 Skip to channel 2 10010011
10011001 Skip to channel 3 10011011
10100001 Skip to channel 4 10100011
10101001 Skip to channel 5 10101011
10110001 Skip to channel 6 10110011
10111001 Skip to channel 7 10111011
11000001 Skip to channel 8 11000011
11001001 Skip to channel 9 11001011
11010001 Skip to channel 10 11010011
11011001 Skip to channel 11 11011011
11100001 Skip to channel 12 11100011
LEAVE|REREAD
The magnetic tape handling options LEAVE and REREAD allow you to specify the
action to be taken when the end of a magnetic tape volume is reached, or when a
data set on a magnetic tape volume is closed. The LEAVE option prevents the tape
from being rewound. The REREAD option rewinds the tape to allow reprocessing
of the data set. If you do not specify either of these, the action at end-of-volume or
on closing of a data set is controlled by the DISP parameter of the associated DD
statement.
LEAVE
REREAD
If a data set is first read or written forward and then read backward in the same
program, specify the LEAVE option to prevent rewinding when the file is closed
(or, with a multivolume data set, when volume switching occurs).
The effects of the LEAVE and REREAD options are summarized in Table 19.
Table 19. Effect of LEAVE and REREAD Options
ENVIRONMENT
option
DISP
parameter
Action
REREAD Positions the current volume
to reprocess the data set.
LEAVE Positions the current volume
at the logical end of the data
set.
Table 19. Effect of LEAVE and REREAD Options (continued)
ENVIRONMENT
option
DISP
parameter
Action
Neither
REREAD nor
LEAVE
PASS
DELETE
KEEP,
CATLG,
UNCATLG
Positions the volume at the
end of the data set.
Rewinds the current volume.
Rewinds and unloads the
current volume.
Creating a data set with record I/O
When you create a consecutive data set, you must open the associated file for
SEQUENTIAL OUTPUT. You can use either the WRITE or LOCATE statement to
write records. Table 17 on page 311 shows the statements and options for creating a
consecutive data set.
When creating a data set, you must identify it to the operating system in a DD
statement. The following paragraphs, summarized in Table 20, tell what essential
information you must include in the DD statement and discuss some of the
optional information you can supply.
Table 20. Creating a consecutive data set with record I/O: essential parameters of the DD statement
Storage device
When required What you must
state Parameters
All Always Output device
Block size
1
UNIT= or SYSOUT=
or
VOLUME=REF=
DCB=(BLKSIZE=...
Direct access
only
Always Storage space
required
SPACE=
Direct access Data set to be used by another
job step but not required at
end of job
Data set to be kept after end
of job
Data set to be on particular
device
Disposition
Disposition
Name of data set
Volume serial
number
DISP=
DISP=
DSNAME=
VOLUME=SER=
or
VOLUME=REF=
Notes:
1
Or you could specify the block size in your PL/I program by using the ENVIRONMENT attribute.
When you create a consecutive data set you must specify:
v The name of data set to be associated with your PL/I file. A data set with
consecutive organization can exist on any type of device.
v The record length. You can specify the record length using the RECSIZE option
of the ENVIRONMENT attribute, of the DD_DDNAME environment variable, or
of the TITLE option of the OPEN statement.
For files associated with the terminal device (stdout: or stderr:), PL/I uses a
default record length of 120 when the RECSIZE option is not specified.
Accessing and updating a data set with record I/O
Once you create a consecutive data set, you can open the file that accesses it for
sequential input, for sequential output, or, for data sets on direct access devices, for
updating. See Figure 31 on page 318 for an example of a program that accesses and
updates a consecutive data set. If you open the file for output, and extend the data
set by adding records at the end, you must specify DISP=MOD in the DD
statement. If you do not, the data set will be overwritten. If you open a file for
updating, you can update only records in their existing sequence, and if you want
to insert records, you must create a new data set. Table 17 on page 311 shows the
statements and options for accessing and updating a consecutive data set.
When you access a consecutive data set by a SEQUENTIAL UPDATE file, you
must retrieve a record with a READ statement before you can update it with a
REWRITE statement; however, every record that is retrieved need not be rewritten.
A REWRITE statement will always update the last record read.
Consider the following:
READ FILE(F) INTO(A);
.
.
.
READ FILE(F) INTO(B);
.
.
.
REWRITE FILE(F) FROM(A);
The REWRITE statement updates the record that was read by the second READ
statement. The record that was read by the first statement cannot be rewritten after
the second READ statement has been executed.
You cannot update a consecutive data set on magnetic tape except by adding
records at the end. To replace or insert records, you must read the data set and
write the updated records into a new data set.
You can read a consecutive data set on magnetic tape forward only. Reading
backward is not supported.
To access a data set, you must identify it to the operating system in a DD
statement. Table 21 on page 317 summarizes the DD statement parameters needed
to access a consecutive data set.
Table 21. Accessing a consecutive data set with record I/O: essential parameters of the DD
statement
Parameters What you must state When required
DSNAME=
DISP=
Name of data set
Disposition of data
set
Always
UNIT= or
VOLUME=REF=
Input device If data set not cataloged (all devices)
VOLUME=SER= Volume serial number If data set not cataloged (direct access)
DCB=(BLKSIZE= Block size
1
If data set does not have standard labels
Note:
1
Or you could specify the block size in your PL/I program by using the
ENVIRONMENT attribute.
The following paragraphs indicate the essential information you must include in
the DD statement, and discuss some of the optional information you can supply.
The discussions do not apply to data sets in the input stream.
If the data set is cataloged, you need to supply only the following information in
the DD statement:
v The name of the data set (DSNAME parameter). The operating system will
locate the information describing the data set in the system catalog, and, if
necessary, will request the operator to mount the volume containing it.
v Confirmation that the data set exists (DISP parameter). If you open the data set
for output with the intention of extending it by adding records at the end, code
DISP=MOD; otherwise, opening the data set for output will result in it being
overwritten.
If the data set is not cataloged, you must additionally specify the device that will
read the data set and, direct access devices, give the serial number of the volume
that contains the data set (UNIT and VOLUME parameters).
Example of consecutive data sets
Creating and accessing consecutive data sets are illustrated in the program in
Figure 31 on page 318. The program merges the contents of two data sets, in the
input stream, and writes them onto a new data set, &&TEMP; each of the original
data sets contains 15-byte fixed-length records arranged in EBCDIC collating
sequence. The two input files, INPUT1 and INPUT2, have the default attribute
BUFFERED, and locate mode is used to read records from the associated data sets
into the respective buffers. Access of based variables in the buffers should not be
attempted after the file has been closed.
//EXAMPLE JOB
//PLI.SYSIN DD *
%PROCESS LIST;
MERGE: PROC OPTIONS(MAIN);
DCL (INPUT1, /* FIRST INPUT FILE */
INPUT2, /* SECOND INPUT FILE */
OUT ) FILE RECORD SEQUENTIAL; /* RESULTING MERGED FILE*/
DCL SYSPRINT FILE PRINT; /* NORMAL PRINT FILE */
DCL INPUT1_EOF BIT(1) INIT('0'B); /* EOF FLAG FOR INPUT1 */
DCL INPUT2_EOF BIT(1) INIT('0'B); /* EOF FLAG FOR INPUT2 */
DCL OUT_EOF BIT(1) INIT('0'B); /* EOF FLAG FOR OUT */
DCL TRUE BIT(1) INIT('1'B); /* CONSTANT TRUE */
DCL FALSE BIT(1) INIT('0'B); /* CONSTANT FALSE */
DCL ITEM1 CHAR(15) BASED(A); /* ITEM FROM INPUT1 */
DCL ITEM2 CHAR(15) BASED(B); /* ITEM FROM INPUT2 */
DCL INPUT_LINE CHAR(15); /* INPUT FOR READ INTO */
DCL A POINTER; /* POINTER VAR */
DCL B POINTER; /* POINTER VAR */
ON ENDFILE(INPUT1) INPUT1_EOF = TRUE;
ON ENDFILE(INPUT2) INPUT2_EOF = TRUE;
ON ENDFILE(OUT) OUT_EOF = TRUE;
OPEN FILE(INPUT1) INPUT,
FILE(INPUT2) INPUT,
FILE(OUT) OUTPUT;
READ FILE(INPUT1) SET(A); /* PRIMING READ */
READ FILE(INPUT2) SET(B);
DO WHILE ((INPUT1_EOF = FALSE) & (INPUT2_EOF = FALSE));
IF ITEM1 > ITEM2 THEN
DO;
WRITE FILE(OUT) FROM(ITEM2);
PUT FILE(SYSPRINT) SKIP EDIT('1>2', ITEM1, ITEM2)
(A(5),A,A);
END;
ELSE
DO;
PUT FILE(SYSPRINT) SKIP EDIT('1<2', ITEM1, ITEM2)
(A(5),A,A);
READ FILE(INPUT1) SET(A);
END;
END;
Figure 31. Merge Sortcreating and accessing a consecutive data set (Part 1 of 2)
The program in Figure 32 on page 320 uses record-oriented data transmission to
print the table created by the program in Figure 28 on page 302.
DO WHILE (INPUT1_EOF = FALSE); /* INPUT2 IS EXHAUSTED */
PUT FILE(SYSPRINT) SKIP EDIT('1', ITEM1) (A(2),A);
READ FILE(INPUT1) SET(A);
END;
DO WHILE (INPUT2_EOF = FALSE); /* INPUT1 IS EXHAUSTED */
PUT FILE(SYSPRINT) SKIP EDIT('2', ITEM2) (A(2),A);
END;
CLOSE FILE(INPUT1), FILE(INPUT2), FILE(OUT);
PUT FILE(SYSPRINT) PAGE;
OPEN FILE(OUT) SEQUENTIAL INPUT;
READ FILE(OUT) INTO(INPUT_LINE); /* DISPLAY OUT FILE */
DO WHILE (OUT_EOF = FALSE);
PUT FILE(SYSPRINT) SKIP EDIT(INPUT_LINE) (A);
READ FILE(OUT) INTO(INPUT_LINE);
END;
CLOSE FILE(OUT);
END MERGE;
/*
//GO.INPUT1 DD *
AAAAAA
CCCCCC
EEEEEE
GGGGGG
IIIIII
/*
//GO.INPUT2 DD *
BBBBBB
DDDDDD
FFFFFF
HHHHHH
JJJJJJ
KKKKKK
/*
//GO.OUT DD DSN=&&TEMP,DISP=(NEW,DELETE),UNIT=SYSDA,
// DCB=(RECFM=FB,BLKSIZE=150,LRECL=15),SPACE=(TRK,(1,1))
Figure 31. Merge Sortcreating and accessing a consecutive data set (Part 2 of 2)
%PROCESS LIST;
PRT: PROC OPTIONS(MAIN);
DCL TABLE FILE RECORD INPUT SEQUENTIAL;
DCL PRINTER FILE RECORD OUTPUT SEQL
ENV(V BLKSIZE(102) CTLASA);
DCL LINE CHAR(94) VAR;
DCL TABLE_EOF BIT(1) INIT('0'B); /* EOF FLAG FOR TABLE */
DCL TRUE BIT(1) INIT(1B); /* CONSTANT TRUE */
DCL FALSE BIT(1) INIT('0'B); /* CONSTANT FALSE */
ON ENDFILE(TABLE) TABLE_EOF = TRUE;
OPEN FILE(TABLE),
FILE(PRINTER);
READ FILE(TABLE) INTO(LINE); /* PRIMING READ */
DO WHILE (TABLE_EOF = FALSE);
WRITE FILE(PRINTER) FROM(LINE);
READ FILE(TABLE) INTO(LINE);
END;
CLOSE FILE(TABLE),
FILE(PRINTER);
END PRT;
Figure 32. Printing record-oriented data transmission
Chapter 11. Defining and using regional data sets
This chapter covers regional data set organization, data transmission statements,
and ENVIRONMENT options that define regional data sets. How to create and
access regional data sets for each type of regional organization is then discussed.
A data set with regional organization is divided into regions, each of which is
identified by a region number, and each of which can contain one record or more
than one record, depending on the type of regional organization. The regions are
numbered in succession, beginning with zero, and a record can be accessed by
specifying its region number, and perhaps a key, in a data transmission statement.
Regional data sets are confined to direct access devices.
Regional organization of a data set allows you to control the physical placement of
records in the data set, and to optimize the access time for a particular application.
Such optimization is not available with consecutive or indexed organization, in
which successive records are written either in strict physical sequence or in logical
sequence depending on ascending key values; neither of these methods takes full
advantage of the characteristics of direct access storage devices.
You can create a regional data set in a manner similar to a consecutive or indexed
data set, presenting records in the order of ascending region numbers; alternatively,
you can use direct access, in which you present records in random sequence and
insert them directly into preformatted regions. Once you create a regional data set,
you can access it by using a file with the attributes SEQUENTIAL or DIRECT as
well as INPUT or UPDATE. You do not need to specify either a region number or
a key if the data set is associated with a SEQUENTIAL INPUT or SEQUENTIAL
UPDATE file. When the file has the DIRECT attribute, you can retrieve, add,
delete, and replace records at random.
Records within a regional data set are either actual records containing valid data or
dummy records.
The major advantage of regional organization over other types of data set
organization is that it allows you to control the relative placement of records; by
judicious programming, you can optimize record access in terms of device
capabilities and the requirements of particular applications.
Direct access of regional data sets is quicker than that of indexed data sets, but
regional data sets have the disadvantage that sequential processing can present
records in random sequence; the order of sequential retrieval is not necessarily that
in which the records were presented, nor need it be related to the relative key
values.
Table 22 on page 322 lists the data transmission statements and options that you
can use to create and access a regional data set.
Table 22. Statements and options allowed for creating and accessing regional data sets
File
declaration
1
Valid statements,
2
with
options you must include
Other options you
can also include
SEQUENTIAL OUTPUT WRITE FILE(file-reference)
FROM(reference)
KEYFROM(expression);
FROM(file-reference)
SEQUENTIAL INPUT READ FILE(file-reference)
INTO(reference);
IGNORE(expression);
KEYTO(reference)
KEYTO(reference)
SEQUENTIAL UPDATE
3
INTO(reference);
IGNORE(expression);
REWRITE FILE(file-reference);
KEYTO(reference)
KEYTO(reference)
FROM(reference)
DIRECT OUTPUT WRITE FILE(file-reference)
FROM(reference)
DIRECT INPUT READ FILE(file-reference)
INTO(reference)
KEY(expression);
DIRECT UPDATE READ FILE(file-reference)
INTO(reference)
KEY(expression);
FROM(reference)
KEY(expression);
FROM(reference)
DELETE FILE(file-reference)
KEY(expression);
Notes:
1. The complete file declaration would include the attributes FILE, RECORD, and
ENVIRONMENT; if you use any of the options KEY, KEYFROM, or KEYTO, you must
also include the attribute KEYED.
2. The statement READ FILE(file-reference); is equivalent to the statement READ
FILE(file-reference) IGNORE(1);
3. The file must not have the UPDATE attribute when creating new data sets.
Defining REGIONAL(1) data sets using PL/I dynamic allocation
REGIONAL(1) data sets can be defined by using either a DD statement, an
environment variable, or the TITLE option of the OPEN statement.
uppercase. Specify the MVS data set in the following way:
DSN(data-set-name)
You must specify one of the following attributes after the DSN keyword:
OLD
SHR
Defining files for a regional data set
Use a file declaration with the following attributes to define a sequential regional
data set:
SEQUENTIAL
BUFFERED
[KEYED]
Since BUFFERED and UNBUFFERED will be treated the same for REGIONAL(1)
data sets, either option can be specified in the ENV option. For example, the
FROM option is not required on a REWRITE for a SEQUENTIAL UNBUFFERED
file and the LOCATE statement is allowed for OUTPUT SEQUENTIAL data sets
even if UNBUFFERED is specified.
To define a direct regional data set, use a file declaration with the following
attributes:
DIRECT
Default file attributes are shown in Table 14 on page 275. The file attributes are
described in the PL/I Language Reference. Options of the ENVIRONMENT attribute
are discussed below.
The ENVIRONMENT options applicable to regional data sets are:
REGIONAL({1})
F
BLKSIZE(block-size)
SCALARVARYING
Chapter 11. Defining and using regional data sets 323
REGIONAL option
Use the REGIONAL option to define a file with regional organization.
REGIONAL ( 1 )
1 specifies REGIONAL(1)
REGIONAL(1)
specifies that the data set contains F-format records that do not have recorded
keys. Each region in the data set contains only one record; therefore, each
region number corresponds to a relative record within the data set (that is,
region numbers start with 0 at the beginning of the data set).
Although REGIONAL(1) data sets have no recorded keys, you can use
REGIONAL(1) DIRECT INPUT or UPDATE files to process data sets that do
have recorded keys.
BLKSIZE(block-size)
If both RECSIZE and BLKSIZE are specified, they must specify the same value.
REGIONAL(1) organization is most suited to applications where there are no
duplicate region numbers, and where most of the regions will be filled (reducing
wasted space in the data set).
Using keys with REGIONAL data sets
There are two kinds of keys, recorded keys and source keys. A recorded key is a
character string that immediately precedes each record in the data set to identify
that record; its length cannot exceed 255 characters. A source key is the character
value of the expression that appears in the KEY or KEYFROM option of a data
transmission statement to identify the record to which the statement refers. When
you access a record in a regional data set, the source key gives a region number,
and can also give a recorded key.
Unlike the keys for indexed data sets, recorded keys in a regional data set are
never embedded within the record.
Using REGIONAL(1) data sets
In a REGIONAL(1) data set, since there are no recorded keys, the region number
serves as the sole identification of a particular record. The character value of the
source key should represent an unsigned decimal integer that should not exceed
16777215 (although the actual number of records allowed can be smaller,
depending on a combination of record size, device capacity, and limits of your
access method. For direct regional(1) files with fixed format records, the maximum
number of tracks which can be addressed by relative track addressing is 65,536.) If
the region number exceeds this figure, it is treated as modulo 16777216; for
instance, 16777226 is treated as 10. Only the characters 0 through 9 and the blank
character are valid in the source key; leading blanks are interpreted as zeros.
Embedded blanks are not allowed in the number; the first embedded blank, if any,
terminates the region number. If more than 8 characters appear in the source key,
only the rightmost 8 are used as the region number; if there are fewer than 8
characters, blanks (interpreted as zeros) are inserted on the left.
Dummy Records
Records in a REGIONAL(1) data set are either actual records containing valid data
or dummy records. A dummy record in a REGIONAL(1) data set is identified by
the constant (8)'1'B in its first byte. Although such dummy records are inserted in
the data set either when it is created or when a record is deleted, they are not
ignored when the data set is read; your PL/I program must be prepared to
recognize them. You can replace dummy records with valid data. Note that if you
insert (8)'1'B in the first byte, the record can be lost if you copy the file onto a data
set that has dummy records that are not retrieved.
Creating a REGIONAL(1) data set
You can create a REGIONAL(1) data set either sequentially or by direct access.
Table 22 on page 322 shows the statements and options for creating a regional data
set.
When you use a SEQUENTIAL OUTPUT file to create the data set, the opening of
the file causes all tracks on the data set to be cleared, and a capacity record to be
written at the beginning of each track to record the amount of space available on
that track. You must present records in ascending order of region numbers; any
region you omit from the sequence is filled with a dummy record. If there is an
error in the sequence, or if you present a duplicate key, the KEY condition is
raised. When the file is closed, any space remaining at the end of the current
extent is filled with dummy records.
If you use a DIRECT OUTPUT file to create the data set, the whole primary extent
allocated to the data set is filled with dummy records when the file is opened. You
can present records in random order; if you present a duplicate, the existing record
will be overwritten.
For sequential creation, the data set can have up to 15 extents, which can be on
more than one volume. For direct creation, the data set can have only one extent,
and can therefore reside on only one volume.
Example
Creating a REGIONAL(1) data set is illustrated in Figure 33 on page 326. The data
set is a list of telephone numbers with the names of the subscribers to whom they
are allocated. The telephone numbers correspond with the region numbers in the
data set, the data in each occupied region being a subscriber's name.
Accessing and updating a REGIONAL(1) data set
Once you create a REGIONAL(1) data set, you can open the file that accesses it for
SEQUENTIAL INPUT or UPDATE, or for DIRECT INPUT or UPDATE. You can
//EX9 JOB
//STEP1 EXEC IBMZCBG,PARM.PLI='NOP,MAR(1,72)',PARM.BIND='LIST'
//PLI.SYSIN DD *
CRR1: PROC OPTIONS(MAIN);
/* CREATING A REGIONAL(1) DATA SET - PHONE DIRECTORY */
DCL NOS FILE RECORD OUTPUT DIRECT KEYED ENV(REGIONAL(1));
DCL SYSIN FILE INPUT RECORD;
DCL SYSIN_REC BIT(1) INIT('1'B);
DCL 1 CARD,
2 NAME CHAR(20),
2 NUMBER CHAR( 2),
2 CARD_1 CHAR(58);
DCL IOFIELD CHAR(20);
ON ENDFILE (SYSIN) SYSIN_REC = '0'B;
OPEN FILE(NOS);
READ FILE(SYSIN) INTO(CARD);
DO WHILE(SYSIN_REC);
IOFIELD = NAME;
WRITE FILE(NOS) FROM(IOFIELD) KEYFROM(NUMBER);
PUT FILE(SYSPRINT) SKIP EDIT (CARD) (A);
END;
CLOSE FILE(NOS);
END CRR1;
/*
//GO.SYSLMOD DD DSN=&&GOSET,DISP=(OLD,DELETE)
//GO.NOS DD DSN=MYID.NOS,UNIT=SYSDA,SPACE=(20,100),
// DCB=(RECFM=F,BLKSIZE=20,DSORG=DA),DISP=(NEW,KEEP)
//GO.SYSIN DD *
ACTION,G. 12
BAKER,R. 13
BRAMLEY,O.H. 28
CHEESNAME,L. 11
CORY,G. 36
ELLIOTT,D. 85
FIGGINS,E.S. 43
HARVEY,C.D.W. 25
HASTINGS,G.M. 31
KENDALL,J.G. 24
LANCASTER,W.R. 64
MILES,R. 23
NEWMAN,M.W. 40
PITT,W.H. 55
ROLF,D.E. 14
SHEERS,C.D. 21
SURCLIFFE,M. 42
TAYLOR,G.C. 47
WILTON,L.W. 44
WINSTONE,E.M. 37
/*
Figure 33. Creating a REGIONAL(1) data set
open it for OUTPUT only if the existing data set is to be overwritten. Table 22 on
page 322 shows the statements and options for accessing a regional data set.
Sequential access
To open a SEQUENTIAL file that is used to process a REGIONAL(1) data set, use
either the INPUT or UPDATE attribute. You must not include the KEY option in
data transmission statements, but the file can have the KEYED attribute, since you
can use the KEYTO option. If the target character string referenced in the KEYTO
option has more than 8 characters, the value returned (the 8-character region
number) is padded on the left with blanks. If the target string has fewer than 8
characters, the value returned is truncated on the left.
Sequential access is in the order of ascending region numbers. All records are
retrieved, whether dummy or actual, and you must ensure that your PL/I program
recognizes dummy records.
Using sequential input with a REGIONAL(1) data set, you can read all the records
in ascending region-number sequence, and in sequential update you can read and
rewrite each record in turn.
The rules governing the relationship between READ and REWRITE statements for
a SEQUENTIAL UPDATE file that accesses a REGIONAL(1) data set are identical
to those for a consecutive data set. Consecutive data sets are discussed in detail in
Chapter 8, Defining and using consecutive data sets, on page 291.
Direct access
To open a DIRECT file that is used to process a REGIONAL(1) data set you can
use either the INPUT or the UPDATE attribute. All data transmission statements
must include source keys; the DIRECT attribute implies the KEYED attribute.
Use DIRECT UPDATE files to retrieve, add, delete, or replace records in a
REGIONAL(1) data set according to the following conventions:
Retrieval
All records, whether dummy or actual, are retrieved. Your program must
recognize dummy records.
Addition
A WRITE statement substitutes a new record for the existing record (actual
or dummy) in the region specified by the source key.
Deletion
The record you specify by the source key in a DELETE statement is
converted to a dummy record.
Replacement
The record you specify by the source key in a REWRITE statement,
whether dummy or actual, is replaced.
Example
Updating a REGIONAL(1) data set is illustrated in Figure 34 on page 328. This
program updates the data set and lists its contents. Before each new or updated
record is written, the existing record in the region is tested to ensure that it is a
dummy; this is necessary because a WRITE statement can overwrite an existing
record in a REGIONAL(1) data set even if it is not a dummy. Similarly, during the
sequential reading and printing of the contents of the data set, each record is tested
and dummy records are not printed.
//EX10 JOB
//STEP2 EXEC IBMZCBG,PARM.PLI='NOP,MAR(1,72)',PARM.BIND='LIST'
//PLI.SYSIN DD *
ACR1: PROC OPTIONS(MAIN);
/* UPDATING A REGIONAL(1) DATA SET - PHONE DIRECTORY */
DCL NOS FILE RECORD KEYED ENV(REGIONAL(1));
DCL SYSIN FILE INPUT RECORD;
DCL (SYSIN_REC,NOS_REC) BIT(1) INIT('1'B);
DCL 1 CARD,
2 NAME CHAR(20),
2 (NEWNO,OLDNO) CHAR( 2),
2 CARD_1 CHAR( 1),
2 CODE CHAR( 1),
2 CARD_2 CHAR(54);
DCL IOFIELD CHAR(20);
DCL BYTE CHAR(1) DEF IOFIELD;
ON ENDFILE(SYSIN) SYSIN_REC = '0'B;
OPEN FILE (NOS) DIRECT UPDATE;
DO WHILE(SYSIN_REC);
SELECT(CODE);
WHEN('A','C') DO;
IF CODE = 'C' THEN
DELETE FILE(NOS) KEY(OLDNO);
READ FILE(NOS) KEY(NEWNO) INTO(IOFIELD);
IF UNSPEC(BYTE) = (8)'1'B
THEN WRITE FILE(NOS) KEYFROM(NEWNO) FROM(NAME);
ELSE PUT FILE(SYSPRINT) SKIP LIST ('DUPLICATE:',NAME);
END;
WHEN('D') DELETE FILE(NOS) KEY(OLDNO);
OTHERWISE PUT FILE(SYSPRINT) SKIP LIST ('INVALID CODE:',NAME);
END;
END;
CLOSE FILE(SYSIN),FILE(NOS);
OPEN FILE(NOS) SEQUENTIAL INPUT;
ON ENDFILE(NOS) NOS_REC = '0'B;
READ FILE(NOS) INTO(IOFIELD) KEYTO(NEWNO);
DO WHILE(NOS_REC);
IF UNSPEC(BYTE) = (8)'1'B
THEN PUT FILE(SYSPRINT) SKIP EDIT (NEWNO,IOFIELD)(A(2),X(3),A);
PUT FILE(SYSPRINT) SKIP EDIT (IOFIELD) (A);
END;
CLOSE FILE(NOS);
END ACR1;
/*
Figure 34. Updating a REGIONAL(1) data set (Part 1 of 2)
Essential information for creating and accessing regional data sets
To create a regional data set, you must give the operating system certain
information, either in your PL/I program or in the DD statement that defines the
data set. The following paragraphs indicate the essential information, and discuss
some of the optional information you can supply.
You must supply the following information when creating a regional data set:
v Device that will write your data set (UNIT or VOLUME parameter of DD
statement).
v Block size: You can specify the block size either in your PL/I program (in the
BLKSIZE option of the ENVIRONMENT attribute) or in the DD statement
(BLKSIZE subparameter). If you do not specify a record length, unblocked
records are the default and the record length is determined from the block size.
If you do specify a record length, it must be equal to the block size.
If you want to keep a data set (that is, you do not want the operating system to
delete it at the end of your job), the DD statement must name the data set and
indicate how it is to be disposed of (DSNAME and DISP parameters). The DISP
parameter alone will suffice if you want to use the data set in a later step but do
not need it after the end of your job.
If you want your data set stored on a particular direct access device, you must
indicate the volume serial number in the DD statement (SER or REF subparameter
of VOLUME parameter). If you do not supply a serial number for a data set that
you want to keep, the operating system allocates one, informs the operator, and
prints the number on your program listing. All the essential parameters required in
a DD statement for the creation of a regional data set are summarized in Table 23
on page 330; and Table 24 on page 330 lists the DCB subparameters needed. See
your z/OS JCL User's Guide for a description of the DCB subparameters.
You cannot place a regional data set on a system output (SYSOUT) device.
In the DCB parameter, if you specify the DSORG parameter, you must specify the
data set organization as direct by coding DSORG=DA. You cannot specify the
//GO.NOS DD DSN=J44PLI.NOS,DISP=(OLD,DELETE),UNIT=SYSDA,VOL=SER=nnnnnn
//GO.SYSIN DD *
NEWMAN,M.W. 5640 C
GOODFELLOW,D.T. 89 A
MILES,R. 23 D
HARVEY,C.D.W. 29 A
BARTLETT,S.G. 13 A
CORY,G. 36 D
READ,K.M. 01 A
PITT,W.H. 55
ROLF,D.F. 14 D
ELLIOTT,D. 4285 C
HASTINGS,G.M. 31 D
BRAMLEY,O.H. 4928 C
/*
Figure 34. Updating a REGIONAL(1) data set (Part 2 of 2)
DUMMY or DSN=NULLFILE parameters in a DD statement for a regional data set.
Using DSORG=DA may cause message IEC225I to be issued. This message can be
safely ignored.
Table 23. Creating a regional data set: essential parameters of the DD statement
UNIT= or
VOLUME=REF=
SPACE=
DCB=
Output device
1
Storage space required
2
Data control block information:
see Table 24
Always
DISP= Disposition Data set to be used in another
job step but not required in
another job
DISP=
DSNAME=
Disposition
Name of data set
Data set to be kept after end of
job
VOLUME=SER= or
VOLUME=REF=
Volume serial number Data set to be on particular
volume
1
Regional data sets are confined to direct access devices.
2
For sequential access, the data set can have up to 15 extents, which can be on more than
one volume. For creation with DIRECT access, the data set can have only one extent.
To access a regional data set, you must identify it to the operating system in a DD
statement. The following paragraphs indicate the minimum information you must
include in the DD statement; this information is summarized in Table 25 on page
331.
If the data set is cataloged, you need to supply only the following information in
your DD statement:
v The name of the data set (DSNAME parameter). The operating system locates
the information that describes the data set in the system catalog and, if
necessary, requests the operator to mount the volume that contains it.
v Confirmation that the data set exists (DISP parameter).
If the data set is not cataloged, you must, in addition, specify the device that will
read the data set and give the serial number of the volume that contains the data
set (UNIT and VOLUME parameters).
When opening a multiple-volume regional data set for sequential update, the
ENDFILE condition is raised at the end of the first volume.
Table 24. DCB subparameters for a regional data set
Subparameters To specify When required
RECFM=F
BLKSIZE=
DSORG=DA
Record format
1
Block size
1
Data set organization
These are always required
1
Or you can specify the block size in the ENVIRONMENT attribute.
Table 25. Accessing a regional data set: essential parameters of the DD statement
DSNAME=
DISP=
Name of data set
Disposition of data set
Always
UNIT= or
VOLUME=REF=
VOLUME=SER=
Input device
Volume serial number
If data set not cataloged
Chapter 12. Defining and using VSAM data sets
This chapter covers VSAM (the Virtual Storage Access Method) organization for
record-oriented data transmission, VSAM ENVIRONMENT options, compatibility
with other PL/I data set organizations, and the statements you use to load and
access the three types of VSAM data sets that PL/I supportsentry-sequenced,
key-sequenced, and relative record. The chapter is concluded by a series of
examples showing the PL/I statements, Access Method Services commands, and
JCL statements necessary to create and access VSAM data sets.
Enterprise PL/I provides no support for ISAM datasets.
For additional information about the facilities of VSAM, the structure of VSAM
data sets and indexes, the way in which they are defined by Access Method
Services, and the required JCL statements, see the VSAM publications for your
system.
Defining VSAM file using PL/I dynamic allocation
VSAM data sets can be defined by using either a DD statement, an environment
variable, or the TITLE option of the OPEN statement.
uppercase. Specify the MVS data set as follows:
DSN(data-set-name)
You must specify one of the following attributes after the DSN keyword:
OLD
SHR
Using VSAM data sets
How to run a program with VSAM data sets
Before you execute a program that accesses a VSAM data set, you need to know:
v The name of the VSAM data set
v The name of the PL/I file
v Whether you intend to share the data set with other users
Then you can write the required DD statement to access the data set:
//filename DD DSNAME=dsname,DISP=OLD|SHR
For example, if your file is named PL1FILE, your data set named VSAMDS, and
you want exclusive control of the data set, enter:
//PL1FILE DD DSNAME=VSAMDS,DISP=OLD
To share your data set, use DISP=SHR.
Enterprise PL/I has no support for ISAM data sets.
To optimize VSAM's performance by controlling the number of VSAM buffers used
for your data set, see the VSAM publications.
Pairing an Alternate Index Path with a File
When using an alternate index, you simply specify the name of the path in the
DSNAME parameter of the DD statement associating the base data set/alternate
index pair with your PL/I file. Before using an alternate index, you should be
aware of the restrictions on processing; these are summarized in Table 27 on page
339.
Given a PL/I file called PL1FILE and the alternate index path called PERSALPH,
the DD statement required would be:
//PL1FILE DD DSNAME=PERSALPH,DISP=OLD
VSAM organization
PL/I provides support for three types of VSAM data sets:
v Key-sequenced data sets (KSDS)
v Entry-sequenced data sets (ESDS)
v Relative record data sets (RRDS).
These correspond roughly to PL/I indexed, consecutive, and regional data set
organizations, respectively. They are all ordered, and they can all have keys
associated with their records. Both sequential and keyed access are possible with
all three types.
Although only key-sequenced data sets have keys as part of their logical records,
keyed access is also possible for entry-sequenced data sets (using relative-byte
addresses) and relative record data sets (using relative record numbers).
All VSAM data sets are held on direct access storage devices, and a virtual storage
operating system is required to use them.
The physical organization of VSAM data sets differs from those used by other
access methods. VSAM does not use the concept of blocking, and, except for
relative record data sets, records need not be of a fixed length. In data sets with
VSAM organization, the data items are arranged in control intervals, which are in
turn arranged in control areas. For processing purposes, the data items within a
control interval are arranged in logical records. A control interval can contain one
or more logical records, and a logical record can span two or more control
intervals. Concern about blocking factors and record length is largely removed by
VSAM, although records cannot exceed the maximum specified size. VSAM allows
access to the control intervals, but this type of access is not supported by PL/I.
VSAM data sets can have two types of indexesprime and alternate. A prime index
is the index to a KSDS that is established when you define a data set; it always
exists and can be the only index for a KSDS. You can have one or more alternate
indexes on a KSDS or an ESDS. Defining an alternate index for an ESDS enables you
to treat the ESDS, in general, as a KSDS. An alternate index on a KSDS enables a
field in the logical record different from that in the prime index to be used as the
key field. Alternate indexes can be either nonunique, in which duplicate keys are
allowed, or unique, in which they are not. The prime index can never have
duplicate keys.
Any change in a data set that has alternate indexes must be reflected in all the
indexes if they are to remain useful. This activity is known as index upgrade, and is
done by VSAM for any index in the index upgrade set of the data set. (For a KSDS,
the prime index is always a member of the index upgrade set.) However, you must
avoid making changes in the data set that would cause duplicate keys in the prime
index or in a unique alternate index.
Before using a VSAM data set for the first time, you need to define it to the system
with the DEFINE command of Access Method Services, which you can use to
completely define the type, structure, and required space of the data set. This
command also defines the data set's indexes (together with their key lengths and
locations) and the index upgrade set if the data set is a KSDS or has one or more
alternate indexes. A VSAM data set is thus created by Access Method Services.
The operation of writing the initial data into a newly created VSAM data set is
referred to as loading in this publication.
Use the three different types of data sets according to the following purposes:
v Use entry-sequenced data sets for data that you primarily access in the order in
which it was created (or the reverse order).
v Use key-sequenced data sets when you normally access records through keys
within the records (for example, a stock-control file where the part number is
used to access a record).
v Use relative record data sets for data in which each item has a particular number,
and you normally access the relevant record by that number (for example, a
telephone system with a record associated with each number).
You can access records in all types of VSAM data sets either directly by means of a
key, or sequentially (backward or forward). You can also use a combination of the
two ways: Select a starting point with a key and then read forward or backward
from that point.
You can create alternate indexes for key-sequenced and entry-sequenced data sets.
You can then access your data in many sequences or by one of many keys. For
example, you could take a data set held or indexed in order of employee number
and index it by name in an alternate index. Then you could access it in alphabetic
order, in reverse alphabetic order, or directly using the name as a key. You could
also access it in the same kind of combinations by employee number.
Table 26 on page 336 shows how the same data could be held in the three different
types of VSAM data sets and illustrates their respective advantages and
disadvantages.
Table 26. Types and advantages of VSAM data sets
Data set type Method of loading Method of reading Method of updating Pros and cons
Key-Sequenced Sequentially in order
or prime index which
must be unique
KEYED by
specifying key of
record in prime
index
SEQUENTIAL
backward or forward
in order of any index
Positioning by key
followed by
sequential reading
either backward or
forward
KEYED specifying a
unique key in any
index
SEQUENTIAL
following positioning
by unique key
Record deletion
allowed
Record insertion
allowed
Advantages: Complete
access and updating
Disadvantages:
Records must be in
order of prime index
before loading
Uses: For uses where
access will be related
to key
Entry-Sequenced Sequentially (forward
only)
The RBA of each
record can be
obtained and used as
a key
SEQUENTIAL
backward or forward
KEYED using RBA
Positioning by key
followed by
sequential either
backward or forward
New records at end
only
Existing records
cannot have length
changed
Record deletion not
allowed
Advantages: Simple
fast creation
No requirement for a
unique index
Disadvantages:
Limited updating
facilities
Uses: For uses where
data will primarily be
accessed sequentially
Relative Record Sequentially starting
from slot 1
KEYED specifying
number of slot
Positioning by key
followed by
sequential writes
KEYED specifying
numbers as key
Sequential forward
or backward
omitting empty
records
Sequentially starting
at a specified slot and
continuing with next
slot
Keyed specifying
numbers as key
Record deletion
allowed
Record insertion into
empty slots allowed
Advantages: Speedy
access to record by
number
Disadvantages:
Structure tied to
numbering sequences
Fixed length records
Uses: For use where
records will be
accessed by number
Keys for VSAM data sets
All VSAM data sets can have keys associated with their records. For
key-sequenced data sets, and for entry-sequenced data sets accessed via an alternate
index, the key is a defined field within the logical record. For entry-sequenced data
sets, the key is the relative byte address (RBA) of the record. For relative-record data
sets, the key is a relative record number.
Keys for indexed VSAM data sets
Keys for key-sequenced data sets and for entry-sequenced data sets accessed via an
alternate index are part of the logical records recorded on the data set. You define
the length and location of the keys when you create the data set.
The ways you can reference the keys in the KEY, KEYFROM, and KEYTO options
are as described under KEY(expression) Option, KEYFROM(expression)
Option, and KEYTO(reference) Option in Chapter 12 of the PL/I Language
Reference.
Relative byte addresses (RBA)
Relative byte addresses allow you to use keyed access on an ESDS associated with
a KEYED SEQUENTIAL file. The RBAs, or keys, are character strings of length 4,
and their values are defined by VSAM. You cannot construct or manipulate RBAs
in PL/I; you can, however, compare their values in order to determine the relative
positions of records within the data set. RBAs are not normally printable.
You can obtain the RBA for a record by using the KEYTO option, either on a
WRITE statement when you are loading or extending the data set, or on a READ
statement when the data set is being read. You can subsequently use an RBA
obtained in either of these ways in the KEY option of a READ or REWRITE
statement.
Do not use an RBA in the KEYFROM option of a WRITE statement.
VSAM allows use of the relative byte address as a key to a KSDS, but this use is
not supported by PL/I.
Relative record numbers
Records in an RRDS are identified by a relative record number that starts at 1 and
is incremented by 1 for each succeeding record. You can use these relative record
numbers as keys for keyed access to the data set.
Keys used as relative record numbers are character strings of length 8. The
character value of a source key you use in the KEY or KEYFROM option must
represent an unsigned integer. If the source key is not 8 characters long, it is
truncated or padded with blanks (interpreted as zeros) on the left. The value
returned by the KEYTO option is a character string of length 8, with leading zeros
suppressed.
Choosing a data set type
When planning your program, the first decision to be made is which type of data
set to use. There are three types of VSAM data sets and five types of non-VSAM
data sets available to you. VSAM data sets can provide all the function of the other
types of data sets, plus additional function available only in VSAM. VSAM can
usually match other data set types in performance, and often improve upon it.
However, VSAM is more subject to performance degradation through misuse of
function.
The comparison of all eight types of data sets given in Table 15 on page 282 is
helpful; however, many factors in the choice of data set type for a large installation
are beyond the scope of this book.
When choosing between the VSAM data set types, you should base your choice on
the most common sequence in which you will require your data. The following is a
suggested procedure that you can use to help ensure a combination of data sets
and indexes that provide the function you require.
1. Determine the type of data and how it will be accessed.
a. Primarily sequentially favors ESDS.
b. Primarily by key favors KSDS.
c. Primarily by number favors RRDS.
2. Determine how you will load the data set. Note that you must load a KSDS in
key sequence; thus an ESDS with an alternate index path can be a more practical
alternative for some applications.
3. Determine whether you require access through an alternate index path. These
are only supported on KSDS and ESDS. If you require an alternate index path,
determine whether the alternate index will have unique or nonunique keys. Use
of nonunique keys can limit key processing. However, it might also be
impractical to assume that you will use unique keys for all future records; if
you attempt to insert a record with a nonunique key in an index that you have
created for unique keys, it will cause an error.
4. When you have determined the data sets and paths that you require, ensure
that the operations you have in mind are supported. Figure 35 might be
helpful.
Do not try to access a dummy VSAM data set, because you will receive an error
message indicating that you have an undefined file.
Table 28 on page 344, Table 29 on page 347, and Table 30 on page 360 show the
statements allowed for entry-sequenced data sets, indexed data sets, and relative
record data sets, respectively.
SEQUENTIAL KEYED SEQUENTIAL DIRECT
INPUT ESDS ESDS KSDS
KSDS KSDS RRDS
RRDS RRDS Path(U)
Path(N) Path(N)
Path(U) Path(U)
OUTPUT ESDS ESDS KSDS
RRDS KSDS RRDS
RRDS Path(U)
UPDATE ESDS ESDS KSDS
KSDS KSDS RRDS
RRDS RRDS Path(U)
Path(N) Path(N)
Path(U) Path(U)
Key: ESDS Entry-sequenced data set
KSDS Key-sequenced data set
RRDS Relative record data set
Path(N) Alternate index path with nonunique keys
Path(U) Alternate index path with unique keys
You can combine the attributes on the left with those at
the top of the figure for the data sets and paths shown.
For example, only an ESDS and an RRDS can be SEQUENTIAL OUTPUT.
PL/I does not support dummy VSAM data sets.
Figure 35. VSAM data sets and allowed file attributes
Table 27. Processing Allowed on Alternate Index Paths
Base Cluster Type Alternate Index Key Type Processing Restrictions
KSDS Unique key
Nonunique key
As normal KSDS
Limited keyed access
May not modify key of
access.
access.
ESDS Unique key
Nonunique key
As KSDS
Limited keyed access
No deletion.
access.
No deletion.
access.
Defining files for VSAM data sets
You define a sequential VSAM data set by using a file declaration with the
following attributes:
SEQUENTIAL
BUFFERED
[KEYED]
You define a direct VSAM data set by using a file declaration with the following
attributes:
DIRECT
[KEYED]
Table 14 on page 275 shows the default attributes. The file attributes are described
in the PL/I Language Reference. Options of the ENVIRONMENT attribute are
discussed below.
Some combinations of the file attributes INPUT or OUTPUT or UPDATE and
DIRECT or SEQUENTIAL or KEYED SEQUENTIAL are allowed only for certain
types of VSAM data sets. Figure 35 on page 338 shows the compatible
combinations.
Many of the options of the ENVIRONMENT attribute affecting data set structure
are not needed for VSAM data sets. If you specify them, they are either ignored or
are used for checking purposes. If those that are checked conflict with the values
defined for the data set, the UNDEFINEDFILE condition is raised when an attempt
is made to open the file.
The ENVIRONMENT options applicable to VSAM data sets are:
BKWD
BUFND (n)
BUFNI (n)
BUFSP (n)
GENKEY
PASSWORD (password-specification)
REUSE
SCALARVARYING
SKIP
VSAM
GENKEY and SCALARVARYING options have the same effect as they do when
you use them for non-VSAM data sets. Note that under VSAM RLS, options
BUFND, BUFNI and BUFSP are ignored.
The options that are checked for a VSAM data set are RECSIZE and, for a
key-sequenced data set, KEYLENGTH and KEYLOC. Table 14 on page 275 shows
which options are ignored for VSAM. Table 14 on page 275 also shows the required
and default options.
For VSAM data sets, you specify the maximum and average lengths of the records
to the Access Method Services utility when you define the data set. If you include
the RECSIZE option in the file declaration for checking purposes, specify the
maximum record size. If you specify RECSIZE and it conflicts with the values
defined for the data set, the UNDEFINEDFILE condition is raised.
BKWD option
Use the BKWD option to specify backward processing for a SEQUENTIAL INPUT
or SEQUENTIAL UPDATE file associated with a VSAM data set.
BKWD
Sequential reads (that is, reads without the KEY option) retrieve the previous
record in sequence. For indexed data sets, the previous record is, in general, the
record with the next lower key. However, if you are accessing the data set via a
nonunique alternate index, records with the same key are recovered in their normal
sequence. For example, if the records are:
A B C1 C2 C3 D E
where C1, C2, and C3 have the same key, they are recovered in the sequence:
E D C1 C2 C3 B A
When a file with the BKWD option is opened, the data set is positioned at the last
record. ENDFILE is raised in the normal way when the start of the data set is
reached.
Do not specify the BKWD option with either the REUSE option or the GENKEY
option. Also, the WRITE statement is not allowed for files declared with the BKWD
option.
BUFND option
Use the BUFND option to specify the number of data buffers required for a VSAM
data set.
BUFND (n)
n specifies an integer, or a variable with attributes FIXED BINARY(31) STATIC.
Multiple data buffers help performance when the file has the SEQUENTIAL
attribute and you are processing long group of contiguous records sequentially.
BUFNI option
Use the BUFNI option to specify the number of index buffers required for a VSAM
key-sequence data set.
BUFNI (n)
Multiple index buffers help performance when the file has the KEYED attribute.
Specify at least as many index buffers as there are levels in the index.
BUFSP option
Use the BUFSP option to specify, in bytes, the total buffer space required for a
VSAM data set (for both the data and index components).
BUFSP (n)
It is usually preferable to specify BUFNI and BUFND options rather than BUFSP.
GENKEY option
For the description of this option, see GENKEY option key classification on
page 280.
PASSWORD option
When you define a VSAM data set to the system (using the DEFINE command of
Access Method Services), you can associate READ and UPDATE passwords with it.
From that point on, you must include the appropriate password in the declaration
of any PL/I file that you use to access the data set.
PASSWORD ( password-specification )
password-specification
is a character constant or character variable that specifies the password for the
type of access your program requires. If you specify a constant, it must not
contain a repetition factor; if you specify a variable, it must be level-1, element,
static, and unsubscripted.
The character string is padded or truncated to 8 characters and passed to VSAM
for inspection. If the password is incorrect, the system operator is given a number
of chances to specify the correct password. You specify the number of chances to
be allowed when you define the data set. After this number of unsuccessful tries,
the UNDEFINEDFILE condition is raised.
REUSE option
Use the REUSE option to specify that an OUTPUT file associated with a VSAM
data set is to be used as a work file.
REUSE
The data set is treated as an empty data set each time the file is opened. Any
secondary allocations for the data set are released, and the data set is treated
exactly as if it were being opened for the first time.
Do not associate a file that has the REUSE option with a data set that has alternate
indexes or the BKWD option, and do not open it for INPUT or UPDATE.
The REUSE option takes effect only if you specify REUSE in the Access Method
Services DEFINE CLUSTER command.
SKIP option
Use the SKIP option of the ENVIRONMENT attribute to specify that the VSAM
OPTCD "SKP" is to be used whenever possible. It is applicable to key-sequenced
data sets that you access by means of a KEYED SEQUENTIAL INPUT or UPDATE
file.
SKIP
You should specify this option for the file if your program accesses individual
records scattered throughout the data set, but does so primarily in ascending key
order.
Omit this option if your program reads large numbers of records sequentially
without the use of the KEY option, or if it inserts large numbers of records at
specific points in the data set (mass sequential insert).
It is never an error to specify (or omit) the SKIP option; its effect on performance is
significant only in the circumstances described.
VSAM option
You must specify the VSAM option for VSAM data sets.
VSAM
Performance options
You can specify the buffer options in the AMP parameter of the DD statement;
they are explained in your Access Method Services manual.
Defining Files for Alternate Index Paths
VSAM allows you to define alternate indexes on key sequenced and entry
sequenced data sets. This enables you to access key sequenced data sets in a
number of ways other than from the prime index. This also allows you to index
and access entry sequenced data sets by key or sequentially in order of the keys.
Consequently, data created in one form can be accessed in a large number of
different ways. For example, an employee file might be indexed by personnel
number, by name, and also by department number.
When an alternate index has been built, you actually access the data set through a
third object known as an alternate index path that acts as a connection between the
alternate index and the data set.
Two types of alternate indexes are allowedunique key and nonunique key. For a
unique key alternate index, each record must have a different alternate key. For a
nonunique key alternate index, any number of records can have the same alternate
key. In the example suggested above, the alternate index using the names could be
a unique key alternate index (provided each person had a different name). The
alternate index using the department number would be a nonunique key alternate
index because more than one person would be in each department.
In most respects, you can treat a data set accessed through a unique key alternate
index path like a KSDS accessed through its prime index. You can access the
records by key or sequentially, you can update records, and you can add new
records. If the data set is a KSDS, you can delete records, and alter the length of
updated records. Restrictions and allowed processing are shown in Table 27 on
page 339. When you add or delete records, all indexes associated with the data set
are by default altered to reflect the new situation.
In data sets accessed through a nonunique key alternate index path, the record
accessed is determined by the key and the sequence. The key can be used to
establish positioning so that sequential access can follow. The use of the key
accesses the first record with that key. When the data set is read backwards, only
the order of the keys is reversed. The order of the records with the same key
remains the same whichever way the data set is read.
Defining VSAM data sets
Use the DEFINE CLUSTER command of Access Method Services to define and
catalog VSAM data sets. To use the DEFINE command, you need to know:
v The name and password of the master catalog if the master catalog is password
protected
v The name and password of the VSAM private catalog you are using if you are
not using the master catalog
v Whether VSAM space for your data set is available
v The type of VSAM data set you are going to create
v The volume on which your data set is to be placed
v The average and maximum record size in your data set
v The position and length of the key for an indexed data set
v The space to be allocated for your data set
v How to code the DEFINE command
v How to use the Access Method Services program.
When you have the information, you are in a position to code the DEFINE
command and then define and catalog the data set using Access Method Services.
Entry-sequenced data sets
The statements and options allowed for files associated with an ESDS are shown in
Table 28.
Table 28. Statements and options allowed for loading and accessing VSAM entry-sequenced
data sets
File
declaration
1
Valid statements, with
Other options you can
also include
SEQUENTIAL OUTPUT
BUFFERED
FROM(reference);
KEYTO(reference)
SEQUENTIAL INPUT
BUFFERED
INTO(reference);
READ FILE(file-reference);
KEYTO(reference) or
KEY(expression)
3
KEYTO(reference) or
KEY(expression)
3
IGNORE(expression)
SEQUENTIAL UPDATE
BUFFERED
INTO(reference);
2
FROM(reference);
KEYTO(reference) or
KEY(expression)
3
KEYTO(reference) or
KEY(expression)
3
IGNORE(expression)
KEYTO(reference)
FROM(reference)
and/or
KEY(expression)
3
Notes:
1. The complete file declaration would include the attributes FILE, RECORD, and
ENVIRONMENT; if you use either of the options KEY or KEYTO, it must also include
the attribute KEYED.
FILE(file-reference) IGNORE (1);.
3. The expression used in the KEY option must be a relative byte address, previously
obtained by means of the KEYTO option.
Loading an ESDS
When an ESDS is being loaded, the associated file must be opened for
SEQUENTIAL OUTPUT. The records are retained in the order in which they are
presented.
You can use the KEYTO option to obtain the relative byte address of each record as
it is written. You can subsequently use these keys to achieve keyed access to the
data set.
Using a SEQUENTIAL file to access an ESDS
You can open a SEQUENTIAL file that is used to access an ESDS with either the
INPUT or the UPDATE attribute. If you use either of the options KEY or KEYTO,
the file must also have the KEYED attribute.
Sequential access is in the order that the records were originally loaded into the
data set. You can use the KEYTO option on the READ statements to recover the
RBAs of the records that are read. If you use the KEY option, the record that is
recovered is the one with the RBA you specify. Subsequent sequential access
continues from the new position in the data set.
For an UPDATE file, the WRITE statement adds a new record at the end of the
data set. With a REWRITE statement, the record rewritten is the one with the
specified RBA if you use the KEY option; otherwise, it is the record accessed on the
previous READ. You must not attempt to change the length of the record that is
being replaced with a REWRITE statement.
The DELETE statement is not allowed for entry-sequenced data sets.
Defining and loading an ESDS
In Figure 36 on page 346, the data set is defined with the DEFINE CLUSTER
command and given the name PLIVSAM.AJC1.BASE. The NONINDEXED
keyword causes an ESDS to be defined.
The PL/I program writes the data set using a SEQUENTIAL OUTPUT file and a
WRITE FROM statement. The DD statement for the file contains the DSNAME of
the data set given in the NAME parameter of the DEFINE CLUSTER command.
The RBA of the records could have been obtained during the writing for
subsequent use as keys in a KEYED file. To do this, a suitable variable would have
to be declared to hold the key and the WRITE...KEYTO statement used. For
example:
DCL CHARS CHAR(4);
WRITE FILE(FAMFILE) FROM (STRING)
KEYTO(CHARS);
Note that the keys would not normally be printable, but could be retained for
subsequent use.
The cataloged procedure IBMZCBG is used. Because the same program (in
Figure 36 on page 346) can be used for adding records to the data set, it is retained
in a library. This procedure is shown in the next example.
Updating an ESDS
Figure 37 shows the addition of a new record on the end of an ESDS. This is done
by executing again the program shown in Figure 36. A SEQUENTIAL OUTPUT file
is used and the data set associated with it by use of the DSNAME parameter
specifying the name PLIVSAM.AJC1.BASE specified in the DEFINE command
shown in Figure 36.
//OPT9#7 JOB
//STEP1 EXEC PGM=IDCAMS,REGION=512K
//SYSIN DD *
DEFINE CLUSTER -
(NAME(PLIVSAM.AJC1.BASE) -
VOLUMES(nnnnnn) -
NONINDEXED -
RECORDSIZE(80 80) -
TRACKS(2 2))
/*
//STEP2 EXEC IBMZCLG
//PLI.SYSIN DD *
CREATE: PROC OPTIONS(MAIN);
DCL
FAMFILE FILE SEQUENTIAL OUTPUT ENV(VSAM),
IN FILE RECORD INPUT,
STRING CHAR(80),
ON ENDFILE(IN) EOF='1'B;
READ FILE(IN) INTO (STRING);
DO I=1 BY 1 WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (STRING) (A);
WRITE FILE(FAMFILE) FROM (STRING);
END;
PUT SKIP EDIT(I-1,' RECORDS PROCESSED')(A);
END;
/*
//LKED.SYSLMOD DD DSN=HPU8.MYDS(PGMA),DISP=(NEW,CATLG),
// UNIT=SYSDA,SPACE=(CYL,(1,1,1))
//GO.FAMFILE DD DSNAME=PLIVSAM.AJC1.BASE,DISP=OLD
//GO.IN DD *
FRED 69 M
ANDY 70 M
SUZAN 72 F
/*
Figure 36. Defining and loading an entry-sequenced data set (ESDS)
You can rewrite existing records in an ESDS, provided that the length of the record
is not changed. You can use a SEQUENTIAL or KEYED SEQUENTIAL update file
to do this. If you use keys, they can be the RBAs or keys of an alternate index path.
Delete is not allowed for ESDS.
Key-sequenced and indexed entry-sequenced data sets
The statements and options allowed for indexed VSAM data sets are shown in
Table 29. An indexed data set can be a KSDS with its prime index, or either a
KSDS or an ESDS with an alternate index Except where otherwise stated, the
following description applies to all indexed VSAM data sets.
Table 29. Statements and options allowed for loading and accessing VSAM indexed data
sets
File
declaration
1
also include
SEQUENTIAL OUTPUT
BUFFERED
FROM(reference)
FILE(file-reference)
SEQUENTIAL INPUT
BUFFERED
INTO(reference);
2
KEY(expression) or
KEYTO(reference)
KEY(expression) or
KEYTO(reference)
IGNORE(expression)
//OPT9#8 JOB
//STEP1 EXEC PGM=PGMA
//STEPLIB DD DSN=HPU8.MYDS(PGMA),DISP=(OLD,KEEP)
// DD DSN=CEE.SCEERUN,DISP=SHR
//FAMFILE DD DSN=PLIVSAM.AJC1.BASE,DISP=SHR
//IN DD *
JANE 75 F
//
Figure 37. Updating an ESDS
sets (continued)
File
declaration
1
also include
SEQUENTIAL UPDATE
BUFFERED
INTO(reference);
2
FROM(reference)
KEY(expression) or
KEYTO(reference)
KEY(expression) or
KEYTO(reference)
IGNORE(expression)
FROM(reference) and/or
KEY(expression)
KEY(expression)
DIRECT
BUFFERED
INTO(reference)
KEY(expression);
KEY(expression);
DIRECT OUTPUT
BUFFERED
FROM(reference)
DIRECT
BUFFERED
INTO(reference)
KEY(expression);
KEY(expression);
FROM(reference)
KEY(expression);
KEY(expression);
FROM(reference)
sets (continued)
File
declaration
1
also include
Notes:
1. The complete file declaration would include the attributes FILE and RECORD. If you use
any of the options KEY, KEYFROM, or KEYTO, you must also include the attribute
KEYED in the declaration.
3. Do not associate a SEQUENTIAL OUTPUT file with a data set accessed via an alternate
index.
4. Do not associate a DIRECT file with a data set accessed via a nonunique alternate index.
5. DELETE statements are not allowed for a file associated with an ESDS accessed via an
alternate index.
Loading a KSDS or indexed ESDS: When a KSDS is being loaded, you must
open the associated file for KEYED SEQUENTIAL OUTPUT. You must present the
records in ascending key order, and you must use the KEYFROM option. Note that
you must use the prime index for loading the data set; you cannot load a VSAM
data set via an alternate index.
If a KSDS already contains some records, and you open the associated file with the
SEQUENTIAL and OUTPUT attributes, you can add only records at the end of the
data set. The rules given in the previous paragraph apply; in particular, the first
record you present must have a key greater than the highest key present on the
data set.
Figure 38 on page 350 shows the DEFINE command used to define a KSDS. The
data set is given the name PLIVSAM.AJC2.BASE and defined as a KSDS because of
the use of the INDEXED operand. The position of the keys within the record is
defined in the KEYS operand.
Within the PL/I program, a KEYED SEQUENTIAL OUTPUT file is used with a
WRITE...FROM...KEYFROM statement. The data is presented in ascending key
order. A KSDS must be loaded in this manner.
The file is associated with the data set by a DD statement which uses the name
given in the DEFINE command as the DSNAME parameter.
Using a SEQUENTIAL file to access a KSDS or indexed ESDS: You can open a
SEQUENTIAL file that is used to access a KSDS with either the INPUT or the
UPDATE attribute.
For READ statements without the KEY option, the records are recovered in
ascending key order (or in descending key order if the BKWD option is used). You
can obtain the key of a record recovered in this way by means of the KEYTO
option.
//OPT9#12 JOB
// EXEC PGM=IDCAMS,REGION=512K
//SYSIN DD *
DEFINE CLUSTER -
VOLUMES(nnnnnn) -
INDEXED -
TRACKS(3 1) -
KEYS(20 0) -
RECORDSIZE(23 80))
/*
// EXEC IBMZCBG
//PLI.SYSIN DD *
TELNOS: PROC OPTIONS(MAIN);
DCL DIREC FILE RECORD SEQUENTIAL OUTPUT KEYED ENV(VSAM),
CARD CHAR(80),
NAME CHAR(20) DEF CARD POS(1),
NUMBER CHAR(3) DEF CARD POS(21),
OUTREC CHAR(23) DEF CARD POS(1),
OPEN FILE(DIREC) OUTPUT;
GET FILE(SYSIN) EDIT(CARD)(A(80));
DO WHILE (EOF);
WRITE FILE(DIREC) FROM(OUTREC) KEYFROM(NAME);
END;
CLOSE FILE(DIREC);
END TELNOS;
/*
//GO.DIREC DD DSNAME=PLIVSAM.AJC2.BASE,DISP=OLD
//GO.SYSIN DD *
ACTION,G. 162
BAKER,R. 152
BRAMLEY,O.H. 248
CHEESEMAN,D. 141
CORY,G. 336
ELLIOTT,D. 875
FIGGINS,S. 413
HARVEY,C.D.W. 205
HASTINGS,G.M. 391
KENDALL,J.G. 294
LANCASTER,W.R. 624
MILES,R. 233
NEWMAN,M.W. 450
PITT,W.H. 515
ROLF,D.E. 114
SHEERS,C.D. 241
SUTCLIFFE,M. 472
TAYLOR,G.C. 407
WILTON,L.W. 404
WINSTONE,E.M. 307
//
Figure 38. Defining and loading a key-sequenced data set (KSDS)
If you use the KEY option, the record recovered by a READ statement is the one
with the specified key. Such a READ statement positions the data set at the
specified record; subsequent sequential reads will recover the following records in
sequence.
WRITE statements with the KEYFROM option are allowed for KEYED
SEQUENTIAL UPDATE files. You can make insertions anywhere in the data set,
without respect to the position of any previous access. If you are accessing the data
set via a unique index, the KEY condition is raised if an attempt is made to insert a
record with the same key as a record that already exists on the data set. For a
nonunique index, subsequent retrieval of records with the same key is in the order
that they were added to the data set.
REWRITE statements with or without the KEY option are allowed for UPDATE
files. If you use the KEY option, the record that is rewritten is the first record with
the specified key; otherwise, it is the record that was accessed by the previous
READ statement. When you rewrite a record using an alternate index, do not
change the prime key of the record.
Using a DIRECT file to access a KSDS or indexed ESDS: You can open a
DIRECT file that is used to access an indexed VSAM data set with the INPUT,
OUTPUT, or UPDATE attribute. Do not use a DIRECT file to access the data set via
a nonunique index.
If you use a DIRECT OUTPUT file to add records to the data set, and if an attempt
is made to insert a record with the same key as a record that already exists, the
KEY condition is raised.
If you use a DIRECT INPUT or DIRECT UPDATE file, you can read, write, rewrite,
or delete records in the same way as for a KEYED SEQUENTIAL file.
Figure 39 on page 352 shows one method by which a KSDS can be updated using
the prime index.
//OPT9#13 JOB
//PLI.SYSIN DD *
DIRUPDT: PROC OPTIONS(MAIN);
DCL DIREC FILE RECORD KEYED ENV(VSAM),
ONCODE BUILTIN,
OUTREC CHAR(23),
NUMBER CHAR(3) DEF OUTREC POS(21),
NAME CHAR(20) DEF OUTREC,
CODE CHAR(1),
ON KEY(DIREC) BEGIN;
IF ONCODE=51 THEN PUT FILE(SYSPRINT) SKIP EDIT
('NOT FOUND: ',NAME)(A(15),A);
('DUPLICATE: ',NAME)(A(15),A);
END;
OPEN FILE(DIREC) DIRECT UPDATE;
GET FILE(SYSIN) EDIT (NAME,NUMBER,CODE)
(COLUMN(1),A(20),A(3),A(1));
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (' ',NAME,'#',NUMBER,' ',CODE)
(A(1),A(20),A(1),A(3),A(1),A(1));
SELECT (CODE);
WHEN('A') WRITE FILE(DIREC) FROM(OUTREC) KEYFROM(NAME);
WHEN('C') REWRITE FILE(DIREC) FROM(OUTREC) KEY(NAME);
WHEN('D') DELETE FILE(DIREC) KEY(NAME);
OTHERWISE PUT FILE(SYSPRINT) SKIP EDIT
('INVALID CODE: ',NAME) (A(15),A);
END;
GET FILE(SYSIN) EDIT (NAME,NUMBER,CODE)
(COLUMN(1),A(20),A(3),A(1));
END;
CLOSE FILE(DIREC);
OPEN FILE(DIREC) SEQUENTIAL INPUT;
EOF='0'B;
ON ENDFILE(DIREC) EOF='1'B;
READ FILE(DIREC) INTO(OUTREC);
DO WHILE(EOF);
PUT FILE(SYSPRINT) SKIP EDIT(OUTREC)(A);
READ FILE(DIREC) INTO(OUTREC);
END;
CLOSE FILE(DIREC);
END DIRUPDT;
Figure 39. Updating a KSDS (Part 1 of 2)
A DIRECT update file is used and the data is altered according to a code that is
passed in the records in the file SYSIN:
A Add a new record
C Change the number of an existing name
D Delete a record
At the label NEXT, the name, number, and code are read in and action taken
according to the value of the code. A KEY ON-unit is used to handle any incorrect
keys. When the updating is finished (at the label PRINT), the file DIREC is closed
and reopened with the attributes SEQUENTIAL INPUT. The file is then read
sequentially and printed.
The file is associated with the data set by a DD statement that uses the DSNAME
PLIVSAM.AJC2.BASE defined in the Access Method Services DEFINE CLUSTER
command in Figure 38 on page 350.
Methods of updating a KSDS: There are a number of methods of updating a KSDS.
The method shown using a DIRECT file is suitable for the data as it is shown in
the example. For mass sequential insertion, use a KEYED SEQUENTIAL UPDATE
file. This gives faster performance because the data is written onto the data set
only when strictly necessary and not after every write statement, and because the
balance of free space within the data set is retained.
Statements to achieve effective mass sequential insertion are:
DCL DIREC KEYED SEQUENTIAL UPDATE
ENV(VSAM);
WRITE FILE(DIREC) FROM(OUTREC)
KEYFROM(NAME);
The PL/I input/output routines detect that the keys are in sequence and make the
correct requests to VSAM. If the keys are not in sequence, this too is detected and
no error occurs, although the performance advantage is lost.
Alternate Indexes for KSDSs or Indexed ESDSs
Alternate indexes allow you to access KSDSs or indexed ESDSs in various ways,
using either unique or nonunique keys.
/*
//GO.DIREC DD DSNAME=PLIVSAM.AJC2.BASE,DISP=OLD
//GO.SYSIN DD *
NEWMAN,M.W. 516C
GOODFELLOW,D.T. 889A
MILES,R. D
HARVEY,C.D.W. 209A
BARTLETT,S.G. 183A
CORY,G. D
READ,K.M. 001A
PITT,W.H.
ROLF,D.F. D
ELLIOTT,D. 291C
HASTINGS,G.M. D
BRAMLEY,O.H. 439C
/*
Figure 39. Updating a KSDS (Part 2 of 2)
Unique Key Alternate Index Path: Figure 40 shows the creation of a unique key
alternate index path for the ESDS defined and loaded in Figure 36 on page 346.
Using this path, the data set is indexed by the name of the child in the first 15
bytes of the record.
Three Access Method Services commands are used. These are:
DEFINE ALTERNATEINDEX: defines the alternate index as a data set to VSAM.
BLDINDEX: places the pointers to the relevant records in the alternate index.
DEFINE PATH: defines an entity that can be associated with a PL/I file in a DD
statement.
DD statements are required for the INFILE and OUTFILE operands of BLDINDEX
and for the sort files. Care should be taken that the correct names are specified at
the various points.
Nonunique Key Alternate Index Path: Figure 41 on page 355 shows the creation
of a nonunique key alternate index path for an ESDS. The alternate index enables
the data to be selected by the sex of the children. This enables the girls or the boys
to be accessed separately and every member of each group to be accessed by use
of the key.
The three Access Method Services commands used are:
DEFINE ALTERNATEINDEX: defines the alternate index as a data set to VSAM.
BLDINDEX: places the pointers to the relevant records in the alternate index.
DEFINE PATH: defines an entity that can be associated with a PL/I file in a DD
statement.
DD statements are required for the INFILE and OUTFILE operands of BLDINDEX
and for the sort files. Care should be taken that the correct names are specified at
the various points.
//OPT9#9 JOB
//SYSIN DD *
DEFINE ALTERNATEINDEX -
(NAME(PLIVSAM.AJC1.ALPHIND) -
VOLUMES(nnnnnn) -
TRACKS(4 1) -
KEYS(15 0) -
RECORDSIZE(20 40) -
UNIQUEKEY -
RELATE(PLIVSAM.AJC1.BASE))
/*
//DD1 DD DSNAME=PLIVSAM.AJC1.BASE,DISP=SHR
//DD2 DD DSNAME=PLIVSAM.AJC1.ALPHIND,DISP=SHR
//SYSIN DD *
BLDINDEX INFILE(DD1) OUTFILE(DD2)
DEFINE PATH -
(NAME(PLIVSAM.AJC1.ALPHPATH) -
PATHENTRY(PLIVSAM.AJC1.ALPHIND))
//
Figure 40. Creating a Unique Key Alternate Index Path for an ESDS
The fact that the index has nonunique keys is specified by the use of the
NONUNIQUEKEY operand. When creating an index with nonunique keys, be
careful to ensure that the RECORDSIZE you specify is large enough. In a
nonunique alternate index, each alternate index record contains pointers to all the
records that have the associated index key. The pointer takes the form of an RBA
for an ESDS and the prime key for a KSDS. When a large number of records might
have the same key, a large record is required.
Figure 42 on page 356 shows the creation of a unique key alternate index path for
a KSDS. The data set is indexed by the telephone number, enabling the number to
be used as a key to discover the name of the person on that extension. The fact
that keys are to be unique is specified by UNIQUEKEY. Also, the data set will be
able to be listed in numerical order to show which numbers are not used. The
three Access Method Services commands used are:
DEFINE ALTERNATEINDEX: defines the data set that will hold the alternate
index data.
BLDINDEX: places the pointer to the relevant records in the alternate index.
DEFINE PATH: defines the entity that can be associated with a PL/I file in a DD
statement.
DD statements are required for the INFILE and OUTFILE of BLDINDEX and for
the sort files. Be careful not to confuse the names involved.
//OPT9#10 JOB
//SYSIN DD *
/* care must be taken with recordsize */
(NAME(PLIVSAM.AJC1.SEXIND) -
VOLUMES(nnnnnn) -
TRACKS(4 1) -
KEYS(1 37) -
RECORDSIZE(20 400) -
NONUNIQUEKEY -
/*
//DD2 DD DSNAME=PLIVSAM.AJC1.SEXIND,DISP=SHR
//SYSIN DD *
DEFINE PATH -
(NAME(PLIVSAM.AJC1.SEXPATH) -
PATHENTRY(PLIVSAM.AJC1.SEXIND))
//
Figure 41. Creating a Nonunique Key Alternate Index Path for an ESDS
When creating an alternate index with a unique key, you should ensure that no
further records could be included with the same alternate key. In practice, a unique
key alternate index would not be entirely satisfactory for a telephone directory as it
would not allow two people to have the same number. Similarly, the prime key
would prevent one person having two numbers. A solution would be to have an
ESDS with two nonunique key alternate indexes, or to restructure the data format
to allow more than one number per person and to have a nonunique key alternate
index for the numbers.
Detecting Nonunique Alternate Index Keys: If you are accessing a VSAM data
set by means of an alternate index path, the presence of nonunique keys can be
detected by means of the SAMEKEY built-in function. After each retrieval,
SAMEKEY indicates whether any further records exist with the same alternate
index key as the record just retrieved. Hence, it is possible to stop at the last of a
series of records with nonunique keys without having to read beyond the last
record. SAMEKEY (file-reference) returns '1'B if the input/output statement has
completed successfully and the accessed record is followed by another with the
same key; otherwise, it returns '0'B.
Using Alternate Indexes with ESDSs: Figure 43 on page 358 shows the use of
alternate indexes and backward reading on an ESDS. The program has four files:
BASEFLE
reads the base data set forward.
BACKFLE
reads the base data set backward.
ALPHFLE
is the alphabetic alternate index path indexing the children by name.
SEXFILE
//OPT9#14 JOB
//SYSIN DD *
(NAME(PLIVSAM.AJC2.NUMIND) -
VOLUMES(nnnnnn) -
TRACKS(4 4) -
KEYS(3 20) -
RECORDSIZE(24 48) -
UNIQUEKEY -
/*
//DD2 DD DSNAME=PLIVSAM.AJC2.NUMIND,DISP=SHR
//SYSIN DD *
DEFINE PATH -
(NAME(PLIVSAM.AJC2.NUMPATH) -
PATHENTRY(PLIVSAM.AJC2.NUMIND))
//
Figure 42. Creating a unique Key Alternate Index Path for a KSDS
is the alternate index path that corresponds to the sex of the children.
There are DD statements for all the files. They connect BASEFLE and BACKFLE to
the base data set by specifying the name of the base data set in the DSNAME
parameter, and connect ALPHFLE and SEXFILE by specifying the names of the
paths given in Figure 40 on page 354 and Figure 41 on page 355.
The program uses SEQUENTIAL files to access the data and print it first in the
normal order, then in the reverse order. At the label AGEQUERY, a DIRECT file is
used to read the data associated with an alternate index key in the unique alternate
index.
Finally, at the label SPRINT, a KEYED SEQUENTIAL file is used to print a list of
the females in the family, using the nonunique key alternate index path. The
SAMEKEY built-in function is used to read all the records with the same key. The
names of the females will be accessed in the order in which they were originally
entered. This will happen whether the file is read forward or backward. For a
nonunique key path, the BKWD option only affects the order in which the keys are
read; the order of items with the same key remains the same as it is when the file
is read forward.
Deletion: At the end of the example, the Access Method Services DELETE command
is used to delete the base data set. When this is done, the associated alternate
indexes and paths will also be deleted.
Using Alternate Indexes with KSDSs: Figure 44 on page 360 shows the use of a
path with a unique alternate index key to update a KSDS and then to access and
print it in the order of the alternate index.
The alternate index path is associated with the PL/I file by a DD statement that
specifies the name of the path (PLIVSAM.AJC2.NUMPATH, given in the DEFINE
PATH command in Figure 42 on page 356) as the DSNAME.
In the first section of the program, a DIRECT OUTPUT file is used to insert a new
record using the alternate index key. Note that any alteration made with an
alternate index must not alter the prime key or the alternate index key of access of
an existing record. Also, the alternation must not add a duplicate key in the prime
index or any unique key alternate index.
In the second section of the program (at the label PRINTIT), the data set is read in
the order of the alternate index keys using a SEQUENTIAL INPUT file. It is then
printed onto SYSPRINT.
//OPT9#15 JOB
//STEP1 EXEC IBMZCLG
//PLI.SYSIN DD *
READIT: PROC OPTIONS(MAIN);
DCL BASEFLE FILE SEQUENTIAL INPUT ENV(VSAM),
/*File to read base data set forward */
BACKFLE FILE SEQUENTIAL INPUT ENV(VSAM BKWD),
/*File to read base data set backward */
ALPHFLE FILE DIRECT INPUT ENV(VSAM),
/*File to access via unique alternate index path */
SEXFILE FILE KEYED SEQUENTIAL INPUT ENV(VSAM),
/*File to access via nonunique alternate index path */
STRING CHAR(80), /*String to be read into */
1 STRUC DEF (STRING),
2 NAME CHAR(25),
2 DATE_OF_BIRTH CHAR(2),
2 FILL CHAR(10),
2 SEX CHAR(1);
DCL NAMEHOLD CHAR(25),SAMEKEY BUILTIN;
DCL EOF BIT(1) INIT(0B);
/*Print out the family eldest first*/
ON ENDFILE(BASEFLE) EOF=1B;
PUT EDIT(FAMILY ELDEST FIRST)(A);
READ FILE(BASEFLE) INTO (STRING);
DO WHILE(EOF);
PUT SKIP EDIT(STRING)(A);
READ FILE(BASEFLE) INTO (STRING);
END;
CLOSE FILE(BASEFLE);
PUT SKIP(2);
/*Close before using data set from other file not
necessary but good practice to prevent potential
problems*/
EOF=0B;
ON ENDFILE(BACKFLE) EOF=1B;
PUT SKIP(3) EDIT(FAMILY YOUNGEST FIRST)(A);
READ FILE(BACKFLE) INTO(STRING);
DO WHILE(EOF);
READ FILE(BACKFLE) INTO (STRING);
END;
CLOSE FILE(BACKFLE);
PUT SKIP(2);
/*Print date of birth of child specified in the file
SYSIN*/
ON KEY(ALPHFLE) BEGIN;
PUT SKIP EDIT
(NAMEHOLD, NOT A MEMBER OF THE SMITH FAMILY) (A);
GO TO SPRINT;
END;
Figure 43. Alternate Index Paths and Backward Reading with an ESDS (Part 1 of 2)
AGEQUERY:
EOF=0B;
ON ENDFILE(SYSIN) EOF=1B;
GET SKIP EDIT(NAMEHOLD)(A(25));
DO WHILE(EOF);
READ FILE(ALPHFLE) INTO (STRING) KEY(NAMEHOLD);
PUT SKIP (2) EDIT(NAMEHOLD, WAS BORN IN ,
DATE_OF_BIRTH)(A,X(1),A,X(1),A);
GET SKIP EDIT(NAMEHOLD)(A(25));
END;
SPRINT:
CLOSE FILE(ALPHFLE);
PUT SKIP(1);
/*Use the alternate index to print out all the females in the
family*/
ON ENDFILE(SEXFILE) GOTO FINITO;
PUT SKIP(2) EDIT(ALL THE FEMALES)(A);
READ FILE(SEXFILE) INTO (STRING) KEY(F);
DO WHILE(SAMEKEY(SEXFILE));
READ FILE(SEXFILE) INTO (STRING);
END;
FINITO:
END;
/*
//GO.BASEFLE DD DSN=PLIVSAM.AJC1.BASE,DISP=SHR
//GO.BACKFLE DD DSN=PLIVSAM.AJC1.BASE,DISP=SHR
//GO.ALPHFLE DD DSN=PLIVSAM.AJC1.ALPHPATH,DISP=SHR
//GO.SEXFILE DD DSN=PLIVSAM.AJC1.SEXPATH,DISP=SHR
//GO.SYSIN DD *
ANDY
/*
//SYSIN DD *
DELETE -
PLIVSAM.AJC1.BASE
//
Figure 43. Alternate Index Paths and Backward Reading with an ESDS (Part 2 of 2)
Relative-record data sets
The statements and options allowed for VSAM relative-record data sets (RRDS) are
shown in Table 30.
Table 30. Statements and options allowed for loading and accessing VSAM relative-record
data sets
File
declaration
1
also include
SEQUENTIAL OUTPUT
BUFFERED
FROM(reference);
KEYFROM(expression) or
KEYTO(reference)
//OPT9#16 JOB
//STEP1 EXEC IBMZCLG,REGION.GO=256K
//PLI.SYSIN DD *
ALTER: PROC OPTIONS(MAIN);
DCL NUMFLE1 FILE RECORD DIRECT OUTPUT ENV(VSAM),
NUMFLE2 FILE RECORD SEQUENTIAL INPUT ENV(VSAM),
IN FILE RECORD,
STRING CHAR(80),
NAME CHAR(20) DEF STRING,
NUMBER CHAR(3) DEF STRING POS(21),
DATA CHAR(23) DEF STRING,
EOF BIT(1) INIT(0B);
ON KEY (NUMFLE1) BEGIN;
PUT SKIP EDIT(DUPLICATE NUMBER)(A);
END;
ON ENDFILE(IN) EOF=1B;
DO WHILE(EOF);
PUT FILE(SYSPRINT) SKIP EDIT (STRING) (A);
WRITE FILE(NUMFLE1) FROM (STRING) KEYFROM(NUMBER);
END;
CLOSE FILE(NUMFLE1);
EOF=0B;
ON ENDFILE(NUMFLE2) EOF=1B;
READ FILE(NUMFLE2) INTO (STRING);
DO WHILE(EOF);
PUT SKIP EDIT(DATA)(A);
READ FILE(NUMFLE2) INTO (STRING);
END;
PUT SKIP(3) EDIT(****SO ENDS THE PHONE DIRECTORY****)(A);
END;
/*
//GO.IN DD *
RIERA L 123
/*
//NUMFLE1 DD DSN=PLIVSAM.AJC2.NUMPATH,DISP=OLD
//NUMFLE2 DD DSN=PLIVSAM.AJC2.NUMPATH,DISP=OLD
//STEP2 EXEC PGM=IDCAMS,COND=EVEN
//SYSIN DD *
DELETE -
PLIVSAM.AJC2.BASE
//
Figure 44. Using a Unique Alternate Index Path to Access a KSDS
data sets (continued)
File
declaration
1
also include
SEQUENTIAL INPUT
BUFFERED
INTO(reference);
2
KEY(expression) or
KEYTO(reference)
KEY(expression) or
KEYTO(reference)
IGNORE(expression)
SEQUENTIAL UPDATE
BUFFERED
INTO(reference);
2
FROM(reference);
DELETE FILE(file-reference);
KEY(expression) or
KEYTO(reference)
KEY(expression) or
KEYTO(reference)
IGNORE(expression)
KEYFROM(expression) or
KEYTO(reference)
FROM(reference)
and/or
KEY(expression)
KEY(expression)
DIRECT OUTPUT
BUFFERED
FROM(reference)
DIRECT INPUT
BUFFERED
INTO(reference)
KEY(expression);
KEY(expression);
DIRECT UPDATE
BUFFERED
INTO(reference)
KEY(expression);
KEY(expression);
FROM(reference)
KEY(expression);
KEY(expression);
FROM(reference)
data sets (continued)
File
declaration
1
also include
Notes:
1. The complete file declaration would include the attributes FILE and RECORD. If you use
any of the options KEY, KEYFROM, or KEYTO, your declaration must also include the
attribute KEYED.
The UNLOCK statement for DIRECT UPDATE files is ignored if you use it for files
associated with a VSAM RRDS.
Loading an RRDS: When an RRDS is being loaded, you must open the associated
file for OUTPUT. Use either a DIRECT or a SEQUENTIAL file.
For a DIRECT OUTPUT file, each record is placed in the position specified by the
relative record number (or key) in the KEYFROM option of the WRITE statement
(see Keys for VSAM data sets on page 336).
For a SEQUENTIAL OUTPUT file, use WRITE statements with or without the
KEYFROM option. If you specify the KEYFROM option, the record is placed in the
specified slot; if you omit it, the record is placed in the slot following the current
position. There is no requirement for the records to be presented in ascending
relative record number order. If you omit the KEYFROM option, you can obtain
the relative record number of the written record by means of the KEYTO option.
If you want to load an RRDS sequentially, without use of the KEYFROM or
KEYTO options, your file is not required to have the KEYED attribute.
It is an error to attempt to load a record into a position that already contains a
record: if you use the KEYFROM option, the KEY condition is raised; if you omit
it, the ERROR condition is raised.
In Figure 45 on page 363, the data set is defined with a DEFINE CLUSTER
command and given the name PLIVSAM.AJC3.BASE. The fact that it is an RRDS is
determined by the NUMBERED keyword. In the PL/I program, it is loaded with a
DIRECT OUTPUT file and a WRITE...FROM...KEYFROM statement is used.
If the data had been in order and the keys in sequence, it would have been
possible to use a SEQUENTIAL file and write into the data set from the start. The
records would then have been placed in the next available slot and given the
appropriate number. The number of the key for each record could have been
returned using the KEYTO option.
The PL/I file is associated with the data set by the DD statement, which uses as
the DSNAME the name given in the DEFINE CLUSTER command.
Using a SEQUENTIAL file to access an RRDS: You can open a SEQUENTIAL
file that is used to access an RRDS with either the INPUT or the UPDATE
attribute. If you use any of the options KEY, KEYTO, or KEYFROM, your file must
also have the KEYED attribute.
For READ statements without the KEY option, the records are recovered in
ascending relative record number order. Any empty slots in the data set are
skipped.
//OPT9#17 JOB
//SYSIN DD *
DEFINE CLUSTER -
VOLUMES(nnnnnn) -
NUMBERED -
TRACKS(2 2) -
RECORDSIZE(20 20))
/*
//PLI.SYSIN DD *
CRR1: PROC OPTIONS(MAIN);
DCL NOS FILE RECORD OUTPUT DIRECT KEYED ENV(VSAM),
CARD CHAR(80),
NAME CHAR(20) DEF CARD,
NUMBER CHAR(2) DEF CARD POS(21),
IOFIELD CHAR(20),
ON ENDFILE (SYSIN) EOF='1'B;
OPEN FILE(NOS);
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (CARD) (A);
IOFIELD=NAME;
WRITE FILE(NOS) FROM(IOFIELD) KEYFROM(NUMBER);
END;
CLOSE FILE(NOS);
END CRR1;
/*
//GO.NOS DD DSN=PLIVSAM.AJC3.BASE,DISP=OLD
//GO.SYSIN DD *
ACTION,G. 12
BAKER,R. 13
BRAMLEY,O.H. 28
CHEESNAME,L. 11
CORY,G. 36
ELLIOTT,D. 85
FIGGINS.E.S. 43
HARVEY,C.D.W. 25
HASTINGS,G.M. 31
KENDALL,J.G. 24
LANCASTER,W.R. 64
MILES,R. 23
NEWMAN,M.W. 40
PITT,W.H. 55
ROLF,D.E. 14
SHEERS,C.D. 21
SURCLIFFE,M. 42
TAYLOR,G.C. 47
WILTON,L.W. 44
WINSTONE,E.M. 37
//
Figure 45. Defining and loading a relative-record data set (RRDS)
If you use the KEY option, the record recovered by a READ statement is the one
with the relative record number you specify. Such a READ statement positions the
data set at the specified record; subsequent sequential reads will recover the
following records in sequence.
WRITE statements with or without the KEYFROM option are allowed for KEYED
SEQUENTIAL UPDATE files. You can make insertions anywhere in the data set,
regardless of the position of any previous access. For WRITE with the KEYFROM
option, the KEY condition is raised if an attempt is made to insert a record with
the same relative record number as a record that already exists on the data set. If
you omit the KEYFROM option, an attempt is made to write the record in the next
slot, relative to the current position. The ERROR condition is raised if this slot is
not empty.
You can use the KEYTO option to recover the key of a record that is added by
means of a WRITE statement without the KEYFROM option.
REWRITE statements, with or without the KEY option, are allowed for UPDATE
files. If you use the KEY option, the record that is rewritten is the record with the
relative record number you specify; otherwise, it is the record that was accessed by
the previous READ statement.
DELETE statements, with or without the KEY option, can be used to delete records
from the dataset.
Using a DIRECT file to access an RRDS: A DIRECT file used to access an RRDS
can have the OUTPUT, INPUT, or UPDATE attribute. You can read, write, rewrite,
or delete records exactly as though a KEYED SEQUENTIAL file were used.
Figure 46 on page 365 shows an RRDS being updated. A DIRECT UPDATE file is
used and new records are written by key. There is no need to check for the records
being empty, because the empty records are not available under VSAM.
In the second half of the program, starting at the label PRINT, the updated file is
printed out. Again there is no need to check for the empty records as there is in
REGIONAL(1).
The PL/I file is associated with the data sets by a DD statement that specifies the
DSNAME PLIVSAM.AJC3.BASE, the name given in the DEFINE CLUSTER
command in Figure 46 on page 365.
At the end of the example, the DELETE command is used to delete the data set.
//* NOTE: WITH A WRITE STATEMENT AFTER THE DELETE FILE STATEMENT,
//* A DUPLICATE MESSAGE IS EXPECTED FOR CODE 'C' ITEMS
//* WHOSE NEWNO CORRESPONDS TO AN EXISTING NUMBER IN THE LIST,
//* FOR EXAMPLE, ELLIOT.
//* WITH A REWRITE STATEMENT AFTER THE DELETE FILE STATEMENT,
//* A NOT FOUND MESSAGE IS EXPECTED FOR CODE 'C' ITEMS
//* WHOSE NEWNO DOES NOT CORRESPOND TO AN EXISTING NUMBER IN
//* THE LIST, FOR EXAMPLE, NEWMAN AND BRAMLEY.
//OPT9#18 JOB
//PLI.SYSIN DD *
ACR1: PROC OPTIONS(MAIN);
DCL NOS FILE RECORD KEYED ENV(VSAM),NAME CHAR(20),
(NEWNO,OLDNO) CHAR(2),CODE CHAR(1),IOFIELD CHAR(20),
BYTE CHAR(1) DEF IOFIELD, EOF BIT(1) INIT('0'B),
ONCODE BUILTIN;
OPEN FILE(NOS) DIRECT UPDATE;
ON KEY(NOS) BEGIN;
('NOT FOUND:',NAME)(A(15),A);
('DUPLICATE:',NAME)(A(15),A);
END;
GET FILE(SYSIN) EDIT(NAME,NEWNO,OLDNO,CODE)
(COLUMN(1),A(20),A(2),A(2),A(1));
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT (' ',NAME,'#',NEWNO,OLDNO,' ',CODE)
(A(1),A(20),A(1),2(A(2)),X(5),2(A(1)));
SELECT(CODE);
WHEN('A') WRITE FILE(NOS) KEYFROM(NEWNO) FROM(NAME);
WHEN('C') DO;
DELETE FILE(NOS) KEY(OLDNO);
WRITE FILE(NOS) KEYFROM(NEWNO) FROM(NAME);
END;
WHEN('D') DELETE FILE(NOS) KEY(OLDNO);
OTHERWISE PUT FILE(SYSPRINT) SKIP EDIT
('INVALID CODE: ',NAME)(A(15),A);
END;
Figure 46. Updating an RRDS (Part 1 of 2)
Using Files Defined for non-VSAM Data Sets
Using Shared Data Sets
PL/I allows cross-region or cross-system sharing of data sets. The support for this
type of sharing is provided by VSAM. For details about the support, see the
following DFSMS manuals:
v DFSMS: Using Data Sets
v DFSMS: Access Method Services for Catalogs
GET FILE(SYSIN) EDIT(NAME,NEWNO,OLDNO,CODE)
(COLUMN(1),A(20),A(2),A(2),A(1));
END;
CLOSE FILE(NOS);
PRINT:
OPEN FILE(NOS) SEQUENTIAL INPUT;
EOF='0'B;
ON ENDFILE(NOS) EOF='1'B;
DO WHILE (EOF);
PUT FILE(SYSPRINT) SKIP EDIT(NEWNO,IOFIELD)(A(5),A);
END;
CLOSE FILE(NOS);
END ACR1;
/*
//GO.NOS DD DSN=PLIVSAM.AJC3.BASE,DISP=OLD
//GO.SYSIN DD *
NEWMAN,M.W. 5640C
GOODFELLOW,D.T. 89 A
MILES,R. 23D
HARVEY,C.D.W. 29 A
BARTLETT,S.G. 13 A
CORY,G. 36D
READ,K.M. 01 A
PITT,W.H. 55
ROLF,D.F. 14D
ELLIOTT,D. 4285C
HASTINGS,G.M. 31D
BRAMLEY,O.H. 4928C
//STEP3 EXEC PGM=IDCAMS,REGION=512K,COND=EVEN
//SYSIN DD *
DELETE -
PLIVSAM.AJC3.BASE
//
Figure 46. Updating an RRDS (Part 2 of 2)
Part 3. Improving your program
Chapter 13. Improving performance . . . . . 369
Selecting compiler options for optimal performance 369
OPTIMIZE . . . . . . . . . . . . . 369
GONUMBER . . . . . . . . . . . . 369
ARCH. . . . . . . . . . . . . . . 369
REDUCE . . . . . . . . . . . . . . 370
RULES . . . . . . . . . . . . . . 370
IBM/ANS . . . . . . . . . . . . 370
(NO)LAXCTL . . . . . . . . . . . 371
PREFIX . . . . . . . . . . . . . . 371
CONVERSION . . . . . . . . . . . 371
FIXEDOVERFLOW . . . . . . . . . 372
DEFAULT . . . . . . . . . . . . . 372
BYADDR or BYVALUE . . . . . . . . 372
(NON)CONNECTED. . . . . . . . . 373
(NO)DESCRIPTOR . . . . . . . . . 373
(NO)INLINE . . . . . . . . . . . 374
LINKAGE . . . . . . . . . . . . 374
(RE)ORDER . . . . . . . . . . . . 374
NOOVERLAP . . . . . . . . . . . 375
RETURNS(BYVALUE) or
RETURNS(BYADDR) . . . . . . . . . 375
Summary of compiler options that improve
performance. . . . . . . . . . . . . 375
Coding for better performance . . . . . . . 375
DATA-directed input and output . . . . . . 375
Input-only parameters . . . . . . . . . 376
GOTO statements . . . . . . . . . . . 376
String assignments . . . . . . . . . . 376
Loop control variables . . . . . . . . . 377
PACKAGEs versus nested PROCEDUREs . . . 377
Example with nested procedures . . . . . 378
Example with packaged procedures . . . . 378
REDUCIBLE Functions . . . . . . . . . 378
DESCLOCATOR or DESCLIST . . . . . . 379
DEFINED versus UNION . . . . . . . . 379
Named constants versus static variables . . . 379
Example with optimal code but no
meaningful names. . . . . . . . . . 380
Example with meaningful names but not
optimal code . . . . . . . . . . . 380
Example with optimal code AND meaningful
names . . . . . . . . . . . . . . 380
Avoiding calls to library routines . . . . . . 381
Preloading library routines . . . . . . . . 381
Chapter 13. Improving performance
Many considerations for improving the speed of your program are independent of
the compiler that you use and the platform on which it runs. This chapter,
however, identifies those considerations that are unique to the PL/I compiler and
the code it generates.
Selecting compiler options for optimal performance
The compiler options you choose can greatly improve the performance of the code
generated by the compiler; however, like most performance considerations, there
are trade-offs associated with these choices. Fortunately, you can weigh the
trade-offs associated with compiler options without editing your source code
because these options can be specified on the command line or in the configuration
file.
If you want to avoid details, the least complex way to improve the performance of
generated code is to specify the following (nondefault) compiler options:
OPT(2) or OPT(3)
DFT(REORDER)
The following sections describe, in more detail, performance improvements and
trade-offs associated with specific compiler options.
OPTIMIZE
You can specify the OPTIMIZE option to improve the speed of your program;
otherwise, the compiler makes only basic optimization efforts.
Choosing OPTIMIZE(2) directs the compiler to generate code for better
performance. Usually, the resultant code is shorter than when the program is
compiled under NOOPTIMIZE. Sometimes, however, a longer sequence of
instructions runs faster than a shorter sequence. This occurs, for instance, when a
branch table is created for a SELECT statement where the values in the WHEN
clauses contain gaps. The increased number of instructions generated is usually
offset by the execution of fewer instructions in other places.
Choosing OPTIMIZE(3) directs the compiler to generate even better code.
However, when you specify the OPTIMIZE(3) option, the compilation will take
longer (and sometimes much, much longer) than when you specify OPTIMIZE(2).
GONUMBER
Using this option results in a statement number table used for debugging. This
added information can be extremely helpful when debugging, but including
statement number tables increases the size of your executable file. Larger
executable files can take longer to load.
ARCH
Using the highest value in the ARCH option allows the compiler to select from the
largest set of instructions available under z/OS and thus permits it to generate the
most optimal code.
REDUCE
The REDUCE option specifies that the compiler is permitted to reduce an
assignment of a null string to a structure into a simple copy operation - even if
that means padding bytes might be overwritten.
The REDUCE option will cause fewer lines of code to be generated for an
assignment of a null string to a structure, and that will usually mean your
compilation will be quicker and your code will run much faster. However, padding
bytes may be zeroed out.
For instance, in the following structure, there is one byte of padding between
field11 and field12.
dcl
1 sample ext,
5 field11 dec fixed(13),
5 field14 bit(32),
5 field16 bit(32),
5 field17 bin fixed(31);
Now consider the assignment sample = ;
Under the NOREDUCE option, it will cause eight assignments to be generated, but
the padding byte will be unchanged.
However, under REDUCE, the assignment would be reduced to three operations.
RULES
Most of the RULES suboptions affect only the severity with which certain coding
practices, such as not declaring variables, are flagged and have no impact on
performance. However, these suboptions do have an impact on performance.
IBM/ANS
When you use the RULES(IBM) option, the compiler supports scaled FIXED
BINARY and, what is more important for performance, generates scaled FIXED
BINARY results in some operations. Under RULES(ANS), scaled FIXED BINARY is
not supported and scaled FIXED BINARY results are never generated. This means
that the code generated under RULES(ANS) always runs at least as fast as the code
generated under RULES(IBM), and sometimes runs faster.
For example, consider the following code fragment:
dcl (i,j,k) fixed bin(15);
.
.
.
i = j / k;
Under RULES(IBM), the result of the division has the attributes FIXED BIN(31,16).
This means that a shift instruction is required before the division and several more
instructions are needed to perform the assignment.
Under RULES(ANS), the result of the division has the attributes FIXED BIN(15,0).
This means that a shift is not needed before the division, and no extra instructions
are needed to perform the assignment.
Improving performance
(NO)LAXCTL
Under RULES(LAXCTL), a CONTROLLED variable may be declared with constant
extents and yet allocated with different extents.
For instance, under RULES(LAXCTL), you may declare a structure as follows:
dcl
1 a controlled,
2 b char(17),
2 c char(29);
However, you could then allocate it as follows:
allocate
1 a,
2 b char(170),
2 c char(290);
This has disastrous consequences for performance because it means that whenever
the compiler sees a reference to the structure A or to any member of that structure,
the compiler is forced to assume that it knows nothing about the lengths,
dimensions or offsets of the fields in it.
However, the RULES(NOLAXCTL) option disallows this coding practice: under
RULES(NOLAXCTL), if you then want to allocate a CONTROLLED variable with a
variable extent, then that extents must be declared either with an asterisk or with a
non-constant expression. Consequently, under RULES(NOLAXCTL), when a
CONTROLLED variable is declared with constant extents, then the compiler can
generate much better code for any reference to that variable.
PREFIX
This option determines if selected PL/I conditions are enabled by default. The
default suboptions for PREFIX are set to conform to the PL/I language definition;
however, overriding the defaults can have a significant effect on the performance
of your program. The default suboptions are:
CONVERSION
INVALIDOP
FIXEDOVERFLOW
OVERFLOW
INVALIDOP
NOSIZE
NOSTRINGRANGE
NOSTRINGSIZE
NOSUBSCRIPTRANGE
UNDERFLOW
ZERODIVIDE
By specifying the SIZE, STRINGRANGE, STRINGSIZE, or SUBSCRIPTRANGE
suboptions, the compiler generates extra code that helps you pinpoint various
problem areas in your source that would otherwise be hard to find. This extra
code, however, can slow program performance significantly.
CONVERSION
When you disable the CONVERSION condition, some character-to-numeric
conversions are done inline and without checking the validity of the source;
therefore, specifying NOCONVERSION also affects program performance.
FIXEDOVERFLOW
On some platforms, the FIXEDOVERFLOW condition is raised by the hardware
and the compiler does not need to generate any extra code to detect it.
DEFAULT
Using the DEFAULT option, you can select attribute defaults. As is true with the
PREFIX option, the suboptions for DEFAULT are set to conform to the PL/I
language definition. Changing the defaults in some instances can affect
performance.
Some of the suboptions, such IBM/ANS and ASSIGNABLE/NONASSIGNABLE,
have no effect on program performance. But other suboptions can affect
performance to varying degrees and, if applied inappropriately, can make your
program invalid. The more important of these suboptions are:
BYADDR or BYVALUE
When the DEFAULT(BYADDR) option is in effect, arguments are passed by
reference (as required by PL/I) unless an attribute in an entry declaration indicates
otherwise. As arguments are passed by reference, the address of the argument is
passed from one routine (calling routine) to another (called routine) as the variable
itself is passed. Any change made to the argument while in the called routine is
reflected in the calling routine when it resumes execution.
Program logic often depends on passing variables by reference. Passing a variable
by reference, however, can hinder performance in two ways:
1. Every reference to that parameter requires an extra instruction.
2. Since the address of the variable is passed to another routine, the compiler is
forced to make assumptions about when that variable might change and
generate very conservative code for any reference to that variable.
Consequently, you should pass parameters by value using the BYVALUE
suboption whenever your program logic allows. Even if you use the BYADDR
attribute to indicate that one parameter should be passed by reference, you can use
the DEFAULT(BYVALUE) option to ensure that all other parameters are passed by
value.
A BYVALUE argument should be one that could reasonably be passed in a register.
Hence its type should be one of
v REAL FIXED BIN
v REAL FLOAT
v POINTER
v OFFSET
v HANDLE
v LIMITED ENTRY
v FILE
v ORDINAL
v CHAR(1)
v WCHAR(1)
v ALIGNED BIT(n) with n less than or equal to 8
Moreover, special care is required when using BYVALUE parameters with
assembler code as any BYVALUE parameter that does not require a multiple of 4
bytes will be widened so that 4 bytes are used in the parameter list.
If a procedure receives and modifies only one parameter that is passed by
BYADDR, consider converting the procedure to a function that receives that
parameter by value. The function would then end with a RETURN statement
containing the updated value of the parameter.
Procedure with BYADDR parameter:
a: proc( parm1, parm2, ..., parmN );
dcl parm1 byaddr ...;
dcl parm2 byvalue ...;
.
.
.
dcl parmN byvalue ...;
/* program logic */
end;
Faster, equivalent function with BYVALUE parameter:
a: proc( parm1, parm2, ..., parmN )
returns( ... /* attributes of parm1 */ );
.
.
.
dcl parmN byvalue ...;
/* program logic */
return( parm1 );
end;
(NON)CONNECTED
The DEFAULT(NONCONNECTED) option indicates that the compiler assumes
that any aggregate parameters are NONCONNECTED. References to elements of
NONCONNECTED aggregate parameters require the compiler to generate code to
access the parameter's descriptor, even if the aggregate is declared with constant
extents.
The compiler does not generate these instructions if the aggregate parameter has
constant extents and is CONNECTED. Consequently, if your application never
passes nonconnected parameters, your code is more optimal if you use the
DEFAULT(CONNECTED) option.
(NO)DESCRIPTOR
The DEFAULT(DESCRIPTOR) option indicates that, by default, a descriptor is
passed for any string, area, or aggregate parameter; however, the descriptor is used
only if the parameter has nonconstant extents or if the parameter is an array with
the NONCONNECTED attribute. In this case, the instructions and space required
to pass the descriptor provide no benefit and incur substantial cost (the size of a
structure descriptor is often greater than size of the structure itself). Consequently,
by specifying DEFAULT(NODESCRIPTOR) and using OPTIONS(DESCRIPTOR)
only as needed on PROCEDURE statements and ENTRY declarations, your code
runs more optimally.
(NO)INLINE
The suboption NOINLINE indicates that procedures and begin blocks should not
be inlined.
Inlining occurs only when you specify optimization.
Inlining user code eliminates the overhead of the function call and linkage, and
also exposes the function's code to the optimizer, resulting in faster code
performance. Inlining produces the best results when the overhead for the function
is nontrivial, for example, when functions are called within nested loops. Inlining
is also beneficial when the inlined function provides additional opportunities for
optimization, such as when constant arguments are used.
For programs containing many procedures that are not nested:
v If the procedures are small and only called from a few places, you can increase
performance by specifying INLINE.
v If the procedures are large and called from several places, inlining duplicates
code throughout the program. This increase in the size of the program might
offset any increase of speed. In this case, you might prefer to leave NOINLINE
as the default and specify OPTIONS(INLINE) only on individually selected
procedures.
When you use inlining, you need more stack space. When a function is called, its
local storage is allocated at the time of the call and freed when it returns to the
calling function. If that same function is inlined, its storage is allocated when the
function that calls it is entered, and is not freed until that calling function ends.
Ensure that you have enough stack space for the local storage of the inlined
functions.
LINKAGE
This suboption tells the compiler the default linkage to use when the LINKAGE
suboption of the OPTIONS attribute or option for an entry has not been specified.
The compiler supports various linkages, each with its unique performance
characteristics. When you invoke an ENTRY provided by an external entity (such
as an operating system), you must use the linkage previously defined for that
ENTRY.
As you create your own applications, however, you can choose the linkage
convention. The OPTLINK linkage is strongly recommended because it provides
significantly better performance than other linkage conventions.
(RE)ORDER
The DEFAULT(ORDER) option indicates that the ORDER option is applied to
every block, meaning that variables in that block referenced in ON-units (or blocks
dynamically descendant from ON-units) have their latest values. This effectively
prohibits almost all optimization on such variables. Consequently, if your program
logic allows, use DEFAULT(REORDER) to generate superior code.
NOOVERLAP
The DEFAULT(NOOVERALP) option lets the compiler assume that the source and
target in an assignment do not overlap, and it can therefore generate smaller and
faster code.
However, if you this option, you must insure that the source and target in
assignment do not overlap. For example, under the DEFAULT( NOOVERLAP )
option, the assignment in this example would be invalid:
dcl c char(20);
substr(c,2,5) = substr(c,1,5);
RETURNS(BYVALUE) or RETURNS(BYADDR)
When the DEFAULT(RETURNS(BYVALUE)) option is in effect, the BYVALUE
attribute is applied to all RETURNS description lists that do not specify BYADDR.
This means that these functions return values in registers, when possible, in order
to produce the most optimal code.
Summary of compiler options that improve performance
In summary, the following options (if appropriate for your application) can
improve performance:
OPTIMIZE(3)
ARCH(9)
REDUCE
RULES(ANS NOLAXCTL)
DEFAULT with the following suboptions
BYVALUE
CONNECTED
NODESCRIPTOR
INLINE
LINKAGE(OPTLINK)
REORDER
NOOVERLAP
RETURNS(BYVALUE)
Coding for better performance
As you write code, there is generally more than one correct way to accomplish a
given task. Many important factors influence the coding style you choose,
including readability and maintainability. The following sections discuss choices
that you can make while coding that potentially affect the performance of your
program.
DATA-directed input and output
Using GET DATA and PUT DATA statements for debugging can prove very
helpful. When you use these statements, however, you generally pay the price of
decreased performance. This cost to performance is usually very high when you
use either GET DATA or PUT DATA without a variable list.
Many programmers use PUT DATA statements in their ON ERROR code as
illustrated in the following example:
on error
begin;
on error system;
.
.
.
put data;
.
.
.
end;
In this case, the program would perform more optimally by including a list of
selected variables with the PUT DATA statement.
The ON ERROR block in the previous example contained an ON ERROR system
statement before the PUT DATA statement. This prevents the program from getting
caught in an infinite loop if an error occurs in the PUT DATA statement (which
could occur if any variables to be listed contained invalid FIXED DECIMAL
values) or elsewhere in the ON ERROR block.
Input-only parameters
If a procedure has a BYADDR parameter which it uses as input only, it is best to
declare that parameter as NONASSIGNABLE (rather than letting it get the default
attribute of ASSIGNABLE). If that procedure is later called with a constant for that
parameter, the compiler can put that constant in static storage and pass the address
of that static area.
This practice is particularly useful for strings and other parameters that cannot be
passed in registers (input-only parameters that can be passed in registers are best
declared as BYVALUE).
In the following declaration, for instance, the first parameter to getenv is an
input-only CHAR VARYINGZ string:
dcl getenv entry( char(*) varyingz nonasgn byaddr,
pointer byaddr )
returns( native fixed bin(31) optional )
If this function is invoked with the string 'IBM_OPTIONS', the compiler can pass
the address of that string rather than assigning it to a compiler-generated
temporary storage area and passing the address of that area.
GOTO statements
A GOTO statement that uses either a label in another block or a label variable
severely limits optimizations that the compiler might perform. If a label array is
initialized and declared AUTOMATIC, either implicitly or explicitly, any GOTO to
an element of that array will hinder optimization. However, if the array is declared
as STATIC, the compiler assumes the CONSTANT attribute for it and no
optimization is hindered.
String assignments
When one string is assigned to another, the compiler ensures that:
v The target has the correct value even if the source and target overlap.
v The source string is truncated if it is longer than the target.
This assurance comes at the price of some extra instructions. The compiler
attempts to generate these extra instructions only when necessary, but often you, as
the programmer, know they are not necessary when the compiler cannot be sure.
For instance, if the source and target are based character strings and you know
they cannot overlap, you could use the PLIMOVE built-in function to eliminate the
extra code the compiler would otherwise be forced to generate.
In the example which follows, faster code is generated for the second assignment
statement:
dcl based_Str char(64) based( null() );
dcl target_Addr pointer;
dcl source_Addr pointer;
target_Addr->based_Str = source_Addr->based_Str;
call plimove( target_Addr, source_Addr, stg(based_Str) );
If you have any doubts about whether the source and target might overlap or
whether the target is big enough to hold the source, you should not use the
PLIMOVE built-in.
Loop control variables
Program performance improves if your loop control variables are one of the types
in the following list. You should rarely, if ever, use other types of variables.
FIXED BINARY with zero scale factor
FLOAT
ORDINAL
HANDLE
POINTER
OFFSET
Performance also improves if loop control variables are not members of arrays,
structures, or unions. The compiler issues a warning message when they are. Loop
control variables that are AUTOMATIC and not used for any other purpose give
you the optimal code generation.
If a loop control variable is a FIXED BINARY, performance is best if it has
precision 31 and is SIGNED.
Performance is decreased if your program depends not only on the value of a loop
control variable, but also on its address. For example, the performance is decreased
if the ADDR built-in function is applied to the variable or if the variable is passed
by BYADDR to another routine.
PACKAGEs versus nested PROCEDUREs
Calling nested procedures requires that an extra hidden parameter (the backchain
pointer) is passed. As a result, the fewer nested procedures that your application
contains, the faster it runs.
To improve the performance of your application, you can convert a
mother-daughter pair of nested procedures into level-1 sister procedures inside of a
package. This conversion is possible if your nested procedure does not rely on any
of the automatic and internal static variables declared in its parent procedures.
If procedure b in Example with nested procedures does not use any of the
variables declared in a, you can improve the performance of both procedures by
reorganizing them into the package illustrated in Example with packaged
procedures.
Example with nested procedures
a: proc;
dcl (i,j,k) fixed bin;
dcl ib based fixed bin;
.
.
.
call b( addr(i) );
.
.
.
b: proc( px );
dcl px pointer;
display( px->ib );
end;
end;
Example with packaged procedures
p: package exports( a );
dcl ib based fixed bin;
a: proc;
dcl (i,j,k) fixed bin;
.
.
.
call b( addr(i) );
.
.
.
end;
b: proc( px );
dcl px pointer;
display( px->ib );
end;
end p;
REDUCIBLE Functions
REDUCIBLE indicates that a procedure or entry need not be invoked multiple
times if the argument(s) stays unchanged, and that the invocation of the procedure
has no side effects.
For example, a user-written function that computes a result based on unchanging
data should be declared REDUCIBLE. A function that computes a result based on
changing data, such as a random number or time of day, should be declared
IRREDUCIBLE.
In the following example, f is invoked only once since REDUCIBLE is part of the
declaration. If IRREDUCIBLE had been used in the declaration, f would be invoked
twice.
dcl (f) entry options( reducible ) returns( fixed bin );
select;
when( f(x) < 0 )
.
.
.
when( f(x) > 0 )
.
.
.
otherwise
.
.
.
end;
DESCLOCATOR or DESCLIST
When the DEFAULT(DESCLOCATOR) option is in effect, the compiler passes
arguments requiring descriptors (such as strings and structures) via a descriptor
locator in much the same way that the old compiler did. More information on
descriptors and how they are passed is available in Chapter 23, PL/I descriptors,
on page 541.
This option allows you to invoke an entry point that is not always passed all of the
arguments that it declares.
This option also allows you to continue the somewhat unwise programming
practice of passing a structure and receiving it as a pointer.
However, the code generated by the compiler for DEFAULT(DESCLOCATOR) may,
in some situations, perform less well than that for DEFAULT(DESCLIST).
DEFINED versus UNION
The UNION attribute is more powerful than the DEFINED attribute and provides
more function. In addition, the compiler generates better code for union references.
In the following example, the pair of variables b3 and b4 perform the same
function as b1 and b2, but the compiler generates more optimal code for the pair in
the union.
dcl b1 bit(31);
dcl b2 bit(16) def b1;
dcl
1 * union,
2 b3 bit(32),
2 b4 bit(16);
Code that uses UNIONs instead of the DEFINED attribute is subject to less
misinterpretation. Variable declarations in unions are in a single location making it
easy to realize that when one member of the union changes, all of the others
change also. This dynamic change is less obvious in declarations that use
DEFINED variables since the declare statements can be several lines apart.
Named constants versus static variables
You can define named constants by declaring a variable with the VALUE attribute.
If you use static variables with the INITIAL attribute and you do not alter the
variable, you should declare the variable a named constant using the VALUE
attribute. The compiler does not treat NONASSIGNABLE scalar STATIC variables
as true named constants.
The compiler generates better code whenever expressions are evaluated during
compilation, so you can use named constants to produce efficient code with no loss
in readability. For example, identical object code is produced for the two usages of
the VERIFY built-in function in the following example:
dcl numeric char value(0123456789);
jx = verify( string, numeric );
jx = verify( string, 0123456789 );
The following examples illustrate how you can use the VALUE attribute to get
optimal code without sacrificing readability.
Example with optimal code but no meaningful names
dcl x bit(8) aligned;
select( x );
when( 01b4 )
.
.
.
when( 02b4 )
.
.
.
when( 03b4 )
.
.
.
end;
Example with meaningful names but not optimal code
dcl ( a1 init( 01b4)
,a2 init( 02b4)
,a3 init( 03b4)
,a4 init( 04b4)
,a5 init( 05b4)
) bit(8) aligned static nonassignable;
select( x );
when( a1 )
.
.
.
when( a2 )
.
.
.
when( a3 )
.
.
.
end;
Example with optimal code AND meaningful names
dcl ( a1 value( 01b4)
,a2 value( 02b4)
,a3 value( 03b4)
,a4 value( 04b4)
,a5 value( 05b4)
) bit(8);
select( x );
when( a1 )
.
.
.
when( a2 )
.
.
.
when( a3 )
.
.
.
end;
Avoiding calls to library routines
The bitwise operations (prefix NOT, infix AND, infix OR, and infix EXCLUSIVE
OR) are often evaluated by calls to library routines. These operations are, however,
handled without a library call if either of the following conditions is true:
v Both operands are bit(1)
v Both operands are aligned and have the same constant length.
For certain assignments, expressions, and built-in function references, the compiler
generates calls to library routines. If you avoid these calls, your code generally
runs faster.
To help you determine when the compiler generates such calls, the compiler
generates a message whenever a conversion is done using a library routine.
When your code refers to a member of a BASED structure using REFER, the
compiler often has to generate one or more calls to a library routine to map the
structure at run-time. These calls can be expensive, and so when the compiler
makes these calls, it will issue a message so that you can locate these potential
hot-spots in your code.
If you do have code that uses BASED structures with REFER and which the
compiler flags with this message, you may be able to get better performance by
passing the structure to a subroutine that declares a corresponding structure with *
extents. This will cause the structure to be mapped once at the CALL statement,
but there will no further remappings when it is accessed in the called subroutine.
Preloading library routines
The PL/I library contains one RMODE 24 routine that is used for low-level system
i/o functions: IBMPOIOA. If your code does RECORD i/o or uses SYSPRINT as a
STREAM OUTPUT file (without compiling with the STDSYS option), then you will
significantly improve your performance if you preload this routine or put it into
the (E)LPA.
Part 4. Using interfaces to other products
Chapter 14. Using the Sort program . . . . . 385
Preparing to use Sort . . . . . . . . . . . 385
Choosing the type of Sort . . . . . . . . 386
Specifying the sorting field . . . . . . . . 389
Specifying the records to be sorted . . . . . 390
Maximum record lengths . . . . . . . 391
Determining storage needed for Sort . . . . 391
Main storage . . . . . . . . . . . 391
Auxiliary storage . . . . . . . . . . 392
Calling the Sort program . . . . . . . . . 392
Example 1 . . . . . . . . . . . . . 393
Example 2 . . . . . . . . . . . . . 394
Example 3 . . . . . . . . . . . . . 394
Example 4 . . . . . . . . . . . . . 394
Example 5 . . . . . . . . . . . . . 394
Determining whether the Sort was successful 394
Establishing data sets for Sort . . . . . . . 395
Sort work data sets . . . . . . . . . 395
Input data set . . . . . . . . . . . 395
Output data set. . . . . . . . . . . 396
Checkpoint data set . . . . . . . . . 396
Sort data input and output . . . . . . . . . 396
Data input and output handling routines . . . . 396
E15Input handling routine (Sort Exit E15) . . 397
E35Output handling routine (Sort Exit E35) 400
Calling PLISRTA example . . . . . . . . 401
Calling PLISRTB example . . . . . . . . 402
Calling PLISRTC example . . . . . . . . 403
Calling PLISRTD example . . . . . . . . 404
Sorting variable-length records example . . . 406
Chapter 15. ILC with C. . . . . . . . . . 409
Equivalent data types . . . . . . . . . . 409
Simple type equivalence. . . . . . . . . 409
Struct type equivalence . . . . . . . . . 410
Enum type equivalence . . . . . . . . . 410
File type equivalence . . . . . . . . . . 411
Using C functions . . . . . . . . . . . . 411
Matching simple parameter types . . . . . 412
Matching string parameter types . . . . . . 415
Functions returning ENTRYs . . . . . . . 416
Linkages . . . . . . . . . . . . . . . 417
Sharing output and input . . . . . . . . . 419
Sharing output . . . . . . . . . . . . 419
Sharing input . . . . . . . . . . . . 420
Using the ATTACH statement . . . . . . . 420
Redirecting C standard streams . . . . . . 420
Summary. . . . . . . . . . . . . . . 420
Chapter 16. Interfacing with Java . . . . . . 423
Java Native Interface (JNI) . . . . . . . . . 423
Calling PL/I program from Java . . . . . . 424
JNI Sample Program #1 - 'Hello World' . . . . . 424
Step 1: Writing the Java Program . . . . . 424
Step 2: Compiling the Java Program. . . . 425
Step 3: Writing the PL/I Program . . . . 425
Step 4: Compiling and Linking the PL/I
Program . . . . . . . . . . . . . 427
Step 5: Running the Sample Program . . . 427
JNI Sample Program #2 - Passing a String . . . . 427
Program . . . . . . . . . . . . . 431
JNI Sample Program #3 - Passing an Integer . . . 432
Program . . . . . . . . . . . . . 435
JNI Sample Program #4 - Java Invocation API . . 436
Program . . . . . . . . . . . . . 440
Determining equivalent Java and PL/I data types 440
Chapter 14. Using the Sort program
The compiler provides an interface called PLISRTx (x = A, B, C, or D) that allows
you to make use of the IBM-supplied Sort programs.
To use the Sort program with PLISRTx, you must:
1. Include a call to one of the entry points of PLISRTx, passing it the information
on the fields to be sorted. This information includes the length of the records,
the maximum amount of storage to use, the name of a variable to be used as a
return code, and other information required to carry out the sort.
2. Specify the data sets required by the Sort program in JCL DD statements.
When used from PL/I, the Sort program sorts records of all normal lengths on a
large number of sorting fields. Data of most types can be sorted into ascending or
descending order. The source of the data to be sorted can be either a data set or a
user-written PL/I procedure that the Sort program will call each time a record is
required for the sort. Similarly, the destination of the sort can be a data set or a
PL/I procedure that handles the sorted records.
Using PL/I procedures allows processing to be done before or after the sort itself,
thus allowing a complete sorting operation to be handled completely by a call to
the sort interface. It is important to understand that the PL/I procedures handling
input or output are called from the Sort program itself and will effectively become
part of it.
PL/I can operate with DFSORT
or a program with the same interface. DFSORT is

a release of the program product 5740-SM1. DFSORT has many built-in features
you can use to eliminate the need for writing program logic (for example,
INCLUDE, OMIT, OUTREC and SUM statement plus the many ICETOOL
operators). See DFSORT Application Programming Guide for details and Getting
Started with DFSORT for a tutorial.
The following material applies to DFSORT. Because you can use programs other
than DFSORT, the actual capabilities and restrictions vary. For these capabilities
and restrictions, see DFSORT Application Programming Guide, or the equivalent
publication for your sort product.
To use the Sort program you must include the correct PL/I statements in your
source program and specify the correct data sets in your JCL.
Preparing to use Sort
Before using Sort, you must determine the type of sort you require, the length and
format of the sorting fields in the data, the length of your data records, and the
amount of auxiliary and main storage you will allow for sorting.
To determine the PLISRTx entry point that you will use, you must decide the
source of your unsorted data, and the destination of your sorted data. You must
choose between data sets and PL/I subroutines. Using data sets is simpler to
understand and gives faster performance. Using PL/I subroutines gives you more
flexibility and more function, enabling you to manipulate or print the data before it
is sorted, and to make immediate use of it in its sorted form. If you decide to use
an input or output handling subroutine, you will need to read Data input and
output handling routines on page 396.
The entry points and the source and destination of data are as follows:
Entry point Source Destination
PLISRTA Data set Data set
PLISRTB Subroutine Data set
PLISRTC Data set Subroutine
PLISRTD Subroutine Subroutine
Having determined the entry point you are using, you must now determine the
following things about your data set:
v The position of the sorting fields; these can be either the complete record or any
part or parts of it
v The type of data these fields represent, for example, character or binary
v Whether you want the sort on each field to be in ascending or descending order
v Whether you want equal records to be retained in the order of the input, or
whether their order can be altered during sorting
Specify these options on the SORT statement, which is the first argument to
PLISRTx. After you have determined these, you must determine two things about
the records to be sorted:
v Whether the record format is fixed or varying
v The length of the record, which is the maximum length for varying
Specify these on the RECORD statement, which is the second argument to
PLISRTx.
Finally, you must decide on the amount of main and auxiliary storage you will
allow for the Sort program. For further details, see Determining storage needed
for Sort on page 391.
Choosing the type of Sort
If you want to make the best use of the Sort program, you must understand
something of how it works. In your PL/I program you specify a sort by using a
CALL statement to the sort interface subroutine PLISRTx. This subroutine has four
entry points: x=A, B, C, and D. Each specifies a different source for the unsorted
data and destination for the data when it has been sorted. For example, a call to
PLISRTA specifies that the unsorted data (the input to sort) is on a data set, and
that the sorted data (the output from sort) is to be placed on another data set. The
CALL PLISRTx statement must contain an argument list giving the Sort program
information about the data set to be sorted, the fields on which it is to be sorted,
the amount of space available, the name of a variable into which Sort will place a
return code indicating the success or failure of the sort, and the name of any
output or input handling procedure that can be used.
The sort interface routine builds an argument list for the Sort program from the
information supplied by the PLISRTx argument list and the choice of PLISRTx
entry point. Control is then transferred to the Sort program. If you have specified
an output- or input-handling routine, this will be called by the Sort program as
many times as is necessary to handle each of the unsorted or sorted records. When
the sort operation is complete, the Sort program returns to the PL/I calling
procedure communicating its success or failure in a return code, which is placed in
one of the arguments passed to the interface routine. The return code can then be
tested in the PL/I routine to discover whether processing should continue.
Figure 47 on page 388 is a simplified flowchart showing this operation.
Within the Sort program itself, the flow of control between the Sort program and
input- and output-handling routines is controlled by return codes. The Sort
Figure 47. Flow of control for Sort program
program calls these routines at the appropriate point in its processing. (Within the
Sort program, and its associated documentation, these routines are known as user
exits. The routine that passes input to be sorted is the E15 sort user exit. The
routine that processes sorted output is the E35 sort user exit.) From the routines,
Sort expects a return code indicating either that it should call the routine again, or
that it should continue with the next stage of processing.
The important points to remember about Sort are: (1) it is a self-contained program
that handles the complete sort operation, and (2) it communicates with the caller,
and with the user exits that it calls, by means of return codes.
The remainder of this chapter gives detailed information on how to use Sort from
PL/I. First the required PL/I statements are described, and then the data set
requirements. The chapter finishes with a series of examples showing the use of
the four entry points of the sort interface routine.
Specifying the sorting field
The SORT statement is the first argument to PLISRTx. The syntax of the SORT
statement must be a character string expression that takes the form:
'bSORTbFIELDS=(start1,length1,form1,seq1,
...startn,lengthn,formn,seqn)[,other options]b'
For example:
' SORT FIELDS=(1,10,CH,A) '
b represents one or more blanks. Blanks shown are mandatory. No other blanks
are allowed.
start,length,form,seq
defines a sorting field. You can specify any number of sorting fields, but there
is a limit on the total length of the fields. If more than one field is to be sorted
on, the records are sorted first according to the first field, and then those that
are of equal value are sorted according to the second field, and so on. If all the
sorting values are equal, the order of equal records will be arbitrary unless you
use the EQUALS option. (See later in this definition list.) Fields can overlay
each other.
For DFSORT (5740-SM1), the maximum total length of the sorting fields is
restricted to 4092 bytes and all sorting fields must be within 4092 bytes of the
start of the record. Other sort products might have different restrictions.
start is the starting position within the record. Give the value in bytes
except for binary data where you can use a byte.bit notation. The
first byte in a string is considered to be byte 1, the first bit is bit 0.
(Thus the second bit in byte 2 is referred to as 2.1.) For varying length
records you must include the 4-byte length prefix, making 5 the first
byte of data.
length is the length of the sorting field. Give the value in bytes except for
binary where you can use byte.bit notation. The length of sorting
fields is restricted according to their data type.
form is the format of the data. This is the format assumed for the purpose of
sorting. All data passed between PL/I routines and Sort must be in the
form of character strings. The main data types and the restrictions on
their length are shown below. Additional data types are available for
special-purpose sorts. See the DFSORT Application Programming Guide,
or the equivalent publication for your sort product.
Code Data type and length
CH character 14096
ZD zoned decimal, signed 132
PD packed decimal, signed 132
FI fixed point, signed 1256
BI binary, unsigned 1 bit to 4092 bytes
FL floating-point, signed 1256
The sum of the lengths of all fields must not exceed 4092 bytes.
seq is the sequence in which the data will be sorted as follows:
A ascending (that is, 1,2,3,...)
D descending (that is, ...,3,2,1)
Note: You cannot specify E, because PL/I does not provide a method
of passing a user-supplied sequence.
other options
You can specify a number of other options, depending on your Sort program.
You must separate them from the FIELDS operand and from each other by
commas. Do not place blanks between operands.
FILSZ=y
specifies the number of records in the sort and allows for optimization by
Sort. If y is only approximate, E should precede y.
SKIPREC=y
specifies that y records at the start of the input file are to be ignored before
sorting the remaining records.
CKPT or CHKPT
specifies that checkpoints are to be taken. If you use this option, you must
provide a SORTCKPT data set and DFSORT's 16NCKPT=NO installation
option must be specified.
EQUALS|NOEQUALS
specifies whether the order of equal records will be preserved as it was in
the input (EQUALS) or will be arbitrary (NOEQUALS). You could improve
sort performance by using the NOEQUALS. The default option is chosen
when Sort is installed. The IBM-supplied default is NOEQUALS.
DYNALLOC=(d,n)
(OS/VS Sort only) specifies that the program dynamically allocates
intermediate storage.
d is the device type (3380, etc.).
n is the number of work areas.
Specifying the records to be sorted
Use the RECORD statement as the second argument to PLISRTx. The syntax of the
RECORD statement must be a character string expression which, when evaluated,
takes the syntax shown below:
'bRECORDbTYPE=rectype[,LENGTH=(L1,[,,L4,L5])]b'
For example:
' RECORD TYPE=F,LENGTH=(80) '
b represents one or more blanks. Blanks shown are mandatory. No other blanks
are allowed.
TYPE
specifies the type of record as follows:
F fixed length
V varying length EBCDIC
D varying length ASCII
Even when you use input and output routines to handle the unsorted and
sorted data, you must specify the record type as it applies to the work data
sets used by Sort.
If varying length strings are passed to Sort from an input routine (E15 exit),
you should normally specify V as a record format. However, if you specify F,
the records are padded to the maximum length with blanks.
LENGTH
specifies the length of the record to be sorted. You can omit LENGTH if you
use PLISRTA or PLISRTC, because the length will be taken from the input data
set. Note that there is a restriction on the maximum and minimum length of
the record that can be sorted (see below). For varying length records, you must
include the four-byte prefix.
L1 is the length of the record to be sorted. For VSAM data sets sorted as
varying records it is the maximum record size+4.
,, represent two arguments that are not applicable to Sort when called
from PL/I. You must include the commas if the arguments that follow
are used.
L4 specifies the minimum length of record when varying length records
are used. If supplied, it is used by Sort for optimization purposes.
L5 specifies the modal (most common) length of record when varying
length records are used. If supplied, it is used by Sort for optimization
purposes.
Maximum record lengths
The length of a record can never exceed the maximum length specified by the user.
The maximum record length for variable length records is 32756 bytes and for
fixed length records it is 32760 bytes.
Determining storage needed for Sort
Main storage
Sort requires both main and auxiliary storage. The minimum main storage for
DFSORT is 88K bytes, but for best performance, more storage (on the order of 1
megabyte or more) is recommended. DFSORT can take advantage of storage above
16M virtual or extended architecture processors. Under z/OS, DFSORT can also
take advantage of expanded storage. You can specify that Sort use the maximum
amount of storage available by passing a storage parameter in the following
manner:
DCL MAXSTOR FIXED BINARY (31,0);
UNSPEC(MAXSTOR)='00000000'B||UNSPEC('MAX');
CALL PLISRTA
(' SORT FIELDS=(1,80,CH,A) ',
' RECORD TYPE=F,LENGTH=(80) ',
MAXSTOR,
RETCODE,
'TASK');
If files are opened in E15 or E35 exit routines, enough residual storage should be
allowed for the files to open successfully.
Auxiliary storage
Calculating the minimum auxiliary storage for a particular sorting operation is a
complicated task. To achieve maximum efficiency with auxiliary storage, use direct
access storage devices (DASDs) whenever possible. For more information on
improving program efficiency, see the DFSORT Application Programming Guide,
particularly the information about dynamic allocation of workspace which allows
DFSORT to determine the auxiliary storage needed and allocate it for you.
If you are interested only in providing enough storage to ensure that the sort will
work, make the total size of the SORTWK data sets large enough to hold three sets
of the records being sorted. (You will not gain any advantage by specifying more
than three if you have enough space in three data sets.)
However, because this suggestion is an approximation, it might not work, so you
should check the sort manuals. If this suggestion does work, you will probably
have wasted space.
Calling the Sort program
When you have determined the points described above, you are in a position to
write the CALL PLISRTx statement. You should do this with some care; for the
entry points and arguments to use, see Table 31.
Table 31. The entry points and arguments to PLISRTx (x = A, B, C, or D)
Entry points Arguments
PLISRTA
Sort input: data set
Sort output: data set
(sort statement,record statement,storage,return code
[,data set prefix,message level, sort technique])
PLISRTB
Sort input: PL/I subroutine
Sort output: data set
(sort statement,record statement,storage,return code,input routine
[,data set prefix,message level,sort technique])
PLISRTC
Sort input: data set
Sort output: PL/I subroutine
(sort statement,record statement,storage,return code,output routine
[,data set prefix,message level,sort technique])
PLISRTD
Sort input: PL/I subroutine
Sort output: PL/I subroutine
(sort statement,record statement,storage,return code,input routine,output
routine[,data set prefix,message level,sort technique])
Sort statement Character string expression containing the Sort program SORT statement. Describes
sorting fields and format. See Specifying the sorting field on page 389.
Record statement Character string expression containing the Sort program RECORD statement.
Describes the length and record format of data. See Specifying the records to be
sorted on page 390.
Storage Fixed binary expression giving maximum amount of main storage to be used by
the Sort program. Must be >88K bytes for DFSORT. See also Determining storage
needed for Sort on page 391.
Return code Fixed binary variable of precision (31,0) in which Sort places a return code when it
has completed. The meaning of the return code is:
0=Sort successful
16=Sort failed
20=Sort message data set missing
Table 31. The entry points and arguments to PLISRTx (x = A, B, C, or D) (continued)
Entry points Arguments
Input routine (PLISRTB and PLISRTD only.) Name of the PL/I external or internal procedure
used to supply the records for the Sort program at sort exit E15.
Output routine (PLISRTC and PLISRTD only.) Name of the PL/I external or internal procedure to
which Sort passes the sorted records at sort exit E35.
Data set prefix Character string expression of four characters that replaces the default prefix of
'SORT' in the names of the sort data sets SORTIN, SORTOUT, SORTWKnn and
SORTCNTL, if used. Thus if the argument is TASK, the data sets TASKIN,
TASKOUT, TASKWKnn, and TASKCNTL can be used. This facility enables multiple
invocations of Sort to be made in the same job step. The four characters must start
with an alphabetic character and must not be one of the reserved names PEER,
BALN, CRCX, OSCL, POLY, DIAG, SYSC, or LIST. You must code a null string for
this argument if you require either of the following arguments but do not require
this argument.
Message level Character string expression of two characters indicating how Sort's diagnostic
messages are to be handled, as follows:
NO No messages to SYSOUT
AP All messages to SYSOUT
CP Critical messages to SYSOUT
SYSOUT will normally be allocated to the printer, hence the use of the mnemonic
letter P. Other codes are also allowed for certain of the Sort programs. For
further details on these codes, see DFSORT Application Programming Guide. You
must code a null string for this argument if you require the following argument
but you do not require this argument.
Sort technique (This is not used by DFSORT; it appears for compatibility reasons only.) Character
string of length 4 that indicates the type of sort to be carried out, as follows:
PEER Peerage sort
BALN Balanced
CRCX Criss-cross sort
OSCL Oscillating
POLY Polyphase sort
Normally the Sort program will analyze the amount of space available and choose
the most effective technique without any action from you. You should use this
argument only as a bypass for sorting problems or when you are certain that
performance could be improved by another technique. See DFSORT Application
Programming Guide for further information.
The examples below indicate the form that the CALL PLISRTx statement normally
takes.
Example 1
A call to PLISRTA sorting 80-byte records from SORTIN to SORTOUT using
1048576 (1 megabyte) of storage, and a return code, RETCODE, declared as FIXED
BINARY (31,0).
CALL PLISRTA (' SORT FIELDS=(1,80,CH,A) ',
1048576,
RETCODE);
Example 2
This example is the same as example 1 except that the input, output, and work
data sets are called TASKIN, TASKOUT, TASKWK01, and so forth. This might
occur if Sort was being called twice in one job step.
1048576,
RETCODE,
'TASK');
Example 3
This example is the same as example 1 except that the sort is to be undertaken on
two fields. First, bytes 1 to 10 which are characters, and then, if these are equal,
bytes 11 and 12 which contain a binary field, both fields are to be sorted in
ascending order.
CALL PLISRTA (' SORT FIELDS=(1,10,CH,A,11,2,BI,A) ',
1048576,
RETCODE);
Example 4
This example shows a call to PLISRTB. The input is to be passed to Sort by the
PL/I routine PUTIN, the sort is to be carried out on characters 1 to 10 of an 80
byte fixed length record. Other information as above.
CALL PLISRTB (' SORT FIELDS=(1,10,CH,A) ',
1048576,
RETCODE,
PUTIN);
Example 5
This example shows a call to PLISRTD. The input is to be supplied by the PL/I
routine PUTIN and the output is to be passed to the PL/I routine PUTOUT. The
record to be sorted is 84 bytes varying (including the length prefix). It is to be
sorted on bytes 1 through 5 of the data in ascending order, then if these fields are
equal, on bytes 6 through 10 in descending order. (Note that the 4-byte length
prefix is included so that the actual values used are 5 and 10 for the starting
points.) If both these fields are the same, the order of the input is to be retained.
(The EQUALS option does this.)
CALL PLISRTD (' SORT FIELDS=(5,5,CH,A,10,5,CH,D),EQUALS ',
' RECORD TYPE=V,LENGTH=(84) ',
1048576,
RETCODE,
PUTIN, /*input routine (sort exit E15)*/
PUTOUT); /*output routine (sort exit E35)*/
Determining whether the Sort was successful
When the sort is completed, Sort sets a return code in the variable named in the
fourth argument of the call to PLISRTx. It then returns control to the statement that
follows the CALL PLISRTx statement. The value returned indicates the success or
failure of the sort as follows:
0 Sort successful
16 Sort failed
20 Sort message data set missing
You must declare the variable to which the return code is passed as FIXED
BINARY (31,0). It is standard practice to test the value of the return code after the
CALL PLISRTx statement and take appropriate action according to the success or
failure of the operation.
For example (assuming the return code was called RETCODE):
IF RETCODE=0 THEN DO;
PUT DATA(RETCODE);
SIGNAL ERROR;
END;
If the job step that follows the sort depends on the success or failure of the sort,
you should set the value returned in the Sort program as the return code from the
PL/I program. This return code is then available for the following job step. The
PL/I return code is set by a call to PLIRETC. You can call PLIRETC with the value
returned from Sort thus:
CALL PLIRETC(RETCODE);
You should not confuse this call to PLIRETC with the calls made in the input and
output routines, where a return code is used for passing control information to
Sort.
Establishing data sets for Sort
If DFSORT was installed in a library not known to the system, you must specify
the DFSORT library in a JOBLIB or STEPLIB DD statement.
When you call Sort, certain sort data sets must not be open. These are:
SYSOUT
A data set (normally the printer) on which messages from the Sort program
will be written.
Sort work data sets
SORTWK01SORTWK32
Note: If you specify more than 32 sort work data sets, DFSORT will only use
the first 32.
****WK01****WK32
From 1 to 32 working data sets used in the sorting process. These must be
direct access. For a discussion of space required and number of data sets, see
Determining storage needed for Sort on page 391.
**** represents the four characters that you can specify as the data set prefix
argument in calls to PLISRTx. This allows you to use data sets with prefixes
other than SORT. They must start with an alphabetic character and must not be
the names PEER, BALN, CRCX, OSCL, POLY, SYSC, LIST, or DIAG.
Input data set
SORTIN
****IN
The input data set used when PLISRTA and PLISRTC are called.
See ****WK01****WK32 above for a detailed description.
Output data set
SORTOUT
****OUT
The output data set used when PLISRTA and PLISRTB are called.
Checkpoint data set
SORTCKPT
Data set used to hold checkpoint data, if CKPT or CHKPT option was used in
the SORT statement argument and DFSORT's 16NCKPT=NO installation option
was specified. For information on this program DD statement, see DFSORT
Application Programming Guide.
DFSPARM SORTCNTL
Data set from which additional or changed control statements can be read
(optional). For additional information on this program DD statement, see
DFSORT Application Programming Guide.
Sort data input and output
The source of the data to be sorted is provided either directly from a data set or
indirectly by a routine (Sort Exit E15) written by the user. Similarly, the destination
of the sorted output is either a data set or a routine (Sort Exit E35) provided by the
user.
PLISRTA is the simplest of all of the interfaces because it sorts from data set to
data set. An example of a PLISRTA program is in Figure 51 on page 401. Other
interfaces require either the input handling routine or the output handling routine,
or both.
Data input and output handling routines
The input handling and output handling routines are called by Sort when
PLISRTB, PLISRTC, or PLISRTD is used. They must be written in PL/I, and can be
either internal or external procedures. If they are internal to the routine that calls
PLISRTx, they behave in the same way as ordinary internal procedures in respect
of scope of names. The input and output procedure names must themselves be
known in the procedure that makes the call to PLISRTx.
The routines are called individually for each record required by Sort or passed
from Sort. Therefore, each routine must be written to handle one record at a time.
Variables declared as AUTOMATIC within the procedures will not retain their
values between calls. Consequently, items such as counters, which need to be
retained from one call to the next, should either be declared as STATIC or be
declared in the containing block.
The E15 and E35 sort exits must not be MAIN procedures.
E15Input handling routine (Sort Exit E15)
Input routines are normally used to process the data in some way before it is
sorted. You can use input routines to print the data, as shown in the Figure 52 on
page 402 and Figure 54 on page 404, or to generate or manipulate the sorting fields
to achieve the correct results.
The input handling routine is used by Sort when a call is made to either PLISRTB
or PLISRTD. When Sort requires a record, it calls the input routine which should
return a record in character string format, with a return code of 12. This return
code means that the record passed is to be included in the sort. Sort continues to
call the routine until a return code of 8 is passed. A return code of 8 means that all
records have already been passed, and that Sort is not to call the routine again. If a
record is returned when the return code is 8, it is ignored by Sort.
The data returned by the routine must be a character string. It can be fixed or
varying. If it is varying, you should normally specify V as the record format in the
RECORD statement which is the second argument in the call to PLISRTx. However,
you can specify F, in which case the string will be padded to its maximum length
with blanks. The record is returned with a RETURN statement, and you must
specify the RETURNS attribute in the PROCEDURE statement. The return code is
set in a call to PLIRETC. A flowchart for a typical input routine is shown in
Figure 48 on page 398.
Skeletal code for a typical input routine is shown in Figure 49 on page 399.
Input Handling Subroutine Output Handling Subroutine
START START
END
END
LAST
RECORD
ALREADY
SENT
Your code to
process record
CALL
PLIRETC(12)
RETURN
RECORD
RECEIVE
RECORD
PARAMETER
Your code to
process record
CALL
PLIRETC(4)
CALL
PLIRETC(8)
YES
NO
Figure 48. Flowcharts for input and output handling subroutines
Examples of an input routine are given in Figure 52 on page 402 and Figure 54 on
page 404.
In addition to the return codes of 12 (include current record in sort) and 8 (all
records sent), Sort allows the use of a return code of 16. This ends the sort and
causes Sort to return to your PL/I program with a return code of 16Sort failed.
Note: A call to PLIRETC sets a return code that will be passed by your PL/I
program, and will be available to any job steps that follow it. When an
output handling routine has been used, it is good practice to reset the return
code with a call to PLIRETC after the call to PLISRTx to avoid receiving a
nonzero completion code. By calling PLIRETC with the return code from
Sort as the argument, you can make the PL/I return code reflect the success
or failure of the sort. This practice is shown in Figure 53 on page 403.
E15: PROC RETURNS (CHAR(80));
/*---------------------------------------------------------------*/
/*RETURNS attribute must be used specifying length of data to be */
/* sorted, maximum length if varying strings are passed to Sort. */
/*---------------------------------------------------------------*/
DCL STRING CHAR(80); /*--------------------------------------------*/
/*A character string variable will normally be*/
/* required to return the data to Sort */
/*--------------------------------------------*/
IF LAST_RECORD_SENT THEN
DO;
/*---------------------------------------------------------------*/
/*A test must be made to see if all the records have been sent, */
/*if they have, a return code of 8 is set up and control returned*/
/*to Sort */
/*---------------------------------------------------------------*/
CALL PLIRETC(8); /*-------------------------------------------*/
/* Set return code of 8, meaning last record */
/* already sent. */
/*-------------------------------------------*/
RETURN();
END;
ELSE
DO;
/*------------------------------------------------*/
/* If another record is to be sent to Sort, do the*/
/* necessary processing, set a return code of 12 */
/* by calling PLIRETC, and return the data as a */
/* character string to Sort */
/*------------------------------------------------*/
****(The code to do your processing goes here)
CALL PLIRETC (12); /*--------------------------------------*/
/* Set return code of 12, meaning this */
/* record is to be included in the sort */
/*--------------------------------------*/
RETURN (STRING); /*Return data with RETURN statement*/
END;
END; /*End of the input procedure*/
Figure 49. Skeletal code for an input procedure
E35Output handling routine (Sort Exit E35)
Output handling routines are normally used for any processing that is necessary
after the sort. This could be to print the sorted data, as shown in Figure 53 on page
403 and Figure 54 on page 404, or to use the sorted data to generate further
information. The output handling routine is used by Sort when a call is made to
PLISRTC or PLISRTD. When the records have been sorted, Sort passes them, one at
a time, to the output handling routine. The output routine then processes them as
required. When all the records have been passed, Sort sets up its return code and
returns to the statement after the CALL PLISRTx statement. There is no indication
from Sort to the output handling routine that the last record has been reached. Any
end-of-data handling must therefore be done in the procedure that calls PLISRTx.
The record is passed from Sort to the output routine as a character string, and you
must declare a character string parameter in the output handling subroutine to
receive the data. The output handling subroutine must also pass a return code of 4
to Sort to indicate that it is ready for another record. You set the return code by a
call to PLIRETC.
The sort can be stopped by passing a return code of 16 to Sort. This will result in
Sort returning to the calling program with a return code of 16Sort failed.
The record passed to the routine by Sort is a character string parameter. If you
specified the record type as F in the second argument in the call to PLISRTx, you
should declare the parameter with the length of the record. If you specified the
record type as V, you should declare the parameter as adjustable, as in the
following example:
DCL STRING CHAR(*);
Figure 55 on page 406 shows a program that sorts varying length records.
A flowchart for a typical output handling routine is given in Figure 48 on page
398. Skeletal code for a typical output handling routine is shown in Figure 50.
You should note that a call to PLIRETC sets a return code that will be passed by
your PL/I program, and will be available to any job steps that follow it. When you
have used an output handling routine, it is good practice to reset the return code
with a call to PLIRETC after the call to PLISRTx to avoid receiving a nonzero
completion code. By calling PLIRETC with the return code from Sort as the
argument, you can make the PL/I return code reflect the success or failure of the
sort. This practice is shown in the examples at the end of this chapter.
E35: PROC(STRING); /*The procedure must have a character string
parameter to receive the record from Sort*/
DCL STRING CHAR(80); /*Declaration of parameter*/
(Your code goes here)
CALL PLIRETC(4); /*Pass return code to Sort indicating that the next
sorted record is to be passed to this procedure.*/
END E35; /*End of procedure returns control to Sort*/
Figure 50. Skeletal code for an output handling procedure
Calling PLISRTA example
After each time that the PL/I input- and output-handling routines communicate
the return-code information to the Sort program, the return-code field is reset to
zero; therefore, it is not used as a regular return code other than its specific use for
the Sort program.
For details on handling conditions, especially those that occur during the input-
and output-handling routines, see z/OS Language Environment Programming Guide.
//OPT14#7 JOB ...
//PLI.SYSIN DD *
EX106: PROC OPTIONS(MAIN);
DCL RETURN_CODE FIXED BIN(31,0);
1048576
RETURN_CODE);
SELECT (RETURN_CODE);
WHEN(0) PUT SKIP EDIT
('SORT COMPLETE RETURN_CODE 0') (A);
('SORT FAILED, RETURN_CODE 16') (A);
('SORT MESSAGE DATASET MISSING ') (A);
OTHER PUT SKIP EDIT (
'INVALID SORT RETURN_CODE = ', RETURN_CODE) (A,F(2));
END /* select */;
CALL PLIRETC(RETURN_CODE);
/*set PL/I return code to reflect success of sort*/
END EX106;
//GO.SORTIN DD *
003329HOOKER S.W. RIVERDALE, SATCHWELL LANE, BACONSFIELD
002886BOOKER R.R. ROTORUA, LINKEDGE LANE, TOBLEY
003077ROOKER & SON, LITTLETON NURSERIES, SHOLTSPAR
059334HOOK E.H. 109 ELMTREE ROAD, GANNET PARK, NORTHAMPTON
073872HOME TAVERN, WESTLEIGH
000931FOREST, IVER, BUCKS
/*
//GO.SYSPRINT DD SYSOUT=A
//GO.SORTOUT DD SYSOUT=A
//GO.SYSOUT DD SYSOUT=A
//GO.SORTWK01 DD UNIT=SYSDA,SPACE=(CYL,2)
/*
Figure 51. PLISRTAsorting from input data set to output data set
Calling PLISRTB example
//OPT14#8 JOB ...
//PLI.SYSIN DD *
CALL PLISRTB (' SORT FIELDS=(7,74,CH,A) ',
1048576
RETURN_CODE,
E15X);
SELECT(RETURN_CODE);
OTHER PUT SKIP EDIT
('INVALID RETURN_CODE = ',RETURN_CODE)(A,F(2));
END /* select */;
E15X: /* INPUT HANDLING ROUTINE GETS RECORDS FROM THE INPUT
STREAM AND PUTS THEM BEFORE THEY ARE SORTED*/
PROC RETURNS (CHAR(80));
DCL SYSIN FILE RECORD INPUT,
INFIELD CHAR(80);
ON ENDFILE(SYSIN) BEGIN;
PUT SKIP(3) EDIT ('END OF SORT PROGRAM INPUT')(A);
CALL PLIRETC(8); /* signal that last record has
already been sent to sort*/
INFIELD = ;
GOTO ENDE15;
END;
READ FILE (SYSIN) INTO (INFIELD);
PUT SKIP EDIT (INFIELD)(A(80)); /*PRINT INPUT*/
CALL PLIRETC(12); /* request sort to include current
record and return for more*/
ENDE15:
RETURN(INFIELD);
END E15X;
END EX107;
/*
//GO.SYSIN DD *
/*
//*
//GO.SORTCNTL DD *
OPTION DYNALLOC=(3380,2),SKIPREC=2
/*
Figure 52. PLISRTBsorting from input handling routine to output data set
Calling PLISRTC example
//OPT14#9 JOB ...
//PLI.SYSIN DD *
CALL PLISRTC (' SORT FIELDS=(7,74,CH,A) ',
1048576
RETURN_CODE,
E35X);
OTHER PUT SKIP EDIT
('INVALID RETURN_CODE = ', RETURN_CODE) (A,F(2));
END /* select */;
CALL PLIRETC (RETURN_CODE);
E35X: /* output handling routine prints sorted records*/
PROC (INREC);
DCL INREC CHAR(80);
PUT SKIP EDIT (INREC) (A);
CALL PLIRETC(4); /*request next record from sort*/
END E35X;
END EX108;
/*
//GO.STEPLIB DD DSN=SYS1.SORTLINK,DISP=SHR
//GO.SORTIN DD *
/*
//GO.SORTCNTL DD *
OPTION DYNALLOC=(3380,2),SKIPREC=2
/*
Figure 53. PLISRTCsorting from input data set to output handling routine
Calling PLISRTD example
//OPT14#10 JOB ...
//PLI.SYSIN DD *
CALL PLISRTD (' SORT FIELDS=(7,74,CH,A) ',
1048576
RETURN_CODE,
E15X,
E35X);
OTHER PUT SKIP EDIT
('INVALID RETURN_CODE = ', RETURN_CODE) (A,F(2));
END /* select */;
E15X: /* Input handling routine prints input before sorting*/
PROC RETURNS(CHAR(80));
DCL INFIELD CHAR(80);
PUT SKIP(3) EDIT ('END OF SORT PROGRAM INPUT. ',
'SORTED OUTPUT SHOULD FOLLOW')(A);
CALL PLIRETC(8); /* Signal end of input to sort*/
INFIELD = ;
GOTO ENDE15;
END;
GET FILE (SYSIN) EDIT (INFIELD) (A(80));
PUT SKIP EDIT (INFIELD)(A);
CALL PLIRETC(12); /*Input to sort continues*/
ENDE15:
RETURN(INFIELD);
END E15X;
E35X: /* Output handling routine prints the sorted records*/
PROC (INREC);
DCL INREC CHAR(80);
PUT SKIP EDIT (INREC) (A);
NEXT: CALL PLIRETC(4); /* Request next record from sort*/
END E35X;
END EX109;
Figure 54. PLISRTDsorting from input handling routine to output handling routine (Part 1 of
2)
/*
//GO.SYSIN DD *
/*
Figure 54. PLISRTDsorting from input handling routine to output handling routine (Part 2 of
2)
Sorting variable-length records example
//OPT14#11 JOB ...
//PLI.SYSIN DD *
/* PL/I EXAMPLE USING PLISRTD TO SORT VARIABLE-LENGTH
RECORDS */
CALL PLISRTD (' SORT FIELDS=(11,14,CH,A) ',
' RECORD TYPE=V,LENGTH=(84,,,24,44) ',
/*NOTE THAT LENGTH IS MAX AND INCLUDES
4 BYTE LENGTH PREFIX*/
1048576
RETURN_CODE,
PUTIN,
PUTOUT);
WHEN(0) PUT SKIP EDIT (
'SORT COMPLETE RETURN_CODE 0') (A);
'SORT FAILED, RETURN_CODE 16') (A);
'SORT MESSAGE DATASET MISSING ') (A);
OTHER PUT SKIP EDIT (
'INVALID RETURN_CODE = ', RETURN_CODE)
(A,F(2));
END /* SELECT */;
/*SET PL/I RETURN CODE TO REFLECT SUCCESS OF SORT*/
PUTIN: PROC RETURNS (CHAR(80) VARYING);
/*OUTPUT HANDLING ROUTINE*/
/*NOTE THAT VARYING MUST BE USED ON RETURNS ATTRIBUTE
WHEN USING VARYING LENGTH RECORDS*/
DCL STRING CHAR(80) VAR;
PUT SKIP EDIT ('END OF INPUT')(A);
CALL PLIRETC(8);
STRING = ;
GOTO ENDPUT;
END;
GET EDIT(STRING)(A(80));
I=INDEX(STRING||' ',' ')-1;/*RESET LENGTH OF THE*/
STRING = SUBSTR(STRING,1,I); /* STRING FROM 80 TO */
/* LENGTH OF TEXT IN */
/* EACH INPUT RECORD.*/
Figure 55. Sorting varying-length records using input and output handling routines (Part 1 of
2)
PUT SKIP EDIT(I,STRING) (F(2),X(3),A);
CALL PLIRETC(12);
ENDPUT: RETURN(STRING);
END;
PUTOUT:PROC(STRING);
/*OUTPUT HANDLING ROUTINE OUTPUT SORTED RECORDS*/
DCL STRING CHAR (*);
/*NOTE THAT FOR VARYING RECORDS THE STRING
PARAMETER FOR THE OUTPUT-HANDLING ROUTINE
SHOULD BE DECLARED ADJUSTABLE BUT CANNOT BE
DECLARED VARYING*/
PUT SKIP EDIT(STRING)(A); /*PRINT THE SORTED DATA*/
CALL PLIRETC(4);
END; /*ENDS PUTOUT*/
END;
/*
//GO.SYSIN DD *
/*
//*
Figure 55. Sorting varying-length records using input and output handling routines (Part 2 of
2)
Chapter 15. ILC with C
This chapter describes some aspects of InterLanguage Communication (ILC)
between PL/I and C. The examples illustrate how to use many of the data types
common to both languages and should enable you to write PL/I code that either
calls or is called by C.
Equivalent data types
The table Table 32 lists the common C and PL/I data type equivalents.
Table 32. C and PL/I Type Equivalents
C type Matching PL/I type
char[...] char(...) varyingz
wchar[...] wchar(...) varyingz
signed char fixed bin(7)
unsigned char unsigned fixed bin(8)
short fixed bin(15)
unsigned short unsigned fixed bin(16)
int fixed bin(31)
unsigned int unsigned fixed bin(32)
long long fixed bin(63)
unsigned long long unsigned fixed bin(64)
float float bin(21)
double float bin(53)
long double float bin(p) (p >= 54)
enum ordinal
typedef define alias
struct define struct
union define union
struct * handle
Simple type equivalence
So, for example, the following illustrates the translation of the simple typedef for
time_t from the C header file time.h:
typedef long time_t;
define alias time_t fixed bin(31);
Figure 56. Simple type equivalence
Struct type equivalence
The following example illustrates the translation of the simple struct for tm from
the C header file time.h:
Enum type equivalence
The following example illustrates the translation of the simple enum __device_t
from the C header file stdio.h:
struct tm {
int tm_sec;
int tm_min;
int tm_hour;
int tm_mday;
int tm_mon;
int tm_year;
int tm_wday;
int tm_yday;
int tm_isdst;
};
define structure
1 tm
,2 tm_sec fixed bin(31)
,2 tm_min fixed bin(31)
,2 tm_hour fixed bin(31)
,2 tm_mday fixed bin(31)
,2 tm_mon fixed bin(31)
,2 tm_year fixed bin(31)
,2 tm_wday fixed bin(31)
,2 tm_yday fixed bin(31)
,2 tm_isdst fixed bin(31)
;
Figure 57. Sample struct type equivalence
File type equivalence
A C file declaration depends on the platform, but it often starts as follows:
All we want is a pointer (or token) for a file, so we can finesse this translation
with:
Using C functions
Let's say we wanted to write a program to read a file and dump it as formatted
hex - using the C functions fopen and fread.
The code for this program is straightforward:
typedef enum {
__disk = 0,
__terminal = 1,
__printer = 2,
__tape = 3,
__tdq = 5,
__dummy = 6,
__memory = 8,
__hfs = 9,
__hiperspace = 10
} __device_t;
define ordinal __device_t (
__disk value(0)
, __terminal value(1)
, __printer value(2)
, __tape value(3)
, __tdq value(4)
, __dummy value(5)
, __memory value(8)
, __hfs value(9)
, __hiperspace value(10)
);
Figure 58. Sample enum type equivalence
struct __file {
unsigned char *__bufPtr;
... } FILE;
Figure 59. Start of the C declaration for its FILE type
define struct 1 file;
define alias file_Handle handle file;
Figure 60. PL/I equivalent for a C file
Chapter 15. ILC with C 411
Most of the declarations in the INCLUDE file filedump are obvious:
Matching simple parameter types
It would be easy to mistranslate the declarations for the C functions. For instance,
you could take the following declaration for the C function fread:
filedump:
proc(fn) options(noexecops main);
dcl fn char(*) var;
%include filedump;
file = fopen( fn, rb );
if file = sysnull() then
do;
display( file could not be opened );
return;
end;
do forever;
unspec(buffer) = b;
read_In = fread( addr(buffer), 1, stg(buffer), file );
if read_In = 0 then
leave;
display( heximage(addr(buffer),16, ) ||
|| translate(buffer,(32).,unprintable) );
if read_In < stg(buffer) then
leave;
end;
call fclose( file );
end filedump;
Figure 61. Sample code to use fopen and fread to dump a file
define struct 1 file;
define alias file_Handle handle file;
define alias size_t unsigned fixed bin(32);
define alias int signed fixed bin(31);
dcl file type(file_Handle);
dcl read_In fixed bin(31);
dcl buffer char(16);
dcl unprintable char(32) value( substr(collate(),1,32) );
Figure 62. Declarations for filedump program
and translate it to:
On some platforms, this would not link successfully because C names are case
sensitive. In order to prevent this kind of linker problem, it is best to specify the
name in mixed case using the extended form of the external attribute. So, for
instance, the declare for fread would be better as:
But this wouldn't run right, because while PL/I parameters are byaddr by default,
C parameters are byvalue by default; so we fix that by adding the byvalue attribute
to the parameters:
But note how the return value is set in Figure 67 on page 414: a fourth parameter
(the address of the temporary _temp5) is passed to the function fread, which is then
expected to place the return code in the integer at that address. This is the
convention for how values are returned when the byaddr attribute applies to
returns, and PL/I uses this convention by default.
size_t fread( void *,
size_t,
size_t,
FILE *);
Figure 63. C declaration of fread
dcl fread ext
entry( pointer,
type size_t,
type size_t,
type file_Handle )
returns( type size_t );
Figure 64. First incorrect declaration of fread
dcl fread ext(fread)
entry( pointer,
type size_t,
type size_t,
type file_Handle )
Figure 65. Second incorrect declaration of fread
entry( pointer byvalue,
type size_t byvalue,
type file_Handle byvalue )
Figure 66. Third incorrect declaration of fread
This wouldn't run right, because C return values are byvalue. So, we fix that with
one more byvalue attribute.
Note how the return value is set now in Figure 69: no extra address is passed, and
the return value is simply returned in register 15.
* read_In = fread( addr(buffer), 1, stg(buffer), file );
*
L r4,FILE(,r13,176)
L r1,fread(,r5,12)
LA r2,_temp5(,r13,420)
LA r8,BUFFER(,r13,184)
L r15,&EPA_&WSA(,r1,8)
ST r0,_CEECAA_(,r12,500)
LA r1,#MX_TEMP1(,r13,152)
ST r8,#MX_TEMP1(,r13,152)
LA r8,1
BALR r14,r15
L r0,_temp5(,r13,420)
ST r0,READ_IN(,r13,180)
Figure 67. Code generated for RETURNS BYADDR
entry( pointer byvalue,
type file_Handle byvalue )
returns( type size_t byvalue );
Figure 68. Correct declaration of fread
* read_In = fread( addr(buffer), 1, stg(buffer), file );
*
L r2,FILE(,r13,176)
L r1,fread(,r5,12)
LA r7,BUFFER(,r13,184)
ST r0,_CEECAA_(,r12,500)
LA r7,1
BALR r14,r15
LR r0,r15
ST r0,READ_IN(,r13,180)
Figure 69. Code generated for RETURNS BYVALUE
Matching string parameter types
Now that we have translated fread correctly, we might try this translation for fopen:
But C really has no strings, only pointers, and these pointers would be passed
byvalue; so the strings should be byaddr:
But PL/I passes descriptors with strings and C doesn't understand them, so they
must be suppressed. We can do this by adding options(nodescriptor) to the
declaration:
This will work, but isn't optimal since the parameters are input-only; if the
parameter is a constant, the nonassignable attribute will prevent a copy being made
and passed. Hence, the best translation of the declaration of fopen is:
At this point, the declare for the fclose function presents few surprises except
perhaps for the optional attribute in the returns specification. This attribute allows
us to invoke the fclose function via a CALL statement and not have to dispose of
the return code. But note that if the file were an output file, the return code on
fclose should always be checked since the last buffer may be written out only
when the file is closed and that write could fail for lack of space.
dcl fopen ext(fopen)
entry( char(*) varyingz byvalue,
char(*) varyingz byvalue )
returns( byvalue type file_handle );
Figure 70. First incorrect declaration of fopen
entry( char(*) varyingz byaddr,
char(*) varyingz byaddr )
returns( byvalue type file_handle );
Figure 71. Second incorrect declaration of fopen
entry( char(*) varyingz byaddr,
char(*) varyingz byaddr )
returns( byvalue type file_handle )
options ( nodescriptor );
Figure 72. Correct declaration of fopen
entry( char(*) varyingz nonasgn byaddr,
char(*) varyingz nonasgn byaddr )
returns( byvalue type file_handle )
Figure 73. Optimal, correct declaration of fopen
Now, on z/OS UNIX, we can compile and run the programs with the commands:
This would produce the following output:
Functions returning ENTRYs
The C quicksort function qsort takes a compare routine. For instance, to sort an
array of integers, the following function (which use the byvalue attribute twice - for
the reasons discussed above) could be used:
And the C qsort function could be used with this compare routine to sort an array
of integers, as in the following code fragment:
dcl fclose ext(fclose)
entry( type file_handle byvalue )
returns( optional type int byvalue )
Figure 74. Declaration of fclose
pli -qdisplay=std filedump.pli
filedump filedump.pli
Figure 75. Commands to compile and run filedump
15408689 938584A4 94977A40 97999683 . filedump: proc
4D86955D 409697A3 899695A2 4D959685 (fn) options(noe
A7858396 97A24094 8189955D 5E151540 xecops main);..
Figure 76. Output of running filedump
comp2:
proc( key, element )
returns( fixed bin(31) byvalue );
dcl (key, element) pointer byvalue;
dcl word based fixed bin(31);
select;
when( key->word < element->word )
return( -1 );
when( key->word = element->word )
return( 0 );
otherwise
return( +1 );
end;
end;
Figure 77. Sample compare routine for C qsort function
But since C function pointers are not the same as PL/I ENTRY variables, the C
qsort function must not be declared as simply:
Recall that a PL/I ENTRY variable may point to a nested function (and thus
requires a backchain address as well as an entry point address). But a C function
pointer is limited in pointing to a non-nested function only and so, a PL/I ENTRY
variable and a C function pointer do not even use the amount of storage.
However, a C function pointer is equivalent to the new PL/I type: a LIMITED
ENTRY. and hence the C qsort function could be declared as:
Linkages
On z/OS, there are two crucial facts about linkages:
v IBM C, JAVA and Enterprise PL/I use the same linkage by default.
v This linkage is not the system linkage.
For a traditional PL/I application where all parameters are byaddr, the differences
between the code generated when a function has the default linkage and when it
has the system linkage would usually not matter. But if the parameters are byvalue
(as they usually are in C and JAVA), the differences can break your code.
dcl a(1:4) fixed bin(31) init(19,17,13,11);
put skip data( a );
call qsort( addr(a), dim(a), stg(a(1)), comp2 );
put skip data( a );
Figure 78. Sample code to use C qsort function
dcl qsort ext(qsort)
entry( pointer,
fixed bin(31),
fixed bin(31),
entry returns( byvalue fixed bin(31) )
)
options( byvalue nodescriptor );
Figure 79. Incorrect declaration of qsort
dcl qsort ext(qsort)
entry( pointer,
fixed bin(31),
fixed bin(31),
limited entry
returns( byvalue fixed bin(31) )
)
options( byvalue nodescriptor );
Figure 80. Correct declaration of qsort
In fact, there is only a small difference if the parameters are byaddr. In Figure 81,
the only difference between the code generated for a function with the default
linkage and for one with the system linkage is that the high-order bit is turned on
for the last parameter of the system linkage call.
This difference would be transparent to most programs.
But, there is a big difference if the parameters are byvalue rather than byaddr. In
Figure 82 on page 419, for the function with the default linkage, register 1 points to
the values of the integers passed, while for the function with the system linkage,
register 1 points to the addresses of those values.
This difference would not be transparent to most programs.
dcl dftv ext entry( fixed bin(31) byaddr
,fixed bin(31) byaddr );
dcl sysv ext entry( fixed bin(31) byaddr
,fixed bin(31) byaddr )
options( linkage(system) );
* call dfta( n, m );
*
LA r0,M(,r13,172)
LA r2,N(,r13,168)
L r15,=V(DFTV)(,r3,126)
BALR r14,r15
*
* call sysa( n, m );
*
LA r0,M(,r13,172)
LA r2,N(,r13,168)
O r0,=X80000000
L r15,=V(SYSV)(,r3,130)
BALR r14,r15
Figure 81. Code when parameters are BYADDR
Sharing output and input
The following topics contain the information about sharing output and input with
a C program by specifying the STDSYS option.
For more information and limitations about the STDSYS option, see the following
manuals:
v STDSYS on page 131 in Chapter 1, Using compiler options and facilities, on
page 5.
v Stream I/O with unprintable characters in Enterprise PL/I for z/OS Compiler and
Run-Time Migration Guide.
Sharing output
If you want to share SYSPRINT with a C program, you must compile your PL/I
code with the STDSYS option.
By default, DISPLAY statements use WTOs to display their output. If you specify
the DISPLAY(STD) compiler option, DISPLAY statements will use the C puts
function to display their output. This can be particularly useful under z/OS UNIX.
Behavior of the standard C stream for sharing output under MVS batch, TSO
batch, IMS batch, and IMS interactive is as follows:
dcl dftv ext entry( fixed bin(31) byvalue
,fixed bin(31) byvalue );
dcl sysv ext entry( fixed bin(31) byvalue
,fixed bin(31) byvalue )
options( linkage(system) );
* call dftv( n, m );
*
L r2,N(,r13,168)
L r0,M(,r13,172)
L r15,=V(DFTV)(,r3,174)
BALR r14,r15
*
* call sysv( n, m );
*
L r1,N(,r13,168)
L r0,M(,r13,172)
ST r0,#wtemp_1(,r13,176)
LA r0,#wtemp_1(,r13,176)
ST r1,#wtemp_2(,r13,180)
LA r2,#wtemp_2(,r13,180)
O r0,=X80000000
L r15,=V(SYSV)(,r3,178)
BALR r14,r15
Figure 82. Code when parameters are BYVALUE
stdout goes first to DD:SYSPRINT. If DD:SYSPRINT does not exist, stdout looks
for DD:SYSTERM and then DD:SYSERR. If neither of these files exist, the library
opens a sysout=* data set using DD SYSPRINT and sends the stdout stream to it.
Sharing input
To share SYSIN with a C program, you must compile the application with the
STDSYS option and open SYSIN as an input stream file or avoid using the DD
names that are reserved by the C Library. You can also copy SYSIN to a temporary
data set in a prior job step and use that as SYSIN in your PL/I job step; it can be
shared when it is not allocated to an instream file.
An input stream file can be opened only once when it is allocated to the SYSIN DD
in JCL.
Behavior of the standard C stream for sharing input under MVS batch, TSO batch,
IMS batch, and IMS interactive is as follows:
stdin goes to DD:SYSIN. If DD:SYSIN does not exist, all read operations from
stdin fails.
Using the ATTACH statement
The ATTACH statement in a PL/I program uses the underlying pthread library to
perform thread management. When the ATTACH statement is issued, the pthread
library tries to allocate standard C stream files.
After the ATTACH statement is issued, any attempt to open SYSIN by the
application fails when SYSIN is an instream file that is allocated to the SYSIN DD in
JCL (for example, //SYSIN DD * or //SYSIN DD DATA). A SYSIN OPEN FAILED error
is issued.
Redirecting C standard streams
Use the STDSYS option in all mixed PL/I and C applications and all
multithreading applications.
If you compile your program with the NOSTDSYS option, it might conflict with
the use of the SYSIN DD name and the SYSPRINT DD name in C.
For example, if your program has the ATTACH statement, it starts the C
environment directly. Starting the C environment causes the SYSIN and SYSPRINT
streams to be opened and closed independent of your PL/I application programs.
In same cases, SYSIN might fail to open or the SYSPRINT data set might be
overwritten.
Summary
What we have learned in this chapter:
v C is case sensitive.
v Parameters should be BYVALUE.
v Return values should be BYVALUE.
v String parameters should be BYADDR.
v Arrays and structures should also be BYADDR.
v No descriptors should be passed.
v Input-only strings should be NONASSIGNABLE.
v C function pointers map to LIMITED ENTRYs.
v The IBM C compilers and the IBM PL/I compilers use the same default linkage
(and it matters).
Chapter 16. Interfacing with Java
This chapter gives a brief description of Java and the Java Native Interface (JNI)
and explains why you might be interested in using it with PL/I. A simple Java -
PL/I application will be described and information on compatibility between the
two languages will also be discussed. Instructions on how to build and run the
Java - PL/I sample applications assume the work is being done in the z/OS UNIX
System Services environment of z/OS.
Before you can communicate with Java from PL/I you need to have Java installed
on your z/OS system. Contact your local System Administrator for more
information on how to set up your z/OS Java environment.
These sample programs have been compiled and tested with Java 2 Version 1.4.2.
To determine the level of Java in your z/OS UNIX System Services environment
enter this command from the command line:
java -version
The active Java version will then be displayed and will look something like:
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2)
Classic VM (build 1.4.2, J2RE 1.4.2 IBM z/OS Persistent Reusable VM build cm142
Java Native Interface (JNI)
Java is an object-oriented programming language invented by Sun Microsystems
and provides a powerful way to make Internet documents interactive.
The Java Native Interface (JNI) is the Java interface to native programming
languages and is part of the Java Development Kits. By writing programs that use
the JNI, you ensure that your code is portable across many platforms.
The JNI allows Java code that runs within a Java Virtual Machine (JVM) to operate
with applications and libraries written in other languages, such as PL/I. In
addition, the Invocation API allows you to embed a Java Virtual Machine into your
native PL/I applications.
Java is a fairly complete programming language; however, there are situations in
which you want to call a program written in another programming language. You
would do this from Java with a method call to a native language, known as a
native method.
Some reasons to use native methods may include the following:
v The native language has a special capability that your application needs and that
the standard Java class libraries lack.
v You already have many existing applications in your native language and you
wish to make them accessible to a Java application.
v You wish to implement a intensive series of complicated calculations in your
native language and have your Java applications call these functions.
v You or your programmers have a broader skill set in your native language and
you do not wish to loose this advantage.
Programming through the JNI lets you use native methods to do many different
operations. A native method can:
v utilize Java objects in the same way that a Java method uses these objects.
v create Java objects, including arrays and strings, and then inspect and use these
objects to perform its tasks.
v inspect and use objects created by Java application code.
v update Java objects that it created or were passed to it, and these updated
objects can then be made available to the Java application.
Finally, native methods can also easily call already existing Java methods,
capitalizing on the functionality already incorporated in the Java programming
framework. In these ways, both the native language side and the Java side of an
application can create, update, and access Java objects and then share these objects
between them.
Calling PL/I program from Java
When a PL/I program is called from a Java program, all files opened in PL/I must
be explicitly closed in PL/I before PL/I returns control to Java for the last time.
Similarly, all modules that are fetched in the PL/I program must be released.
JNI Sample Program #1 - 'Hello World'
Writing Java Sample Program #1
The first sample program is yet another variation of the "Hello World!" program.
The "Hello World!" program has one Java class, callingPLI.java. The native method,
written in PL/I, is contained in hiFromPLI.pli. Here is a brief overview of the steps
for creating this sample program:
1. Write a Java program that defines a class containing a native method, loads the
native load library, and calls the native method.
2. Compile the Java program to create a Java class.
3. Write a PL/I program that implements the native method and displays the
"Hello!" text.
4. Compile and link the PL/I program.
5. Run the Java program which calls the native method in the PL/I program.
Step 1: Writing the Java Program
Declare the Native Method: All methods, whether Java methods or native
methods, must be declared within a Java class. The only difference in the
declaration of a Java method and a native method is the keyword native. The native
keyword tells Java that the implementation of this method will be found in a
native library that will be loaded during the execution of the program. The
declaration of the native method looks like this:
public native void callToPLI();
In the above statement, the void means that there is no return value expected from
this native method call. The empty parentheses in the method name callToPLI( ),
means that there are no parameters being passed on the call to the native method.
Load the Native Library: A step that loads the native library must be included so
the native library will be loaded at execution time. The Java statement that loads
the native library looks like this:
static {
System.loadLibrary("hiFromPLI");
}
In the above statement, the Java System method System.loadLibrary(...) is called to
find and load the native library. The PL/I shared library, libhiFromPLI.so, will
created during the step that compiles and links the PL/I program.
Write the Java Main Method: The callingPLI class also includes a main method to
instantiate the class and call the native method. The main method instantiates
callingPLI and calls the callToPLI() native method.
The complete definition of the callingPLI class, including all the points addressed
above in this section, looks like this:
public class callingPLI {
public native void callToPLI();
static {
System.loadLibrary("hiFromPLI");
}
public static void main(String[] argv) {
callingPLI callPLI = new callingPLI();
callPLI.callToPLI();
System.out.println("And Hello from Java, too!");
}
}
Step 2: Compiling the Java Program
Use the Java compiler to compile the callingPLI class into an executable form. The
command would look like this:
javac callingPLI.java
Step 3: Writing the PL/I Program
The PL/I implementation of the native method looks much like any other PL/I
subroutine.
Useful PL/I Compiler Options: The sample program contains a series of
*PROCESS statements that define the important compiler options.
*Process Limits( Extname( 100 ) ) Margins( 1, 100 ) ;
*Process Display(Std) Dllinit Extrn(Short);
*Process Rent Default( ASCII IEEE );
Here is a brief description of them and why they are useful:
Extname(100)
Allows for longer, Java style, external names.
Margins(1,100)
Extending the margins gives you more room for Java style names and
identifiers.
Display(Std)
Writes the "Hello World" text to stdout, not via WTOs. In the z/OS UNIX
environment WTOs would not be seen by the user.
Dllinit
Includes the initialization coded needed for creating a DLL.
Chapter 16. Interfacing with Java 425
Extrn(Short)
EXTRNs are emitted only for those constants that are referenced. This option is
necessary for Enterprise PL/I V3R3 and later.
Default( ASCII IEEE );
ASCII specifies that CHARACTER and PICTURE data is held in ASCII - the
form in which it is held by JAVA
IEEE specifies that FLOAT data is held in IEEE format - the form in which it is
held by JAVA
RENT
The RENT option insures that code is reentrant even if it writes on static
variables.
Correct Form of PL/I Procedure Name and Procedure Statement: The PL/I
procedure name must conform to the Java naming convention in order to be
located by the Java Class Loader at execution time. The Java naming scheme
consists of three parts. The first part identifies the routine to the Java environment,
the second part is the name of the Java class that defines the native method, and
the third part is the name of the native method itself.
Here is a breakdown of PL/I procedure name Java_callingPLI_callToPLI in the
sample program:
Java
All native methods resident in dynamic libraries must begin with Java
_callingPLI
The name of the Java class that declares the native method
_callToPLI
The name of the native method itself.
Note: There is an important difference between coding a native method in PL/I
and in C. The javah tool, which is shipped with the JDK, generates the form
of the external references required for C programs. When you write your
native methods in PL/I and follow the rules above for naming your PL/I
external references, performing the javah step is not necessary for PL/I
native methods.
In addition, the following options need to be specified in the OPTIONS option on
the PROCEDURE statement:
v FromAlien
v NoDescriptor
v ByValue
The complete procedure statement for the sample program looks like this:
Java_callingPLI_callToPLI:
Proc( JNIEnv , MyJObject )
External( "Java_callingPLI_callToPLI" )
Options( FromAlien NoDescriptor ByValue );
JNI Include File: The two PL/I include files which contain the PL/I definitions of
the Java Native interface are ibmzjni.inc which in turn includes ibmzjnim.inc. These
include files are included with this statement:
%include ibmzjni;
The ibmzjni and ibmzjnim include files are provided in the PL/I SIBMZSAM data
set.
The Complete PL/I Procedure: For completeness, here is the entire PL/I program
that defines the native method:
PliJava_Demo: Package Exports(*);
Java_callingPLI_callToPLI:
Proc( JNIEnv , MyJObject )
External( "Java_callingPLI_callToPLI" )
%include ibmzjni;
Dcl myJObject Type jObject;
Display(Hello from Enterprise PL/I!);
End;
Step 4: Compiling and Linking the PL/I Program
Compiling the PL/I Program: Compile the PL/I sample program with the
following command:
pli -c hiFromPLI.pli
Linking the Shared Library: Link the resulting PL/I object deck into a shared
library with this command:
c89 -o libhiFromPLI.so hiFromPLI.o
Be sure to include the lib prefix on the name of the PL/I shared library or the Java
class loader will not find it.
Step 5: Running the Sample Program
Run the Java - PL/I sample program with this command:
java callingPLI
The output of the sample program will look like this:
Hello from Enterprise PL/I!
And Hello from Java, too!
The first line written from the PL/I native method. The second line is from the
calling Java class after returning from the PL/I native method call.
JNI Sample Program #2 - Passing a String
This sample program passes a string back and forth between Java and PL/I. See
Figure 83 on page 429 for the complete listing of the jPassString.java program. The
Java portion has one Java class, jPassString.java. The native method, written in
PL/I, is contained in passString.pli. Much of the information from the first sample
program applies to this sample program as well. Only new or different aspects will
be discussed for this sample program.
Declare the Native Method: The native method for this sample program looks
like this:
public native void pliShowString();
Load the Native Library: The Java statement that loads the native library for this
sample program looks like this:
static {
System.loadLibrary("passString");
}
Write the Java Main Method: The jPassString class also includes a main method
to instantiate the class and call the native method. The main method instantiates
jPassString and calls the pliShowString() native method.
This sample program prompts the user for a string and reads that value in from
the command line. This is done within a try/catch statement as shown in Figure 83
on page 429.
The command to compile the Java code would look like this:
javac jPassString.java
// Read a string, call PL/I, display new string upon return
import java.io.*;
public class jPassString{
/* Field to hold Java string */
String myString;
/* Load the PL/I native library */
static {
System.loadLibrary("passString");
}
/* Declare the PL/I native method */
public native void pliShowString();
/* Main Java class */
public static void main(String[] arg) {
System.out.println(" ");
/* Instantiate Java class and initialize string */
jPassString myPassString = new jPassString();
myPassString.myString = " ";
/* Prompt user for a string */
try {
BufferedReader in = new BufferedReader(
new InputStreamReader(System.in));
/* Process until quit received */
while (!myPassString.myString.equalsIgnoreCase("quit")) {
System.out.println(
"From Java: Enter a string or quit to quit.");
System.out.print("Java Prompt > ");
/* Get string from command line */
myPassString.myString = in.readLine();
if (!myPassString.myString.equalsIgnoreCase("quit"))
{
/* Call PL/I native method */
myPassString.pliShowString();
/* Return from PL/I and display new string */
System.out.println(
"From Java: String set by PL/I is: "
+ myPassString.myString );
}
}
} catch (IOException e) {
}
}
}
Figure 83. Java Sample Program #2 - Passing a String
All of the information about writing the PL/I "Hello World" sample program
applies to this program as well.
procedure name for this program would be Java_jPassString_pliShowString.
Java_jPassString_pliShowString:
Proc( JNIEnv , myjobject )
external( "Java_jPassString_pliShowString" )
%include ibmzjni;
set.
The Complete PL/I Procedure: The complete PL/I program is shown in Figure 84
on page 431. This sample PL/I program makes several calls through the JNI.
Upon entry, a reference to the calling Java Object, myObject is passed into the PL/I
procedure. The PL/I program will use this reference to get information from the
calling object. The first piece of information is the Class of the calling object which
is retrieved using the GetObjectClass JNI function. This Class value is then used by
the GetFieldID JNI function to get the identity of the Java string field in the Java
object that we are interested in. This Java field is further identified by providing
the name of the field, myString, and the JNI field descriptor, Ljava/lang/String;,
which identifies the field as a Java String field. The value of the Java string field is
then retrieved using the GetObjectField JNI function. Before PL/I can use the Java
string value, it must be unpacked into a form that PL/I can understand. The
GetStringUTFChars JNI function is used to convert the Java string into a PL/I
varyingz string which is then displayed by the PL/I program.
After displaying the retrieved Java string, the PL/I program prompts the user for a
PL/I string to be used to update the string field in the calling Java object. The
PL/I string value is converted to a Java string using the NewString JNI function.
This new Java string is then used to update the string field in the calling Java
object using the SetObjectField JNI function.
When the PL/I program ends control is returned to Java, where the newly updated
Java string is displayed by the Java program.
following command:
pli -c passString.pli
plijava_demo: package exports(*);
Java_passString_pliShowString:
Proc( JNIEnv , myJObject )
external( "Java_jPassString_pliShowString" )
%include ibmzjni;
Dcl myBool Type jBoolean;
Dcl myClazz Type jClass;
Dcl myFID Type jFieldID;
Dcl myJString Type jString;
Dcl newJString Type jString;
Dcl myID Char(9) Varz static init( myString );
Dcl mySig Char(18) Varz static
init( Ljava/lang/String; );
Dcl pliStr Char(132) Varz Based(pliStrPtr);
Dcl pliReply Char(132) Varz;
Dcl pliStrPtr Pointer;
Dcl nullPtr Pointer;
Display( );
/* Get information about the calling Class */
myClazz = GetObjectClass(JNIEnv, myJObject);
/* Get Field ID for String field from Java */
myFID = GetFieldID(JNIEnv, myClazz, myID, mySig );
/* Get the Java String in the string field */
myJString = GetObjectField(JNIEnv, myJObject, myFID );
/* Convert the Java String to a PL/I string */
pliStrPtr = GetStringUTFChars(JNIEnv, myJString, myBool );
Display(From PLI: String retrieved from Java is: || pliStr );
Display(From PLI: Enter a string to be returned to Java: )
reply(pliReply);
/* Convert the new PL/I string to a Java String */
newJString = NewString(JNIEnv, trim(pliReply), length(pliReply) );
/* Change the Java String field to the new string value */
nullPtr = SetObjectField(JNIEnv, myJObject, myFID, newJString);
End;
end;
Figure 84. PL/I Sample Program #2 - Passing a String
c89 -o libpassString.so passString.o
Be sure to include the lib prefix on the name or the PL/I shared library or the Java
java jPassString
The output of the sample program, complete with the prompts for user input from
both Java and PL/I, will look like this:
>java jPassString
From Java: Enter a string or quit to quit.
Java Prompt > A string entered in Java
From PLI: String retrieved from Java is: A string entered in Java
From PLI: Enter a string to be returned to Java:
A string entered in PL/I
From Java: String set by PL/I is: A string entered in PL/I
From Java: Enter a string or quit to quit.
Java Prompt > quit
>
JNI Sample Program #3 - Passing an Integer
This sample program passes an integer back and forth between Java and PL/I. See
Figure 85 on page 433 for the complete listing of the jPassInt.java program. The
Java portion has one Java class, jPassInt.java. The native method, written in PL/I, is
contained in passInt.pli. Much of the information from the first sample program
applies to this sample program as well. Only new or different aspects will be
discussed for this sample program.
Declare the Native Method: The native method for this sample program looks
like this:
public native void pliShowInt();
Load the Native Library: The Java statement that loads the native library for this
sample program looks like this:
static {
System.loadLibrary("passInt");
}
Write the Java Main Method: The jPassInt class also includes a main method to
instantiate the class and call the native method. The main method instantiates
jPassInt and calls the pliShowInt() native method.
This sample program prompts the user for an integer and reads that value in from
the command line. This is done within a try/catch statement as shown in Figure 85
on page 433.
// Read an integer, call PL/I, display new integer upon return
import java.io.*;
import java.lang.*;
public class jPassInt{
/* Fields to hold Java string and int */
int myInt;
String myString;
/* Load the PL/I native library */
static {
System.loadLibrary("passInt");
}
/* Declare the PL/I native method */
public native void pliShowInt();
/* Main Java class */
public static void main(String[] arg) {
/* Instantiate Java class and initialize string */
jPassInt pInt = new jPassInt();
pInt.myInt = 1024;
pInt.myString = " ";
/* Prompt user for an integer */
try {
BufferedReader in = new BufferedReader(
new InputStreamReader(System.in));
/* Process until quit received */
while (!pInt.myString.equalsIgnoreCase("quit")) {
System.out.println
("From Java: Enter an Integer or quit to quit.");
System.out.print("Java Prompt > ");
/* Get string from command line */
pInt.myString = in.readLine();
if (!pInt.myString.equalsIgnoreCase("quit"))
{
/* Set int to integer value of String */
pInt.myInt = Integer.parseInt( pInt.myString );
/* Call PL/I native method */
pInt.pliShowInt();
/* Return from PL/I and display new string */
System.out.println
("From Java: Integer set by PL/I is: " + pInt.myInt );
}
}
} catch (IOException e) {
}
}
}
Figure 85. Java Sample Program #3 - Passing an Integer
The command to compile the Java code would look like this:
javac jPassInt.java
All of the information about writing the PL/I "Hello World" sample program
applies to this program as well.
procedure name for this program would be Java_jPassInt_pliShowInt.
Java_passNum_pliShowInt:
external( "Java_jPassInt_pliShowInt" )
%include ibmzjni;
set.
The Complete PL/I Procedure: The complete PL/I program is shown in Figure 86
Upon entry, a reference to the calling Java object, myObject, is passed into the PL/I
procedure. The PL/I program will use this reference to get information from the
calling object. The first piece of information is the Class of the calling object which
is retrieved using the GetObjectClass JNI function. This Class value is then used by
the GetFieldID JNI function to get the identity of the Java integer field in the Java
object that we are interested in. This Java field is further identified by providing
the name of the field, myInt, and the JNI field descriptor, I, which identifies the
field as an integer field. The value of the Java integer field is then retrieved using
the GetIntField JNI function which is then displayed by the PL/I program.
After displaying the retrieved Java integer, the PL/I program prompts the user for
a PL/I integer to be used to update the integer field in the calling Java object. The
PL/I integer value is then used to update the integer field in the calling Java object
using the SetIntField JNI function.
When the PL/I program ends, control is returned to Java, where the newly
updated Java integer is displayed by the Java program.
following command:
pli -c passInt.pli
c89 -o libpassInt.so passInt.o
Be sure to include the lib prefix on the name or the PL/I shared library or the Java
plijava_demo: package exports(*);
Java_passNum_pliShowInt:
external( "Java_jPassInt_pliShowInt" )
%include ibmzjni;
Dcl myClazz Type jClass;
Dcl myFID Type jFieldID;
Dcl myJInt Type jInt;
dcl rtnJInt Type jInt;
Dcl pliReply Char(132) Varz;
Dcl nullPtr Pointer;
Display( );
/* Get information about the calling Class */
myClazz = GetObjectClass(JNIEnv, myJObject);
/* Get Field ID for int field from Java */
myFID = GetFieldID(JNIEnv, myClazz, "myInt", "I");
/* Get Integer value from Java */
myJInt = GetIntField(JNIEnv, myJObject, myFID);
display(From PLI: Integer retrieved from Java is: || trim(myJInt) );
display(From PLI: Enter an integer to be returned to Java: )
reply(pliReply);
rtnJInt = pliReply;
/* Set Integer value in Java from PL/I */
nullPtr = SetIntField(JNIEnv, myJObject, myFID, rtnJInt);
End;
end;
Figure 86. PL/I Sample Program #3 - Passing an Integer
java jPassInt
The output of the sample program, complete with the prompts for user input from
both Java and PL/I, will look like this:
>java jPassInt
From Java: Enter an Integer or quit to quit.
Java Prompt > 12345
From PLI: Integer retrieved from Java is: 12345
From PLI: Enter an integer to be returned to Java:
54321
From Java: Integer set by PL/I is: 54321
From Java: Enter an Integer or quit to quit.
Java Prompt > quit
>
JNI Sample Program #4 - Java Invocation API
This sample program is a little different from the previous samples. In this sample
PL/I invokes Java first, through the Java Invocation API, creating an embedded
Java Virtual Machine (JVM). PL/I then calls a Java method passing it a string
which the Java method then displays.
The PL/I sample program is named createJVM.pli and the Java method it calls is
contained in javaPart.java.
Since this sample does not use a PL/I native method there is no need to declare
one. Instead, the Java portion for this sample is just a simple Java method.
Write the Java Main Method: The javaPart class contains only one statement. This
statement prints out a short 'Hello World...' from Java then appends the string that
was passed to it from the PL/I program. The entire class is shown in Figure 87.
The command to compile the Java code looks like this:
javac javaPart.java
Most of the information about writing the PL/I "Hello World" sample program
applies to this program as well. However, since in this sample PL/I is calling Java,
there are some additional points to consider.
// Receive a string from PL/I then display it after saying "Hello"
public class javaPart {
public static void main(String[] args) {
System.out.println("From Java - Hello World... " + args[0]);
}
}
Figure 87. Java Sample Program #4 - Receiving and printing a String
Correct Form of PL/I Procedure Name and Procedure Statement: Since in this
sample the PL/I program is calling Java, the PL/I program is MAIN. There is no
need to be concerned about the external name of this PL/I program because it is
not referenced.
createJVM: Proc Options(Main);
the Java Native interface are ibmzjni.inc which in turn includes ibmzjnim.inc. Even
though in this sample PL/I is calling Java, these include files are still necessary.
These include files are included with this statement:
%include ibmzjni;
set.
Linking the PL/I program with the Java library: Since this PL/I sample program
calls Java we need to prepare it to link to the Java library. The Java libraries are
linked with XPLINK and the PL/I modules are not. PL/I can still link to and call
XPLINK libraries but must use the PLIXOPT variable to specify the XPLINK=ON
run-time option. This is what the declare of the PLIXOPT variable would look like:
Dcl PLIXOPT Char(40) Varying Ext Static Init( XPLINK(ON)e );
For a description of using PLIXOPT, see z/OS Language Environment Programming
Guide.
Using the Java Invocation API: This PL/I sample program calls the Java
Invocation API JNI_CreateJavaVM to create its own embedded JVM. This API
requires certain structures to be set up and initialized correctly as shown in
Figure 88 on page 439. First, JNI_GetDefaultJavaVMInitArgs is called to get the
default initilization options. Next, these default options are modified with the
addition of the java.class.path information. Finally, JNI_CreateJavaVM is called to
create the embedded JVM.
The Complete PL/I Program: The complete PL/I program is shown in Figure 88
In this sample the reference to the Java object, a newly created JVM in this case, is
not passed in but is instead returned from the call to the JNI_CreateJavaVM API.
The PL/I program will use this reference to get information from the JVM. The
first piece of information is the Class containing the Java method we wish to call.
This Class is found by using the FindClass JNI function. The Class value is then
used by the GetStaticMethodID JNI function to get the identity of the Java method
that will be called.
Before we call this Java method we need to convert the PL/I string into a format
that Java understands. The PL/I program holds the string in ASCII format. Java
strings are stored in UTF format. In addition, Java strings are not really strings as
PL/I programmers understand them but are themselves a Java class and can only
be modified through methods. We create a Java string by using the NewStringUTF
JNI function. This function returns a Java object called myJString that contains the
PL/I string converted to UTF. Next we create a Java object array by calling the
NewObjectArray JNI function passing it the reference to the myJString object. This
function returns a reference to a Java object array containing the string we want
the Java method to display.
Now the Java method can be called with the CallStaticVoidMethod JNI function and
will then display the string passed to it. After displaying the string, the PL/I
program destroys the embedded JVM with the DestroyJavaVM JNI function and the
PL/I program completes.
The complete source of the PL/I program is in Figure 88 on page 439.
*Process Limits( Extname( 100 ) ) Margins( 1, 100 );
*Process Margins( 1, 100 ) ;
*Process Display(STD) Rent;
*Process Default( ASCII ) Or(|);
createJVM: Proc Options(Main);
%include ibmzjni;
Dcl myJObjArray Type jobjectArray;
Dcl myClass Type jclass;
Dcl myMethodID Type jmethodID;
Dcl myJString Type jstring;
Dcl myRC Fixed Bin(31) Init(0);
Dcl myPLIStr Char(50) Varz
Init(... a PLI string in a Java Virtual Machine!);
Dcl OptStr Char(1024) Varz;
Dcl myNull Pointer;
Dcl VM_Args Like JavaVMInitArgs;
Dcl myOptions Like JavaVMOption;
Dcl PLIXOPT Char(40) Varying Ext Static Init( XPLINK(ON)e );
Display(From PL/I - Beginning execution of createJVM...);
/* Important difference between J1.3 and J1.4: */
/* J1.3 requires VM_Args.version = JNI_VERSION_1_2; */
/* J1.4 requires VM_Args.version = JNI_VERSION_1_4; */
VM_Args.version = JNI_VERSION_1_4;
myRC = JNI_GetDefaultJavaVMInitArgs( addr(VM_Args) );
OptStr = "-Djava.class.path=.:";
myOptions.theOptions = addr(OptStr);
VM_Args.nOptions = 1;
VM_Args.JavaVMOption = addr(myOptions);
/* Create the Java VM */
myrc = JNI_CreateJavaVM(
addr(JavaVM),
addr(JNIEnv),
addr(VM_Args) );
/* Get the Java Class for the javaPart class */
myClass = FindClass(JNIEnv, "javaPart");
/* Get static method ID */
myMethodID = GetStaticMethodID(JNIEnv, myClass,"main",
"([Ljava/lang/String;)V" );
/* Create a Java String Object from the PL/I string. */
myJString = NewStringUTF(JNIenv, myPLIStr);
myJObjArray = NewObjectArray(JNIEnv, 1,
FindClass(JNIEnv,"java/lang/String"), myJString);
Display(From PL/I - Calling Java method in new JVM from PL/I...);
Display( );
myNull = CallStaticVoidMethod(JNIEnv, myClass,
myMethodID, myJObjArray );
/* destroy the Java VM */
Display( );
Display(From PL/I - Destroying the new JVM from PL/I...);
myRC = DestroyJavaVM( JavaVM ); /* rc = -1 is OK */
end;
Figure 88. PL/I Sample Program #4 - Calling the Java Invocation API
following command:
pli -c createJVM.pli
Linking the PL/I program to the Java Library: Link the resulting PL/I object
deck with the Java library with this command:
c89 -o createJVM createJVM.o $JAVA_HOME/bin/classic/libjvm.x
Notice the reference to the $JAVA_HOME environment variable. This variable
should point to the directory that your Java 1.4 product is installed in. For
example, to set up this variable in your environment you would use the following
command:
export JAVA_HOME="/usr/lpp/java/J1.4"
In this case the Java 1.4 product is assumed to be installed in /usr/lpp/java/J1.4.
Run the PL/I - Java sample program with this command:
createJVM
The output of the sample program will look like this:
From PL/I - Beginning execution of createJVM...
From PL/I - Calling Java method in new JVM from PL/I...
From Java - Hello World... ... a PLI string in a Java Virtual Machine!
From PL/I - Destroying the new JVM from PL/I...
Determining equivalent Java and PL/I data types
When you communicate with Java from PL/I you will need to match the data
types between the two programming languages. This table shows Java primitive
types and their PL/I equivalents:
Table 33. Java Primitive Types and PL/I Native Equivalents
Java Type PL/I Type Size in Bits
boolean jboolean 8, unsigned
byte jbyte 8
char jchar 16, unsigned
short jshort 16
int jint 32
long jlong 64
float jfloat 21
double jdouble 53
void jvoid n/a
Part 5. Specialized programming tasks
Chapter 17. Using the PLISAXA and PLISAXB
XML parsers . . . . . . . . . . . . . 443
Overview. . . . . . . . . . . . . . . 443
The PLISAXA built-in subroutine. . . . . . . 444
The PLISAXB built-in subroutine . . . . . . . 444
end_of_document . . . . . . . . . . . 446
start_of_element . . . . . . . . . . . 446
attribute_name . . . . . . . . . . . . 446
attribute_predefined_reference. . . . . . . 447
attribute_character_reference . . . . . . . 447
end_of_element. . . . . . . . . . . . 447
content_predefined_reference . . . . . . . 448
content_character_reference. . . . . . . . 448
comment . . . . . . . . . . . . . . 448
unknown_attribute_reference . . . . . . . 448
unknown_content_reference . . . . . . . 448
start_of_prefix_mapping. . . . . . . . . 448
end_of_prefix_mapping . . . . . . . . . 448
exception. . . . . . . . . . . . . . 448
Supported EBCDIC code pages . . . . . . 450
Supported ASCII code pages . . . . . . . 450
Using a number . . . . . . . . . . 451
Using an alias . . . . . . . . . . . 451
Exceptions . . . . . . . . . . . . . . 451
Example . . . . . . . . . . . . . . . 452
Continuable exception codes . . . . . . . . 464
Terminating exception codes . . . . . . . . 467
Chapter 18. Using the PLISAXC and PLISAXD
XML parsers . . . . . . . . . . . . . 473
Overview. . . . . . . . . . . . . . . 473
The PLISAXC built-in subroutine. . . . . . . 474
The PLISAXD built-in subroutine. . . . . . . 474
end_of_document . . . . . . . . . . . 476
start_of_element . . . . . . . . . . . 476
attribute_name . . . . . . . . . . . . 476
end_of_element. . . . . . . . . . . . 476
comment . . . . . . . . . . . . . . 477
namespace_declare . . . . . . . . . . 477
end_of_input . . . . . . . . . . . . 477
unresolved_reference . . . . . . . . . . 478
exception. . . . . . . . . . . . . . 478
Differences in the events . . . . . . . . 479
Supported code pages . . . . . . . . . 481
Using a number . . . . . . . . . . 481
Using an alias . . . . . . . . . . . 482
Exceptions . . . . . . . . . . . . . . 482
Parsing XML documents with validation . . . . 482
XML schema . . . . . . . . . . . . 482
Creating an OSR . . . . . . . . . . . 483
Example with a simple document . . . . . . 484
Example of using the PLISAXC built-in
subroutine . . . . . . . . . . . . . 484
Example of using the PLISAXD built-in
subroutine . . . . . . . . . . . . . 494
Chapter 19. Using PLIDUMP. . . . . . . . 507
PLIDUMP usage notes . . . . . . . . . . 508
Locating variables in the PLIDUMP output . . . 509
Locating AUTOMATIC variables . . . . . . 509
Locating STATIC variables . . . . . . . . 510
Locating CONTROLLED variables . . . . . 511
Under NORENT WRITABLE . . . . . . 511
Under NORENT NOWRITABLE(FWS) . . . 512
Under NORENT NOWRITABLE(PRV) . . . 513
Saved Compilation Data. . . . . . . . . . 514
Copyright . . . . . . . . . . . . . 514
Timestamp . . . . . . . . . . . . . 514
Saved options string . . . . . . . . . . 515
Chapter 20. Interrupts and attention processing 521
Using ATTENTION ON-units . . . . . . . . 522
Interaction with a debugging tool . . . . . . 522
Chapter 21. Using the Checkpoint/Restart
facility . . . . . . . . . . . . . . . 523
Requesting a checkpoint record . . . . . . . 523
Defining the checkpoint data set . . . . . . 524
Requesting a restart . . . . . . . . . . . 525
Automatic restart after a system failure. . . . 525
Automatic restart within a program . . . . . 525
Getting a deferred restart . . . . . . . . 525
Modifying checkpoint/restart activity . . . . 526
Chapter 22. Using user exits . . . . . . . 527
Procedures performed by the compiler user exit 527
Activating the compiler user exit . . . . . . . 528
The IBM-supplied compiler exit, IBMUEXIT . . 528
Customizing the compiler user exit . . . . . 528
Modifying SYSUEXIT . . . . . . . . . 529
Writing your own compiler exit . . . . . . 529
Structure of global control blocks . . . . . . 529
Writing the initialization procedure . . . . . 530
Writing the message filtering procedure . . . 531
Writing the termination procedure . . . . . 533
Example of suppressing SQL messages . . . . . 533
Chapter 23. PL/I descriptors. . . . . . . . 541
Passing an argument . . . . . . . . . . . 541
Argument passing by descriptor list . . . . . 541
Argument passing by descriptor-locator . . . 542
CMPAT(V*) descriptors . . . . . . . . . . 542
CMPAT(LE) descriptors . . . . . . . . . . 544
Chapter 17. Using the PLISAXA and PLISAXB XML parsers
The PLISAXx (x = A or B) built-in subroutines provide basic XML parsing
capability which allows programs to consume inbound XML documents, check
them for well-formedness, and react to their contents.
These subroutines do not provide XML generation, which must instead be
accomplished by PL/I program logic or by using the XMLCHAR built-in function.
PLISAXA and PLISAXB have no special environmental requirements. They execute
in all the principal runtime environments, including CICS, IMS, MQ Series, z/OS
batch and TSO.
PLISAXA and PLISAXB do have some important limits: they have no support for
XML name spaces, no support for Unicode UTF-8 documents, and they require
that the entire XML document be passed to them (either in a buffer or a file) before
they do any parsing of it. The PLISAXC and PLISAXD built-in subroutines do not
have these limits; the next chapter discusses PLISAXC and PLISAXD in detail.
Overview
There are two major types of interfaces for XML parsing: event-based and
tree-based.
For an event-based API, the parser reports events to the application through
callbacks. Such events include the start of the document, the beginning of an
element, and so on. The application provides handlers to deal with the events
reported by the parser. The Simple API for XML or SAX is an example of an
industry-standard event-based API.
For a tree-based API (such as the Document Object Model or DOM), the parser
translates the XML into an internal tree-based representation. Interfaces are
provided to navigate the tree.
IBM PL/I provides a SAX-like event-based interface for parsing XML documents.
The parser invokes an application-supplied handler for parser events, passing
references to the corresponding document fragments.
The parser has the following characteristics:
v It provides high-performance, but non-standard interfaces.
v It supports XML files encoded in either Unicode UTF-16 or any of several
single-byte code pages listed in the section Coded character sets for XML
documents on page 449.
v The parser is non-validating, but does partially check well-formedness.
XML documents have two levels of conformance: well-formedness and validity,
both of which are defined in the XML standard, which you can find at
http://www.w3c.org/XML/. Recapitulating these definitions, an XML document is
well-formed if it complies with the basic XML grammar, and with a few specific
rules, such as the requirement that the names on start and end element tags must
match. A well-formed XML document is also valid if it has an associated document
type declaration (DTD) and if it complies with the constraints expressed in the
DTD.
The XML parser is non-validating, but does partially check for well-formedness
errors, and generates exception events if it discovers any.
For each parser event, you must provide a PL/I function that accepts the
appropriate parameters and returns the appropriate return value - as in the
example code below. Note in particular that the return value must be returned
BYVALUE. Also, these functions must all use the OPTLINK linkage. You can use
the DEFAULT(LINKAGE(OPTLINK)) option to specify this linkage, or you can
specify it on the individual PROCEDUREs and ENTRYs via the
OPTIONS(LINKAGE(OPTLINK)) attribute.
The PLISAXA built-in subroutine
The PLISAXA built-in subroutine allows you to invoke the XML parser for an XML
document residing in a buffer in your program.
PLISAXA(e,p,x,n )
,c
e An event structure
p A pointer value or "token" that the parser will pass back to the event functions
x The address of the buffer containing the input XML
n The number of bytes of data in that buffer
c A numeric expression specifying the purported codepage of that XML
Note that if the XML is contained in a CHARACTER VARYING or a WIDECHAR
VARYING string, then the ADDRDATA built-in function should be used to obtain
the address of the first data byte.
Also note that if the XML is contained in a WIDECHAR string, the value for the
number of bytes is twice the value returned by the LENGTH built-in function.
The PLISAXB built-in subroutine
The PLISAXB built-in subroutine allows you to invoke the XML parser for an XML
document residing in a file.
PLISAXB(e,p,x )
,c
x A character string expression specifying the input file
c A numeric expression specifying the purported codepage of that XML
Under batch, the character string specifying the input file should have the form
'file://dd:ddname', where ddname is the name of the DD statement specifying the
file.
Under z/OS UNIX, the character string specifying the input file should have the
form 'file://filename', where filenameis the name of a z/OS UNIX file.
Under both batch and z/OS UNIX, the character string specifying the input file
should have no leading or trailing blanks.
The input XML file must be less than 2G in size. Moreover, since the parser will
read the entire file into memory, the REGION for your program must be large
enough for the parser to obtain a piece of storage that can contain all of the
document.
The SAX event structure
The event structure is a structure consisting of 24 LIMITED ENTRY variables
which point to functions that the parser will invoke for various "events".
All these ENTRYs must use the OPTLINK linkage.
The descriptions below of each event refer to the example of an XML document in
Figure 89. In these descriptions, the term "XML text" refers to the string based on
the pointer and length passed to the event.
In the order of their appearance in this structure, the parser may recognize the
following events:
start_of_document
This event occurs once, at the beginning of parsing the document. The parser
passes the address and length of the entire document, including any line-control
characters, such as LF (Line Feed) or NL (New Line). For the above example, the
document is 305 characters in length.
version_information
This event occurs within the optional XML declaration for the version information.
The parser passes the address and length of the text containing the version value,
"1.0" in the example above.
xmlDocument =
<?xml version="1.0" standalone="yes"?>
|| 
|| <sandwich>
|| <bread type="baker"s best"/>
|| <?spread please use real mayonnaise ?>
|| <meat>Ham & turkey</meat>
|| <filling>Cheese, lettuce, tomato, etc.</filling>
|| '<![CDATA[We should add a <relish> element in future!]]>
|| </sandwich>
|| junk;
Figure 89. Sample XML document
Chapter 17. Using the PLISAXA and PLISAXB XML parsers 445
encoding_declaration
This event occurs within the XML declaration for the optional encoding
declaration. The parser passes the address and length of the text containing the
encoding value.
standalone_declaration
This event occurs within the XML declaration for the optional standalone
standalone value, "yes" in the example above.
document_type_declaration
This event occurs when the parser finds a document type declaration. Document
type declarations begin with the character sequence "<!DOCTYPE" and end with a
">" character, with some fairly complicated grammar rules describing the content
in between. The parser passes the address and length of the text containing the
entire declaration, including the opening and closing character sequences, and is
the only event where XML text includes the delimiters. The example above does
not have a document type declaration.
end_of_document
This event occurs once, when document parsing has completed.
start_of_element
This event occurs once for each element start tag or empty element tag. The parser
passes the address and length of the text containing the element name. For the first
start_of_element event during parsing of the example, this would be the string
"sandwich".
attribute_name
This event occurs for each attribute in an element start tag or empty element tag,
after recognizing a valid name. The parser passes the address and length of the
text containing the attribute name. The only attribute name in the example is
"type".
attribute_characters
This event occurs for each fragment of an attribute value. The parser passes the
address and length of the text containing the fragment. An attribute value
normally consists of a single string only, even if it is split across lines:
<element attribute="This attribute value is
split across two lines"/>
The attribute value might consist of multiple pieces, however. For instance, the
value of the "type" attribute in the "sandwich" example at the beginning of the
section consists of three fragments: the string "baker", the single character "'" and
the string "s best". The parser passes these fragments as three separate events. It
passes each string, "baker" and "s best" in the example, as attribute_characters
events, and the single character "'" as an attribute_predefined_reference event,
described next.
attribute_predefined_reference
This event occurs in attribute values for the five pre-defined entity references "&",
"'", ">", "<" and "'". The parser passes a CHAR(1) or WIDECHAR(1) value that
contains one of "&", "'", ">", "<" or '"', respectively.
attribute_character_reference
This event occurs in attribute values for numeric character references (Unicode
code points or "scalar values") of the form "&#dd;" or "&#xhh;", where "d" and "h"
represent decimal and hexadecimal digits, respectively. The parser passes a FIXED
BIN(31) value that contains the corresponding integer value.
end_of_element
This event occurs once for each element end tag or empty element tag when the
parser recognizes the closing angle bracket of the tag. The parser passes the
address and length of the text containing the element name.
start_of_CDATA_section
This event occurs at the start of a CDATA section. CDATA sections begin with the
string <![CDATA[ and end with the string ]], and are used to "escape" blocks
of text containing characters that would otherwise be recognized as XML markup.
The parser passes the address and length of the text containing the opening
characters <![CDATA[. The parser passes the content of a CDATA section
between these delimiters as a single content-characters event. In the preceding
example, the content-characters event is passed the text "We should add a <relish>
element in future!".
end_of_CDATA_section
This event occurs when the parser recognizes the end of a CDATA section. The
parser passes the address and length of the text containing the closing character
sequence, ]].
content_characters
This event represents the "heart" of an XML document: the character data between
element start and end tags. The parser passes the address and length of the text
containing the this data, which usually consists of a single string only, even if it is
split across lines:
<element1>This character content is
split across two lines</element1>
If the content of an element includes any references or other elements, the
complete content may comprise several segments. For instance, the content of the
"meat" element in the example consists of the string "Ham ", the character "&" and
the string " turkey". Notice the trailing and leading spaces, respectively, in these
two string fragments. The parser passes these three content fragments as separate
events. It passes the string content fragments, "Ham " and " turkey", as
content_characters events, and the single "&" character as a
content_predefined_reference event. The parser also uses the content_characters
event to pass the text of CDATA sections to the application.
content_predefined_reference
This event occurs in element content for the five pre-defined entity references "&",
"'", ">", "<" and "'". The parser passes a CHAR(1) or WIDECHAR(1) value that
contains one of "&", "'", ">", "<" or '"', respectively.
content_character_reference
This event occurs in element content for numeric character references (Unicode
code points or "scalar values") of the form "&#dd;" or "&#xhh;", where "d" and "h"
represent decimal and hexadecimal digits, respectively. The parser passes a FIXED
BIN(31) value that contains the corresponding integer value.
processing_instruction
Processing instructions (PIs) allow XML documents to contain special instructions
for applications. This event occurs when the parser recognizes the name following
the PI opening character sequence, "<?". The event also covers the data following
the processing instruction (PI) target, up to but not including the PI closing
character sequence, "?>". Trailing, but not leading white space characters in the
data are included. The parser passes the address and length of the text containing
the target, "spread" in the example, and the address and length of the text
containing the data, "please use real mayonnaise" in the example.
comment
This event occurs for any comments in the XML document. The parser passes the
address and length of the text between the opening and closing comment
delimiters, "", respectively. In the example, the text of the only
comment is "This document is just an example".
unknown_attribute_reference
This event occurs within attribute values for entity references other than the five
pre-defined entity references, listed for the event attribute_predefined_character.
The parser passes the address and length of the text containing the entity name.
unknown_content_reference
This event occurs within element content for entity references other than the five
pre-defined entity references listed for the content_predefined_character event. The
parser passes the address and length of the text containing the entity name.
start_of_prefix_mapping
This event is currently not generated.
end_of_prefix_mapping
This event is currently not generated.
exception
The parser generates this event when it detects an error in processing the XML
document.
Parameters to the event functions
All of these functions must return a BYVALUE FIXED BIN(31) value that is a
return code to the parser. For the parser to continue normally, this value should be
zero.
All of these functions will be passed as the first argument a BYVALUE POINTER
that is the token value passed originally as the second argument to the built-in
function.
With the following exceptions, all of the functions will also be passed a BYVALUE
POINTER and a BYVALUE FIXED BIN(31) that supply the address and length of
the text element for the event. The following functions and events are different:
end_of_document
No argument other than the user token is passed.
attribute_predefined_reference
In addition to the user token, one additional argument is passed: a BYVALUE
CHAR(1) or, for a UTF-16 document, a BYVALUE WIDECHAR(1) that holds
the value of the predefined character.
content_predefined_reference
CHAR(1) or, for a UTF-16 document, a BYVALUE WIDECHAR(1) that holds
the value of the predefined character.
attribute_character_reference
FIXED BIN(31) that holds the value of the numeric reference.
content_character_reference
FIXED BIN(31) that holds the value of the numeric reference.
In addition to the user token, four additional arguments are passed:
1. a BYVALUE POINTER that is the address of the target text
2. a BYVALUE FIXED BIN(31) that is the length of the target text
3. a BYVALUE POINTER that is the address of the data text
4. a BYVALUE FIXED BIN(31) that is the length of the data text
exception
In addition to the user token, three additional arguments are passed:
1. a BYVALUE POINTER that is the address of the offending text
2. a BYVALUE FIXED BIN(31) that is the byte offset of the offending text
within the document
3. a BYVALUE FIXED BIN(31) that is the value of the exception code
Coded character sets for XML documents
The PLISAX built-in subroutine supports only XML documents in WIDECHAR
encoded using Unicode UTF-16, or in CHARACTER encoded using one of the
explicitly supported single-byte character sets listed below. The parser uses up to
three sources of information about the encoding of your XML document, and
signals an exception XML event if it discovers any conflicts between these sources:
1. The parser determines the basic encoding of a document by inspecting its
initial characters.
2. If step 1 succeeds, the parser then looks for any encoding declaration.
3. Finally, it refers to the codepage value on the PLISAX built-in subroutine call. If
this parameter was omitted, it defaults to the value provided by the
CODEPAGE compiler option value that you specified explicitly or by default.
If the XML document begins with an XML declaration that includes an encoding
declaration specifying one of the supported code pages listed below, the parser
honors the encoding declaration if it does not conflict with either the basic
document encoding or the encoding information from the PLISAX built-in
subroutine. If the XML document does not have an XML declaration at all, or if the
XML declaration omits the encoding declaration, the parser uses the encoding
information from the PLISAX built-in subroutine to process the document, as long
as it does not conflict with the basic document encoding.
Supported EBCDIC code pages
In the following table, the first number is for the Euro Country Extended Code
Page (ECECP), and the second is for Country Extended Code Page (CECP).
CCSID Description
01047 Latin 1 / Open Systems
01140, 00037 USA, Canada, etc.
01141, 00273 Austria, Germany
01142, 00277 Denmark, Norway
01143, 00278 Finland, Sweden
01144, 00280 Italy
01145, 00284 Spain, Latin America (Spanish)
01146, 00285 UK
01147, 00297 France
01148, 00500 International
01149, 00871 Iceland
Supported ASCII code pages
CCSID Description
00813 ISO 8859-7 Greek / Latin
00819 ISO 8859-1 Latin 1 / Open Systems
00920 ISO 8859-9 Latin 5 (ECMA-128, Turkey TS-5881)
Specifying the code page
If your document does not include an encoding declaration in the XML
declaration, or does not have an XML declaration at all, the parser uses the
encoding information provided by the PLISAX built-in subroutine call in
conjunction with the basic encoding of the document.
You can also specify the encoding information for the document in the XML
declaration, with which most XML documents begin. An example of an XML
declaration that includes an encoding declaration is:
<?xml version="1.0" encoding="ibm-1140"?>
If your XML document includes an encoding declaration, ensure that it is
consistent with the encoding information provided by the PLISAX built-in
subroutine and with the basic encoding of the document. If there is any conflict
between the encoding declaration, the encoding information provided by the
PLISAX built-in subroutine and the basic encoding of the document, the parser
signals an exception XML event.
Specify the encoding declaration as follows:
Using a number
You can specify the CCSID number (with or without any number of leading
zeroes), prefixed by any of the following (in any mixture of upper or lower case):
IBM_
IBM-
CP
CP_
CP-
CCSID_
CCSID-
Using an alias
You can use any of the following supported aliases (in any mixture of lower and
upper case):
Code page Supported aliases
037 EBCDIC-CP-US, EBCDIC-CP-CA, EBCDIC-CP-WT,
EBCDIC-CP-NL
500 EBCDIC-CP-BE, EBCDIC-CP-CH
813 ISO-8859-7, ISO_8859-7
819 ISO-8859-1, ISO_8859-1
920 ISO-8859-9, ISO_8859-9
1200 UTF-16
Exceptions
For most exceptions, the XML text contains the part of the document that was
parsed up to and including the point where the exception was detected. For
encoding conflict exceptions, which are signaled before parsing begins, the length
of the XML text is either zero or the XML text contains just the encoding
declaration value from the document. The example above contains one item that
causes an exception event, the superfluous "junk" following the "sandwich"
element end tag.
There are two kinds of exceptions:
1. Exceptions that allow you to continue parsing optionally. Continuable
exceptions have exception codes in the range 1 through 99, 100,001 through
165,535, or 200,001 to 265,535. The exception event in the example above has an
exception number of 1 and thus is continuable.
2. Fatal exceptions, which don't allow continuation. Fatal exceptions have
exception codes greater than 99 (but less than 100,000).
Returning from the exception event function with a non-zero return code normally
causes the parser to stop processing the document, and return control to the
program that invoked the PLISAXA or PLISAXB built-in subroutine.
For continuable exceptions, returning from the exception event function with a
zero return code requests the parser to continue processing the document, although
further exceptions might subsequently occur. See section 2.5.6.1, "Continuable
exceptions" for details of the actions that the parser takes when you request
continuation.
A special case applies to exceptions with exception numbers in the ranges 100,001
through 165,535 and 200,001 through 265,535. These ranges of exception codes
indicate that the document's CCSID (determined by examining the beginning of
the document, including any encoding declaration) is not identical to the CCSID
value provided (explicitly or implicitly) by the PLISAXA or PLISAXB built-in
subroutine, even if both CCSIDs are for the same basic encoding, EBCDIC or
ASCII.
For these exceptions, the exception code passed to the exception event contains the
document's CCSID, plus 100,000 for EBCDIC CCSIDs, or 200,000 for ASCII
CCSIDs. For instance, if the exception code contains 101,140, the documents
CCSID is 01140. The CCSID value provided by the PLISAXA or PLISAXB built-in
subroutine is either set explicitly as the last argument on the call or implicitly
when the last argument is omitted and the value of the CODEPAGE compiler
option is used.
Depending on the value of the return code after returning from the exception event
function for these CCSID conflict exceptions, the parser takes one of three actions:
1. If the return code is zero, the parser proceeds using the CCSID provided by the
built-in subroutine.
2. If the return code contains the document's CCSID (that is, the original
exception code value minus 100,000 or 200,000), the parser proceeds using the
documents CCSID. This is the only case where the parser continues after a
non-zero value is returned from one of the parsing events.
3. Otherwise, the parser stops processing the document, and returns control to the
PLISAXA or PLISAXB built-in subroutine which will raise the ERROR
condition.
Example
The following example illustrates the use of the PLISAXA built-in subroutine and
uses the example XML document cited above:
saxtest: package exports(saxtest);
define alias event
limited entry( pointer, pointer, fixed bin(31) )
options( byvalue linkage(optlink) );
define alias event_end_of_document
limited entry( pointer )
define alias event_predefined_ref
limited entry( pointer, char(1) )
options( byvalue linkage(optlink) nodescriptor );
define alias event_character_ref
limited entry( pointer, fixed bin(31) )
define alias event_pi
limited entry( pointer, pointer, fixed bin(31),
pointer, fixed bin(31) )
define alias event_exception
fixed bin(31) )
Figure 90. PLISAXA coding example - type declarations
saxtest: proc options( main );
dcl
1 eventHandler static
,2 e01 type event
init( start_of_document )
,2 e02 type event
init( version_information )
,2 e03 type event
init( encoding_declaration )
,2 e04 type event
init( standalone_declaration )
,2 e05 type event
init( document_type_declaration )
,2 e06 type event_end_of_document
init( end_of_document )
,2 e07 type event
init( start_of_element )
,2 e08 type event
init( attribute_name )
,2 e09 type event
init( attribute_characters )
,2 e10 type event_predefined_ref
init( attribute_predefined_reference )
,2 e11 type event_character_ref
init( attribute_character_reference )
,2 e12 type event
init( end_of_element )
,2 e13 type event
init( start_of_CDATA )
,2 e14 type event
init( end_of_CDATA )
,2 e15 type event
init( content_characters )
,2 e16 type event_predefined_ref
init( content_predefined_reference )
,2 e17 type event_character_ref
init( content_character_reference )
,2 e18 type event_pi
init( processing_instruction )
,2 e19 type event
init( comment )
,2 e20 type event
init( unknown_attribute_reference )
,2 e21 type event
init( unknown_content_reference )
,2 e22 type event
init( start_of_prefix_mapping )
,2 e23 type event
init( end_of_prefix_mapping )
,2 e24 type event_exception
init( exception )
;
Figure 91. PLISAXA coding example - event structure
dcl token char(8);
dcl xmlDocument char(4000) var;
xmlDocument =
|| <sandwich>
|| '<![CDATA[We should add a <relish> element in future!]]>'.
|| </sandwich>
|| junk;
call plisaxa( eventHandler,
addr(token),
addrdata(xmlDocument),
length(xmlDocument) );
end;
Figure 92. PLISAXA coding example - main routine
dcl chars char(32000) based;
start_of_document:
proc( userToken, xmlToken, TokenLength )
dcl userToken pointer;
dcl xmlToken pointer;
dcl tokenLength fixed bin(31);
put skip list( lowercase( procname() )
|| length= || tokenlength );
return(0);
end;
version_information:
|| < || substr(xmltoken->chars,1,tokenlength ) || > );
return(0);
end;
encoding_declaration:
return(0);
end;
Figure 93. PLISAXA coding example - event routines (Part 1 of 8)
standalone_declaration:
return(0);
end;
document_type_declaration:
return(0);
end;
end_of_document:
proc( userToken )
put skip list( lowercase( procname() ) );
return(0);
end;
start_of_element:
return(0);
end;
attribute_name:
return(0);
end;
attribute_characters:
return(0);
end;
attribute_predefined_reference:
proc( userToken, reference )
dcl reference char(1);
|| || hex(reference ) );
return(0);
end;
attribute_character_reference:
dcl reference fixed bin(31);
|| < || hex(reference ) );
return(0);
end;
end_of_element:
return(0);
end;
start_of_CDATA:
return(0);
end;
end_of_CDATA:
return(0);
end;
content_characters:
return(0);
end;
content_predefined_reference:
dcl reference char(1);
|| || hex(reference ) );
return(0);
end;
content_character_reference:
dcl reference fixed bin(31);
|| < || hex(reference ) );
return(0);
end;
processing_instruction:
proc( userToken, piTarget, piTargetLength,
piData, piDataLength )
dcl piTarget pointer;
dcl piTargetLength fixed bin(31);
dcl piData pointer;
dcl piDataLength fixed bin(31);
|| < || substr(piTarget->chars,1,piTargetLength ) || > );
return(0);
end;
comment:
return(0);
end;
unknown_attribute_reference:
return(0);
end;
unknown_content_reference:
return(0);
end;
The preceding program would produce the following output:
start_of_prefix_mapping:
return(0);
end;
end_of_prefix_mapping:
return(0);
end;
exception:
proc( userToken, xmlToken, currentOffset, errorID )
dcl currentOffset fixed bin(31);
dcl errorID fixed bin(31);
|| errorid = || errorid );
return(0);
end;
end;
Continuable exception codes
For each value of the exception code parameter passed to the exception event
(listed under the heading "Number"), the following table describes the exception,
and the actions that the parser takes when you request it to continue after the
exception. In these descriptions, the term "XML text" refers to the string based on
the pointer and length passed to the event.
Table 34. Continuable Exceptions
Number Description Parser Action on Continuation
1 The parser found an invalid character while
scanning white space outside element content.
The parser generates a content_characters event
with XML text containing the (single) invalid
character. Parsing continues at the character
after the invalid character.
2 The parser found an invalid start of a
processing instruction, element, comment or
document type declaration outside element
content.
The parser generates a content_characters event
with the XML text containing the 2- or
3-character invalid initial character sequence.
Parsing continues at the character after the
invalid sequence.
start_of_dcoument length= 305
version_information <1.0>
standalone_declaration <yes>
comment <This document is just an example>
start_of_element <sandwich>
start_of_element <bread>
attribute_name <type>
attribute_characters <baker>
attribute_predefined_reference 7D
attribute_characters <s best>
end_of_element <bread>
processing_instruction <spread>
start_of_element <meat>
content_characters <Ham >
content_predefined_reference 50
content_characters < turkey>
end_of_element <meat>
start_of_element <filling>
content_characters <Cheese, lettuce, tomato, etc.>
end_of_element <filling>
start_of_cdata <<![CDATA[>
content_characters <We should add a <relish> element in future!>
end_of_cdata <]]>
end_of_element <sandwich>
exception errorid = 1
content_characters <j>
content_characters <u>
content_characters <n>
content_characters <k>
end_of_document
Figure 94. PLISAXA coding example - program output
Table 34. Continuable Exceptions (continued)
3 The parser found a duplicate attribute name. The parser generates an attribute_name event
with the XML text containing the duplicate
attribute name.
4 The parser found the markup character "<" in
an attribute value.
Prior to generating the exception event, the
parser generates an attribute_characters event
for any part of the attribute value prior to the
"<" character. After the exception event, the
parser generates an attribute_characters event
with XML text containing "<". Parsing then
continues at the character after the "<".
5 The start and end tag names of an element did
not match.
The parser generates an end_of_element event
with XML text containing the mismatched end
name.
6 The parser found an invalid character in
element content.
The parser includes the invalid character in
XML text for the subsequent content_characters
event.
7 The parser found an invalid start of an element,
comment, processing instruction or CDATA
section in element content.
parser generates a content_characters event for
any part of the content prior to the "<" markup
character. After the exception event, the parser
generates a content_characters event with XML
text containing 2 characters: the "<" followed by
the invalid character. Parsing continues at the
character after the invalid character.
8 The parser found in element content the
CDATA closing character sequence ]] without
the matching opening character sequence
<![CDATA[.
parser generates a content_characters event for
any part of the content prior to the ]]
character sequence. After the exception event,
the parser generates a content_characters event
with XML text containing the 3-character
sequence ]]. Parsing continues at the
character after this sequence.
9 The parser found an invalid character in a
comment.
XML text for the subsequent comment event.
10 The parser found in a comment the character
sequence "--" not followed by ">".
The parser assumes that the "--" character
sequence terminates the comment, and
generates a comment event. Parsing continues
at the character after the "--" sequence.
11 The parser found an invalid character in a
processing instruction data segment.
XML text for the subsequent
processing_instruction event.
12 A processing instruction target name was "xml"
in lower-case, upper-case or mixed-case.
The parser generates a processing_instruction
event with XML text containing "xml" in the
original case.
13 The parser found an invalid digit in a
hexadecimal character reference (of the form
&#xdddd;).
The parser generates an attribute_characters or
content_characters event with XML text
containing the invalid digit. Parsing of the
reference continues at the character after this
invalid digit.
14 The parser found an invalid digit in a decimal
character reference (of the form &#dddd;).
The parser generates an attribute_characters or
content_characters event with XML text
containing the invalid digit. Parsing of the
reference continues at the character after this
invalid digit.
15 The encoding declaration value in the XML
declaration did not begin with lower- or
upper-case A through Z
The parser generates the encoding event with
XML text containing the encoding declaration
value as it was specified.
16 A character reference did not refer to a legal
XML character.
The parser generates an
attribute_character_reference or
content_character_reference event with
XML-NTEXT containing the single Unicode
character specified by the character reference.
17 The parser found an invalid character in an
entity reference name.
The parser includes the invalid character in the
unknown_attribute_reference or
unknown_content_reference event.
18 The parser found an invalid character in an
attribute value.
attribute_characters event.
50 The document was encoded in EBCDIC, and
the CODEPAGE compiler option specified a
supported EBCDIC code page, but the
document encoding declaration did not specify
a recognizable encoding.
The parser uses the encoding specified by the
CODEPAGE compiler option.
the document encoding declaration specified a
supported EBCDIC encoding, but the parser
does not support the code page specified by the
document encoding declaration.
document encoding declaration specified an
ASCII encoding.
document encoding declaration specified a
supported Unicode encoding.
Unicode encoding that the parser does not
support.
document encoding declaration specified an
encoding that the parser does not support.
56 The document was encoded in ASCII, and the
CODEPAGE compiler option specified a
supported ASCII code page, but the document
encoding declaration did not specify a
recognizable encoding.
supported ASCII encoding, but the parser does
not support the code page specified by the
document encoding declaration.
encoding declaration specified a supported
EBCDIC encoding.
encoding declaration specified a supported
Unicode encoding.
encoding declaration specified a Unicode
encoding that the parser does not support.
encoding declaration specified an encoding that
the parser does not support.
100,001
through
165,535
The document was encoded in EBCDIC, and
the encodings specified by the CODEPAGE
compiler option and the document encoding
declaration are both supported EBCDIC code
pages, but are not the same. The exception code
contains the CCSID for the encoding
declaration plus 100,000.
If you return zero from the exception event, the
parser uses the encoding specified by the
CODEPAGE compiler option. If you return the
CCSID from the document encoding declaration
(by subtracting 100,000 from the exception
code), the parser uses this encoding.
200,001
through
265,535
The document was encoded in ASCII, and the
encodings specified by the CODEPAGE
compiler option and the document encoding
declaration are both supported ASCII code
pages, but are not the same. The exception code
contains the CCSID for the encoding
declaration plus 200,000.
If you return zero from the exception event, the
parser uses the encoding specified by the
CODEPAGE compiler option. If you return the
CCSID from the document encoding declaration
(by subtracting 200,000 from the exception
code), the parser uses this encoding.
Terminating exception codes
Table 35. Terminating Exceptions
Number Description
100 The parser reached the end of the document while scanning the start of the XML
declaration.
Table 35. Terminating Exceptions (continued)
Number Description
101 The parser reached the end of the document while looking for the end of the XML
declaration.
102 The parser reached the end of the document while looking for the root element.
103 The parser reached the end of the document while looking for the version information
in the XML declaration.
104 The parser reached the end of the document while looking for the version information
value in the XML declaration.
106 The parser reached the end of the document while looking for the encoding declaration
value in the XML declaration.
108 The parser reached the end of the document while looking for the standalone
declaration value in the XML declaration.
109 The parser reached the end of the document while scanning an attribute name.
110 The parser reached the end of the document while scanning an attribute value.
111 The parser reached the end of the document while scanning a character reference or
entity reference in an attribute value.
112 The parser reached the end of the document while scanning an empty element tag.
113 The parser reached the end of the document while scanning the root element name.
114 The parser reached the end of the document while scanning an element name.
115 The parser reached the end of the document while scanning character data in element
content.
116 The parser reached the end of the document while scanning a processing instruction in
element content.
117 The parser reached the end of the document while scanning a comment or CDATA
section in element content.
118 The parser reached the end of the document while scanning a comment in element
content.
119 The parser reached the end of the document while scanning a CDATA section in
element content.
120 The parser reached the end of the document while scanning a character reference or
entity reference in element content.
121 The parser reached the end of the document while scanning after the close of the root
element.
122 The parser found a possible invalid start of a document type declaration.
123 The parser found a second document type declaration.
124 The first character of the root element name was not a letter, '_' or ':'.
125 The first character of the first attribute name of an element was not a letter, '_' or ':'.
126 The parser found an invalid character either in or following an element name.
127 The parser found a character other than '=' following an attribute name.
128 The parser found an invalid attribute value delimiter.
130 The first character of an attribute name was not a letter, '_' or ':'.
131 The parser found an invalid character either in or following an attribute name.
132 An empty element tag was not terminated by a '>' following the '/'.
133 The first character of an element end tag name was not a letter, '_' or ':'.
134 An element end tag name was not terminated by a '>'.
Number Description
135 The first character of an element name was not a letter, '_' or ':'.
136 The parser found an invalid start of a comment or CDATA section in element content.
137 The parser found an invalid start of a comment.
138 The first character of a processing instruction target name was not a letter, '_' or ':'.
139 The parser found an invalid character in or following a processing instruction target
name.
140 A processing instruction was not terminated by the closing character sequence '?>'.
141 The parser found an invalid character following '&' in a character reference or entity
reference.
142 The version information was not present in the XML declaration.
143 'version' in the XML declaration was not followed by a '='.
144 The version declaration value in the XML declaration is either missing or improperly
delimited.
145 The version information value in the XML declaration specified a bad character, or the
start and end delimiters did not match.
146 The parser found an invalid character following the version information value closing
delimiter in the XML declaration.
147 The parser found an invalid attribute instead of the optional encoding declaration in the
XML declaration.
148 'encoding' in the XML declaration was not followed by a '='.
149 The encoding declaration value in the XML declaration is either missing or improperly
delimited.
150 The encoding declaration value in the XML declaration specified a bad character, or the
start and end delimiters did not match.
151 The parser found an invalid character following the encoding declaration value closing
delimiter in the XML declaration.
152 The parser found an invalid attribute instead of the optional standalone declaration in
the XML declaration.
153 'standalone' in the XML declaration was not followed by a '='.
154 The standalone declaration value in the XML declaration is either missing or improperly
delimited.
155 The standalone declaration value was neither 'yes' nor 'no' only.
156 The standalone declaration value in the XML declaration specified a bad character, or
the start and end delimiters did not match.
157 The parser found an invalid character following the standalone declaration value
closing delimiter in the XML declaration.
158 The XML declaration was not terminated by the proper character sequence '?>', or
contained an invalid attribute.
159 The parser found the start of a document type declaration after the end of the root
element.
160 The parser found the start of an element after the end of the root element.
300 The document was encoded in EBCDIC, but the CODEPAGE compiler option specified
a supported ASCII code page.
Unicode.
Number Description
an unsupported code page.
303 The document was encoded in EBCDIC, but the CODEPAGE compiler option is
unsupported and the document encoding declaration was either empty or contained an
unsupported alphabetic encoding alias.
unsupported and the document did not contain an encoding declaration.
unsupported and the document encoding declaration did not specify a supported
EBCDIC encoding.
306 The document was encoded in ASCII, but the CODEPAGE compiler option specified a
supported EBCDIC code page.
307 The document was encoded in ASCII, but the CODEPAGE compiler option specified
Unicode.
308 The document was encoded in ASCII, but the CODEPAGE compiler option did not
specify a supported EBCDIC code page, ASCII or Unicode.
309 The CODEPAGE compiler option specified a supported ASCII code page, but the
document was encoded in Unicode.
310 The CODEPAGE compiler option specified a supported EBCDIC code page, but the
311 The CODEPAGE compiler option specified an unsupported code page, but the
312 The document was encoded in ASCII, but both the encodings provided externally and
within the document encoding declaration are unsupported.
313 The document was encoded in ASCII, but the CODEPAGE compiler option is
unsupported and the document did not contain an encoding declaration.
314 The document was encoded in ASCII, but the CODEPAGE compiler option is
unsupported and the document encoding declaration did not specify a supported ASCII
encoding.
315 The document was encoded in UTF-16 Little Endian, which the parser does not support
on this platform.
316 The document was encoded in UCS4, which the parser does not support.
317 The parser cannot determine the document encoding. The document may be damaged.
318 The document was encoded in UTF-8, which the parser does not support.
319 The document was encoded in UTF-16 Big Endian, which the parser does not support
on this platform.
501, 502, 503 An internal error occurred in PLISAX(A|B). Contact IBM support.
500 Memory allocation failed for the PLISAXA internal data structures. Increase the amount
of storage available to the application program.
520 Memory allocation failed for the PLISAXB internal data structures. Increase the amount
of storage available to the application program.
521 An internal error occurred in PLISAX(A|B). Contact IBM support.
523 PLISAXB encountered a file I/O error.
524 Memory allocation failed in PLISAXB while attempting to cache the XML document
from the file system. Increase the amount of storage available to the application
program.
525 An unsupported URI scheme was specified to PLISAXB.
Number Description
526 The XML document provided to PLISAXB was less than the minimum of 4 characters
or was too large.
527, 560 An internal error occurred in PLISAX(A|B). Contact IBM support.
561 No event handler was specified to either PLISAX(A|B).
562, 563, 580, 581 An internal error occurred in PLISAX(A|B). Contact IBM support.
600
to
99,999
Internal error. Report the error to your service representative.
Chapter 18. Using the PLISAXC and PLISAXD XML parsers
The PLISAXC and PLISAXD built-in subroutines provide basic XML parsing
capability, which allows programs to consume inbound XML documents, check
them for well-formedness, and react to their contents.
The XML parser used by PLISAXC is non-validating, but does partially check for
well-formedness errors, and generates exception events if it discovers any.
The PLISAXD built-in subroutine provides XML parsing with validation capability.
It determines whether an inbound XML documentation conforms to a set of rules
specified in an inbound XML schema.
The PLISAXC and PLISAXD subroutines do not provide XML generation, which
must instead be accomplished by PL/I program logic or by using the XMLCHAR
built-in function.
PLISAXC and PLISAXD have no special environmental requirements except that it
is not supported in AMODE 24. It executes in all the principal runtime
environments, including CICS, IMS, MQ Series, z/OS batch, and TSO.
Because the PLISAXC and PLISAXD built-in subroutines and the PLISAXA and
PLISAXB built-in subroutines do have much similarity, some of the discussion
below repeats material from the previous chapter.
Overview
There are two major types of interfaces for XML parsing: event-based and
tree-based.
For an event-based API, the parser reports events to the application through
callbacks. Such events include the start of the document, the beginning of an
element, and so on. The application provides handlers to deal with the events
reported by the parser. The Simple API for XML or SAX is an example of an
industry-standard event-based API.
For a tree-based API (such as the Document Object Model or DOM), the parser
translates the XML into an internal tree-based representation. Interfaces are
provided to navigate the tree.
IBM PL/I compiler provides, by using PLISAXC or PLISAXD, a SAX-like
event-based interface for parsing XML documents. The parser invokes an
application-supplied handler for parser events, passing references to the
corresponding document fragments.
The parser has the following characteristics:
v It provides high-performance, but non-standard interfaces.
v It supports XML files encoded in either Unicode UTF-16, UTF-8 or any of several
single-byte code pages listed in the section Coded character sets for XML
documents on page 480.
XML documents have two levels of conformance: well-formedness and validity,
both of which are defined in the XML standard, which you can find at
http://www.w3c.org/XML/. Recapitulating these definitions, an XML document is
well-formed if it complies with the basic XML grammar, and with a few specific
rules, such as the requirement that the names on start and end element tags must
match. A well-formed XML document is also valid if it has an associated document
type declaration (DTD) and if it complies with the constraints expressed in the
DTD.
For each parser event, you must provide a PL/I function that accepts the
appropriate parameters and returns the appropriate return value - as in the
example code below. Note in particular that for these functions
v the return value must be returned BYVALUE
v the linkage used must be the OPTLINK linkage. You can use the
DEFAULT(LINKAGE(OPTLINK)) option to specify this linkage, or you can
specify it on the individual PROCEDUREs and ENTRYs by using the
OPTIONS(LINKAGE(OPTLINK)) attribute.
The PLISAXC built-in subroutine
The PLISAXC built-in subroutine allows you to invoke the XML parser for an XML
document residing in one or more buffers in your program.
PLISAXC(e,p,x,n )
,c
x The address of the initial buffer containing the input XML
c A numeric expression specifying the codepage of that XML
Note that if the XML is contained in a CHARACTER VARYING or a WIDECHAR
VARYING string, then the ADDRDATA built-in function should be used to obtain
the address of the first data byte.
Also note that if the XML is contained in a WIDECHAR string, the value for the
number of bytes is twice the value returned by the LENGTH built-in function.
The PLISAXD built-in subroutine
The PLISAXD built-in subroutine allows you to invoke the XML parser with the
validation capability. Both the XML document and the Optimized Schema
Representation (OSR) file are in one or more buffers in your program.
PLISAXD(e,p,x,n,o )
,c
p A pointer value or "token" that the parser passes back to the event functions
x The address of the initial buffer containing the input XML
o The address of the buffer containing the input OSR
c A numeric expression specifying the codepage of the XML document
If the XML is contained in a CHARACTER VARYING or a WIDECHAR VARYING
string, the ADDRDATA built-in function should be used to obtain the address of
the first data byte.
If the XML is contained in a WIDECHAR string, the value for the number of bytes
is twice the value returned by the LENGTH built-in function.
Note: An OSR is the preprocessed version of a schema. For more information
about OSR, see XML System Services Users Guide and Reference.
The SAX event structure
The event structure is a structure consisting of 19 LIMITED ENTRY variables,
which point to functions that the parser invokes for various "events".
All of these ENTRYs must use the OPTLINK linkage.
All of these ENTRYs have a first (and sometimes the only) parameter: the user
token passed by the program to PLISAXC and PLISAXD.
The descriptions below of these 19 events refer to the example of an XML
document in Figure 95. In these descriptions, the term "XML text" refers to the
string based on the pointer and length passed to the event.
Depending on the contents of the XML documents, the parser might recognize the
following events:
start_of_document
This event occurs once, at the beginning of parsing the document. The parser
passes no parameters to this event (except the user token).
version_information
This event occurs within the optional XML declaration for the version information.
The parser passes the address and length of the text containing the version value,
"1.0" in the previous example.
xmlDocument =
|| <sandwich>
|| '<![CDATA[We should add a <relish> element in future!]]>
|| </sandwich>;
Figure 95. Sample XML document
Chapter 18. Using the PLISAXC and PLISAXD XML parsers 475
encoding_declaration
This event occurs within the XML declaration for the optional encoding
encoding value.
standalone_declaration
This event occurs within the XML declaration for the optional standalone
standalone value, "yes" in the previous example.
document_type_declaration
This event occurs when the parser finds a document type declaration. Document
type declarations begin with the character sequence "<!DOCTYPE" and end with a
">" character, with some fairly complicated grammar rules describing the content
in between. The parser passes the address and length of the text containing the
entire declaration, including the opening and closing character sequences, and is
the only event where XML text includes the delimiters. The example above does
not have a document type declaration.
end_of_document
This event occurs once, when document parsing has completed. The parser passes
no parameters to this event (except the user token).
start_of_element
This event occurs once for each element start tag or empty element tag. The parser
passes the address and length of the text containing the element name as well as
any applicable namespace information. For the first start_of_element event during
parsing of the example, this would be the string "sandwich".
attribute_name
This event occurs for each attribute in an element start tag or empty element tag,
after recognizing a valid name. The parser passes the address and length of the
text containing the attribute name as well as any applicable namespace
information. The only attribute name in the example is "type".
attribute_characters
This event occurs for each attribute value. The parser passes the address and
length of the text containing the fragment. An attribute value normally consists of
a single string only, even if it is split across lines:
<element attribute="This attribute value is
split across two lines"/>
end_of_element
This event occurs once for each element end tag or empty element tag when the
parser recognizes the closing angle bracket of the tag. The parser passes the
address and length of the text containing the element name as well as any
applicable namespace information.
start_of_CDATA_section
This event occurs at the start of a CDATA section. CDATA sections begin with the
string <![CDATA[ and end with the string ]], and are used to "escape" blocks
of text containing characters that would otherwise be recognized as XML markup.
The parser passes no parameters to this event (except the user token). After this
event, the parser passes the content of the CDATA section between these delimiters
as one or more content-characters events. In the preceding example, the
content-characters event is passed the text "We should add a <relish> element in
future!".
end_of_CDATA_section
This event occurs when the parser recognizes the end of a CDATA section. The
parser passes no parameters to this event (except the user token).
content_characters
This event represents the "heart" of an XML document: the character data between
element start and end tags. The parser passes the address and length of the text
containing the data, which usually consists of a single string only, even if it is split
across lines:
<element1>This character content is
split across two lines</element1>
The parser also passes a flag byte which indicates if the next event will provide
additional characters that form part of the content. This can be true when there is a
lot of data between the start and end tags.
The parser also uses the content_characters event to pass the text of CDATA
sections to the application.
Processing instructions (PIs) allow XML documents to contain special instructions
for applications. This event occurs when the parser recognizes the name following
the PI opening character sequence "<?". The event also covers the data following
the processing instruction (PI) target, up to but not including the PI closing
character sequence "?>". Trailing, but not leading white space characters in the data
are included. The parser passes the address and length of the text containing the
target, "spread" in the example, and the address and length of the text containing
the data, "please use real mayonnaise" in the example.
comment
This event occurs for any comments in the XML document. The parser passes the
address and length of the text between the opening and closing comment
delimiters, "", respectively. In the example, the text of the only
comment is "This document is just an example".
namespace_declare
This event occurs for any namespace declarations in the XML document. The
parser passes the address and length of the namespace prefix (if any) as well as the
address and length of the namespace uri. If there is no namespace prefix, the
passed length will be zero and the value of the address should not be used. There
is no corresponding event in the PLIXSAXA and PLISAXB built-in subroutines.
end_of_input
This event occurs whenever the parser reaches the end of the current input buffer.
The parser passes (along with the BYVALUE user token) two BYADDR parameters:
the address and length of the next buffer for it to process. Note that this and the
content character events are the only events that have any BYADDR parameters,
but this is the only event that has parameters that the called event should change.
There is no corresponding event in the PLIXSAXA and PLISAXB built-in
subroutines, and it is this event that allows PLISAXC and PLISAXD to parse an
XML document of arbitrary size.
unresolved_reference
This event occurs for any unresolved references in the XML document. The parser
passes the address and length of the unresolved reference.
exception
The parser generates this event when it detects an error in processing the XML
document.
Parameters to the event functions
All of these functions must return a BYVALUE FIXED BIN(31) value that is a
return code to the parser. If any value other than zero is returned, the parser will
terminate.
All of these functions are passed as the first argument a BYVALUE POINTER. This
pointer is the token value that is passed originally as the second argument to the
built-in function.
With the following exceptions, all of the functions will also be passed a BYVALUE
POINTER and a BYVALUE FIXED BIN(31) that supply the address and length of
the text element for the event. The following functions and events are different:
start_of_document
end_of_document
start_of_CDATA
end_of_CDATA
start_of_element
In addition to the usual three parameters, four additional arguments are
passed:
1. a BYVALUE POINTER that is the address of the namespace prefix
2. a BYVALUE FIXED BIN(31) that is the length of the namespace prefix
3. a BYVALUE POINTER that is the address of the namespace uri
4. a BYVALUE FIXED BIN(31) that is the length of the namespace uri
end_of_element
passed:
attribute_name
passed:
namespace_declare
In addition to the user token, four additional arguments passed:
content_characters
In addition to the usual three parameters, one additional argument is passed:
v a BYVALUE ALIGNED BIT(8) flag byte that indicates
if more content characters will be presented in the next event - this is true
if the first bit is on, that is, this is true if this field anded with '80'BX is
non-null
if there are no characters that need to be escaped if converted back to
XML - this is true if the second bit is on, that is, this is true if this field
anded with '40'BX is non-null
Note that this entry must also be declared with OPTIONS(NODESCRIPTOR).
end_of_input
In addition to the user token, two additional arguments are passed:
1. a BYADDR POINTER that is the address of the next input buffer
2. a BYADDR FIXED BIN(31) that is the length of the next input buffer
In addition to the user token, four additional arguments are passed:
1. a BYVALUE POINTER that is the address of the target text
2. a BYVALUE FIXED BIN(31) that is the length of the target text
3. a BYVALUE POINTER that is the address of the data text
4. a BYVALUE FIXED BIN(31) that is the length of the data text
exception
In addition to the user token, three additional arguments are passed:
1. a BYVALUE FIXED BIN(31) that is the byte offset of the offending text
within the document
2. a BYVALUE FIXED BIN(31) that is the return code for the exception
3. a BYVALUE FIXED BIN(31) that is the reason code for the exception
Differences in the events
The following events are part of the PLISAXA and PLISAXB event structure, but
are not in PLISAXC and PLISAXD:
v attribute_predefined_reference
v attribute_character_reference
v content_predefined_reference
v content_character_reference
v unknown_attribute_reference
v unknown_content_reference
v start_of_prefix_mapping
v end_of_prefix_mapping
The following events are not part of the PLISAXA and PLISAXB event structure,
but are in PLISAXC and PLISAXD:
v namespace_declare
v unresolved_reference
v end_of_input
Some of the events that are common to PLISAXA and PLISAXB, PLISAXC and
PLISAXD are passed different parameters (apart from the omnipresent user token):
start_of_document
is passed no parameters
start_of_element
is passed namespace data as well
end_of_element
attribute_name
content_characters
is passed a flag byte as well
exception
is passed a return and reason code instead of an error id
start_of_cdata
end_of_cdata
Coded character sets for XML documents
The PLISAXC and PLISAXD built-in subroutines support only XML documents in
WIDECHAR encoded using Unicode UTF-16 or in CHARACTER encoded using
either UTF-8 or one of the explicitly supported single-byte character sets listed
below. The parser uses up to three sources of information about the encoding of
your XML document, and signals an exception XML event if it discovers any
conflicts between these sources:
1. The parser determines the basic encoding of a document by inspecting its
initial characters.
2. If step 1 succeeds, the parser then looks for any encoding declaration.
3. Finally, it refers to the codepage value on the PLISAXC or PLISAXD built-in
subroutine call. If this parameter was omitted, it defaults to the value provided
by the CODEPAGE compiler option value that you specified explicitly or by
default.
If the XML document begins with an XML declaration that includes an encoding
declaration specifying one of the supported code pages listed below, the parser
honors the encoding declaration if it does not conflict with either the basic
document encoding or the encoding information from the PLISAXC or PLISAXD
built-in subroutines. If the XML document does not have an XML declaration at all,
or if the XML declaration omits the encoding declaration, the parser uses the
encoding information from the PLISAXC or PLISAXD built-in subroutine to
process the document, as long as it does not conflict with the basic document
encoding.
Supported code pages
In the following table, the first number is for the Euro Country Extended Code
Page (ECECP), and the second is for Country Extended Code Page (CECP).
CCSID Description
01208 Unicode UTF-8
01047 Latin 1 / Open Systems
01140, 00037 USA, Canada, etc.
01141, 00273 Austria, Germany
01142, 00277 Denmark, Norway
01143, 00278 Finland, Sweden
01144, 00280 Italy
01145, 00284 Spain, Latin America (Spanish)
01146, 00285 UK
01147, 00297 France
01148, 00500 International
01149, 00871 Iceland
Specifying the code page
If your document does not include an encoding declaration in the XML
declaration, or does not have an XML declaration at all, the parser uses the
encoding information provided by the PLISAXC or PLISAXD built-in subroutine
call in conjunction with the basic encoding of the document.
You can also specify the encoding information for the document in the XML
declaration, with which most XML documents begin. An example of an XML
declaration that includes an encoding declaration is:
<?xml version="1.0" encoding="ibm-1140"?>
If your XML document includes an encoding declaration, ensure that it is
consistent with the encoding information provided by the PLISAXC or PLISAXD
built-in subroutine and with the basic encoding of the document. If there is any
conflict between the encoding declaration, the encoding information provided by
the PLISAXC or PLISAXD built-in subroutine and the basic encoding of the
document, the parser signals an exception XML event.
Specify the encoding declaration as follows:
Using a number
You can specify the CCSID number (with or without any number of leading
zeroes), prefixed by any of the following (in any mixture of upper or lower case):
IBM_
IBM-
CP
CP_
CP-
CCSID_
CCSID-
Using an alias
You can use any of the following supported aliases (in any mixture of lower and
upper case):
Code page Supported aliases
037 EBCDIC-CP-US, EBCDIC-CP-CA, EBCDIC-CP-WT,
EBCDIC-CP-NL
500 EBCDIC-CP-BE, EBCDIC-CP-CH
813 ISO-8859-7, ISO_8859-7
819 ISO-8859-1, ISO_8859-1
920 ISO-8859-9, ISO_8859-9
1200 UTF-16
Exceptions
If an exception event occurs, the reason and return codes passed to it are those
from the XML System Services parser, and the documentation provided with that
parser explains what these return and reason codes mean.
Parsing XML documents with validation
The PLISAXD built-in subroutine not only parses XML documents in the same
manner as PLISAXC, but also determines whether an inbound XML document
conforms to a set of rules specified in an inbound XML schema.
The inbound schema used for XML validation when using the PLISAXD built-in
subroutine must be in a preprocessed format known as an Optimized Schema
Representation (OSR).
The following information contains the description of the XML schema and the
way to build an OSR for an XML schema. For an example of using PLISAXD
built-in subroutine to parse XML documents with validation, see Example of
using the PLISAXD built-in subroutine on page 494.
XML schema
An XML schema is a mechanism, defined by the W3C, for describing and
constraining the structure and content of XML documents. Through its support for
datatypes and namespaces, an XML schema has the potential to provide the
standard structure for XML elements and attributes. Therefore, an XML schema,
which is itself expressed in XML, can effectively define a class of XML documents
of a given type, for example, stock item.
The following sample XML document describes an item for stock keeping
purposes:
|| 
|| <stockItem itemNumber="453-SR">
|| <itemName>Stainless steel rope thimbles</itemName>
|| <quantityOnHand>23</quantityOnHand>
|| <stockItem>;
The stock keeping example document above is both well formed and valid
according to the following schema called stock.xsd. (The numbers that precede
each line are not part of the schema, but are used in the explanation after the
schema.)
1. <xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema">
2.
3. <xsd:element name="stockItem" type="stockItemType"/>
4.
5. <xsd:complexType name="stockItemType">
6. <xsd:sequence>
7. <xsd:element name="itemName" type="xsd:string" minOccurs="0"/>
8. <xsd:element name="quantityOnHand">
9. <xsd:simpleType>
10. <xsd:restriction base="xsd:nonNegativeInteger">
11. <xsd:maxExclusive value="100"/>
12. </xsd:restriction>
13. </xsd:simpleType>
14. </xsd:element>
15. </xsd:sequence>
16. <xsd:attribute name="itemNumber" type="SKU" use="required"/>
17. </xsd:complexType>
18.
19. <xsd:simpleType name="SKU">
20. <xsd:restriction base="xsd:string">
21. <xsd:pattern value="\d{3}-[A-Z]{2}"/>
22. </xsd:restriction>
23. </xsd:simpleType>
24.
25. </xsd:schema>
The schema declares (line 3) that the root element is stockItem, which has a
mandatory itemNumber attribute (line 16) of type SKU, and includes a sequence
(lines 6 - 15) of other elements:
v An optional itemName element of type string (line 7)
v A required quantityOnHand element that has a constrained range of 1 - 99 based
on the type nonNegativeInteger (lines 8 - 14)
Type declarations can be inline and unnamed, as in lines 9 - 13, which include the
maxExclusive facet to specify the legal values for the quantityOnHand element.
For the itemNumber attribute, by contrast, the named type SKU is declared separately
in lines 19 - 23, which include a pattern facet that uses regular expression syntax to
specify that the legal values for that type consist of (in order): 3 digits, a
hyphen-minus, then two uppercase letters.
Creating an OSR
To generate a schema in the OSR format from a text-form schema, use the z/OS
UNIX command xsdosrg, which invokes the OSR generator provided by z/OS
System Service.
For example, to convert the text-form schema in the stock.xsd file to a schema in
preprocessed format in the stock.osr file, you can use the following z/OS UNIX
command:
xsdosrg -v o /u/HLQ/xml/stock.osr /u/HLQ/xml/stock.xsd
where /u/HLQ/xml/ is the directory in which the stock.osr and stock.xsd files are
located.
If you want to copy the generated OSR file into a PDS, use the z/OS UNIX cp
command. To specify an MVS data set name, precede the name with double
slashes (//). For example, to copy the HFS file stock.osr into a fixed block record
format PDS called HLQ.XML.OSR with record length 80, you can use the following
command:
cp p /u/HLQ/xml/stock.osr //HLQ.XML.OSR(STOCK)
To omit the fully qualified name of the PDS, you can eliminate the single quotation
mark in the cp statement such as:
cp p /u/HLQ/xml/stock.osr //XML.OSR(STOCK)
For more information, see the related reference in XML System Services User's Guide
and Reference.
Example with a simple document
The following two examples illustrate the use of the PLISAXC and PLISAXD
built-in subroutines.
Example of using the PLISAXC built-in subroutine
The following example illustrates the use of the PLISAXC built-in subroutine and
uses the example XML document in the preceding example. This example is
essentially the same as that used in the previous chapter: it does not use
namespaces, and all the input is passed when PLISAXC is first invoked (and as a
result, the end_of_input event should not be invoked).
saxtest: package exports(saxtest);
define alias event
define alias event_with_flag
bit(8) aligned )
options( nodescriptor byvalue linkage(optlink) );
define alias event_with_namespace
pointer, fixed bin(31),
define alias event_without_data
define alias event_namespace_dcl
limited entry( pointer, fixed bin(31),
fixed bin(31),
fixed bin(31) )
define alias event_end_of_input
limited entry( pointer,
pointer byaddr,
fixed bin(31) byaddr )
Figure 96. PLISAXC coding example - type declarations
saxtest: proc options( main );
dcl
,2 e01 type event_without_data
,2 e02 type event
,2 e03 type event
,2 e04 type event
,2 e05 type event
,2 e07 type event_with_namespace
,2 e09 type event
,2 e13 type event_with_flag
,2 e15 type event
init( comment )
,2 e16 type event_namespace_dcl
init( namespace_declare )
,2 e17 type event_end_of_input
init( end_of_input )
,2 e18 type event
init( unresolved_reference )
init( exception )
;
Figure 97. PLISAXC coding example - event structure
dcl token char(8);
dcl xmlDocument char(4000) var;
xmlDocument =
|| <sandwich>
|| <bread type="baker's best"/>
|| '<![CDATA[We should add a <relish> element in future!]]>'.
|| </sandwich>
|| ;
call plisaxc( eventHandler,
addr(token),
addrdata(xmlDocument),
length(xmlDocument) );
end;
Figure 98. PLISAXC coding example - main routine
dcl chars char(32000) based;
start_of_document:
proc( userToken )
|| length= || tokenlength );
return(0);
end;
return(0);
end;
return(0);
end;
return(0);
end;
Figure 99. PLISAXC coding example - event routines (Part 1 of 6)
return(0);
end;
namespace_declare:
proc( userToken, nsPrefix, nsPrefixLength,
nsUri, nsUriLength )
dcl nsPrefix pointer;
dcl nsPrefixLength fixed bin(31);
dcl nsUri pointer;
dcl nsUriLength fixed bin(31);
put skip list( prefix =
|| < || substr(nsPrefix->chars,1,nsPrefixlength ) || > );
put skip list( Uri =
|| < || substr(nsUri->chars,1,nsUrilength ) || > );
return(0);
end;
end_of_document:
proc( userToken )
return(0);
end;
start_of_element:
proc( userToken, xmlToken, TokenLength,
nsPrefix, nsPrefixLength,
dcl nsUri pointer;
return(0);
end;
attribute_name:
dcl nsUri pointer;
return(0);
end;
end_of_element:
dcl nsUri pointer;
return(0);
end;
content_characters:
proc( userToken, xmlToken, TokenLength, flags )
options( nodescriptor, byvalue linkage(optlink) );
dcl flags bit(8) aligned;
if flags = b then;
else
put skip list( !!flags = || flags );
return(0);
end;
return(0);
end;
start_of_CDATA:
proc( userToken )
return(0);
end;
end_of_CDATA:
proc( userToken )
return(0);
end;
proc( userToken,
piTarget, piTargetLength,
dcl piData pointer;
|| < || substr(piTarget->chars,1,piTargetLength ) || > );
return(0);
end;
comment:
return(0);
end;
unresolved_reference:
return(0);
end;
exception:
proc( userToken, currentOffset, return_code, reason_code )
dcl return_code fixed bin(31);
dcl reason_code fixed bin(31);
|| return_code = || hex(return_code)
|| , reason_code = || hex(reason_code)
|| , offset = || currentOffset );
return(0);
end;
The preceding program would produce the following output:
end_of_input:
proc( userToken, addr_xml, length_xml )
dcl addr_xml byaddr pointer;
dcl length_xml byaddr fixed bin(31);
return(1);
end;
end;
start_of_document
version_information <1.0>
standalone_declaration <yes>
comment <This document is just an example>
start_of_element <sandwich>
prefix = <>
Uri = <>
start_of_element <bread>
prefix = <>
Uri = <>
attribute_name <type>
prefix = <>
Uri = <>
attribute_characters <baker's best>
end_of_element <bread>
prefix = <>
Uri = <>
processing_instruction <spread>
piData = <please use real mayonnaise >
start_of_element <meat>
prefix = <>
Uri = <>
content_characters <Ham & turkey>
end_of_element <meat>
prefix = <>
Uri = <>
start_of_element <filling>
prefix = <>
Uri = <>
content_characters <Cheese, lettuce, tomato, etc.>
!!flags = 01000000
end_of_element <filling>
prefix = <>
Uri = <>
start_of_cdata
content_characters <We should add a <relish> element in future >
end_of_cdata
end_of_element <sandwich>
prefix = <>
Uri = <>
end_of_document
Figure 100. PLISAXC coding example - program output
Example of using the PLISAXD built-in subroutine
The following example illustrates the use of the PLISAXD built-in subroutine and
uses the example XML document and the XML schema cited in the previous
examples.
This example includes the validation of 8 different XML files against the same
stock.osr schema. In the following output, you can see which XML documents in
the saxdtest program fail validation against the schema.
The PLISAXD built-in subroutine requires the XML schema file to be read into a
buffer. The OSR file in the following example is in a PDS. The initial size of the
OSR buffer is set to 4096. If you have a larger OSR file, you can increase the initial
size of the OSR buffer accordingly.
If the inbound schema file were in an HFS file instead, you could use the following
code to read the OSR file into the buffer:
dcl osrin file input stream environment(u);
dcl fileddint builtin;
dcl fileread builtin;
/* Read the HFS OSR file into buffer*/
open file(osrin);
osr_length = fileddint( osrin, filesize);
osr_ptr = allocate(osr_length);
rc = fileread(osrin,osr_ptr,osr_length);
To run a program by using an OSR in a PDS, you can specify the following DD
statement in the JCL:
//OSRIN DD DSN=HLQ.XML.OSR(STOCK),DISP=SHR
If the associated ddname OSRIN is an HFS file, then use the following JCL
statement instead:
//OSRIN DD PATH=/u/HLQ/xml/stock.osr
saxdtest: package exports(saxdtest);
/************************************************************/
/* saxdtest: Test PL/I XML validation support */
/* expected output: */
/* */
/* SAXDTEST: PL/I XML Validation sample */
/* SAXDTEST: Document Successfully parsed */
/* SAXDTEST: Document Successfully parsed */
/* Invalid: missing attribute itemNumber. */
/* exception return_code =00000018, reason_code =8613 */
/* Invalid: unexpected attribute warehouse. */
/* Invalid: illegal attribute value 123-Ab. */
/* Invalid: missing element quantityOnHand. */
/* Invalid: unexpected element comment. */
/* Invalid: out-of-range element value 100 */
/* */
/************************************************************/
define alias event
define alias event_with_flag
bit(8) aligned )
options( nodescriptor byvalue linkage(optlink) );
Figure 101. PLISAXD coding example - event routines (Part 1 of 10)
define alias event_with_namespace
pointer, fixed bin(31),
define alias event_without_data
define alias event_namespace_dcl
limited entry( pointer, fixed bin(31),
fixed bin(31),
fixed bin(31) )
define alias event_end_of_input
limited entry( pointer,
pointer byaddr,
fixed bin(31) byaddr )
saxdtest: proc options( main );
dcl
,2 e02 type event
,2 e03 type event
,2 e04 type event
,2 e05 type event
,2 e09 type event
,2 e13 type event_with_flag
,2 e15 type event
init( comment )
,2 e16 type event_namespace_dcl
init( namespace_declare )
,2 e17 type event_end_of_input
init( end_of_input )
,2 e18 type event
init( unresolved_reference )
init( exception )
;
dcl token char(45);
dcl rc fixed bin(31);
dcl i fixed bin(31);
dcl xml_document(8) char(300) var;
dcl xml_valid_msg(8) char(45) var;
dcl osr_ptr pointer;
dcl record char(80);
dcl osr_index fixed bin;
dcl osr_buf_tail fixed bin;
dcl temp_osr pointer;
dcl buf_size fixed bin(31) init(4096);
dcl rec_size fixed bin(31);
dcl osr_length fixed bin(31);
dcl buf_length fixed bin(31);
dcl eof fixed bin(31) init(0);
dcl osrin file input;
on endfile(osrin) begin;
eof = 1;
end;
/* read the entire PDS osr file into the buffer */
put skip list (SAXDTEST: PL/I XML Validation sample );
osr_length = buf_size;
osr_index = 0;
osr_buf_tail = 0;
rec_size = length(record);
osr_ptr = allocate(osr_length);
do while (eof = 0 );
read file(osrin) into(record);
osr_buf_tail += rec_size;
if osr_buf_tail > buf_size then
do;
buf_length = osr_length;
osr_length +=buf_size;
temp_osr = allocate(osr_length);
call plimove(temp_osr, osr_ptr, buf_length);
call plifree(osr_ptr);
osr_ptr = temp_osr;
osr_buf_tail = rec_size;
call plimove(osr_ptr+osr_index, addr(record), rec_size);
osr_index += rec_size;
end;
else
do;
call plimove(osr_ptr+osr_index, addr(record), rec_size);
osr_index +=rec_size;
end;
end;
/* Valid XMLFILE */
xml_document(1) = <stockItem itemNumber="453-SR">
|| </stockItem>;
xml_valid_msg(1) = Valid XMLFILE ;
/* Valid: the ITEMNAME element can be omitted */
|| </stockItem>;
xml_valid_msg(2) = Valid: the ITEMNAME element can be omitted.;
/* Invalid: missing attribute itemNumber */
xml_document(3) = <stockItem>
|| </stockItem>;
xml_valid_msg(3) = Invalid: missing attribute itemNumber.;
/* Invalid: unexpected attribute warehouse */
xml_document(4) = <stockItem itemNumber="453-SR" warehouse="NY">
|| </stockItem>;
xml_valid_msg4) = Invalid: unexpected attribute warehouse.;
/* Invalid: illegal attribute value 123-Ab */
xml_document(5) = <stockItem itemNumber="123-Ab">
|| </stockItem>;
xml_valid_msg(5) = Invalid: illegal attribute value 123-Ab.;
/* Invalid: missing element quantityOnHand */
xml_document(6) = <stockItem itemNumber="074-UN">
|| </stockItem>;
xml_valid_msg(6) = Invalid: missing element quantityOnHand.;
/* Invalid: unexpected element comment */
|| <commnet>Nylon bristles</comment>
|| </stockItem>;
xml_valid_msg(7) = Invalid: unexpected element comment.;
/* Invalid: out-of-range element value 100 */
xml_document(8) = <stockItem itemNumber="123-AB">
|| <itemName>Paintbrush</itemName>
|| </stockItem>;
xml_valid_msg(8) = Invalid: out-of-range element value 100;
do i = 1 to hbound(xml_document);;
token = xml_valid_msg(i);
call plisaxd( eventHandler,
addr(token),
addrdata(xml_document(i)),
length(xml_document(i)),
osr_ptr,
37 );
end;
close file(osrin);
call plifree(osr_ptr);
end;
start_of_document:
proc( userToken )
return(0);
end;
return(0);
end;
return(0);
end;
return(0);
end;
return(0);
end;
end_of_document:
proc( userToken )
put skip list( SAXDTEST: Document Successfully parsed );
return(0);
end;
start_of_element:
dcl nsUri pointer;
return(0);
end;
attribute_name:
proc( userToken, xmlToken, tokenLength,
dcl nsUri pointer;
return(0);
end;
return(0);
end;
end_of_element:
dcl nsUri pointer;
return(0);
end;
start_of_CDATA:
proc( userToken )
return(0);
end;
end_of_CDATA:
proc( userToken )
return(0);
end;
content_characters:
proc( userToken, xmlToken, TokenLength, flags )
options( nodescriptor, byvalue linkage(optlink) );
dcl flags bit(8) aligned;
if flags = b then;
else
return(0);
end;
proc( userToken, piTarget, piTargetLength,
dcl piData pointer;
return(0);
end;
comment:
return(0);
end;
namespace_declare:
proc( userToken, nsPrefix, nsPrefixLength,
dcl nsUri pointer;
return(0);
end;
unresolved_reference:
return(0);
end;
exception:
proc( userToken, currentOffset, return_code, reason_code )
dcl return_code fixed bin(31);
dcl reason_code fixed bin(31);
dcl validmsg char(45) based;
put skip list( userToken -> validmsg);
|| return_code = || hex(return_code)
|| , reason_code = || substr(hex(reason_code),5,4));
return(0);
end;
end_of_input:
proc( userToken, addr_xml, length_xml )
dcl addr_xml byaddr pointer;
dcl length_xml byaddr fixed bin(31);
return(0);
end;
The following output shows the result of the sample program. For those
documents that are not valid, the PLISAXD built-in subroutine invokes the XML
exception event with the return code and reason code listed. For a detailed
description of each return code and reason code, see XML System Services Users
Guide and Reference.
SAXDTEST: PL/I XML Validation sample
SAXDTEST: Document Successfully parsed
SAXDTEST: Document Successfully parsed
Invalid: missing attribute itemNumber.
exception return_code =00000018, reason_code =8613
Invalid: unexpected attribute warehouse.
Invalid: illegal attribute value 123-Ab.
Invalid: missing element quantityOnHand.
Invalid: unexpected element comment.
Invalid: out-of-range element value 100
Figure 102. Output from PLISAXD sample
Chapter 19. Using PLIDUMP
This section provides information about dump options and the syntax used to call
PLIDUMP, and describes PL/I-specific information included in the dump that can
help you debug your routine.
Note: PLIDUMP conforms to National Language Support standards.
Figure 103 shows an example of a PL/I routine calling PLIDUMP to produce a
z/OS Language Environment dump. In this example, the main routine PLIDMP
calls PLIDMPA, which then calls PLIDMPB. The call to PLIDUMP is made in
routine PLIDMPB.
The syntax and options for PLIDUMP are shown below.
PLIDUMP (character-string-expression 1,character-string-expression 2)
character-string-expression 1
is a dump options character string consisting of one or more of the following:
A Requests information relevant to all tasks in a multitasking program.
B BLOCKS (PL/I hexadecimal dump).
C Continue. The routine continues after the dump.
E Exit from current task of a multitasking program. Program continues to
run after requested dump is completed.
F FILES.
%PROCESS MAP GOSTMT SOURCE STG LIST OFFSET LC(101);
PLIDMP: PROC OPTIONS(MAIN) ;
Declare (H,I) Fixed bin(31) Auto;
Declare Names Char(17) Static init('Bob Teri Bo Jason');
H = 5; I = 9;
Put skip list('PLIDMP Starting');
Call PLIDMPA;
PLIDMPA: PROC;
Declare (a,b) Fixed bin(31) Auto;
a = 1; b = 3;
Put skip list('PLIDMPA Starting');
Call PLIDMPB;
PLIDMPB: PROC;
Declare 1 Name auto,
2 First Char(12) Varying,
2 Last Char(12) Varying;
First = 'Teri';
Last = 'Gillispy';
Put skip list('PLIDMPB Starting');
Call PLIDUMP('TBFC','PLIDUMP called from procedure PLIDMPB');
Put Data;
End PLIDMPB;
End PLIDMPA;
End PLIDMP;
Figure 103. Example PL/I routine calling PLIDUMP
H STORAGE.
This includes all Language Environment storage, and hence all the
BASED and CONTROLLED storage acquired via ALLOCATE
statements.
Note: A ddname of CEESNAP should be specified with the H option
to produce a SNAP dump of a PL/I routine, but if this is
omitted, Language Environment will issue a message but still
produce a dump with much very useful information.
K BLOCKS (when running under CICS). The Transaction Work Area is
included.
NB NOBLOCKS.
NF NOFILES.
NH NOSTORAGE.
NK NOBLOCKS (when running under CICS).
NT NOTRACEBACK.
O Only information relevant to the current task in a multitasking
program.
S Stop. The enclave is terminated with a dump.
T TRACEBACK.
T, F, and C are the default options.
character-string-expression 2
is a user-identified character string up to 80 characters long that is printed as
the dump header.
PLIDUMP usage notes
If you use PLIDUMP, the following considerations apply:
v If a routine calls PLIDUMP a number of times, use a unique user-identifier for
each PLIDUMP invocation. This simplifies identifying the beginning of each
dump.
v A DD statement with the ddname PLIDUMP, PL1DUMP, or CEEDUMP can be
used to define the data set for the dump.
v The data set defined by the PLIDUMP, PL1DUMP, or CEEDUMP DD statement
should specify a logical record length (LRECL) of at least 133 to prevent dump
records from wrapping. If SYSOUT is used as the target in any one of these
DDs, you must specify MSGFILE(SYSOUT,FBA,133,0) or
MSGFILE(SYSOUT,VBA,137,0) to ensure that the lines are not wrapped.
v When you specify the H option in a call to PLIDUMP, the PL/I library issues an
OS SNAP macro to obtain a dump of virtual storage. The first invocation of
PLIDUMP results in a SNAP identifier of 0. For each successive invocation, the
ID is increased by one to a maximum of 256, after which the ID is reset to 0.
v Support for SNAP dumps using PLIDUMP is only provided under z/OS. SNAP
dumps are not produced in a CICS environment.
If the SNAP is not successful, the CEE3DMP DUMP file displays the message:
Snap was unsuccessful
If the SNAP is successful, CEE3DMP displays the message:
Snap was successful; snap ID = nnn
where nnn corresponds to the SNAP identifier described above. An
unsuccessful SNAP does not increment the identifier.
v If you want the program unit name, program unit address and program unit
offset to be listed correctly in the dump traceback table, you need to make sure
that your PL/I program unit is compiled with a compile-time option other than
TEST(NONE,NOSYM). For example, you can specify the option as
TEST(NOSYM,NOHOOK,BLOCK).
If you want to ensure portability across system platforms, use PLIDUMP to
generate a dump of your PL/I routine.
Locating variables in the PLIDUMP output
To find variables in the PLIDUMP output, you should compile your program with
the MAP option. The MAP option will cause the compiler to add to the listing a
table showing the offset within AUTOMATIC and STATIC storage of all level-1
variables that are AUTOMATIC or STATIC.
To find a variable that is an element in a structure, it is also useful if you compile
your program with the AGGREGATE option. This option will cause the compiler
to add to the listing a table showing the offsets of all the elements of all the
structures in your program.
Locating AUTOMATIC variables
To find an AUTOMATIC variable in the dump, you should find its offset within
automatic using the output from the MAP option (and if necessary the
AGGREGATE option). If PLIDUMP has been invoked with the B option, the dump
output will contain a hex dump of the "Dynamic save area" (or DSA) for each
block. This is the automatic storage for that block.
For example, consider the following simple program:
Compiler Source
Line.File
2.0 test: proc options(main);
3.0
4.0 dcl a fixed bin(31);
5.0 dcl b fixed bin(31);
6.0
7.0 on error
8.0 begin;
9.0 call plidump(TFBC);
10.0 end;
11.0
12.0 a = 0;
13.0 b = 29;
14.0 b = 17 / a;
The result of the compiler MAP option for this program looks like this, except that
there is actually one more column on the right and the columns are actually
spaced much further apart:
* * * * * S T O R A G E O F F S E T L I S T I N G * * * * *
IDENTIFIER DEFINITION ATTRIBUTES
A 1-0:4 Class = automatic, Location = 160 : 0xA0(r13),
B 1-0:5 Class = automatic, Location = 164 : 0xA4(r13),
Chapter 19. Using PLIDUMP 509
So, A is located at hex A0 off of register 13 and B is located at hex A4 off of
register 13, where register 13 points to the DSA.
Since in this program PLIDUMP is called with the B option, it will include a
hexadecimal dump of automatic storage for each block in the current calling chain.
This will look like (again with the right columns cutoff):
Dynamic save area (TEST): 0AD963C8
+000000 0AD963C8 10000000 0AD96188 00000000 00000000
+000020 0AD963E8 00000000 00000000 00000000 00000000
+000040 0AD96408 00000000 00000000 00000000 0AD96518
+000060 0AD96428 00000000 00000000 00000000 00000000
+000080 0AD96448 - +00009F 0AD96467 same as above
+0000A0 0AD96468 00000000 0000001D 00100000 00000000
+0000C0 0AD96488 0B300000 0A700930 0AD963C8 00000000
+0000E0 0AD964A8 00000000 00000000 00000000 00000000
+000100 0AD964C8 0AA47810 0A70E6D0 0AD96540 0AD960F0
+000120 0AD964E8 00000001 0A70F4F8 0AD96318 00000000
+000140 0AD96508 00000000 00000000 00000000 00000000
Since A is at hex offset A0 and B is at hex offset A4 in AUTOMATIC, the dump
shows that A and B have the (expected) hex values of 00000000 and 0000001D
respectively.
Please note that under the compiler options OPT(2) and OPT(3), some variables,
particularly FIXED BIN and POINTER scalar variables, may never be allocated
storage and thus could not be found in the dump output.
Locating STATIC variables
If you compiled your code with the RENT option, static variables are located in the
WSA (the Writeable Static Area) for the current load module. The offset of a
variable within the WSA can be found from the output of the MAP option, and the
WSA is held in the Language Environment control block called the CAA. The
value of the WSA is also listed in the Language Environment dump.
However, if you compiled your code with the NORENT option, EXTERNAL
STATIC is found as usual (using the linker listing and the output of the compiler's
MAP option). INTERNAL STATIC will be dumped as part of the Language
Environment dump (if PLIDUMP was called with the B option).
Please also note that unlike the older PL/I compilers, the address of static is not
dedicated to any one register.
For example, consider the program above with the variables changed to STATIC:
Compiler Source
Line.File
3.0
4.0 dcl a fixed bin(31) static;
5.0 dcl b fixed bin(31) static;
6.0
7.0 on error
8.0 begin;
9.0 call plidump(TFBC);
10.0 end;
11.0
12.0 a = 0;
13.0 b = 29;
14.0 b = 17 / a;
When compiled with the NORENT option, the result of the compiler MAP option
for this program looks like this, except that there is actually one more column on
the right and the columns are actually spaced much further apart:
A 1-0:4 Class = static, Location = 0 : 0x0 + CSECT ***TEST2
B 1-0:5 Class = static, Location = 4 : 0x4 + CSECT ***TEST2
So, A is located at hex offset 00 into the static CSECT for the compilation unit
TEST while B is located at hex offset 04.
Since in this program PLIDUMP is called with the B option, it will include a
hexadecimal dump of static storage for each compilation in the current calling
chain. This will look like (again with the right columns cutoff):
Static for procedure TEST Timestamp: 2004.08.12
+000000 0FC00AA0 00000000 0000001D 0FC00DC8 0FC00AC0
+000020 0FC00AC0 0FC00AA8 00444042 00A3AE01 0FC009C8
+000040 0FC00AE0 6E3BFFE0 00000000 00000000 00000000
+000060 0FC00B00 00000000 00000000 00000000 00000000
+000000 0AD963C8 10000000 0AD96188 00000000 00000000
So, A at hex offset 00 has the (expected) hex value 00000000, and B at hex offset 04
has the (also expected) hex value 0000001D or the decimal value 29.
Locating CONTROLLED variables
CONTROLLED variables are essentially LIFO stacks, and each CONTROLLED
variable has an "anchor" that points to the top of that stack. The key to locating a
CONTROLLED variable is to locate this anchor, and its location depends on the
compiler options as described below.
In the rest of this discussion of CONTROLLED variables, the program source will
be the same program as above, but with the storage class changed to
CONTROLLED:
Compiler Source
Line.File
3.0
4.0 dcl a fixed bin(31) controlled;
5.0 dcl b fixed bin(31) controlled;
6.0
7.0 on error
8.0 begin;
9.0 call plidump(TFBHC);
10.0 end;
11.0
12.0 allocate a, b;
13.0 a = 0;
14.0 b = 29;
15.0 b = 17 / a;
Under NORENT WRITABLE
The result of the compiler MAP option looks like this, except that again there is
actually one more column on the right and the columns are actually spaced much
further apart:
A 1-0:4 Class = static, Location = 8 : 0x8 + CSECT ***TEST2
B 1-0:5 Class = static, Location = 12 : 0xC + CSECT ***TEST2
Note: that these lines describe the location of the anchors for A and B (not the
location of A and B themselves). So, As anchor is located at hex 08 into the static
CSECT for the compilation unit TEST while Bs anchor is located at hex 0C.
If PLIDUMP is called with the B option, it will include a hexadecimal dump of
static storage for each compilation in the current calling chain. This will look like
(again with the right columns cutoff):
Static for procedure TEST Timestamp: . . .
+000000 0FC00A88 0FC00DB0 0FC00AA8 102B8A30 102B8A50
+000020 0FC00AA8 0FC00A88 00444042 00A3AE01 0FC009B0
+000040 0FC00AC8 6E3BFFE0 00000000 00000000 00000000
So, As anchor is at 102B8A30 and Bs is at 102B8A50. But since these are
CONTROLLED variables, their storage was obtained via ALLOCATE statements,
and hence these addresses point into heap storage. But If PLIDUMP is called with
the H option, it will include a hexadecimal dump of heap storage. This will look
like (again with the right columns cutoff):
Enclave Storage:
Initial (User) Heap
+000000 102B7018 C8C1D5C3 0FC0F990 0FC0F990 00000000
. . .
+001A00 102B8A18 102B7018 00000020 0FC00A90 00000014
00000000 00000000 00000000 00000000
+001A20 102B8A38 102B7018 00000020 0FC00A94 00000014
00000000 00000000 0000001D 00000000
Since As anchor was at 102B8A30, A has the hex value 00000000, and since Bs
anchor was at 102B8A50, B has the (expected) hex value 0000001D.
Under NORENT NOWRITABLE(FWS)
The result of the compiler MAP option under these options looks like this, except
that again there is actually one more column on the right and the columns are
actually spaced much further apart:
A 1-0:4 Class = automatic, Location = 236 : 0xEC(r13)
B 1-0:5 Class = automatic, Location = 240 : 0xF0(r13)
Note: under these options, there is an extra level of indirection in locating
CONTROLLED variables, and hence the lines above describe the locations of the
addresses of the anchors for A and B. So, the address of As anchor is located at
hex EC into the automatic for the block TEST while Bs is located at hex F0.
Since PLIDUMP is called with the B option, it will include a hexadecimal dump of
automatic storage for each block in the current calling chain. This will look like
(again with the right columns cutoff):
Dynamic save area (TEST): 102973C8
+000000 102973C8 10000000 10297188 00000000 8FC007DA
....
+0000E0 102974A8 0FC00998 00000000 00000000 102B8A40
102B8A28 10297030 102977D0 8FDF3D7E
So, the address of As anchor is 102B8A40 and Bs is 102B8A28.
Since PLIDUMP was also called with the H option, it will include a hexadecimal
dump of heap storage. This will look like (again with the right columns cutoff):
Enclave Storage:
Initial (User) Heap
+000000 102B7018 C8C1D5C3 0FC0F990 0FC0F990 00000000
. . .
+001A00 102B8A18 102B7018 00000018 00000000 0FC00A78
102B8A80 00000000 102B7018 00000018
+001A20 102B8A38 102B8A20 0FC00A74 102B8A60 00000000
102B7018 00000020 102B8A40 00000014
+001A40 102B8A58 00000000 00000000 00000000 00000000
102B7018 00000020 102B8A28 00000014
+001A60 102B8A78 00000000 00000000 0000001D 00000000
00000000 00000000 00000000 00000000
Since Bs anchor address was at 102B8A28, Bs anchor is at 102B8A80, and B has,
as expected, the hex value 0000001D or decimal 29.
Under NORENT NOWRITABLE(PRV)
The MAP listing when compiled with these options would look like this:
***TEST3 1-0:4 Class = ext def, Location = CSECT ***TEST3
***TEST4 1-0:5 Class = ext def, Location = CSECT ***TEST4
_PRV_OFFSETS 1-0:1 Class = static, Location = 8 : 0x8 + CSECT ***TEST2
The key here is the last line in this output: _PRV_OFFSETS is a static table that
holds the offset into the PRV table for each CONTROLLED variable. This static
table is generated only if the MAP option is specified.
To interpret this table, the compiler will also produce, immediately after the block
names table, another, usually small, listing, which for our program would look
like:
PRV Offsets
Number Offset Name
1 8 A
1 C B
This table lists the hex offset within the runtime _PRV_OFFSETS table for each of
the named CONTROLLED variables. The block number (in the first column) can
be used to distinguish variables with the same name but declared in different
blocks.
Since the _PRV_OFFSETS table is in static storage (at hex offset 8) and since
PLIDUMP was called with the B option, it will appear in the dump output, which
would look like this:
Static for procedure TEST Timestamp: . . .
+000000 10908EC8 02020240 00000005 6DD7D9E5 6DD6C6C6
00000000 00000004 D00000A0 00100000
+000020 10908EE8 6E3BFFE0 00000000 00000000 00000000
00000000 90010000 00000000 00000000
So, the offset of A in the PRV table is 0, and the offset of B in the PRV table is 4.
Note also the eyecatcher "_PRV_OFF" that occupies the first 8 bytes of the
_PRV_OFFSETs table.
The PRV table is always located at offset 4 within the CAA which, since PLIDUMP
was called with the H option, will be in the dump output and which looks like:
Control Blocks Associated with the Thread:
CAA: 0A7107D0
+000000 0A7107D0 00000800 0ADB7DE0 0AD97018 0ADB7018
00000000 00000000 00000000 00000000
So, the address of the PRV table is 0ADB7DE0, and it will also be in the dump
output amongst the HEAP storage:
Enclave Storage:
Initial (User) Heap
+000000 102B7018 C8C1D5C3 0FC0F990 0FC0F990 00000000
. . .
+000DC0 0ADB7DD8 00000000 00000000 0ADB8A38 0ADB8A58
0ADB7018 00000488 00000000 00000000
So, the PRV table contains 0ADB8A38 0ADB8A58 etc, and since, as derived from
the _PRV_OFFSETS table, As offset into the PRV table is 0 and Bs offset is 4, these
are also the addresses of A and B respectively.
These addresses will also appear in the HEAP storage in the dump:
Enclave Storage:
Initial (User) Heap
+000000 102B7018 C8C1D5C3 0FC0F990 0FC0F990 00000000
. . .
+001A00 0ADB8A18 00000000 00000000 0ADB7018 00000020
00000000 00000014 0A7107D4 00000000
+001A20 0ADB8A38 00000000 00000000 0ADB7018 00000020
00000004 00000014 0A7107D4 00000000
+001A40 0ADB8A58 0000001D 00000000 00000000 00000000
00000000 00000000 00000000 00000000
So, since As address is 0ADB8A38, its hex value is, as expected, 00000000, and
since Bs address is 0ADB8A58, its hex value is, also as expected, 0000001D.
Saved Compilation Data
During a compilation, the compiler saves various information about the
compilation in the load module. This information can be very useful in debugging
and in future migration efforts. This chapter describes the information saved.
Copyright
If you specify the COPYRIGHT compiler option, the compiler will save the
COPYRIGHT string as a CHARACTER VARYING string placed immediately
before the timestamp data in the object. This string will be followed by as many
blanks as necessary so that the string plus these blanks will occupy a multiple of 4
bytes.
Timestamp
The compiler saves in every load module a "timestamp", which is a 20-byte
character string of the form "YYYYMMDDHHMISSVNRNML" which records the
date and time of the compilation as well as the version of the compiler that
produced this string. The elements of the string have the following meanings:
YYYY
the year of the compilation
MM the month of the compilation
DD the day of the compilation
HH the hour of the compilation
MI the minute of the compilation
SS the second of the compilation
VN the version number of the compiler
RN the release number of the compiler
ML the maintenance level of the compiler
The timestamp can be located from the PPA2: at offset 12 in the PPA2 is a
four-byte integer giving the offset (possibly negative) to the timestamp from the
address of the PPA2.
The PPA2, in turn, can be located from the PPA1: at offset 4 in the PPA1 is a
four-byte integer giving the offset (possibly negative) to the PPA2 from the entry
point address corresponding to that PPA1.
The PPA1, in turn, can be located from the entry point address for a block: at offset
12 from the entry point address is a four-byte integer giving the offset (possibly
negative) to the PPA1 from the entry point address.
Saved options string
The compiler also stores in the load module a 32-byte string that records the
compiler options used in building the load module.
A PL/I declaration for the saved options string is given in Figure 104. For most of
the fields in the structure, the meaning of the field is obvious given its name, but a
few of the fields need some explanation:
v sos_words holds the number of the bytes in the structure divided by 4.
v sos_version is the version number for this structure. It is not a compiler version
number.
v The size of the structure and what fields have been set depends on the version
number.
dcl
1 sos based
,2 sos_words fixed bin(8) unsigned
,2 sos_version fixed bin(8) unsigned
,2 sos_arch fixed bin(8) unsigned
,2 sos_tune fixed bin(8) unsigned
,2 sos_currency char(1)
,2 sos_optlevel bit(4) /* set with version >= 2 */
,2 sos_scheduler bit(1) /* set with version >= 5 */
,2 sos_nowritable_prv bit(1) /* set with version >= 4 */
,2 sos_noblockedio bit(1) /* set with version >= 3 */
,2 sos_optimize bit(1)
,2 sos_window fixed bin(15)
,2 sos_codepage fixed bin(31)
,2 sos_limits_intname fixed bin(8) unsigned
,2 sos_limits_extname fixed bin(8) unsigned
,2 sos_limits_fixbinp1 fixed bin(8) unsigned
,2 sos_limits_fixbinp2 fixed bin(8) unsigned
,2 sos_limits_fixdecp1 fixed bin(8) unsigned
,2 sos_limits_fixdecp2 /* set with version >= 4 */
fixed bin(8) unsigned
Figure 104. Declare for the saved options string (Part 1 of 4)
,2 sos_flags1
,3 sos_check_stg bit(1)
,3 sos_compact bit(1)
,3 sos_csect bit(1)
,3 sos_dbcs bit(1)
,3 sos_display_wto bit(1)
,3 sos_extrn_full bit(1)
,3 sos_graphic bit(1)
,3 sos_check_conform bit(1) /* set with version >= 6 */
,2 sos_flags2
,3 sos_interrupt bit(1)
,3 sos_reduce bit(1)
,3 sos_norent bit(1)
,3 sos_respect_date bit(1)
,3 sos_rules_ans bit(1)
,3 sos_stdsys bit(1)
,3 sos_nowritable bit(1)
,3 sos_wchar_big bit(1)
,2 sos_flags3
,3 sos_cmpat bit(4)
,3 sos_system bit(4)
,2 sos_flags4
,3 sos_dllinit bit(1)
,3 sos_xinfo_def bit(1)
,3 sos_xinfo_xml bit(1)
,3 sos_static_full bit(1)
,3 sos_backreg_5 bit(1)
,3 sos_noresexp bit(1) /* set with version >= 2 */
,3 sos_bifprec bit(2) /* set with version >= 2 */
/* 01 => bifprec(15) */
/* 10 => bifprec(31) */
,2 sos_test
,3 sos_test_hooks bit(4)
,3 sos_test_sym bit(1)
,3 sos_test_nohook bit(1) /* set with version >= 5 */
,3 sos_test_separate bit(1) /* set with version >= 7 */
,3 sos_Static_Length bit(1) /* set with version >= 2 */
,2 sos_float
,3 sos_afp bit(1)
,3 sos_dft_nobin1arg bit(1) /* set with version >>= 7 */
,3 sos_dec_forcedsign bit(1) /* set with version >= 6 */
,3 sos_dec_nofoflonasgn bit(1) /* set with version >= 6 */
,3 sos_prectype bit(2) /* set with version >= 5 */
,3 sos_floatinmath bit(2) /* set with version >= 2 */
,2 sos_usage
,3 sos_ans_round bit(1)
,3 sos_ans_unspec bit(1)
,3 sos_common bit(1) /* set with version >= 6 */
,3 sos_initauto bit(1) /* set with version >= 5 */
,3 sos_initbased bit(1) /* set with version >= 5 */
,3 sos_initctl bit(1) /* set with version >= 5 */
,3 sos_initstatic bit(1) /* set with version >= 5 */
,3 sos_stringofg_is_c bit(1) /* set with version >= 5 */
,2 sos_default
,3 sos_ans bit(1)
,3 sos_asgn bit(1)
,3 sos_byaddr bit(1)
,3 sos_conn bit(1)
,3 sos_descriptor bit(1)
,3 sos_ebcdic bit(1)
,3 sos_nonnative bit(1)
,3 sos_nonnativeaddr bit(1)
,3 sos_inline bit(1)
,3 sos_reorder bit(1)
,3 sos_evendec bit(1)
,3 sos_null370 bit(1)
,3 sos_recursive bit(1)
,3 sos_desclctr bit(1)
,3 sos_ret_byaddr bit(1)
,3 sos_initfill bit(1)
,3 sos_initfill_char char(1)
,3 sos_short_ieee bit(1)
,3 sos_dummy_unal bit(1)
,3 sos_retcode bit(1)
,3 sos_unaligned bit(1)
,3 sos_ordinal_max bit(1)
,3 sos_overlap bit(1)
,3 sos_hex bit(1)
,3 sos_e_hex bit(1)
,3 sos_linkage fixed bin(8) unsigned
,2 sos_prefix
,3 sos_size bit(1)
,3 sos_stringrange bit(1)
,3 sos_stringsize bit(1)
,3 sos_subrg bit(1)
,3 sos_fofl bit(1)
,3 sos_ofl bit(1)
,3 sos_invalidop bit(1)
,3 sos_ufl bit(1)
,3 sos_zdiv bit(1)
,3 sos_conv bit(1)
,3 * bit(1)
,3 sos_gn_in_sidefile bit(1) /* set with version >= 12 */
,3 sos_nosepname bit(1) /* set with version >= 8 */
,3 sos_csectcut bit(3) /* set with version >= 5 */
The possible values for the sos_cmpat field are given by these declares:
dcl sos_cmpat_le bit(4) value(0000b);
dcl sos_cmpat_v1 bit(4) value(0001b);
The possible values for the sos_system field are given by these declares:
dcl sos_system_mvs bit(4) value(0001b);
dcl sos_system_tso bit(4) value(0010b);
dcl sos_system_cics bit(4) value(0011b);
dcl sos_system_ims bit(4) value(0100b);
dcl sos_system_os bit(4) value(0101b);
The possible values for the sos_test_hooks field are given by these declares:
dcl sos_test_hooks_none bit(4) value(0000b);
dcl sos_test_hooks_block bit(4) value(0001b);
dcl sos_test_hooks_stmt bit(4) value(0011b);
dcl sos_test_hooks_path bit(4) value(0101b);
dcl sos_test_hooks_all bit(4) value(0111b);
The possible values for the sos_linkage field are given by these declares:
dcl sos_linkage_optlink fixed bin(8) unsigned value(1);
dcl sos_linkage_system fixed bin(8) unsigned value(2);
The possible values for the sos_bifprec field are given by these declares:
dcl sos_bifprec_15 bit(2) value(01b);
dcl sos_bifprec_31 bit(2) value(10b);
The possible values for the sos_floatinmath field are given by these declares:
,2 sos_extension01
,3 sos_hgpr bit(1) /* set with version >= 10 */
,3 sos_hgpr_preserve bit(1) /* set with version >= 10 */
,3 sos_goff bit(1) /* set with version >= 10 */
,3 sos_dec_foflonmult bit(1) /* set with version >= 10 */
,3 sos_usage_hex_cstg bit(1) /* set with version >= 10 */
,3 sos_usage_substr_loose /* set with version >= 10 */
bit(1)
,3 sos_nonullstraddr bit(1) /* set with version >= 10 */
,3 sos_nullstrptr_sysnull
bit(1) /* set with version >= 11 */
,3 sos_ignore_display bit(1) /* set with version >= 11 */
,3 sos_ignore_put bit(1) /* set with version >= 11 */
,3 sos_onsnap_stringsize /* set with version >= 11 */
bit(1)
,3 sos_onsnap_stringrange /* set with version >= 11 */
bit(1)
,3 sos_rtcheck_nullptr /* set with version >= 11 */
bit(1)
,3 sos_initauto_short bit(1) /* set with version >= 11 */
,3 sos_pp_cics bit(1) /* set with version >= 11 */
,3 sos_pp_sql bit(1) /* set with version >= 11 */
,3 sos_cuname_offset fixed bin(16) unsigned
,2 sos_extension02
,3 sos_xref_in_sidefile /* set with version >= 13 */
bit(1)
,3 sos_unroll
,3 sos_nolaxreturn
,3 * bit(29) /* reserved */
dcl sos_floatinmath_asis bit(2) value(00b);
dcl sos_floatinmath_long bit(2) value(10b);
dcl sos_floatinmath_extndd bit(2) value(11b);
The saved options string is located after the timestamp in one of two ways:
1. if the service option has been specified, the string specified in the service
option follows immediately after the timestamp as a character varying string.
Then the saved options string follows after the service string as a second
character varying string.
2. if the service option has not been specified, the saved options string follows
immediately after the timestamp as a character varying string.
The length of the varying string that holds the saved options string may be longer
than the size of the saved options string itself.
The presence (or absence) of the service string is indicated in the PPA2 by the flag
byte at decimal offset 20 in the PPA2: if the result of anding this byte with '20'bx is
not zero, then the service string is present.
In some earlier releases of the PL/I compiler, the compiler did not place a saved
options string in the load module. The presence (or absence) of the saved options
string is indicated in the PPA2 by the flag byte at decimal offset 20 in the PPA2: if
the result of anding this byte with '02'bx is not zero, then the saved options string
is present.
Chapter 20. Interrupts and attention processing
To enable a PL/I program to recognize attention interrupts, two operations must
be possible:
v You must be able to create an interrupt. This is done in different ways
depending upon both the terminal you use and the operating system.
v Your program must be prepared to respond to the interrupt. You can write an
ON ATTENTION statement in your program so that the program receives
control when the ATTENTION condition is raised.
Note: If the program has an ATTENTION ON-unit that you want invoked, you
must compile the program with either of the following:
The INTERRUPT option (supported only in TSO)
A TEST option other than NOTEST or TEST(NONE,NOSYM).
Compiling this way causes INTERRUPT(ON) to be in effect, unless you
explicitly specify INTERRUPT(OFF) in PLIXOPT.
You can find the procedure used to create an interrupt in the IBM instruction
manual for the operating system and terminal that you are using.
There is a difference between the interrupt (the operating system recognized your
request) and the raising of the ATTENTION condition.
An interrupt is your request that the operating system notify the running program.
If a PL/I program was compiled with the INTERRUPT compile-time option,
instructions are included that test an internal interrupt switch at discrete points in
the program. The internal interrupt switch can be set if any program in the load
module was compiled with the INTERRUPT compile-time option.
The internal switch is set when the operating system recognizes that an interrupt
request was made. The execution of the special testing instructions (polling) raises
the ATTENTION condition. If a debugging tool hook (or a CALL PLITEST) is
encountered before the polling occurs, the debugging tool can be given control
before the ATTENTION condition processing starts.
Polling ensures that the ATTENTION condition is raised between PL/I statements,
rather than within the statements.
Figure 105 on page 522 shows a skeleton program, an ATTENTION ON-unit, and
several situations where polling instructions will be generated. In the program
polling will occur at:
v LABEL1
v Each iteration of the DO
v The ELSE PUT SKIP ... statement
v Block END statements
Using ATTENTION ON-units
You can use processing within the ATTENTION ON-unit to terminate potentially
endless looping in a program.
Control is given to an ATTENTION ON-unit when polling instructions recognize
that an interrupt has occurred. Normal return from the ON-unit is to the statement
following the polling code.
Interaction with a debugging tool
If the program has the TEST(ALL) or TEST(ERROR) run-time option in effect, an
interrupt causes the debugging tool to receive control the next time a hook is
encountered. This might be before the program's polling code recognizes that the
interrupt occurred.
Later, when the ATTENTION condition is raised, the debugging tool receives
control again for condition processing.
%PROCESS INTERRUPT;
.
.
.
ON ATTENTION
BEGIN;
DCL X FIXED BINARY(15);
PUT SKIP LIST ('Enter 1 to terminate, 0 to continue.');
GET SKIP LIST (X);
IF X = 1 THEN
STOP;
ELSE
PUT SKIP LIST ('Attention was ignored');
END;
.
.
.
LABEL1:
IF EMPNO ...
.
.
.
DO I = 1 TO 10;
.
.
.
END;
.
.
.
Figure 105. Using an ATTENTION ON-unit
Chapter 21. Using the Checkpoint/Restart facility
This chapter describes the PL/I Checkpoint/Restart feature which provides a
convenient method of taking checkpoints during the execution of a long-running
program in a batch environment.
At points specified in the program, information about the current status of the
program is written as a record on a data set. If the program terminates due to a
system failure, you can use this information to restart the program close to the
point where the failure occurred, avoiding the need to rerun the program
completely.
This restart can be either automatic or deferred. An automatic restart is one that
takes place immediately (provided the operator authorizes it when requested by a
system message). A deferred restart is one that is performed later as a new job.
You can request an automatic restart from within your program without a system
failure having occurred.
PL/I Checkpoint/Restart uses the Advanced Checkpoint/Restart Facility of the
operating system. This facility is described in the books listed in Bibliography on
page 599.
To use checkpoint/restart you must do the following:
v Request, at suitable points in your program, that a checkpoint record is written.
This is done with the built-in subroutine PLICKPT.
v Provide a data set on which the checkpoint record can be written.
v Also, to ensure the desired restart activity, you might need to specify the RD
parameter in the EXEC or JOB statement (see z/OS JCL Reference).
Note: You should be aware of the restrictions affecting data sets used by your
program. These are detailed in the Bibliography on page 599.
Requesting a checkpoint record
Each time you want a checkpoint record to be written, you must invoke, from your
PL/I program, the built-in subroutine PLICKPT.
CALL PLICKPT
( ddname )
, check-id
, org
, code
The four arguments are all optional. If you do not use an argument, you need not
specify it unless you specify another argument that follows it in the given order. In
this case, you must specify the unused argument as a null string (''). The following
paragraphs describe the arguments.
ddname
is a character string constant or variable specifying the name of the DD
statement defining the data set that is to be used for checkpoint records. If you
omit this argument, the system will use the default ddname SYSCHK.
check-id
is a character string constant or variable specifying the name that you want to
assign to the checkpoint record so that you can identify it later. If you omit this
argument, the system will supply a unique identification and print it at the
operator's console.
org
is a character string constant or variable with the attributes CHARACTER(2)
whose value indicates, in operating system terms, the organization of the
checkpoint data set. PS indicates sequential (that is, CONSECUTIVE)
organization; PO represents partitioned organization. If you omit this
argument, PS is assumed.
code
is a variable with the attributes FIXED BINARY (31), which can receive a
return code from PLICKPT. The return code has the following values:
0 A checkpoint has been successfully taken.
4 A restart has been successfully made.
8 A checkpoint has not been taken. The PLICKPT statement should be
checked.
12 A checkpoint has not been taken. Check for a missing DD statement, a
hardware error, or insufficient space in the data set. A checkpoint will
fail if taken while a DISPLAY statement with the REPLY option is still
incomplete.
16 A checkpoint has been taken, but ENQ macro calls are outstanding and
will not be restored on restart. This situation will not normally arise for
a PL/I program.
Defining the checkpoint data set
You must include a DD statement in the job control procedure to define the data
set in which the checkpoint records are to be placed. This data set can have either
CONSECUTIVE or partitioned organization. You can use any valid ddname. If you
use the ddname SYSCHK, you do not need to specify the ddname when invoking
PLICKPT.
You must specify a data set name only if you want to keep the data set for a
deferred restart. The I/O device can be any direct access device.
To obtain only the last checkpoint record, then specify status as NEW (or OLD if
the data set already exists). This will cause each checkpoint record to overwrite the
previous one.
To retain more than one checkpoint record, specify status as MOD. This will cause
each checkpoint record to be added after the previous one.
If the checkpoint data set is a library, check-id is used as the member-name.
Thus a checkpoint will delete any previously taken checkpoint with the same
name.
For direct access storage, you should allocate enough primary space to store as
many checkpoint records as you will retain. You can specify an incremental space
allocation, but it will not be used. A checkpoint record is approximately 5000 bytes
longer than the area of main storage allocated to the step.
No DCB information is required, but you can include any of the following, where
applicable:
OPTCD=W, OPTCD=C, RECFM=UT
These subparameters are described in the z/OS JCL User's Guide.
Requesting a restart
A restart can be automatic or deferred. You can make automatic restarts after a
system failure or from within the program itself. The system operator must
authorize all automatic restarts when requested by the system.
Automatic restart after a system failure
If a system failure occurs after a checkpoint has been taken, the automatic restart
will occur at the last checkpoint if you have specified RD=R (or omitted the RD
parameter) in the EXEC or JOB statement.
If a system failure occurs before any checkpoint has been taken, an automatic
restart, from the beginning of the job step, can still occur if you have specified
RD=R in the EXEC or JOB statement.
After a system failure occurs, you can still force automatic restart from the
beginning of the job step by specifying RD=RNC in the EXEC or JOB statement. By
specifying RD=RNC, you are requesting an automatic step restart without
checkpoint processing if another system failure occurs.
Automatic restart within a program
You can request a restart at any point in your program. The rules for the restart are
the same as for a restart after a system failure. To request the restart, you must
execute the statement:
CALL PLIREST;
To effect the restart, the compiler terminates the program abnormally, with a
system completion code of 4092. Therefore, to use this facility, the system
completion code 4092 must not have been deleted from the table of eligible codes
at system generation.
Getting a deferred restart
To ensure that automatic restart activity is canceled, but that the checkpoints are
still available for a deferred restart, specify RD=NR in the EXEC or JOB statement
when the program is first executed.
RESTART = ( stepname )
,
check-id
If you subsequently require a deferred restart, you must submit the program as a
new job, with the RESTART parameter in the JOB statement. Use the RESTART
parameter to specify the job step at which the restart is to be made and, if you
want to restart at a checkpoint, the name of the checkpoint record.
Chapter 21. Using the Checkpoint/Restart facility 525
For a restart from a checkpoint, you must also provide a DD statement that defines
the data set containing the checkpoint record. The DD statement must be named
SYSCHK. The DD statement must occur immediately before the EXEC statement
for the job step.
Modifying checkpoint/restart activity
You can cancel automatic restart activity from any checkpoints taken in your
program by executing the statement:
CALL PLICANC;
However, if you specified RD=R or RD=RNC in the JOB or EXEC statement,
automatic restart can still take place from the beginning of the job step.
Also, any checkpoints already taken are still available for a deferred restart.
You can cancel any automatic restart and the taking of checkpoints, even if they
were requested in your program, by specifying RD=NC in the JOB or EXEC
statement.
Chapter 22. Using user exits
PL/I provides a number of user exits that allow you to customize the PL/I
product to suit your needs. The PL/I products supply default exits and the
associated source files.
If you want the exits to perform functions that are different from those supplied by
the default exits, we recommend that you modify the supplied source files as
appropriate.
At times, it is useful to be able to tailor the compiler to meet the needs of your
organization. For example, you might want to suppress certain messages or alter
the severity of others. You might want to perform a specific function with each
compilation, such as logging statistical information about the compilation into a
file. A compiler user exit handles this type of function.
With PL/I, you can write your own user exit or use the exit provided with the
product, either 'as is' or modified, depending on what you want to do with it. The
user exit source code provided with the product can be seen in Figure 16 on page
241.
The purpose of this chapter is to describe:
v Procedures that the compiler user exit supports
v How to activate the compiler user exit
v IBMUEXIT, the IBM-supplied compiler user exit
v Requirements for writing your own compiler user exit.
Procedures performed by the compiler user exit
The compiler user exit performs three specific procedures:
v Initialization
v Interception and filtering of compiler messages
v Termination
As illustrated in Figure 106, the compiler passes control to the initialization
procedure, the message filter procedure, and the termination procedure. Each of
these three procedures, in turn, passes control back to the compiler when the
requested procedure is completed.
Each of the three procedures is passed two different control blocks:
v A global control block that contains information about the compilation. This is
passed as the first parameter. For specific information on the global control
block, see Structure of global control blocks on page 529.
v A function-specific control block that is passed as the second parameter. The
content of this control block depends upon which procedure has been invoked.
For detailed information, see Writing the initialization procedure on page 530,
Writing the message filtering procedure on page 531, and Writing the
termination procedure on page 533.
Activating the compiler user exit
In order to activate the compiler user exit, you must specify the EXIT compile-time
option. For more information on the EXIT option, see EXIT on page 46.
The EXIT compile-time option allows you to specify a user-option-string which
specifies the DDname for the user exit input file. If you do not specify a string,
SYSUEXIT is used as the DDname for the user exit input file.
The user-option-string is passed to the user exit functions in the global control
block which is discussed in Structure of global control blocks on page 529. See
the field Uex_UIB_User_char_str in the section Structure of global control
blocks on page 529 for additional information.
The IBM-supplied compiler exit, IBMUEXIT
IBM supplies you with the sample compiler user exit, IBMUEXIT, which filters
messages for you. It monitors messages and, based on the message number that
you specify, suppresses the message or changes the severity of the message.
Customizing the compiler user exit
As was mentioned earlier, you can write your own compiler user exit or simply
use the one shipped with the compiler. In either case, the name of the fetchable file
for the compiler user exit must be IBMUEXIT.
Figure 106. PL/I compiler user exit procedures
This section describes how to:
v Modify the user exit input file for customized message filtering
v Create your own compiler user exit
Modifying SYSUEXIT
Rather than spending the time to write a completely new compiler user exit, you
can simplify modify the user exit input file.
Edit the file to indicate which message numbers you want to suppress, and which
message number severity levels you want to change. A sample file is shown in
Figure 107.
The first two lines are header lines and are ignored by IBMUEXIT. The remaining
lines contain input separated by a variable number of blanks.
Each column of the file is relevant to the compiler user exit:
v The first column should contain the letters 'IBM' in single quotation marks for
all compiler messages to which you want the exit to apply.
v The second column contains the four digit message number.
v The third column shows the new message severity. Severity -1 indicates that the
severity should be left as the default value.
v The fourth column indicates whether or not the message is to be suppressed. A
'1' indicates the message is to be suppressed, and a '0' indicates that it should be
printed.
v The comment field, found in the last column, is for your information, and is
ignored by IBMUEXIT.
Writing your own compiler exit
To write your own user exit, you can use IBMUEXIT (see the source in Figure 16.)
as a model. As you write the exit, make sure it covers the areas of initialization,
message filtering, and termination.
As noted in that section, the compiler user exit must be compiled with the RENT
option and linked as a DLL.
Structure of global control blocks
The global control block is passed to each of the three user exit procedures
(initialization, filtering, and termination) whenever they are invoked. The following
code and accompanying explanations describe the contents of each field in the
global control block.
Dcl
1 Uex_UIB native based( null() ),
Fac Id Msg No Severity Suppress Comment
+--------+--------+----------+----------+--------------------------------
IBM 1042 -1 1 String spans multiple lines
IBM 1044 -1 1 FIXED BIN 7 mapped to 1 byte
IBM 1047 8 0 Order inhibits optimization
IBM 1052 -1 1 Nodescriptor with * extent arg
IBM 1059 0 0 Select without OTHERWISE
IBM 1169 0 1 Precision of result determined
Figure 107. Example of an user exit input file
Chapter 22. Using user exits 529
2 Uex_UIB_Exit_token pointer, /* for user exits use */
2 Uex_UIB_Exit_Routs, /* exit entries set at
initialization */
*, *, *, * )
limited entry (
*, /* to Uex_UIB */
*, /* to a request area */
);
Data Entry Fields
v Uex_UIB_ Length: Contains the length of the control block in bytes. The value is
storage (Uex_UIB).
v Uex_UIB_Exit_token: Used by the user exit procedure. For example, the
initialization may set it to a data structure which is used by both the message
filter, and the termination procedures.
v Uex_UIB_User_char_str: Points to an optional character string, if you specify it.
For example, in pli filename (EXIT (string))...fn can be a character string
up to thirty-one characters in length.
v Uex_UIB_char_len: Contains the length of the string pointed to by the
User_char_str. The compiler sets this value.
v Uex_UIB_Filename_str: Contains the name of the source file that you are
compiling, and includes the drive and subdirectories as well as the filename. The
compiler sets this value.
v Uex_UIB_Filename_len: Contains the length of the name of the source file
pointed to by the Filename_str. The compiler sets this value.
v Uex_UIB_return_code: Contains the return code from the user exit procedure.
The user sets this value.
v Uex__UIB_reason_code: Contains the procedure reason code. The user sets this
value.
v Uex_UIB_Exit_Routs: Contains the exit entries set up by the initialization
procedure.
v Uex_UIB_Termination: Contains the entry that is to be called by the compiler at
termination time. The user sets this value.
v Uex_UIB_Message_Filter: Contains the entry that is to be called by the compiler
whenever a message needs to be generated. The user sets this value.
Writing the initialization procedure
Your initialization procedure should perform any initialization required by the exit,
such as opening files and allocating storage. The initialization procedure-specific
control block is coded as follows:
2 Uex_ISA_Length_fixed bin(31); /* storage(Uex_ISA) * /
The global control block syntax for the initialization procedure is discussed in the
section Structure of global control blocks on page 529.
Upon completion of the initialization procedure, you should set the return/reason
codes to the following:
0/0
Continue compilation
4/n
Reserved for future use
8/n
12/n
16/n
Abort compilation
Writing the message filtering procedure
The message filtering procedure permits you to either suppress messages or alter
the severity of messages. You can increase the severity of any of the messages but
you can decrease the severity only of ERROR (severity code 8) or WARNING
(severity code 4) messages.
The procedure-specific control block contains information about the messages. It is
used to pass information back to the compiler indicating how a particular message
should be handled.
The following is an example of a procedure-specific message filter control block:
Dcl 1 Uex_MFX native based( null() ),
2 Uex_MFX_Length fixed bin(31),
2 Uex_MFX_Facility_Id char(3), /* of component writing
message */
2 * char(1),
2 Uex_MFX_Message_no fixed bin(31),
2 Uex_MFX_Severity fixed bin(15),
2 Uex_MFX_New_Severity fixed bin(15), /* set by exit proc */
2 Uex_MFX_Inserts fixed bin(15),
2 Uex_MFX_Inserts_Data( 6 refer(Uex_MFX_Inserts) ),
3 Uex_MFX_Ins_Type fixed bin(7),
3 Uex_MFX_Ins_Type_Data union unaligned,
4 * char(8),
4 Uex_MFX_Ins_Bin8 fixed bin(63),
4 Uex_MFX_Ins_Bin fixed bin(31),
4 Uex_MFX_Ins_Str,
5 Uex_MFX_Ins_Str_Len fixed bin(15),
5 Uex_MFX_Ins_Str_Addr pointer,
4 Uex_MFX_Ins_Series,
5 Uex_MFX_Ins_Series_Sep char(1),
5 Uex_MFX_Ins_Series_Addr pointer;
Data Entry Fields
v Uex_MFX_Length: Contains the length of the control block in bytes. The value is
storage (Uex_MFX).
v Uex_MFX_Facility_Id: Contains the ID of the facility; for the compiler, the ID is
IBM. The compiler sets this value.
v Uex_MFX_Message_no: Contains the message number that the compiler is
going to generate. The compiler sets this value.
v Uex_MFX_Severity: Contains the severity level of the message; it can be from
one to fifteen characters in length. The compiler sets this value.
v Uex_MFX_New_Severity: Contains the new severity level of the message; it can
be from one to fifteen characters in length. The user sets this value.
v Uex_MFX_Inserts: Contains the number of inserts for the message; it can range
from zero to six. The compiler sets this value.
v Uex_MFX_Inserts_Data: Contains fields to describe each of the inserts. The
compiler sets these values.
v Uex_MFX_Ins_Type: Contains the type of the insert. The possible insert types
are:
Uex_Ins_Type_Xb31: Is used for a FIXED BIN(31) and has the value 1.
Uex_Ins_Type_Char: Is used for a CHAR string and has the value 2.
Uex_Ins_Type_Series: Is used for a series of CHAR strings and has the value
3.
Uex_Ins_Type_Xb63: Is used for a FIXED BIN(63) and has the value 4.
The compiler sets this value.
v Uex_MFX_Ins_Bin: Contains the integer value for an insert that has integer
type. The compiler sets this value.
v Uex_MFX_Ins_Str_Len: Contains the length (in bytes) for an insert that has
character type. The compiler sets this value.
v Uex_MFX_Ins_Str_Addr: Contains the address of the character string for an
insert that has character type. The compiler sets this value.
v Uex_MFX_Ins_Series_Sep: Contains the character that should be inserted
between each element for an insert that has series type. Typically, this is a blank,
period or comma. The compiler sets this value.
v Uex_MFX_Ins_Series_Addr: Contains the address of the series of varying
character strings for an insert that has series type. The address points to a FIXED
BIN(31) field holding the number of strings to concatenate followed by the
adrresses of those strings. The compiler sets this value.
Upon completion of the message filtering procedure, set the return/reason codes to
one of the following:
0/0
Continue compilation, output message
0/1
Continue compilation, do not output message
4/n
8/n
16/n
Abort compilation
Writing the termination procedure
You should use the termination procedure to perform any cleanup required, such
as closing files. You might also want to write out final statistical reports based on
information collected during the error message filter procedures and the
initialization procedures.
The termination procedure-specific control block is coded as follows:
Dcl 1 Uex_ISA native based,
2 Uex_ISA_Length_fixed bin(31); /* storage(Uex_ISA) */
The global control block syntax for the termination procedure is discussed in
Structure of global control blocks on page 529. Upon completion of the
termination procedure, set the return/reason codes to one of the following:
0/0
Continue compilation
4/n
8/n
12/n
16/n
Abort compilation
Example of suppressing SQL messages
Figure 108 on page 534 shows how to modify the user exit to examine the message
inserts and suppress two SQL informational messages and one SQL warning
message.
*Process dft(nodescriptor connected);
*Process or(|) not(!);
*Process limits(extname(31)) rent;
/********************************************************************/
/* */
/* NAME - IBMUEXIT.PLI */
/* */
/* DESCRIPTION */
/* User-exit sample program. */
/* */
/* Licensed Materials - Property of IBM */
/* 5639-A83, 5639-A24 (C) Copyright IBM Corp. 1992,2011. */
/* All Rights Reserved. */
/* US Government Users Restricted Rights-- Use, duplication or */
/* disclosure restricted by GSA ADP Schedule Contract with */
/* IBM Corp. */
/* */
/* DISCLAIMER OF WARRANTIES */
/* The following enclosed code is sample code created by IBM */
/* Corporation. This sample code is not part of any standard */
/* IBM product and is provided to you solely for the purpose of */
/* assisting you in the development of your applications. The */
/* code is provided "AS IS", without warranty of any kind. */
/* IBM shall not be liable for any damages arising out of your */
/* use of the sample code, even if IBM has been advised of the */
/* possibility of such damages. */
/* */
/********************************************************************/
Figure 108. Suppressing SQL messages (Part 1 of 7)
/********************************************************************/
/* */
/* During initialization, IBMUEXIT is called. It reads */
/* information about the messages being screened from a text */
/* file and stores the information in a linked list. IBMUEXIT */
/* also sets up the entry points for the message filter service */
/* and termination service. */
/* */
/* For each message generated by the compiler, the compiler */
/* calls the message filter registered by IBMUEXIT. The filter */
/* looks the message up in the linked list previously created */
/* to see if it is one for which some action should be taken. */
/* */
/* The termination service is called at the end of the compile */
/* but does nothing. It could be enhanced to generates reports */
/* or do other cleanup work. */
/* */
/********************************************************************/
pack: package exports(*);
Dcl
1 Uex_UIB native Based( null() ),
2 Uex_UIB_Exit_token pointer, /* for user exits use */
2 Uex_UIB_Exit_Routs, /* exit entries setat
initialization */
*, *, *, * )
limited entry (
*, /* to Uex_UIB */
* /* to a request area */
);
/*******************************************************************/
/* */
/* request area for initialization exit */
/* */
/*******************************************************************/
2 Uex_ISA_Length fixed bin(31);
/*******************************************************************/
/* */
/* request area for message_filter exit */
/* */
/*******************************************************************/
Dcl 1 Uex_MFX based( null() ),
2 Uex_MFX_Length fixed bin(31),
2 Uex_MFX_Facility_Id char(3),
2 Uex_MFX_Version fixed bin(7),
2 Uex_MFX_Message_no fixed bin(31),
2 Uex_MFX_Severity fixed bin(15),
2 Uex_MFX_New_Severity fixed bin(15),
2 Uex_MFX_Inserts fixed bin(15),
2 Uex_MFX_Inserts_Data( 6 ),
3 Uex_MFX_Ins_Type fixed bin(7),
3 Uex_MFX_Ins_Type_Data union unaligned,
4 * char(8),
4 Uex_MFX_Ins_Bin fixed bin(31),
4 Uex_MFX_Ins_Str,
5 Uex_MFX_Ins_Str_Len fixed bin(15),
5 Uex_MFX_Ins_Str_Addr pointer,
4 Uex_MFX_Ins_Series,
5 Uex_MFX_Ins_Series_Sep char(1),
5 Uex_MFX_Ins_Series_Addr pointer;
dcl uex_Ins_Type_Xb31 fixed bin(15) value(1);
dcl uex_Ins_Type_Char fixed bin(15) value(2);
dcl uex_Ins_Type_Series fixed bin(15) value(3);
/*******************************************************************/
/* */
/* request area for terminate exit */
/* */
/*******************************************************************/
Dcl 1 Uex_TSA native based( null() ),
2 Uex_TSA_Length fixed bin(31);
/*******************************************************************/
/* */
/* severity codes */
/* */
/*******************************************************************/
dcl uex_Severity_Normal fixed bin(15) value(0);
dcl uex_Severity_Warning fixed bin(15) value(4);
dcl uex_Severity_Error fixed bin(15) value(8);
dcl uex_Severity_Severe fixed bin(15) value(12);
dcl uex_Severity_Unrecoverable fixed bin(15) value(16);
/*******************************************************************/
/* */
/* return codes */
/* */
/*******************************************************************/
dcl uex_Return_Normal fixed bin(15) value(0);
dcl uex_Return_Warning fixed bin(15) value(4);
dcl uex_Return_Error fixed bin(15) value(8);
dcl uex_Return_Severe fixed bin(15) value(12);
dcl uex_Return_Unrecoverable fixed bin(15) value(16);
/*******************************************************************/
/* */
/* reason codes */
/* */
/*******************************************************************/
dcl uex_Reason_Output fixed bin(15) value(0);
dcl uex_Reason_Suppress fixed bin(15) value(1);
dcl header pointer;
dcl
1 message_item native based,
2 message_Info,
3 facid char(3),
3 msgno fixed bin(31),
3 newsev fixed bin(15),
3 reason fixed bin(31),
3 variable char(31) var,
2 link pointer;
ibmuexit: proc ( ue, ia ) options( fetchable );
dcl 1 ia like uex_Isa byaddr;
dcl sysuexit file stream input env(recsize(80));
dcl next pointer;
dcl title_Str char(8) var;
dcl eof bit(1);
on error
begin;
on error system;
call plidump(TFBHS );
end;
on undefinedfile(sysuexit)
begin;
put edit (** User exit unable to open exit file )
(A) skip;
put skip;
signal error;
end;
if ue.uex_Uib_User_Char_Len = 0 then
do;
open file(sysuexit);
end;
else
do;
title_Str
= substr( ue.uex_Uib_User_Char_Str->based_Chars,
1, ue.uex_Uib_User_Char_Len );
open file(sysuexit) title(title_Str);
end;
/***************************************************************/
/* */
/* save the address of the message filter so that it will */
/* be invoked by the compiler */
/* */
/***************************************************************/
ue.Uex_UIB_Message_Filter = message_filter;
/***************************************************************/
/* */
/* set the pointer to the linked list to null */
/* */
/* then allocate the first message record */
/* */
/***************************************************************/
header = sysnull();
allocate message_item set(next);
/******************************************************************/
/* */
/* skip header lines and read the file */
/* */
/* the file is expected to start with a header line and */
/* then a line with a scale and then the data lines, for example,*/
/* it could look like the 5 lines below starting with "Fac Id" */
/* */
/* Fac Id Msg No Severity Suppress Insert */
/* +-------+-------+---------+---------+------------------------ */
/* IBM 3259 0 1 DSNH527 */
/* IBM 3024 0 1 DSNH4760 */
/* IBM 3024 0 1 DSNH050 */
/* */
/******************************************************************/
eof = 0b;
on endfile(sysuexit)
eof = 1b;
get file(sysuexit) list(next->message_info) skip(3);
do while( eof = 0b );
/*************************************************************/
/* */
/* put message information in linked list */
/* */
/*************************************************************/
next->link = header;
header = next;
/*************************************************************/
/* */
/* read next data line */
/* */
/*************************************************************/
allocate message_item set(next);
get file(sysuexit) skip;
get file(sysuexit) list(next->message_info);
end;
/***************************************************************/
/* */
/* free the last message record allocated and close the file */
/* */
/***************************************************************/
free next->message_Item;
close file(sysuexit);
end;
message_Filter: proc ( ue, mf );
dcl 1 mf like uex_Mfx byaddr;
dcl next pointer;
dcl jx fixed bin(31);
dcl insert char(256) var;
on error
begin;
on error system;
call plidump(TFBHS );
end;
/***************************************************************/
/* */
/* by default, leave the reason code etc unchanged */
/* */
/***************************************************************/
ue.uex_Uib_Reason_Code = uex_Reason_Output;
ue.uex_Uib_Return_Code = 0;
mf.uex_Mfx_New_Severity = mf.uex_Mfx_Severity;
/***************************************************************/
/* */
/* save the first insert if it has character type */
/* */
/***************************************************************/
insert = *;
if mf.Uex_MFX_Length < stg(mf) then;
else
if mf.Uex_MFX_Inserts = 0 then;
else
do jx = 1 to mf.Uex_MFX_Inserts;
select( mf.Uex_MFX_Ins_Type(jx) );
when( uex_Ins_Type_Char )
do;
if jx = 1 then
insert =
substr( mf.Uex_MFX_Ins_Str_Addr(jx)->based_Chars,
1,mf.Uex_MFX_Ins_Str_Len(jx));
end;
otherwise;
end;
end;
/***************************************************************/
/* */
/* search list for matching error message */
/* */
/***************************************************************/
search_list:
do next = header repeat( next->link ) while( next !=sysnull() );
if next->msgno = mf.uex_Mfx_Message_No
& next->facid = mf.Uex_Mfx_Facility_Id then
do;
if next->variable = * then
leave search_list;
if next->variable
= substr(insert,1,length(next->variable)) then
leave search_list;
end;
end;
/***************************************************************/
/* */
/* if list exhausted, then */
/* no match was found */
/* else */
/* filter the message according to the match found */
/* */
/***************************************************************/
if next = sysnull() then;
else
do;
/***********************************************************/
/* */
/* filter error based on information in table */
/* */
/***********************************************************/
ue.uex_Uib_Reason_Code = next->reason;
if next->newsev < 0 then;
else
mf.uex_Mfx_New_Severity = next->newsev;
end;
end;
exitterm: proc ( ue, ta );
dcl 1 ta like uex_Tsa byaddr;
ue.uex_Uib_return_Code = 0;
ue.uex_Uib_reason_Code = 0;
end;
end;
Chapter 23. PL/I descriptors
This chapter describes PL/I parameter passing conventions between PL/I routines
at run time. For additional information about Language Environment run-time
environment considerations, other than descriptors, see z/OS Language Environment
Programming Guide. This includes run-time environment conventions and assembler
macros supporting these conventions.
Passing an argument
When a string, an array, or a structure is passed as an argument, the compiler
passes a descriptor for that argument unless the called routine is declared with
OPTIONS(NODESCRIPTOR). There are two methods for passing such descriptors:
v By descriptor list
v By descriptor locator
The following key features should be noted about each of these two methods:
v When arguments are passed with a descriptor list
The number of arguments passed is one greater than the number of
arguments specified if any of the arguments needs a descriptor.
An argument passed with a descriptor can be received as a pointer passed by
value (BYVALUE).
The compiler uses this method when the DEFAULT(DESCLIST) compiler
option is in effect.
v When arguments are passed by descriptor locator
The number of arguments passed always matches the number of arguments
specified.
An argument passed with a descriptor can be received as a pointer passed by
address (BYADDR).
The compiler uses this method when the DEFAULT(DESCLOCATOR)
compiler option is in effect.
Argument passing by descriptor list
When arguments and their descriptors are passed with a descriptor list, an extra
argument is passed whenever at least one argument needs a descriptor. This extra
argument is a pointer to a list of pointers. The number of entries in this list equals
the number of arguments passed. For arguments that don't require a descriptor, the
corresponding pointer in the descriptor list is set to SYSNULL. For arguments that
do require a descriptor, the corresponding pointer in the descriptor list is set to the
address of that argument's descriptor.
So, for example, suppose the routine sample is declared as
declare sample entry( fixed bin(31), varying char(*) )
options( byaddr descriptor );
Then, if sample is called as in the following statement:
call sample( 1, test );
The following three arguments are passed to the routine:
v Address of a fixed bin(31) temporary with the value 1
v Address of a varying char(4) temporary with the value test
v Address of a descriptor list consisting of the following:
SYSNULL()
Address of the descriptor for a varying char(4) string
Argument passing by descriptor-locator
When arguments and their descriptors are passed by descriptor-locator, whenever
an argument requires a descriptor, the address of a locator/descriptor for it is
passed instead.
Except for strings, the locator/descriptor is a pair of pointers. The first pointer is
the address of the data; the second pointer is the address of the descriptor. For
strings, under CMPAT(LE), the locator/descriptor is still such a pair of pointers.
But under the other CMPAT options, the locator/descriptor consists of the address
of the string and then the string descriptor itself.
So, for example, suppose the routine sample is declared again as
declare sample entry( fixed bin(31), varying char(*) )
options( byaddr descriptor );
Then, if sample is called as in the following statement
call sample( 1, test );
The following two arguments are passed to the routine:
v the address of a fixed bin(31) temporary with the value 1
v the address of a locator-descriptor consisting of the following:
the address of a varying char(4) temporary with the value test
Under CMPAT(LE), the address of the CMPAT(LE) descriptor for a varying
char(4) string
Under CMPAT(V*), the CMPAT(V*) descriptor for a varying char(4) string
CMPAT(V*) descriptors
Unlike LE descriptors, the CMPAT(V*) descriptors are not self-describing.
However, the string descriptors are the same for all CMPAT(V*) options, and they
also share the same codepage encoding as the LE string descriptors.
String descriptors
In a string descriptor, the first 2 bytes specify the maximum length for the string.
This maximum length is always held in native format.
The third byte contains various flags (to indicate, for example, if the string length
in a VARYING string is held in littleendian or bigendian format or if the data in a
WIDECHAR string is held in littleendian or bigendian format).
In a string descriptor for a nonvarying bit string, the fourth byte gives the bit
offset.
In a string descriptor for a CHARACTER string, the fourth byte encodes the
compiler CODEPAGE option.
The declare for a string descriptor is:
declare
1 dso_string based( null() ),
2 dso_string_length fixed bin(15),
2 dso_string_flags,
3 dso_string_is_varying bit(1),
3 dso_string_is_varyingz bit(1),
3 dso_string_has_nonnative_len bit(1), /* for varying */
3 dso_string_is_ascii bit(1), /* for char */
3 dso_string_has_nonnative_data bit(1), /* for wchar */
3 * bit(1), /* reserved, 0b */
2 * union,
3 dso_String_Codepage ordinal ccs_Codepage_Enum,
3 dso_string_bitofs fixed bin(8) unsigned,
2 dso_string_end char(0);
The possible values for the codepage encoding are defined via:
define ordinal
ccs_Codepage_Enum
( ccs_Codepage_01047 value(1)
,ccs_Codepage_01140
,ccs_Codepage_01141
,ccs_Codepage_01142
,ccs_Codepage_01143
,ccs_Codepage_01144
,ccs_Codepage_01145
,ccs_Codepage_01146
,ccs_Codepage_01147
,ccs_Codepage_01148
,ccs_Codepage_01149
,ccs_Codepage_00819
,ccs_Codepage_00813
,ccs_Codepage_00920
,ccs_Codepage_00037
,ccs_Codepage_00273
,ccs_Codepage_00277
,ccs_Codepage_00278
,ccs_Codepage_00280
,ccs_Codepage_00284
,ccs_Codepage_00285
,ccs_Codepage_00297
,ccs_Codepage_00500
,ccs_Codepage_00871
,ccs_Codepage_01026
,ccs_Codepage_01155
) unsigned prec(8);
Array descriptors
In the following declares, the upper bound for the arrays is declared as 15, but it
should be understood that the actual upper bound will always match the number
of dimensions in the array it describes.
The declare for a CMPAT(V1) array descriptor is:
declare
1 dso_v1 based( null() ),
2 dso_v1_rvo fixed bin(31), /* relative virtual origin */
2 dso_v1_data(1:15),
3 dso_v1_stride fixed bin(31), /* multiplier */
3 dso_v1_hbound fixed bin(15), /* hbound */
3 dso_v1_lbound fixed bin(15); /* lbound */
Chapter 23. PL/I descriptors 543
declare
declare
CMPAT(LE) descriptors
Every LE descriptor starts with a 4-byte field. The first byte specifies the descriptor
type (scalar, array, structure or union). The remaining three bytes are zero unless
they are set by the particular descriptor type.
The declare for a descriptor header is:
declare
1 dsc_Header based( sysnull() ),
2 dsc_Type fixed bin(8) unsigned,
2 dsc_Datatype fixed bin(8) unsigned,
2 * fixed bin(8) unsigned,
2 * fixed bin(8) unsigned;
The possible values for the dsc_Type field are:
declare
dsc_Type_Unset fixed bin(8) value(0),
dsc_Type_Element fixed bin(8) value(2),
dsc_Type_Array fixed bin(8) value(3),
dsc_Type_Structure fixed bin(8) value(4),
dsc_Type_Union fixed bin(8) value(4);
String descriptors
In a string descriptor, the second byte of the header indicates the string type (bit,
character or graphic as well as nonvarying, varying or varyingz).
In a string descriptor for a nonvarying bit string, the third byte of the header gives
the bit offset.
In a string descriptor for a CHARACTER string, the third byte of the header
encodes the compiler CODEPAGE option.
In a string descriptor for a varying string, the fourth byte has a bit indicating if the
string length is held in nonnative format.
In a string descriptor for a character string, the fourth byte also has a bit indicating
if the string data is in EBCDIC.
The declare for a string descriptor is:
declare
1 dsc_String based( sysnull() ),
2 dsc_String_Header,
3 * fixed bin(8) unsigned,
3 dsc_String_Type fixed bin(8) unsigned,
3 * union,
4 dsc_String_Codepage ordinal ccs_Codepage_Enum,
4 dsc_String_BitOfs fixed bin(8) unsigned,
3 *,
4 dsc_String_Has_Nonnative_Len bit(1),
4 dsc_String_Is_Ebcdic bit(1),
4 dsc_String_Has_Nonnative_Data bit(1),
4 * bit(5),
2 dsc_String_Length fixed bin(31); /* max length of string */
The possible values for the dsc_String_Type field are:
declare
dsc_String_Type_Unset fixed bin(8) value(0),
dsc_String_Type_Char_Nonvarying fixed bin(8) value(2),
dsc_String_Type_Char_Varyingz fixed bin(8) value(3),
dsc_String_Type_Char_Varying2 fixed bin(8) value(4),
dsc_String_Type_Bit_Nonvarying fixed bin(8) value(6),
dsc_String_Type_Bit_Varying2 fixed bin(8) value(7),
dsc_String_Type_Graphic_Nonvarying fixed bin(8) value(9),
dsc_String_Type_Graphic_Varyingz fixed bin(8) value(10),
dsc_String_Type_Graphic_Varying2 fixed bin(8) value(11),
dsc_String_Type_Widechar_Nonvarying fixed bin(8) value(13),
dsc_String_Type_Widechar_Varyingz fixed bin(8) value(14),
dsc_String_Type_Widechar_Varying2 fixed bin(8) value(15);
Array descriptors
The declare for an array descriptor is:
declare
1 dsc_Array based( sysnull() ),
2 dsc_Array_Header like dsc_Header,
2 dsc_Array_EltLen fixed bin(31), /* Length of array element */
2 dsc_Array_Rank fixed bin(31), /* Count of dimensions */
2 dsc_Array_RVO fixed bin(31), /* Relative virtual origin */
2 dsc_Array_Data( 1: 1 refer(dsc_Array_Rank) ),
3 dsc_Array_LBound fixed bin(31), /* LBound */
3 dsc_Array_Extent fixed bin(31), /* HBound - LBound + 1 */
3 dsc_Array_Stride fixed bin(31); /* Multiplier */
Chapter 23. PL/I descriptors 545
Part 6. Appendixes
Appendix. SYSADATA message information
When you specify the MSG suboption of the XINFO compile-time option, the
compiler generates a SYSADATA file that contains:
v Counter records
v Literal records
v File records
v Message records
You should note that the records in the file are not necessarily produced in the
order listed above, for example, literal and file records might be interleaved. If you
are writing code that reads a SYSADATA file, you should not rely on the order of
the records in the file except for the following exceptions:
v Counter records are the first records in the file.
v Each literal record precedes any reference to the literal it defines.
v Each file record precedes any reference to the file it describes.
Understanding the SYSADATA file
The SYSADATA file is a sequential binary file. Under z/OS batch, the compiler
writes the SYSADATA records to the file specified by the SYSADATA DD
statement, and that file must not be a member of a PDS. On all other systems, the
compiler writes to a file with the extension "adt".
Each record in the file contains a header. This 8-byte header has fields that are the
same for all records in the file:
Compiler
A number representing the compiler that produced the data. For PL/I, the
number is 40.
Edition number
The edition number of the compiler that produced the data. For this product, it
is the number 2.
SYSADATA level
A number representing the level of SYSADATA that this file format represents.
For this product, it is the number 4.
The header also has some fields that vary from record to record:
v Record type
v Whether or not the record is continued onto the next record
Possible record types are encoded as an ordinal value as shown in Figure 109 on
page 550.
The declare for the header part of a record is shown in Figure 110.
Summary Record
The record with type Xin_Rect_Sum, the summary record, will be the first record
in the file and its declare is shown in Figure 111 on page 551
Define ordinal xin_Rect
(Xin_Rect_Msg value(50), /* Message record */
Xin_Rect_Fil value(57), /* File record */
Xin_Rect_Sum value(61), /* Summary record */
Xin_Rect_Src value(63), /* Source record */
Xin_Rect_Tok value(64), /* Token record */
Xin_Rect_Sym value(66), /* Symbol record */
Xin_Rect_Lit value(67), /* Literal record */
Xin_Rect_Syn value(69), /* Syntax record */
Xin_Rect_Ord_Type value(80), /* ordinal type record */
Xin_Rect_Ord_Elem value(81), /* ordinal element record */
Xin_Rect_Ctr value(82) ) /* counter record */
prec(15);
Figure 109. Record types encoded as an ordinal value
Dcl
1 Xin_Hdr Based( null() ), /* Header portion */
/* */
2 Xin_Hdr_Prod /* Language code */
fixed bin(8) unsigned, /* */
/* */
2 Xin_Hdr_Rect /* Record type */
unal ordinal xin_Rect, /* */
/* */
2 Xin_Hdr_Level /* SYSADATA level */
/* */
2 * union, /* */
3 xin_Hdr_Flags bit(8), /* flags */
3 *, /* */
4 * bit(6), /* Reserved */
4 Xin_Hdr_Little_Endian /* ints are little endian */
bit(1), /* */
4 Xin_Hdr_Cont bit(1), /* Record continued in next rec */
/* */
2 Xin_Hdr_Edition /* compiler "edition" */
/* */
2 Xin_Hdr_Fill bit(32), /* reserved */
/* */
2 Xin_Hdr_Data_Len /* length of data part */
/* */
2 Xin_Hdr_End char(0); /* */
Figure 110. Declare for the header part of a record
Counter records
Each counter record specifies, for a subsequent record type, how many records of
that type the file contains and how many bytes those records occupy.
Literal records
Each literal record assigns a number, called a literal index, that is used by later
records to refer to the characters named in this particular record.
Dcl
1 Xin_Sum Based( null() ), /* summary record */
/* */
2 Xin_Sum_Hdr /* standard header */
like Xin_Hdr, /* */
/* */
2 Xin_Sum_Max_Severity /* max severity from compiler */
/* */
2 Xin_Sum_Left_Margin /* left margin */
/* */
2 Xin_Sum_Right_Margin /* right margin */
/* */
2 xin_Sum_Rsrvd(15) /* reserved */
fixed bin(32) unsigned; /* */
Figure 111. Declare for a summary record
Dcl
1 xin_Ctr Based( null() ), /* counter/size record */
/* */
2 xin_Ctr_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Ctr_Rect /* record type */
unal ordinal xin_Rect, /* */
/* */
2 * /* */
/* */
2 xin_Ctr_Count /* count of that record type */
/* */
2 xin_Ctr_Size /* size used */
Figure 112. Declare for a counter record
Appendix. SYSADATA message information 551
File records
Each file record assigns a number, called a file index, that is used by later records
to refer to the file described by this record. The described file may be the primary
PL/I source file or an INCLUDEd file. Each file record specifies a literal index for
the fully qualified name of the file.
For an INCLUDEd file, each file record also contains the file index and source line
number from whence the INCLUDE request came. (For primary source files, these
fields are zero.)
Message records
Each message record describes a message issued during the compilation. Message
records are not generated for suppressed messages.
Each message record contains:
v The file index and source line number for the file and line to which the message
is attributed. If the message pertains to the compilation as a whole, these fields
are zero.
Dcl
1 xin_Lit Based( null() ), /* literal record */
/* */
2 xin_Lit_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Lit_Inx /* adata index for literal */
/* */
2 xin_Lit_Len /* length of literal */
/* */
2 xin_Lit_Val char(2000); /* literal value */
Figure 113. Declare for a literal record
Dcl
1 xin_Fil Based( null() ), /* file record */
/* */
2 xin_Fil_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Fil_File_Id /* file id from whence it */
fixed bin(31) unsigned, /* was INCLUDEd */
/* */
2 xin_Fil_Line_No /* line no within that file */
/* */
2 xin_Fil_Id /* id assigned to this file */
/* */
2 xin_Fil_Name /* literal index of the */
fixed bin(31) unsigned; /* fully qualified file name */
Figure 114. Declare for a file record
v The identifier (e.g. IBM1502) and severity associated with the message.
v The text of the message.
The declare for a message record is as follows:
Understanding SYSADATA symbol information
When you specify the SYM suboption of the XINFO compile-time option, the
compiler generates a SYSADATA file that contains the following information in
addition to the records generated for the MSG suboption:
v Ordinal type records
v Ordinal element records
v Symbol records
Symbol records are not generated for built-in functions, generic variables or
variables with non-constant extents.
Ordinal type records
Each ordinal type record assigns a number, called an ordinal type index, that is
used by later records to refer to an ordinal type described by this record. The name
of the type is indicated by a literal index. Each ordinal type record contains the file
index and source line number for the file and line in which the ordinal type was
declared.
Each ordinal type record contains:
v A count of the number of values defined by that type
v The precision associated with the type
v Bits indicating if it is signed or unsigned
Dcl
1 xin_Msg Based( null() ), /* message record */
/* */
2 xin_Msg_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Msg_File_Id /* file id */
/* */
2 xin_Msg_Line_No /* line no within file */
/* */
2 xin_Msg_Id /* identifier (i.e. IBM1502) */
char(16), /* */
/* */
2 xin_Msg_Severity /* severity (0, 4, 8, 12 or 16) */
fixed bin(15) signed, /* */
/* */
2 xin_Msg_Length /* length of message */
/* */
2 Xin_Msg_Text /* actual message */
char( 100 refer(xin_Msg_Length) );
Figure 115. Declare for a message record
Ordinal element records
Each ordinal type record is immediately followed by a series of records (as many
as specified by the ordinal type count) that describes the values named by that
ordinal.
Each ordinal element record assigns a number, called an ordinal element index,
that is used by later records to refer to an ordinal element described by this record.
The name of the element is indicated by a literal index. Each ordinal element
record contains the file index and source line number for the file and line in which
the ordinal element was declared.
Additionally, each ordinal element record contains
v The ordinal type index of the ordinal type to which it belongs
v The value for that element
declare /* */
1 xin_Ord_Type based( null() ), /* */
/* */
2 xin_Ord_Type_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Ord_Type_File_Id /* file id */
/* */
2 xin_Ord_Type_Line_No /* line no within file */
/* */
2 xin_Ord_Type_Id /* identifying number */
fixed bin(31), /* */
/* */
2 xin_Ord_Type_Count /* count of elements */
/* */
2 xin_Ord_Type_Prec /* precision for ordinal */
/* */
2 *, /* */
3 xin_Ordinal_Type_Signed /* signed attribute applies */
bit(1), /* */
3 xin_Ordinal_Type_Unsigned /* unsigned attribute applies */
bit(1), /* */
3 * /* unused */
bit(6), /* */
/* */
2 * /* unused */
char(2), /* */
/* */
2 xin_Ord_Type_Name /* type name */
fixed bin(31); /* */
Figure 116. Declare for an ordinal type record
Symbol records
Each symbol record assigns a number that is called a symbol index. The index is
used by later records to refer to the symbol described by this record. For example,
the index can be used as the name of a user variable or constant. The name of the
identifier is indicated by a literal index. Each symbol record contains the file index
and source line number for the file and line in which the symbol was declared.
If the identifier is part of a structure or union, the symbol record contains a symbol
index for each of the following:
v The first sibling, if any
v The parent, if any
v The first child, if any
Consider the following structure:
dcl
1 a
, 3 b fixed bin
, 3 c fixed bin
, 3 d
, 5 e fixed bin
, 5 f fixed bin
;
The symbol indices assigned to the elements of the preceding structure would be
as follows:
declare /* */
1 xin_Ord_Elem based( null() ), /* */
/* */
2 xin_Ord_Elem_Hdr /* standard header */
like xin_Hdr, /* */
/* */
2 xin_Ord_Elem_File_Id /* file id */
/* */
2 xin_Ord_Elem_Line_No /* line no within file */
/* */
2 xin_Ord_Elem_Id /* identifying number */
/* */
2 xin_Ord_Elem_Type_Id /* id of ordinal type */
/* */
2 xin_Ord_Elem_Value /* ordinal value */
/* */
2 xin_Ord_Elem_Name /* ordinal name */
fixed bin(31); /* */
Figure 117. Declare for an ordinal element record
Each symbol record also contains a series of bit(1) fields that indicate if various
attributes apply to this variable.
Each symbol record also contains the following elements:
User-given structure level
This is a user-given structure level for the identifier. For the element c of the
structure above, the value is 3. For non-structure members, the value is set to
1.
Logican structure level
The logical structure level for the identifier For the element c of the structure
above, the value is 2. For non-structure members, the value is set to 1.
Dimensions
The number of dimensions declared for the variable not counting any inherited
dimensions.
The number of dimensions for the variable including all inherited dimensions.
Offset
The offset into the outermost parent structure.
Elemental size
Elemental size is in bytes unless the variable is bit aligned, in which case it is
in bits. In either case, this does not factor any in dimensions.
Size
Size in bytes with its dimensions factored in.
Alignment
Identified by the following:
v 0 for bit-aligned
v 7 for byte-aligned
v 15 for halfword-aligned
v 31 for fullword-aligned
v 63 for quadword-aligned
A union within the record is dedicated to describing information that is dependent
on the variable's storage class:
Static variables
If the variable is declared as external with a separate external name (dcl x
ext('y')), the literal index of that name is specified.
Based variables
If the variable is declared as based on another mapped variable that is not an
element of an array, the symbol index of that variable is specified.
symbol index sibling parent child
----- ----- ------- ------ -----
a 1 0 0 2
b 2 3 1 0
c 3 4 1 0
d 4 0 1 5
e 5 6 4 0
f 6 0 4 0
Figure 118. Symbol indices assigned to the elements of a structure
Defined variables
If the variable is declared as defined on another mapped variable that is not an
element of an array, the symbol index of that variable is specified here. If its
position attribute is constant, it is also specified.
The variable's data type is specified by the ordinal shown in Figure 119.
A union within the record is dedicated to describing information that is dependent
on the variable's data type. Most of this information is self-explanatory (for
example, the precision for an arithmetic type) except perhaps for the following:
Picture variables
The literal index of the picture specification is specified.
Entry variables
If the variable has the returns attribute, the symbol index of the returns
description is specified.
Ordinal variables
The ordinal type index is specified.
Typed variables and handles
The symbol index of the underlying type is specified.
String and area variables
The type and value of the extent is specified in addition to the symbol index of
the returns description. The type of the extent is encoded by the values:
declare
( xin_Extent_Constant value(01)
,xin_Extent_Star value(02)
,xin_Extent_Nonconstant value(04)
define
ordinal
xin_Data_Kind
( xin_Data_Kind_Unset
,xin_Data_Kind_Character
,xin_Data_Kind_Bit
,xin_Data_Kind_Graphic
,xin_Data_Kind_Fixed
,xin_Data_Kind_Float
,xin_Data_Kind_Picture
,xin_Data_Kind_Pointer
,xin_Data_Kind_Offset
,xin_Data_Kind_Entry
,xin_Data_Kind_File
,xin_Data_Kind_Label
,xin_Data_Kind_Format
,xin_Data_Kind_Area
,xin_Data_Kind_Task
,xin_Data_Kind_Event
,xin_Data_Kind_Condition
,xin_Data_Kind_Structure
,xin_Data_Kind_Union
,xin_Data_Kind_Descriptor
,xin_Data_Kind_Ordinal
,xin_Data_Kind_Handle
,xin_Data_Kind_Type
) prec(8) unsigned;
Figure 119. Data type of a variable
,xin_Extent_Refer value(08)
,xin_Extent_In_Error value(16)
)
fixed bin;
If the element has any dimensions, the type and values for its lower and upper
bounds are specified at the very end of the record. These fields are not present if
the element has no dimensions. Figure 120 shows the declare for a symbol record.
declare /* */
1 xin_Sym based( null() ), /* */
/* */
2 Xin_Sym_Hdr /* standard header */
like Xin_Hdr, /* */
/* */
2 Xin_Sym_File_Id /* file id */
/* */
2 Xin_Sym_Line_No /* line no within file */
/* */
2 xin_Id /* identifying number */
/* */
2 xin_Sibling /* xin_id of next sibling */
/* */
2 xin_Parent /* xin_id of parent */
/* */
2 xin_Child /* xin_id of first child */
/* */
2 xin_Blk_Id /* blk_id of owning block */
/* */
2 xin_Sym_Tok /* token id of declaring token */
/* */
2 xin_Logical_Level /* logical level in structure */
unsigned fixed bin(08), /* */
/* */
2 xin_Physical_Level /* given level in structure */
/* */
2 xin_Total_Dims /* Total number of dims */
/* */
2 xin_Own_Dims /* count of self-made dims */
/* */
Figure 120. Declare for a symbol record (Part 1 of 7)
2 xin_Attr_Flags union, /* */
/* */
3 * bit(64), /* */
/* */
3 *, /* */
/* */
4 xin_Attr_Automatic /* */
bit(1), /* */
4 xin_Attr_Based /* */
bit(1), /* */
4 xin_Attr_Controlled /* */
bit(1), /* */
4 xin_Attr_Defined /* */
bit(1), /* */
4 xin_Attr_Parameter /* */
bit(1), /* */
4 xin_Attr_Position /* */
bit(1), /* */
4 xin_Attr_Reserved /* */
bit(1), /* */
4 xin_Attr_Static /* */
bit(1), /* */
4 xin_Attr_Condition /* */
bit(1), /* */
4 xin_Attr_Constant /* */
bit(1), /* */
4 xin_Attr_Variable /* */
bit(1), /* */
4 xin_Attr_Internal /* */
bit(1), /* */
4 xin_Attr_External /* */
bit(1), /* */
4 xin_Attr_Abnormal /* */
bit(1), /* */
4 xin_Attr_Normal /* */
bit(1), /* */
4 xin_Attr_Assignable /* */
bit(1), /* */
4 xin_Attr_Nonassignable /* */
bit(1), /* */
4 xin_Attr_Aligned /* */
bit(1), /* */
4 xin_Attr_Unaligned /* */
bit(1), /* */
4 xin_Attr_Descriptor /* */
bit(1), /* */
4 xin_Attr_Value /* */
bit(1), /* */
4 xin_Attr_Byvalue /* */
bit(1), /* */
4 xin_Attr_Byaddr /* */
bit(1), /* */
4 xin_Attr_Connected /* */
bit(1), /* */
4 xin_Attr_Nonconnected /* */
bit(1), /* */
4 xin_Attr_Optional /* */
bit(1), /* */
4 xin_Attr_Native /* */
bit(1), /* */
4 xin_Attr_Nonnative /* */
bit(1), /* */
4 xin_Attr_Initial /* */
bit(1), /* */
4 xin_Attr_Typedef /* */
bit(1), /* */
4 xin_Attr_Builtin /* */
bit(1), /* */
4 xin_Attr_Generic /* */
bit(1), /* */
4 xin_Attr_Date /* */
bit(1), /* */
4 xin_Attr_Noinit /* */
bit(1), /* */
/* */
2 xin_Data_Is /* */
ordinal xin_Data_Kind, /* */
/* */
2 xin_Misc_Flags union, /* */
/* */
3 * bit(8), /* */
/* */
3 *, /* */
/* */
4 xin_Implicit_Dcl /* dcl is implicit */
bit(1), /* */
/* */
4 xin_Contextual_Dcl /* dcl is contextual */
bit(1), /* */
/* */
4 xin_Has_Been_Mapped /* aggregate has been mapped */
bit(1), /* */
/* */
2 xin_Align /* alignment */
/* */
2 xin_Begin_Offset /* bitlocation(sym) */
/* */
2 xin_Offset /* location(sym) */
/* */
2 xin_Size /* length in bytes, with all */
fixed bin(31), /* children and array */
/* elements factored in */
/* */
2 xin_Base_Size /* element length - in bytes */
fixed bin(31), /* unless bit aligned */
/* */
2 xin_Name /* name - id of lit record */
/* */
2 * union, /* */
/* */
3 xin_Static_Data, /* */
/* */
4 xin_Static_Ext /* id of literal specifing its */
fixed bin(31), /* external name */
/* */
3 xin_Based_Data, /* */
/* */
4 xin_Based_On_Id /* xin_Id of basing reference */
fixed bin(31), /* 0 if not simple */
/* */
3 xin_Defined_Data, /* */
/* */
4 xin_Defined_On_Id /* xin_Id of basing reference */
fixed bin(31), /* 0 if not simple */
/* */
4 xin_Defined_Pos /* -1 if not constant */
/* */
3 xin_Parm_Data, /* */
/* */
4 xin_Parm_Index /* index of parm */
fixed bin(31), /* 1 for first, etc */
/* */
2 * union, /* */
/* */
3 xin_Str_Data, /* used for char, bit, graphic */
/* and area, but not used for */
/* picture character or numeric*/
/* */
4 xin_Str_Len_Node /* length as parse tree */
/* */
4 xin_Str_Len /* length: if type is constant */
/* */
4 xin_Str_Len_Type /* type */
/* */
4 *, /* */
5 xin_Str_Varying /* */
bit(1), /* */
5 xin_Str_Nonvarying /* */
bit(1), /* */
5 xin_Str_Varyingz /* */
bit(1), /* */
/* */
4 xin_Str_Date /* index of date literal */
/* */
3 xin_Arith_Data, /* used for fixed and float */
/* */
4 xin_Arith_Precision /* precision */
/* */
4 xin_Arith_Scale_Factor /* scale factor */
signed fixed bin(07), /* */
/* */
4 *, /* */
5 xin_Arith_Binary /* */
bit(1), /* */
5 xin_Arith_Decimal /* */
bit(1), /* */
5 xin_Arith_Fixed /* */
bit(1), /* */
5 xin_Arith_Float /* */
bit(1), /* */
5 xin_Arith_Real /* */
bit(1), /* */
5 xin_Arith_Complex /* */
bit(1), /* */
5 xin_Arith_Signed /* */
bit(1), /* */
5 xin_Arith_Unsigned /* */
bit(1), /* */
5 xin_Arith_Ieee /* */
bit(1), /* */
5 xin_Arith_Hexadec /* */
bit(1), /* */
/* */
4 * /* unused */
/* */
4 * /* unused */
/* */
4 xin_Arith_Date /* index of date literal */
/* */
3 xin_Ordinal_Data, /* used for ordinal */
/* */
4 xin_Ordinal_Type_Id /* type id */
/* */
3 xin_Type_Data, /* used for typed */
/* */
4 xin_Type_Is /* type id */
/* */
3 xin_Pic_Data, /* used for all pictures */
/* */
4 xin_Pic_Ext /* external specification */
/* */
4 *, /* */
5 xin_Pic_Fixed /* */
bit(1), /* */
5 xin_Pic_Float /* */
bit(1), /* */
5 xin_Pic_Character /* */
bit(1), /* */
5 xin_Pic_Real /* */
bit(1), /* */
5 xin_Pic_Complex /* */
bit(1), /* */
5 * /* */
bit(3), /* */
5 * /* */
bit(8), /* */
5 xin_Pic_Prec /* */
unsigned /* */
5 xin_Pic_Scale /* */
signed /* */
/* */
4 * /* unused */
/* */
4 xin_Pic_Date /* index of date literal */
/* */
3 xin_Entry_Data, /* */
/* */
4 xin_Entry_Min /* min number of args */
fixed bin(15), /* allowed when invoked */
/* */
4 xin_Entry_Max /* max number of args */
fixed bin(15), /* allowed when invoked */
/* */
4 xin_Entry_Returns_Id /* xin_Id of returns descriptor*/
/* */
4 xin_Entry_Parms_Id /* xin_Id of first parms */
/* */
4 *, /* */
5 xin_Entry_Returns /* */
bit(1), /* */
5 xin_Entry_Limited /* */
bit(1), /* */
5 xin_Entry_Fetchable /* */
bit(1), /* */
5 xin_Entry_Is_Proc /* */
bit(1), /* */
5 xin_Entry_Is_Secondary /* */
bit(1), /* */
/* */
3 xin_Ptr_Data, /* */
/* */
4 *, /* */
5 xin_Ptr_Segmented /* */
bit(1), /* */
/* */
3 xin_Offset_Data, /* */
/* */
4 xin_Offset_Area /* */
/* */
3 xin_Sym_Bif_Id /* */
ordinal xin_Bif_Kind, /* */
/* */
/* */
3 xin_File_Data, /* */
/* */
4 *, /* */
5 xin_File_Buffered /* */
bit(1), /* */
5 xin_File_Direct /* */
bit(1), /* */
5 xin_File_Exclusive /* */
bit(1), /* */
5 xin_File_Input /* */
bit(1), /* */
5 xin_File_Keyed /* */
bit(1), /* */
5 xin_File_Output /* */
bit(1), /* */
5 xin_File_Print /* */
bit(1), /* */
5 xin_File_Record /* */
bit(1), /* */
5 xin_File_Stream /* */
bit(1), /* */
5 xin_File_Transient /* */
bit(1), /* */
5 xin_File_Unbuffered /* */
bit(1), /* */
5 xin_File_Update /* */
bit(1), /* */
/* */
2 * union, /* */
/* */
3 xin_Value_Id /* id of value lit - if the */
fixed bin(31), /* xin_Attr_Value flag is set */
/* */
3 xin_First_Stmt_Id /* id of first stmt record - */
fixed bin(31), /* if xin_Attr_Entry and */
/* xin_Entry_Is_Proc flags */
/* are both set */
/* */
2 xin_Bounds dim(15), /* */
/* */
3 xin_Lbound_Type /* lbound type */
/* */
3 xin_Hbound_Type /* hbound type */
/* */
3 * /* */
char(2), /* */
/* */
3 xin_Lbound_Node /* expression parse tree */
/* */
3 xin_Hbound_Node /* expression parse tree */
/* */
3 xin_Lbound /* value: if type is constant */
fixed bin(31), /* xin_Id: if type is refer */
/* */
3 xin_Hbound /* value: if type is constant */
fixed bin(31), /* xin_Id: if type is refer */
/* */
2 * char(0);
The definition of the ordinal xin_Bif_Kind can be found in Figure 121
define
ordinal
xin_Bif_Kind
( xin_Bif_Unknown
,xin_bif_abs
,xin_bif_acos
,xin_bif_add
,xin_bif_addr
,xin_bif_all
,xin_bif_allocation
,xin_bif_allocn
,xin_bif_any
,xin_bif_asin
,xin_bif_atan
,xin_bif_atand
,xin_bif_atanh
,xin_bif_bin
,xin_bif_binvalue
,xin_bif_binary
,xin_bif_binaryvalue
,xin_bif_bit
,xin_bif_bool
,xin_bif_ceil
,xin_bif_char
,xin_bif_completion
,xin_bif_complex
,xin_bif_conjg
,xin_bif_copy
,xin_bif_cos
,xin_bif_cosd
,xin_bif_cosh
,xin_bif_count
,xin_bif_cpln
,xin_bif_cplx
,xin_bif_cstg
,xin_bif_currentstorage
,xin_bif_datafield
,xin_bif_date
,xin_bif_datetime
,xin_bif_dec
,xin_bif_decimal
,xin_bif_dim
,xin_bif_divide
,xin_bif_empty
,xin_bif_entryaddr
,xin_bif_erf
,xin_bif_erfc
,xin_bif_exp
,xin_bif_fixed
,xin_bif_float
,xin_bif_floor
,xin_bif_graphic
,xin_bif_hbound
Figure 121. Declare for xin_Bif_Kind (Part 1 of 6)
,xin_bif_high
,xin_bif_imag
,xin_bif_index
,xin_bif_lbound
,xin_bif_length
,xin_bif_lineno
,xin_bif_log
,xin_bif_log10
,xin_bif_log2
,xin_bif_low
,xin_bif_max
,xin_bif_min
,xin_bif_mod
,xin_bif_mpstr
,xin_bif_multiply
,xin_bif_null
,xin_bif_offset
,xin_bif_onchar
,xin_bif_oncode
,xin_bif_oncount
,xin_bif_onfile
,xin_bif_onkey
,xin_bif_onloc
,xin_bif_onsource
,xin_bif_pageno
,xin_bif_plicanc
,xin_bif_plickpt
,xin_bif_plidump
,xin_bif_plirest
,xin_bif_pliretc
,xin_bif_pliretv
,xin_bif_plisrta
,xin_bif_plisrtb
,xin_bif_plisrtc
,xin_bif_plisrtd
,xin_bif_plitest
,xin_bif_pointer
,xin_bif_pointeradd
,xin_bif_pointervalue
,xin_bif_poly
,xin_bif_prec
,xin_bif_precision
,xin_bif_priority
,xin_bif_prod
,xin_bif_ptr
,xin_bif_ptradd
,xin_bif_ptrvalue
,xin_bif_real
,xin_bif_repeat
,xin_bif_round
,xin_bif_samekey
,xin_bif_sign
,xin_bif_sin
,xin_bif_sind
,xin_bif_sinh
,xin_bif_sqrt
,xin_bif_status
,xin_bif_stg
,xin_bif_storage
,xin_bif_string
,xin_bif_substr
,xin_bif_sum
,xin_bif_sysnull
,xin_bif_tan
,xin_bif_tand
,xin_bif_tanh
,xin_bif_time
,xin_bif_translate
,xin_bif_trunc
,xin_bif_unspec
,xin_bif_verify
,xin_bif_days
,xin_bif_daystodate
,xin_bif_acosf
,xin_bif_addrdata
,xin_bif_alloc
,xin_bif_allocate
,xin_bif_allocsize
,xin_bif_asinf
,xin_bif_atanf
,xin_bif_auto
,xin_bif_automatic
,xin_bif_availablearea
,xin_bif_bitloc
,xin_bif_bitlocation
,xin_bif_byte
,xin_bif_cds
,xin_bif_center
,xin_bif_centerleft
,xin_bif_centerright
,xin_bif_centre
,xin_bif_centreleft
,xin_bif_centreright
,xin_bif_character
,xin_bif_charg
,xin_bif_chargraphic
,xin_bif_charval
,xin_bif_checkstg
,xin_bif_collate
,xin_bif_compare
,xin_bif_cosf
,xin_bif_cs
,xin_bif_currentsize
,xin_bif_daystosecs
,xin_bif_dimension
,xin_bif_edit
,xin_bif_endfile
,xin_bif_epsilon
,xin_bif_expf
,xin_bif_exponent
,xin_bif_fileddint
,xin_bif_fileddtest
,xin_bif_fileddword
,xin_bif_fileid
,xin_bif_fileread
,xin_bif_fileseek
,xin_bif_filetell
,xin_bif_filewrite
,xin_bif_gamma
,xin_bif_getenv
,xin_bif_handle
,xin_bif_hex
,xin_bif_heximage
,xin_bif_huge
,xin_bif_iand
,xin_bif_ieor
,xin_bif_inot
,xin_bif_ior
,xin_bif_isigned
,xin_bif_isll
,xin_bif_ismain
,xin_bif_isrl
,xin_bif_iunsigned
,xin_bif_left
,xin_bif_loc
,xin_bif_location
,xin_bif_log10f
,xin_bif_logf
,xin_bif_loggamma
,xin_bif_lower2
,xin_bif_lowercase
,xin_bif_maxexp
,xin_bif_maxlength
,xin_bif_memindex
,xin_bif_memsearch
,xin_bif_memsearchr
,xin_bif_memverify
,xin_bif_memverifyr
,xin_bif_minexp
,xin_bif_offsetadd
,xin_bif_offsetdiff
,xin_bif_offsetsubtract
,xin_bif_offsetvalue
,xin_bif_omitted
,xin_bif_oncondcond
,xin_bif_oncondid
,xin_bif_ongsource
,xin_bif_onsubcode
,xin_bif_onwchar
,xin_bif_onwsource
,xin_bif_ordinalname
,xin_bif_ordinalpred
,xin_bif_ordinalsucc
,xin_bif_packagename
,xin_bif_picspec
,xin_bif_places
,xin_bif_pliascii
,xin_bif_pliebcdic
,xin_bif_plifill
,xin_bif_plifree
,xin_bif_plimove
,xin_bif_pliover
,xin_bif_plisaxa
,xin_bif_plisaxb
,xin_bif_pointerdiff
,xin_bif_pointersubtract
,xin_bif_pred
,xin_bif_present
,xin_bif_procedurename
,xin_bif_procname
,xin_bif_ptrdiff
,xin_bif_ptrsubtract
,xin_bif_putenv
,xin_bif_radix
,xin_bif_raise2
,xin_bif_random
,xin_bif_rank
,xin_bif_rem
,xin_bif_repattern
,xin_bif_replaceby2
,xin_bif_reverse
,xin_bif_right
,xin_bif_scale
,xin_bif_search
,xin_bif_searchr
,xin_bif_secs
,xin_bif_secstodate
,xin_bif_secstodays
,xin_bif_signed
,xin_bif_sinf
,xin_bif_size
,xin_bif_sourcefile
,xin_bif_sourceline
,xin_bif_sqrtf
,xin_bif_subtract
,xin_bif_succ
,xin_bif_system
,xin_bif_tally
,xin_bif_tanf
,xin_bif_threadid
,xin_bif_tiny
,xin_bif_trim
,xin_bif_type
,xin_bif_unallocated
,xin_bif_unsigned
,xin_bif_uppercase
,xin_bif_valid
,xin_bif_validdate
,xin_bif_varglist
,xin_bif_vargsize
,xin_bif_verifyr
,xin_bif_wchar
,xin_bif_wcharval
,xin_bif_weekday
,xin_bif_whigh
,xin_bif_widechar
,xin_bif_wlow
,xin_bif_xmlchar
,xin_bif_y4date
,xin_bif_y4julian
,xin_bif_y4year
Note that the attributes flags reflect the attributes after the compiler has applied all
defaults. So, for example, every numeric variable (including numeric PICTURE
variables) has either the REAL or COMPLEX attribute flag set.
Understanding SYSADATA syntax information
When you specify the SYN suboption of the XINFO compile-time option, the
compiler generates a SYSADATA file that contains the following information in
addition to the records generated for the MSG and SYM suboptions:
v Source records
v Token records
v Syntax records
Source records
Each source record assigns a number, called a source id, that is used by later
records to refer to the source line described by this record. The line may be from a
,xin_bif_fixedbin
,xin_bif_fixeddec
,xin_bif_floatbin
,xin_bif_floatdec
,xin_bif_isfinite
,xin_bif_isnan
,xin_bif_isnormal
,xin_bif_iszero
,xin_bif_memconvert
,xin_bif_memcu12
,xin_bif_memcu14
,xin_bif_memcu21
,xin_bif_memcu24
,xin_bif_memcu41
,xin_bif_memcu42
,xin_bif_online
,xin_bif_onoffset
,xin_bif_plidelete
,xin_bif_plisaxc
,xin_bif_plisaxd
,xin_bif_plitran11
,xin_bif_plitran12
,xin_bif_plitran21
,xin_bif_plitran22
,xin_bif_reg12
,xin_bif_rounddec
,xin_bif_stackaddr
,xin_bif_ulength
,xin_bif_ulength16
,xin_bif_ulength8
,xin_bif_upos
,xin_bif_usubstr
,xin_bif_usurrogate
,xin_bif_uvalid
,xin_bif_uwidth
,xin_bif_onarea
,xin_bif_indicators
,xin_bif_popcnt
,xin_bif_lboundacross
,xin_bif_hboundacross
) prec(16) unsigned;
the primary PL/I source file or an INCLUDEd file, as indicated by the source file
id and linenumber fields in the record. The rest of the record holds the actual data
in the source line.
Token records
Each token record assigns a number, called a token index, that is used by later
records to refer to a token recognized by the PL/I compiler. The record also
identifies the type of the token plus the column and line on which it started and
ended.
Dcl
1 Xin_Src Based( null() ), /* source record */
/* */
2 Xin_Src_Hdr /* standard header */
like Xin_Hdr, /* */
/* */
2 Xin_Src_File_Id /* file id */
/* */
2 Xin_Src_Line_No /* line no within file */
/* */
2 Xin_Src_Id /* id for this source record */
/* */
2 Xin_Src_Length /* length of text */
/* */
2 Xin_Src_Text /* actual text */
char( 137 refer(xin_Src_Length) );
Figure 122. Declare for a source record
The ordinal xin_Tok_Kind identifies the type of the token record.
Syntax records
Each syntax record assigns a number, called a node id, that is used by later records
to refer to other syntax records.
The first syntax record will have kind xin_Syn_Kind_Package, and if the
compilation unit has any procedures, the child node of this record will point to the
first of these procedures. The parent, sibling and child nodes will then provide a
map with the appropriate relationships of all the procedures and begin bocks in
the compilation unit.
Consider the following simple program:
a: proc;
call b;
call c;
b: proc;
Dcl
1 Xin_Tok Based( null() ), /* token record */
/* */
2 Xin_Tok_Hdr /* standard header */
like Xin_Hdr, /* */
/* */
2 Xin_Tok_Inx /* adata index for token */
/* */
2 Xin_Tok_Begin_Line /* starting line no within file */
/* */
2 Xin_Tok_End_Line_Offset /* offset of end line from first */
/* */
2 Xin_Tok_Kind_Value /* token kind */
ordinal xin_Tok_Kind, /* */
/* */
2 Xin_Tok_Rsrvd /* reserved */
/* */
2 Xin_Tok_Begin_Col /* starting column */
/* */
2 Xin_Tok_End_Col /* ending column */
Figure 123. Declare for a token record
Define
ordinal
xin_Tok_Kind
( xin_Tok_Kind_Unset
,xin_Tok_Kind_Lexeme
,xin_Tok_Kind_Comment
,xin_Tok_Kind_Literal
,xin_Tok_Kind_Identifier
,xin_Tok_Kind_Keyword
) prec(8) unsigned;
Figure 124. Declare for the token record kind
end b;
c: proc;
call d;
d: proc;
end d;
end c;
end a;
The node indices assigned to the blocks of the preceding program would be as
follows:
symbol index sibling parent child
----- ----- ------- ------ -----
- 1 0 0 2
a 2 0 1 3
b 3 4 2 0
c 4 0 2 5
d 5 0 4 0
Figure 125. Node indices assigned to the blocks in a program
Dcl
1 Xin_Syn Based( null() ), /* syntax record */
/* */
2 Xin_Syn_Hdr /* standard header */
like Xin_Hdr, /* */
/* */
2 Xin_Syn_Node_Id /* node id */
/* */
2 Xin_Syn_Node_Kind /* node type */
ordinal xin_syn_kind, /* */
/* */
2 Xin_Syn_Node_Exp_Kind /* node sub type */
ordinal xin_exp_kind, /* */
/* */
2 * /* reserved */
/* */
2 Xin_Syn_Parent_Node_Id /* node id of parent */
/* */
2 Xin_Syn_Sibling_Node_Id /* node id of sibling */
/* */
2 Xin_Syn_Child_Node_Id /* node id of child */
/* */
2 xin_Syn_First_Tok /* id of first spanned token */
/* */
2 xin_Syn_Last_Tok /* id of last spanned token */
Figure 126. Declare for a syntax record (Part 1 of 3)
/* */
2 * union, /* qualifier for node */
/* */
3 Xin_Syn_Int_Value /* used if int */
/* */
3 Xin_Syn_Literal_Id /* used if name, number, picture */
/* */
3 Xin_Syn_Node_Lex /* used if lexeme, assignment, */
ordinal xin_Lex_kind, /* infix_op, prefix_op */
/* */
3 Xin_Syn_Node_Voc /* used if keyword, end_for_do */
ordinal xin_Voc_kind, /* */
/* */
3 Xin_Syn_Block_Node /* used if call_begin */
fixed bin(31), /* to hold node of begin block */
/* */
3 Xin_Syn_Bif_Id /* used if bif_rfrnc */
/* */
3 Xin_Syn_Sym_Id /* used if label, unsub_rfrnc, */
fixed bin(32) unsigned, /* subscripted_rfrnc */
/* */
3 Xin_Syn_Proc_Data, /* used if package, proc or begin*/
/* */
4 Xin_Syn_First_Sym /* id of first contained sym */
/* */
4 Xin_Syn_Block_Sym /* id of sym for this block */
The ordinal xin_Syn_Kind identifies the type of the syntax record.
/* */
3 Xin_Syn_Number_Data, /* used if number */
/* */
4 Xin_Syn_Number_Id /* id of literal */
/* */
4 Xin_Syn_Number_Type /* type */
ordinal xin_Number_Kind,/* */
/* */
4 Xin_Syn_Number_Prec /* precision */
/* */
4 Xin_Syn_Number_Scale /* scale factor */
fixed bin(7) signed, /* */
/* */
4 Xin_Syn_Number_Bytes /* bytes it would occupy */
fixed bin(8) unsigned, /* in its internal form */
/* */
3 Xin_Syn_String_Data, /* used if char_string, */
/* bit_string, graphic_string */
/* */
4 Xin_Syn_String_Id /* id of literal */
/* */
4 Xin_Syn_String_Len /* string length in its units */
/* */
3 Xin_Syn_Stmt_Data, /* used if stmt */
/* */
4 Xin_Syn_File_Id /* file id */
/* */
4 Xin_Syn_Line_No /* line no within file */
/* */
2 * char(0); /* */
Consider the following simple program:
a: proc(x);
dcl x char(8);
x = substr(datetime(),1,8);
end;
The node indices assigned to the blocks of the preceding program would be as
follows:
Define
ordinal
xin_Syn_Kind
( xin_Syn_Kind_Unset
,xin_Syn_Kind_Lexeme
,xin_Syn_Kind_Asterisk
,xin_Syn_Kind_Int
,xin_Syn_Kind_Name
,xin_Syn_Kind_Expression
,xin_Syn_Kind_Parenthesized_Expr
,xin_Syn_Kind_Argument_List
,xin_Syn_Kind_Keyword
,xin_Syn_Kind_Proc_Stmt
,xin_Syn_Kind_Begin_Stmt
,xin_Syn_Kind_Stmt
,xin_Syn_Kind_Substmt
,xin_Syn_Kind_Label
,xin_Syn_Kind_Invoke_Begin
,xin_Syn_Kind_Assignment
,xin_Syn_Kind_Assignment_Byname
,xin_Syn_Kind_Do_Fragment
,xin_Syn_Kind_Keyed_List
,xin_Syn_Kind_Iteration_Factor
,xin_Syn_Kind_If_Clause
,xin_Syn_Kind_Else_Clause
,xin_Syn_Kind_Do_Stmt
,xin_Syn_Kind_Select_Stmt
,xin_Syn_Kind_When_Stmt
,xin_Syn_Kind_Otherwise_Stmt
,xin_Syn_Kind_Procedure
,xin_Syn_Kind_Package
,xin_Syn_Kind_Begin_Block
,xin_Syn_Kind_Picture
,xin_Syn_Kind_Raw_Rfrnc
,xin_Syn_Kind_Generic_Desc
) prec(8) unsigned;
Figure 127. Declare for the syntax record kind
The procedure record contains the identifier (in the block_sym field) for the symbol
record for ENTRY A. This symbol record contains, in turn, the node identifier (in
the first_stmt_id field) for the first statement in that procedure.
Note that for the statement records
v the sibling node identifier points to the next statement record, if any
v the child node identifier points to the first element of that statement record
The records for the PROCEDURE statement consists of 4 records:
v a label record
v a keyword record (for the PROCEDURE keyword)
v an expression record (for the parameter X) with expression kind of unsub_rfrnc
and a sym_id for the symbol X
v a lexeme record (for the semicolon)
The records for the assignment statement consists of 2 records:
v an assignment record which has 2 children:
an expression record (for the target X) with expression kind of unsub_rfrnc
and a sym_id for the symbol X
an expression record (for the source) with expression kind of builtin_rfrnc and
a sym_id for the symbol SUBSTR, and this record has itself 3 children:
- an expression record (for the first argument) with expression kind of
builtin_rfrnc and a sym_id for the symbol DATETIME
- an expression record (for the second argument) with expression kind of
number and a literal_id for the value 1
- an expression record (for the third argument) with expression kind of
number and a literal_id for the value 8
The records for the END statement consists of 2 records:
node_kind index sibling parent child
----------- ----- ------- ------ -----
package 1 0 0 2
procedure 2 0 1 0
expression 3 0 0 0
stmt 4 5 2 6
stmt 5 10 2 11
label 6 7 4 0
keyword 7 8 4 0
expression 8 9 4 0
lexeme 9 0 4 0
stmt 10 0 2 18
assignment 11 12 5 13
lexeme 12 0 5 0
expression 13 14 11 0
keyword 18 19 10 0
lexeme 19 0 10 0
Figure 128. Node indices assigned to the syntax records in a program
v a keyword record (for the END keyword)
The ordinal xin_Exp_Kind identifies the type of an expression for a syntax record
that describes an expression. Some of these records will have non-zero child nodes,
for example:
v an infix op, such as a minus for a subtraction, will have a child node that
describes its lefthand operand (and the sibling node of that operand will
describe the righthand operator)
v a prefix op, such as a minus for a negation, will have a child node that describes
its operand
The ordinal xin_Number_Kind identifies the type of a number for a syntax record
that describes a number.
The ordinal xin_Lex_Kind identifies the type of a lexeme for a syntax record that
describes a lexical unit. In these ordinal names,
v "vrule" means "vertical rule" which is used, for instance, as the "or" symbol
Define
ordinal
xin_Exp_Kind
( xin_Exp_Kind_Unset
,xin_Exp_Kind_Bit_String
,xin_Exp_Kind_Char_String
,xin_Exp_Kind_Graphic_String
,xin_Exp_Kind_Number
,xin_Exp_Kind_Infix_Op
,xin_Exp_Kind_Prefix_Op
,xin_Exp_Kind_Builtin_Rfrnc
,xin_Exp_Kind_Entry_Rfrnc
,xin_Exp_Kind_Qualified_Rfrnc
,xin_Exp_Kind_Unsub_Rfrnc
,xin_Exp_Kind_Subscripted_Rfrnc
,xin_Exp_Kind_Type_Func
,xin_Exp_Kind_Widechar_String
) prec(8) unsigned;
Figure 129. Declare for the expression kind
Define
ordinal
xin_Number_Kind
( xin_Number_Kind_Unset
,xin_Number_Kind_Real_Fixed_Bin
,xin_Number_Kind_Real_Fixed_Dec
,xin_Number_Kind_Real_Float_Bin
,xin_Number_Kind_Real_Float_Dec
,xin_Number_Kind_Cplx_Fixed_Bin
,xin_Number_Kind_Cplx_Fixed_Dec
,xin_Number_Kind_Cplx_Float_Bin
,xin_Number_Kind_Cplx_Float_Dec
) prec(8) unsigned;
Figure 130. Declare for the number kind
v "dbl" means "double", so that dbl_Vrule is a doubled vertical rule which is used,
for instance, as the "concatenate" symbol
The ordinal xin_Voc_Kind identifies the keyword for a syntax record that describes
an item from the compiler's "vocabulary".
Define
ordinal
xin_Lex_Kind
( xin_Lex_Undefined
,xin_Lex_Period
,xin_Lex_Colon
,xin_Lex_Semicolon
,xin_Lex_Lparen
,xin_Lex_Rparen
,xin_Lex_Comma
,xin_Lex_Equals
,xin_Lex_Gt
,xin_Lex_Ge
,xin_Lex_Lt
,xin_Lex_Le
,xin_Lex_Ne
,xin_Lex_Lctr
,xin_Lex_Star
,xin_Lex_Dbl_Colon
,xin_Lex_Not
,xin_Lex_Vrule
,xin_Lex_Dbl_Vrule
,xin_Lex_And
,xin_Lex_Dbl_Star
,xin_Lex_Plus
,xin_Lex_Minus
,xin_Lex_Slash
,xin_Lex_Equals_Gt
,xin_Lex_Lparen_Colon
,xin_Lex_Colon_Rparen
,xin_Lex_Plus_Equals
,xin_Lex_Minus_Equals
,xin_Lex_Star_Equals
,xin_Lex_Slash_Equals
,xin_Lex_Vrule_Equals
,xin_Lex_And_Equals
,xin_Lex_Dbl_Star_Equals
,xin_Lex_Dbl_Vrule_Equals
,xin_Lex_Dbl_Slash
) unsigned prec(16);
Figure 131. Declare for the lexeme kind
Define
ordinal
xin_Voc_Kind
( xin_Voc_Undefined
,xin_Voc_a
,xin_Voc_abnormal
,xin_Voc_act
,xin_Voc_activate
,xin_Voc_adata
,xin_Voc_addbuff
,xin_Voc_aggregate
,xin_Voc_aix
,xin_Voc_alias
,xin_Voc_alien
,xin_Voc_aligned
,xin_Voc_all
,xin_Voc_alloc
,xin_Voc_allocate
,xin_Voc_anno
,xin_Voc_ans
,xin_Voc_any
,xin_Voc_anycond
,xin_Voc_anycondition
,xin_Voc_area
,xin_Voc_as
,xin_Voc_ascii
,xin_Voc_asgn
,xin_Voc_asm
,xin_Voc_asmtdli
,xin_Voc_assembler
,xin_Voc_assignable
,xin_Voc_attach
,xin_Voc_attention
,xin_Voc_attn
,xin_Voc_attribute
,xin_Voc_attributes
,xin_Voc_auto
,xin_Voc_automatic
,xin_Voc_b
,xin_Voc_backwards
,xin_Voc_based
,xin_Voc_begin
,xin_Voc_beta
,xin_Voc_bigendian
,xin_Voc_bin
,xin_Voc_binary
,xin_Voc_bind
,xin_Voc_bit
,xin_Voc_bkwd
,xin_Voc_blksize
,xin_Voc_block
,xin_Voc_buf
Figure 132. Declare for the voc kind (Part 1 of 15)
,xin_Voc_buffered
,xin_Voc_buffers
,xin_Voc_bufnd
,xin_Voc_bufni
,xin_Voc_bufoff
,xin_Voc_bufsp
,xin_Voc_build
,xin_Voc_builtin
,xin_Voc_by
,xin_Voc_byaddr
,xin_Voc_byname
,xin_Voc_byvalue
,xin_Voc_c
,xin_Voc_call
,xin_Voc_cdecl
,xin_Voc_cdecl16
,xin_Voc_cee
,xin_Voc_ceetdli
,xin_Voc_cell
,xin_Voc_char
,xin_Voc_character
,xin_Voc_charg
,xin_Voc_chargraphic
,xin_Voc_charset
,xin_Voc_check
,xin_Voc_cics
,xin_Voc_class
,xin_Voc_close
,xin_Voc_cmp
,xin_Voc_cmpat
,xin_Voc_cms
,xin_Voc_cmstpl
,xin_Voc_cobol
,xin_Voc_col
,xin_Voc_column
,xin_Voc_compile
,xin_Voc_complex
,xin_Voc_cond
,xin_Voc_condition
,xin_Voc_conn
,xin_Voc_connected
,xin_Voc_consecutive
,xin_Voc_constant
,xin_Voc_control
,xin_Voc_controlled
,xin_Voc_conv
,xin_Voc_conversion
,xin_Voc_copy
,xin_Voc_count
,xin_Voc_cplx
,xin_Voc_create
,xin_Voc_cs
,xin_Voc_ct
,xin_Voc_ctl
,xin_Voc_ctl360
,xin_Voc_ctlasa
,xin_Voc_currency
,xin_Voc_d
,xin_Voc_data
,xin_Voc_dataonly
,xin_Voc_db
,xin_Voc_dcl
,xin_Voc_deact
,xin_Voc_deactivate
,xin_Voc_debug
,xin_Voc_dec
,xin_Voc_decimal
,xin_Voc_deck
,xin_Voc_declare
,xin_Voc_def
,xin_Voc_default
,xin_Voc_define
,xin_Voc_defined
,xin_Voc_defines
,xin_Voc_delay
,xin_Voc_delete
,xin_Voc_desclist
,xin_Voc_desclocator
,xin_Voc_descriptor
,xin_Voc_descriptors
,xin_Voc_detach
,xin_Voc_dft
,xin_Voc_dim
,xin_Voc_dimension
,xin_Voc_direct
,xin_Voc_directed
,xin_Voc_display
,xin_Voc_dli
,xin_Voc_dllinit
,xin_Voc_do
,xin_Voc_downthru
,xin_Voc_dummydesc
,xin_Voc_duplicate
,xin_Voc_e
,xin_Voc_ebcdic
,xin_Voc_edit
,xin_Voc_alpha
,xin_Voc_else
,xin_Voc_emulate
,xin_Voc_enclave
,xin_Voc_end
,xin_Voc_endf
,xin_Voc_endfile
,xin_Voc_endif
,xin_Voc_endp
,xin_Voc_endpage
,xin_Voc_entry
,xin_Voc_enu
,xin_Voc_env
,xin_Voc_environment
,xin_Voc_error
,xin_Voc_esd
,xin_Voc_evendec
,xin_Voc_event
,xin_Voc_exclusive
,xin_Voc_exec
,xin_Voc_execops
,xin_Voc_execute
,xin_Voc_exit
,xin_Voc_exports
,xin_Voc_ext
,xin_Voc_extchk
,xin_Voc_external
,xin_Voc_externalonly
,xin_Voc_extname
,xin_Voc_extonly
,xin_Voc_f
,xin_Voc_fastcall
,xin_Voc_fastcall16
,xin_Voc_fb
,xin_Voc_fbs
,xin_Voc_fetch
,xin_Voc_fetchable
,xin_Voc_file
,xin_Voc_finish
,xin_Voc_first
,xin_Voc_fixed
,xin_Voc_fixeddec
,xin_Voc_fixedoverflow
,xin_Voc_flag
,xin_Voc_float
,xin_Voc_flow
,xin_Voc_flush
,xin_Voc_fofl
,xin_Voc_forever
,xin_Voc_format
,xin_Voc_fortran
,xin_Voc_free
,xin_Voc_from
,xin_Voc_fromalien
,xin_Voc_fs
,xin_Voc_full
,xin_Voc_g
,xin_Voc_generic
,xin_Voc_genkey
,xin_Voc_get
,xin_Voc_gn
,xin_Voc_go
,xin_Voc_gonumber
,xin_Voc_gostmt
,xin_Voc_goto
,xin_Voc_gr
,xin_Voc_graphic
,xin_Voc_gs
,xin_Voc_halt
,xin_Voc_handle
,xin_Voc_hexadec
,xin_Voc_hexadecimal
,xin_Voc_i
,xin_Voc_ibm
,xin_Voc_ieee
,xin_Voc_if
,xin_Voc_ign
,xin_Voc_ignore
,xin_Voc_imp
,xin_Voc_impl
,xin_Voc_implicit
,xin_Voc_imported
,xin_Voc_imprecise
,xin_Voc_ims
,xin_Voc_in
,xin_Voc_inc
,xin_Voc_incafter
,xin_Voc_incdir
,xin_Voc_include
,xin_Voc_incpath
,xin_Voc_indexarea
,xin_Voc_indexed
,xin_Voc_inherits
,xin_Voc_init
,xin_Voc_initfill
,xin_Voc_initial
,xin_Voc_inline
,xin_Voc_inout
,xin_Voc_input
,xin_Voc_insource
,xin_Voc_instance
,xin_Voc_int
,xin_Voc_inter
,xin_Voc_internal
,xin_Voc_interrupt
,xin_Voc_into
,xin_Voc_invalidop
,xin_Voc_ipa
,xin_Voc_irred
,xin_Voc_irreducible
,xin_Voc_is
,xin_Voc_iterate
,xin_Voc_itrace
,xin_Voc_jpn
,xin_Voc_k
,xin_Voc_key
,xin_Voc_keyed
,xin_Voc_keyfrom
,xin_Voc_keylength
,xin_Voc_keyloc
,xin_Voc_keyto
,xin_Voc_l
,xin_Voc_label
,xin_Voc_langlvl
,xin_Voc_last
,xin_Voc_laxconv
,xin_Voc_laxdcl
,xin_Voc_laxif
,xin_Voc_laxint
,xin_Voc_laxqual
,xin_Voc_lc
,xin_Voc_leave
,xin_Voc_library
,xin_Voc_libs
,xin_Voc_like
,xin_Voc_limited
,xin_Voc_limits
,xin_Voc_line
,xin_Voc_linecount
,xin_Voc_lineno
,xin_Voc_linesize
,xin_Voc_linkage
,xin_Voc_list
,xin_Voc_littleendian
,xin_Voc_lmessage
,xin_Voc_lmsg
,xin_Voc_local
,xin_Voc_localonly
,xin_Voc_locate
,xin_Voc_log
,xin_Voc_loop
,xin_Voc_lowerinc
,xin_Voc_lsfirst
,xin_Voc_m
,xin_Voc_macro
,xin_Voc_main
,xin_Voc_map
,xin_Voc_mar
,xin_Voc_margini
,xin_Voc_margins
,xin_Voc_mask
,xin_Voc_max
,xin_Voc_maxgen
,xin_Voc_maxmem
,xin_Voc_md
,xin_Voc_mdeck
,xin_Voc_member
,xin_Voc_metaclass
,xin_Voc_method
,xin_Voc_methods
,xin_Voc_mi
,xin_Voc_min
,xin_Voc_msfirst
,xin_Voc_msg
,xin_Voc_multi
,xin_Voc_mvs
,xin_Voc_n
,xin_Voc_na
,xin_Voc_nag
,xin_Voc_name
,xin_Voc_names
,xin_Voc_nan
,xin_Voc_native
,xin_Voc_nativeaddr
,xin_Voc_natlang
,xin_Voc_nc
,xin_Voc_ncp
,xin_Voc_nct
,xin_Voc_nd
,xin_Voc_nest
,xin_Voc_new
,xin_Voc_ngn
,xin_Voc_ngr
,xin_Voc_ngs
,xin_Voc_nign
,xin_Voc_nimp
,xin_Voc_nimpl
,xin_Voc_ninc
,xin_Voc_nint
,xin_Voc_nis
,xin_Voc_nm
,xin_Voc_nmd
,xin_Voc_nmi
,xin_Voc_nnum
,xin_Voc_noadata
,xin_Voc_noaggregate
,xin_Voc_noanno
,xin_Voc_noattributes
,xin_Voc_noauto
,xin_Voc_noautomatic
,xin_Voc_nobj
,xin_Voc_nobuild
,xin_Voc_nocee
,xin_Voc_nocharg
,xin_Voc_nochargraphic
,xin_Voc_nocheck
,xin_Voc_nocompile
,xin_Voc_noconv
,xin_Voc_noconversion
,xin_Voc_nocount
,xin_Voc_nodebug
,xin_Voc_nodeck
,xin_Voc_nodef
,xin_Voc_nodescriptor
,xin_Voc_nodescriptors
,xin_Voc_nodirected
,xin_Voc_nodli
,xin_Voc_nodllinit
,xin_Voc_nodummydesc
,xin_Voc_noduplicate
,xin_Voc_noemulate
,xin_Voc_noesd
,xin_Voc_noevendec
,xin_Voc_noexecops
,xin_Voc_noexit
,xin_Voc_noext
,xin_Voc_noextchk
,xin_Voc_nof
,xin_Voc_nofetchable
,xin_Voc_nofixedoverflow
,xin_Voc_noflow
,xin_Voc_nofofl
,xin_Voc_nofromalien
,xin_Voc_nogonumber
,xin_Voc_nogostmt
,xin_Voc_nographic
,xin_Voc_noignore
,xin_Voc_noimplicit
,xin_Voc_noimprecise
,xin_Voc_noinclude
,xin_Voc_noinitfill
,xin_Voc_noinline
,xin_Voc_noinsource
,xin_Voc_nointerrupt
,xin_Voc_noinvalidop
,xin_Voc_noipa
,xin_Voc_nolaxasgn
,xin_Voc_nolaxconv
,xin_Voc_nolaxdcl
,xin_Voc_nolaxif
,xin_Voc_nolaxint
,xin_Voc_nolaxqual
,xin_Voc_nolibs
,xin_Voc_nolist
,xin_Voc_nolock
,xin_Voc_nolog
,xin_Voc_nomacro
,xin_Voc_nomap
,xin_Voc_nomapin
,xin_Voc_nomapout
,xin_Voc_nomargini
,xin_Voc_nomdeck
,xin_Voc_nomsg
,xin_Voc_nonasgn
,xin_Voc_nonassignable
,xin_Voc_nonconn
,xin_Voc_nonconnected
,xin_Voc_none
,xin_Voc_nonest
,xin_Voc_nonlocal
,xin_Voc_nonnative
,xin_Voc_nonnativeaddr
,xin_Voc_nonrecursive
,xin_Voc_nonumber
,xin_Voc_nonvar
,xin_Voc_nonvarying
,xin_Voc_noobject
,xin_Voc_nooffset
,xin_Voc_noofl
,xin_Voc_nooptimize
,xin_Voc_nooptions
,xin_Voc_nooverflow
,xin_Voc_nop
,xin_Voc_nopp
,xin_Voc_nopptrace
,xin_Voc_noprobe
,xin_Voc_noproceed
,xin_Voc_noprofile
,xin_Voc_nopt
,xin_Voc_norb
,xin_Voc_noreserve
,xin_Voc_noretcode
,xin_Voc_normal
,xin_Voc_norunops
,xin_Voc_noscheduler
,xin_Voc_nosemantic
,xin_Voc_nosequence
,xin_Voc_noshort
,xin_Voc_nosize
,xin_Voc_nosnap
,xin_Voc_nosource
,xin_Voc_nosprog
,xin_Voc_nostmt
,xin_Voc_nostorage
,xin_Voc_nostrg
,xin_Voc_nostringrange
,xin_Voc_nostringsize
,xin_Voc_nostrz
,xin_Voc_nosubrg
,xin_Voc_nosubscriptrang
,xin_Voc_nosym
,xin_Voc_nosyntax
,xin_Voc_not
,xin_Voc_noterminal
,xin_Voc_notest
,xin_Voc_notiled
,xin_Voc_notrace
,xin_Voc_noufl
,xin_Voc_nounderflow
,xin_Voc_nowcode
,xin_Voc_nowrite
,xin_Voc_noxref
,xin_Voc_nozdiv
,xin_Voc_nozerodivide
,xin_Voc_npro
,xin_Voc_ns
,xin_Voc_nsem
,xin_Voc_nseq
,xin_Voc_nstg
,xin_Voc_nsyn
,xin_Voc_nterm
,xin_Voc_null370
,xin_Voc_nullsys
,xin_Voc_num
,xin_Voc_number
,xin_Voc_nx
,xin_Voc_obj
,xin_Voc_object
,xin_Voc_of
,xin_Voc_offset
,xin_Voc_ofl
,xin_Voc_on
,xin_Voc_onproc
,xin_Voc_op
,xin_Voc_open
,xin_Voc_opt
,xin_Voc_optimize
,xin_Voc_optional
,xin_Voc_options
,xin_Voc_optlink
,xin_Voc_or
,xin_Voc_order
,xin_Voc_ordinal
,xin_Voc_organization
,xin_Voc_os
,xin_Voc_os2
,xin_Voc_other
,xin_Voc_otherwise
,xin_Voc_out
,xin_Voc_output
,xin_Voc_overflow
,xin_Voc_overrides
,xin_Voc_owns
,xin_Voc_p
,xin_Voc_package
,xin_Voc_page
,xin_Voc_pagesize
,xin_Voc_parameter
,xin_Voc_parents
,xin_Voc_parm
,xin_Voc_pascal
,xin_Voc_pascal16
,xin_Voc_password
,xin_Voc_path
,xin_Voc_pending
,xin_Voc_pentium
,xin_Voc_pic
,xin_Voc_picture
,xin_Voc_plitdli
,xin_Voc_plitest
,xin_Voc_pointer
,xin_Voc_pos
,xin_Voc_position
,xin_Voc_pp
,xin_Voc_pptrace
,xin_Voc_prec
,xin_Voc_precision
,xin_Voc_prefix
,xin_Voc_preproc
,xin_Voc_preview
,xin_Voc_print
,xin_Voc_priority
,xin_Voc_private
,xin_Voc_pro
,xin_Voc_probe
,xin_Voc_proc
,xin_Voc_procedure
,xin_Voc_proceed
,xin_Voc_process
,xin_Voc_profile
,xin_Voc_protected
,xin_Voc_ptr
,xin_Voc_public
,xin_Voc_put
,xin_Voc_r
,xin_Voc_range
,xin_Voc_read
,xin_Voc_real
,xin_Voc_record
,xin_Voc_recsize
,xin_Voc_recursive
,xin_Voc_red
,xin_Voc_reducible
,xin_Voc_reentrant
,xin_Voc_refer
,xin_Voc_refine
,xin_Voc_regional
,xin_Voc_relative
,xin_Voc_release
,xin_Voc_renames
,xin_Voc_reorder
,xin_Voc_repeat
,xin_Voc_reply
,xin_Voc_reread
,xin_Voc_reserve
,xin_Voc_reserved
,xin_Voc_reserves
,xin_Voc_resignal
,xin_Voc_retcode
,xin_Voc_return
,xin_Voc_returns
,xin_Voc_reuse
,xin_Voc_revert
,xin_Voc_rewrite
,xin_Voc_rules
,xin_Voc_runops
,xin_Voc_s
,xin_Voc_s386
,xin_Voc_s486
,xin_Voc_saa
,xin_Voc_saa2
,xin_Voc_saa3
,xin_Voc_scalarvarying
,xin_Voc_scheduler
,xin_Voc_segmented
,xin_Voc_select
,xin_Voc_sem
,xin_Voc_semantic
,xin_Voc_seq
,xin_Voc_seql
,xin_Voc_sequence
,xin_Voc_sequential
,xin_Voc_set
,xin_Voc_short
,xin_Voc_signal
,xin_Voc_signed
,xin_Voc_single
,xin_Voc_sis
,xin_Voc_size
,xin_Voc_sizefrom
,xin_Voc_sizeto
,xin_Voc_skip
,xin_Voc_smessage
,xin_Voc_smsg
,xin_Voc_snap
,xin_Voc_source
,xin_Voc_spill
,xin_Voc_sprog
,xin_Voc_sql
,xin_Voc_static
,xin_Voc_stdcall
,xin_Voc_stg
,xin_Voc_stmt
,xin_Voc_stop
,xin_Voc_storage
,xin_Voc_stream
,xin_Voc_strg
,xin_Voc_string
,xin_Voc_stringrange
,xin_Voc_stringsize
,xin_Voc_struct
,xin_Voc_structure
,xin_Voc_strz
,xin_Voc_subrg
,xin_Voc_subscriptrange
,xin_Voc_suspend
,xin_Voc_sym
,xin_Voc_syn
,xin_Voc_syntax
,xin_Voc_sysin
,xin_Voc_sysparm
,xin_Voc_sysprint
,xin_Voc_system
,xin_Voc_sz
,xin_Voc_task
,xin_Voc_term
,xin_Voc_terminal
,xin_Voc_test
,xin_Voc_then
,xin_Voc_thread
,xin_Voc_tiled
,xin_Voc_time
,xin_Voc_title
,xin_Voc_to
,xin_Voc_total
,xin_Voc_tp
,xin_Voc_trace
,xin_Voc_transient
,xin_Voc_transmit
,xin_Voc_trkofl
,xin_Voc_tso
,xin_Voc_tstack
,xin_Voc_type
,xin_Voc_u
,xin_Voc_uen
,xin_Voc_ufl
,xin_Voc_unal
,xin_Voc_unaligned
,xin_Voc_unbuf
,xin_Voc_unbuffered
,xin_Voc_undefinedfile
,xin_Voc_underflow
,xin_Voc_undf
,xin_Voc_union
,xin_Voc_unlimited
,xin_Voc_unlock
,xin_Voc_unroll
,xin_Voc_unsigned
,xin_Voc_until
,xin_Voc_update
,xin_Voc_upperinc
,xin_Voc_upthru
,xin_Voc_v
,xin_Voc_v1
,xin_Voc_v2
,xin_Voc_value
,xin_Voc_valuelist
,xin_Voc_valuerange
,xin_Voc_var
,xin_Voc_variable
,xin_Voc_varying
,xin_Voc_varyingz
,xin_Voc_varz
,xin_Voc_vb
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in
other countries. Consult your local IBM representative for information on the
products and services currently available in your area. Any reference to an IBM
product, program, or service is not intended to state or imply that only that IBM
product, program, or service may be used. Any functionally equivalent product,
program, or service that does not infringe any IBM intellectual property right may
be used instead. However, it is the users responsibility to evaluate and verify the
operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter
described in this document. The furnishing of this document does not give you
any license to these patents. You can send license inquiries, in writing, to:
IBM Corporation
J74/G4
555 Bailey Avenue
San Jose, CA 95141-1099
U.S.A.
For license inquiries regarding double-byte (DBCS) information, contact the IBM
Intellectual Property Department in your country or send inquiries, in writing, to:
Intellectual Property Licensing
Legal and Intellectual Property Law
IBM Japan, Ltd.
3-2-12, Roppongi, Minato-ku, Tokyo 106-8711
,xin_Voc_vbs
,xin_Voc_virtual
,xin_Voc_vs
,xin_Voc_vsam
,xin_Voc_w
,xin_Voc_wait
,xin_Voc_wcode
,xin_Voc_when
,xin_Voc_while
,xin_Voc_windows
,xin_Voc_winproc
,xin_Voc_wkeep
,xin_Voc_write
,xin_Voc_x
,xin_Voc_xchar
,xin_Voc_xinfo
,xin_Voc_xoptions
,xin_Voc_xref
,xin_Voc_zdiv
,xin_Voc_zerodivide
) unsigned prec(16);
The following paragraph does not apply to the United Kingdom or any other
country where such provisions are inconsistent with local law:
INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES
THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND,
EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED
TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Some states do not allow disclaimer of express or implied warranties in certain
transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors.
Changes are periodically made to the information herein; these changes will be
incorporated in new editions of the publication. IBM may make improvements
and/or changes in the product(s) and/or the program(s) described in this
publication at any time without notice.
Any references in this publication to non-IBM Web sites are provided for
convenience only and do not in any manner serve as an endorsement of those Web
sites. The materials at those Web sites are not part of the materials for this IBM
product and use of those Web sites is at your own risk.
Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of
International Business Machines Corp., registered in many jurisdictions worldwide.
Other product and service names might be trademarks of IBM or other companies.
A current list of IBM trademarks is available on the Web at Copyright and
trademark information at www.ibm.com/legal/copytrade.shtml.
Intel is a registered trademark of Intel Corporation in the United States and other
countries.
Java and all Java-based trademarks and logos are trademarks or registered
trademarks of Oracle and/or its affiliates.
Microsoft, Windows, and Windows NT are trademarks of Microsoft Corporation in
the United States and other countries.
Pentium is a registered trademark of Intel Corporation in the United States and
other countries.
Unicode is a trademark of the Unicode Consortium.
UNIX is a registered trademark of The Open Group in the United States and other
countries.
Other company, product or service names may be the trademarks or service marks
of others.
If you are viewing this information in softcopy, the photographs and color
illustrations may not appear.
Notices 597
Bibliography
PL/I publications
Programming Guide, GI11-9145
Language Reference, SC14-7285
Messages and Codes, GC14-7286
Compiler and Run-Time Migration Guide, GC14-7284
PL/I for MVS & VM
Installation and Customization under MVS, SC26-3119
Compile-Time Messages and Codes, SC26-3229
Diagnosis Guide, SC26-3149
Migration Guide, SC26-3118
Programming Guide, SC26-3113
Reference Summary, SX26-3821
PL/I for AIX
Programming Guide, SC14-7319
Messages and Codes, GC14-7321
Installation Guide, GC14-7322
Related publications
DB2 UDB for z/OS
Administration Guide, SC18-7413
Application Programming and SQL Guide, SC18-7415
Command Reference, SC18-7416
Messages, GC18-9602
Codes, GC18-9603
SQL Reference, SC18-7426
DFSORT
Application Programming Guide, SC33-4035

Installation and Customization, SC33-4034
IMS/ESA
Application Programming: Database Manager, SC26-8015

Application Programming: Database Manager Summary, SC26-8037
Application Programming: Design Guide, SC26-8016
Application Programming: Transaction Manager, SC26-8017
Application Programming: Transaction Manager Summary, SC26-8038
Application Programming: EXEC DL/I Commands for CICS and IMS
, SC26-8018
Application Programming: EXEC DL/I Commands for CICS and IMS Summary,
SC26-8036
TXSeries for Multiplatforms
Encina Administration Guide Volume 2: Server Administration, SC09-4474
Encina SFS Programming Guide, SC09-4483
See also the Information Center: publib.boulder.ibm.com/infocenter/txformp/
v7r1/index.jsp
z/Architecture
Principles of Operation, SA22-7832
z/OS Language Environment
Concepts Guide, SA22-7567
Debugging Guide, GA22-7560
Run-Time Messages, SA22-7566
Customization, SA22-7564
Programming Guide, SA22-7561
Programming Reference, SA22-7562
Run-Time Application Migration Guide, GA22-7565
Writing Interlanguage Communication Applications, SA22-7563
z/OS MVS
JCL Reference, SA22-7597
JCL User's Guide, SA22-7598
System Commands, SA22-7627
z/OS TSO/E
Command Reference, SA22-7782
User's Guide, SA22-7794
z/OS UNIX System Services
z/OS UNIX System Services Command Reference, SA22-7802
z/OS UNIX System Services Programming: Assembler Callable Services Reference,
SA22-7803
z/OS UNIX System Services User's Guide, SA22-7801
Unicode
and character representation

z/OS Support for Unicode: Using Conversion Services, SC33-7050
Glossary
This glossary defines terms for all platforms and
releases of PL/I. It might contain terms that this
manual does not use. If you do not find the terms
for which you are looking, see the index in this
manual or IBM Dictionary of Computing,
SC20-1699.
A
access. To reference or retrieve data.
action specification. In an ON statement, the ON-unit
or the single keyword SYSTEM, either of which
specifies the action to be taken whenever the
appropriate condition is raised.
activate (a block). To initiate the execution of a block.
A procedure block is activated when it is invoked. A
begin-block is activated when it is encountered in the
normal flow of control, including a branch. A package
cannot be activated.
activate (a preprocessor variable or preprocessor entry
point). To make a macro facility identifier eligible for
replacement in subsequent source code. The
%ACTIVATE statement activates preprocessor variables
or preprocessor entry points.
active. The state of a block after activation and before
termination. The state in which a preprocessor variable
or preprocessor entry name is said to be when its value
can replace the corresponding identifier in source
program text. The state in which an event variable is
said to be during the time it is associated with an
asynchronous operation. The state in which a task
variable is said to be when its associated task is
attached. The state in which a task is said to be before
it has been terminated.
actual origin (AO). The location of the first item in the
array or structure.
additive attribute. A file description attribute for
which there are no defaults, and which, if required,
must be stated explicitly or implied by another
explicitly stated attribute. Contrast with alternative
attribute.
adjustable extent. The bound (of an array), the length
(of a string), or the size (of an area) that might be
different for different generations of the associated
variable. Adjustable extents are specified as expressions
or asterisks (or by REFER options for based variables),
which are evaluated separately for each generation.
They cannot be used for static variables.
aggregate. See data aggregate.
aggregate expression. An array, structure, or union
expression.
aggregate type. For any item of data, the specification
whether it is structure, union, or array.
allocated variable. A variable with which main
storage is associated and not freed.
allocation. The reservation of main storage for a
variable. A generation of an allocated variable. The
association of a PL/I file with a system data set, device,
or file.
alignment. The storing of data items in relation to
certain machine-dependent boundaries (for example, a
fullword or halfword boundary).
alphabetic character. Any of the characters A through
Z of the English alphabet and the alphabetic extenders
#, $, and @ (which can have a different graphic
representation in different countries).
alphameric character. An alphabetic character or a
digit.
alternative attribute. A file description attribute that is
chosen from a group of attributes. If none is specified, a
default is assumed. Contrast with additive attribute.
ambiguous reference. A reference that is not
sufficiently qualified to identify one and only one name
known at the point of reference.
area. A portion of storage within which based
variables can be allocated.
argument. An expression in an argument list as part
of an invocation of a subroutine or function.
argument list. A parenthesized list of zero or more
arguments, separated by commas, following an entry
name constant, an entry name variable, a generic name,
or a built-in function name. The list becomes the
parameter list of the entry point.
arithmetic comparison. A comparison of numeric
values. See also bit comparison, character comparison.
arithmetic constant. A fixed-point constant or a
floating-point constant. Although most arithmetic
constants can be signed, the sign is not part of the
constant.
arithmetic conversion. The transformation of a value
from one arithmetic representation to another.
arithmetic data. Data that has the characteristics of
base, scale, mode, and precision. Coded arithmetic data
and pictured numeric character data are included.
arithmetic operators. Either of the prefix operators +
and , or any of the following infix operators: + * /
**
array. A named, ordered collection of one or more
data elements with identical attributes, grouped into
one or more dimensions.
array expression. An expression whose evaluation
yields an array of values.
array of structures. An ordered collection of identical
structures specified by giving the dimension attribute
to a structure name.
array variable. A variable that represents an aggregate
of data items that must have identical attributes.
Contrast with structure variable.
ASCII. American National Standard Code for
Information Interchange.
assignment. The process of giving a value to a
variable.
asynchronous operation. The overlap of an
input/output operation with the execution of
statements. The concurrent execution of procedures
using multiple flows of control for different tasks.
attachment of a task. The invocation of a procedure
and the establishment of a separate flow of control to
execute the invoked procedure (and procedures it
invokes) asynchronously, with execution of the
invoking procedure.
attention. An occurrence, external to a task, that could
cause a task to be interrupted.
attribute. A descriptive property associated with a
name to describe a characteristic represented. A
descriptive property used to describe a characteristic of
the result of evaluation of an expression.
automatic storage allocation. The allocation of storage
for automatic variables.
automatic variable. A variable whose storage is
allocated automatically at the activation of a block and
released automatically at the termination of that block.
B
base. The number system in which an arithmetic
value is represented.
base element. A member of a structure or a union that
is itself not another structure or union.
base item. The automatic, controlled, or static variable
or the parameter upon which a defined variable is
defined.
based reference. A reference that has the based
storage class.
based storage allocation. The allocation of storage for
based variables.
based variable. A variable whose storage address is
provided by a locator. Multiple generations of the same
variable are accessible. It does not identify a fixed
location in storage.
begin-block. A collection of statements delimited by
BEGIN and END statements, forming a name scope. A
begin-block is activated either by the raising of a
condition (if the begin-block is the action specification
for an ON-unit) or through the normal flow of control,
including any branch resulting from a GOTO
statement.
binary. A number system whose only numerals are 0
and 1.
binary digit. See bit.
binary fixed-point value. An integer consisting of
binary digits and having an optional binary point and
optional sign. Contrast with decimal fixed-point value.
binary floating-point value. An approximation of a
real number in the form of a significand, which can be
considered as a binary fraction, and an exponent,
which can be considered as an integer exponent to the
base of 2. Contrast with decimal floating-point value.
bit. A 0 or a 1. The smallest amount of space of
computer storage.
bit comparison. A left-to-right, bit-by-bit comparison
of binary digits. See also arithmetic comparison, character
comparison.
bit string constant. A series of binary digits enclosed
in and followed immediately by the suffix B. Contrast
with character constant. A series of hexadecimal digits
enclosed in single quotes and followed by the suffix B4.
bit string. A string composed of zero or more bits.
bit string operators. The logical operators not and
exclusive-or (), and (&), and or (|).
bit value. A value that represents a bit type.
block. A sequence of statements, processed as a unit,
that specifies the scope of names and the allocation of
storage for names declared within it. A block can be a
package, procedure, or a begin-block.
bounds. The upper and lower limits of an array
dimension.
break character. The underscore symbol ( _ ). It can be
used to improve the readability of identifiers. For
instance, a variable could be called
OLD_INVENTORY_TOTAL instead of
OLDINVENTORYTOTAL.
built-in function. A predefined function supplied by
the language, such as SQRT (square root).
built-in function reference. A built-in function name,
which has an optional argument list.
built-in name. The entry name of a built-in
subroutine.
built-in subroutine. Subroutine that has an entry
name that is defined at compile-time and is invoked by
a CALL statement.
buffer. Intermediate storage, used in input/output
operations, into which a record is read during input
and from which a record is written during output.
C
call. To invoke a subroutine by using the CALL
statement or CALL option.
character comparison. A left-to-right,
character-by-character comparison according to the
collating sequence. See also arithmetic comparison, bit
comparison.
character string constant. A sequence of characters
enclosed in single quotes; for example, 'Shakespeare''s
'Hamlet:''.
character set. A defined collection of characters. See
language character set and data character set. See also
ASCII and EBCDIC.
character string picture data. Picture data that has
only a character value. This type of picture data must
have at least one A or X picture specification character.
Contrast with numeric picture data.
closing (of a file). The dissociation of a file from a
data set or device.
coded arithmetic data. Data items that represent
numeric values and are characterized by their base
(decimal or binary), scale (fixed-point or floating-point),
and precision (the number of digits each can have).
This data is stored in a form that is acceptable, without
conversion, for arithmetic calculations.
combined nesting depth. The deepest level of nesting,
determined by counting the levels of
PROCEDURE/BEGIN/ON, DO, SELECT, and
IF...THEN...ELSE nestings in the program.
comment. A string of zero or more characters used for
documentation that are delimited by /* and */.
commercial character.
v CR (credit) picture specification character
v DB (debit) picture specification character
comparison operator. An operator that can be used in
an arithmetic, string locator, or logical relation to
indicate the comparison to be done between the terms
in the relation. The comparison operators are:
= (equal to)
> (greater than)
< (less than)
>= (greater than or equal to)
<= (less than or equal to)
= (not equal to)
> (not greater than)
< (not less than)
compile time. In general, the time during which a
source program is translated into an object module. In
PL/I, it is the time during which a source program can
be altered, if desired, and then translated into an object
program.
compiler options. Keywords that are specified to
control certain aspects of a compilation, such as: the
nature of the object module generated, the types of
printed output produced, and so forth.
complex data. Arithmetic data, each item of which
consists of a real part and an imaginary part.
composite operator. An operator that consists of more
than one special character, such as <=, **, and /*.
compound statement. A statement that contains other
statements. In PL/I, IF, ON, OTHERWISE, and WHEN
are the only compound statements. See statement body.
concatenation. The operation that joins two strings in
the order specified, forming one string whose length is
equal to the sum of the lengths of the two original
strings. It is specified by the operator ||.
condition. An exceptional situation, either an error
(such as an overflow), or an expected situation (such as
the end of an input file). When a condition is raised
(detected), the action established for it is processed. See
also established action and implicit action.
condition name. Name of a PL/I-defined or
programmer-defined condition.
Glossary 603
condition prefix. A parenthesized list of one or more
condition names prefixed to a statement. It specifies
whether the named conditions are to be enabled or
disabled.
connected aggregate. An array or structure whose
elements occupy contiguous storage without any
intervening data items. Contrast with nonconnected
aggregate.
connected reference. A reference to connected storage.
It must be apparent, prior to execution of the program,
that the storage is connected.
connected storage. Main storage of an uninterrupted
linear sequence of items that can be referred to by a
single name.
constant. An arithmetic or string data item that does
not have a name and whose value cannot change. An
identifier declared with the VALUE attribute. An
identifier declared with the FILE or the ENTRY
attribute but without the VARIABLE attribute.
constant reference. A value reference which has a
constant as its object
contained block, declaration, or source text. All
blocks, procedures, statements, declarations, or source
text inside a begin, procedure, or a package block. The
entire package, procedure, and the BEGIN statement
and its corresponding END statements are not
contained in the block.
containing block. The package, procedure, or
begin-block that contains the declaration, statement,
procedure, or other source text in question.
contextual declaration. The appearance of an
identifier that has not been explicitly declared in a
DECLARE statement, but whose context of use allows
the association of specific attributes with the identifier.
control character. A character in a character set whose
occurrence in a particular context specifies a control
function. One example is the end-of-file (EOF) marker.
control format item. A specification used in
edit-directed transmission to specify positioning of a
data item within the stream or printed page.
control variable. A variable that is used to control the
iterative execution of a DO statement.
controlled parameter. A parameter for which the
CONTROLLED attribute is specified in a DECLARE
statement. It can be associated only with arguments
that have the CONTROLLED attribute.
controlled storage allocation. The allocation of storage
for controlled variables.
controlled variable. A variable whose allocation and
release are controlled by the ALLOCATE and FREE
statements, with access to the current generation only.
control sections. Grouped machine instructions in an
object module.
conversion. The transformation of a value from one
representation to another to conform to a given set of
attributes. For example, converting a character string to
an arithmetic value such as FIXED BINARY (15,0).
cross section of an array. The elements represented by
the extent of at least one dimension of an array. An
asterisk in the place of a subscript in an array reference
indicates the entire extent of that dimension.
current generation. The generation of an automatic or
controlled variable that is currently available by
referring to the name of the variable.
D
data. Representation of information or of value in a
form suitable for processing.
data aggregate. A data item that is a collection of
other data items.
data attribute. A keyword that specifies the type of
data that the data item represents, such as FIXED
BINARY.
data-directed transmission. The type of
stream-oriented transmission in which data is
transmitted. It resembles an assignment statement and
is of the form name = constant.
data item. A single named unit of data.
data list. In stream-oriented transmission, a
parenthesized list of the data items used in GET and
PUT statements. Contrast with format list.
data set. A collection of data external to the program
that can be accessed by reference to a single file name.
A device that can be referenced.
data specification. The portion of a stream-oriented
transmission statement that specifies the mode of
transmission (DATA, LIST, or EDIT) and includes the
data list(s) and, for edit-directed mode, the format
list(s).
data stream. Data being transferred from or to a data
set by stream-oriented transmission, as a continuous
stream of data elements in character form.
data transmission. The transfer of data from a data set
to the program or vice versa.
data type. A set of data attributes.
DBCS. In the character set, each character is
represented by two consecutive bytes.
deactivated. The state in which an identifier is said to
be when its value cannot replace a preprocessor
identifier in source program text. Contrast with active.
debugging. Process of removing bugs from a
program.
decimal. The number system whose numerals are 0
through 9.
decimal digit picture character. The picture
specification character 9.
decimal fixed-point constant. A constant consisting of
one or more decimal digits with an optional decimal
point.
decimal fixed-point value. A rational number
consisting of a sequence of decimal digits with an
assumed position of the decimal point. Contrast with
binary fixed-point value.
decimal floating-point constant. A value made up of
a significand that consists of a decimal fixed-point
constant, and an exponent that consists of the letter E
followed by an optionally signed integer constant not
exceeding three digits.
decimal floating-point value. An approximation of a
real number, in the form of a significand, which can be
considered as a decimal fraction, and an exponent,
which can be considered as an integer exponent to the
base 10. Contrast with binary floating-point value.
decimal picture data. See numeric picture data.
declaration. The establishment of an identifier as a
name and the specification of a set of attributes (partial
or complete) for it. A source of attributes of a particular
name.
default. Describes a value, attribute, or option that is
assumed when none has been specified.
defined variable. A variable that is associated with
some or all of the storage of the designated base
variable.
delimit. To enclose one or more items or statements
with preceding and following characters or keywords.
delimiter. All comments and the following characters:
percent, parentheses, comma, period, semicolon, colon,
assignment symbol, blank, pointer, asterisk, and single
quote. They define the limits of identifiers, constants,
picture specifications, iSUBs, and keywords.
descriptor. A control block that holds information
about a variable, such as area size, array bounds, or
string length.
digit. One of the characters 0 through 9.
dimension attribute. An attribute that specifies the
number of dimensions of an array and indicates the
bounds of each dimension.
disabled. The state of a condition in which no
interrupt occurs and no established action will take
place.
do-group. A sequence of statements delimited by a
DO statement and ended by its corresponding END
statement, used for control purposes. Contrast with
block.
do-loop. See iterative do-group.
dummy argument. Temporary storage that is created
automatically to hold the value of an argument that
cannot be passed by reference.
dump. Printout of all or part of the storage used by a
program as well as other program information, such as
a trace of an error's origin.
E
EBCDIC. (Extended Binary-Coded Decimal
Interchange Code). A coded character set consisting of
8-bit coded characters.
edit-directed transmission. The type of
stream-oriented transmission in which data appears as
a continuous stream of characters and for which a
format list is required to specify the editing desired for
the associated data list.
element. A single item of data as opposed to a
collection of data items such as an array; a scalar item.
element expression. An expression whose evaluation
yields an element value.
element variable. A variable that represents an
element; a scalar variable.
elementary name. See base element.
enabled. The state of a condition in which the
condition can cause an interrupt and then invocation of
the appropriate established ON-unit.
end-of-step message. message that follows the listng
of the job control statements and job scheduler
messages and contains return code indicating success
or failure for each step.
entry constant. The label prefix of a PROCEDURE
statement (an entry name). The declaration of a name
with the ENTRY attribute but without the VARIABLE
attribute.
Glossary 605
entry data. A data item that represents an entry point
to a procedure.
entry expression. An expression whose evaluation
yields an entry name.
entry name. An identifier that is explicitly or
contextually declared to have the ENTRY attribute
(unless the VARIABLE attribute is given) or An
identifier that has the value of an entry variable with
the ENTRY attribute implied.
entry point. A point in a procedure at which it can be
invoked. primary entry point and secondary entry point.
entry reference. An entry constant, an entry variable
reference, or a function reference that returns an entry
value.
entry variable. A variable to which an entry value can
be assigned. It must have both the ENTRY and
VARIABLE attributes.
entry value. The entry point represented by an entry
constant or variable; the value includes the
environment of the activation that is associated with
the entry constant.
environment (of an activation). Information
associated with and used in the invoked block
regarding data declared in containing blocks.
environment (of a label constant). Identity of the
particular activation of a block to which a reference to a
statement-label constant applies. This information is
determined at the time a statement-label constant is
passed as an argument or is assigned to a
statement-label variable, and it is passed or assigned
along with the constant.
established action. The action taken when a condition
is raised. See also implicit action and ON-statement
action.
epilogue. Those processes that occur automatically at
the termination of a block or task.
evaluation. The reduction of an expression to a single
value, an array of values, or a structured set of values.
event. An activity in a program whose status and
completion can be determined from an associated event
variable.
event variable. A variable with the EVENT attribute
that can be associated with an event. Its value indicates
whether the action has been completed and the status
of the completion.
explicit declaration. The appearance of an identifier (a
name) in a DECLARE statement, as a label prefix, or in
a parameter list. Contrast with implicit declaration.
exponent characters. The following picture
specification characters:
1. K and E, which are used in floating-point picture
specifications to indicate the beginning of the
exponent field.
2. F, the scaling factor character, specified with an
integer constant that indicates the number of
decimal positions the decimal point is to be moved
from its assumed position to the right (if the
constant is positive) or to the left (if the constant is
negative).
expression. A notation, within a program, that
represents a value, an array of values, or a structured
set of values. A constant or a reference appearing alone,
or a combination of constants and/or references with
operators.
extended alphabet. The uppercase and lowercase
alphabetic characters A through Z, $, @ and #, or those
specified in the NAMES compiler option.
extent. The range indicated by the bounds of an array
dimension, by the length of a string, or by the size of
an area. The size of the target area if this area were to
be assigned to a target area.
external name. A name (with the EXTERNAL
attribute) whose scope is not necessarily confined only
to one block and its contained blocks.
external procedure. A procedure that is not contained
in any other procedure. A level-2 procedure contained
in a package that is also exported.
external symbol. Name that can be referred to in a
control section other than the one in which it is
defined.
External Symbol Dictionary (ESD). Table containing
all the external symbols that appear in the object
module.
extralingual character. Characters (such as $, @, and #)
that are not classified as alphanumeric or special. This
group includes characters that are determined with the
NAMES compiler option.
F
factoring. The application of one or more attributes to
a parenthesized list of names in a DECLARE statement,
eliminating the repetition of identical attributes for
multiple names.
field (in the data stream). That portion of the data
stream whose width, in number of characters, is
defined by a single data or spacing format item.
field (of a picture specification). Any character-string
picture specification or that portion (or all) of a
numeric character picture specification that describes a
fixed-point number.
file. A named representation, within a program, of a
data set or data sets. A file is associated with the data
set(s) for each opening.
file constant. A name declared with the FILE attribute
but not the VARIABLE attribute.
file description attributes. Keywords that describe the
individual characteristics of each file constant. See also
alternative attribute and additive attribute.
file expression. An expression whose evaluation
yields a value of the type file.
file name. A name declared for a file.
file variable. A variable to which file constants can be
assigned. It has the attributes FILE and VARIABLE and
cannot have any of the file description attributes.
fixed-point constant. See arithmetic constant.
fix-up. A solution, performed by the compiler after
detecting an error during compilation, that allows the
compiled program to run.
floating-point constant. See arithmetic constant.
flow of control. Sequence of execution.
format. A specification used in edit-directed data
transmission to describe the representation of a data
item in the stream (data format item) or the specific
positioning of a data item within the stream (control
format item).
format constant. The label prefix on a FORMAT
statement.
format data. A variable with the FORMAT attribute.
format label. The label prefix on a FORMAT
statement.
format list. In stream-oriented transmission, a list
specifying the format of the data item on the external
medium. Contrast with data list.
fully qualified name. A name that includes all the
names in the hierarchical sequence above the member
to which the name refers, as well as the name of the
member itself.
function (procedure). A procedure that has a
RETURNS option in the PROCEDURE statement. A
name declared with the RETURNS attribute. It is
invoked by the appearance of one of its entry names in
a function reference and it returns a scalar value to the
point of reference. Contrast with subroutine.
function reference. An entry constant or an entry
variable, either of which must represent a function,
followed by a possibly empty argument list. Contrast
with subroutine call.
G
generation (of a variable). The allocation of a static
variable, a particular allocation of a controlled or
automatic variable, or the storage indicated by a
particular locator qualification of a based variable or by
a defined variable or parameter.
generic descriptor. A descriptor used in a GENERIC
attribute.
generic key. A character string that identifies a class of
keys. All keys that begin with the string are members
of that class. For example, the recorded keys 'ABCD',
'ABCE', and 'ABDF', are all members of the classes
identified by the generic keys 'A' and 'AB', and the first
two are also members of the class 'ABC'; and the three
recorded keys can be considered to be unique members
of the classes 'ABCD', 'ABCE', 'ABDF', respectively.
generic name. The name of a family of entry names. A
reference to the generic name is replaced by the entry
name whose parameter descriptors match the attributes
of the arguments in the argument list at the point of
invocation.
group. A collection of statements contained within
larger program units. A group is either a do-group or a
select-group and it can be used wherever a single
statement can appear, except as an on-unit.
H
hex. See hexadecimal digit.
hexadecimal. Pertaining to a numbering system with a
base of sixteen; valid numbers use the digits 0 through
9 and the characters A through F, where A represents
10 and F represents 15.
hexadecimal digit. One of the digits 0 through 9 and
A through F. A through F represent the decimal values
10 through 15, respectively.
I
identifier. A string of characters, not contained in a
comment or constant, and preceded and followed by a
delimiter. The first character of the identifier must be
one of the 26 alphabetic characters and extralingual
characters, if any. The other characters, if any, can
additionally include extended alphabetic, digit, or the
break character.
IEEE. Institute of Electrical and Electronics Engineers.
Glossary 607
implicit. The action taken in the absence of an explicit
specification.
implicit action. The action taken when an enabled
condition is raised and no ON-unit is currently
established for the condition. Contrast with
ON-statement action.
implicit declaration. A name not explicitly declared in
a DECLARE statement or contextually declared.
implicit opening. The opening of a file as the result of
an input or output statement other than the OPEN
statement.
infix operator. An operator that appears between two
operands.
inherited dimensions. For a structure, union, or
element, those dimensions that are derived from the
containing structures. If the name is an element that is
not an array, the dimensions consist entirely of its
inherited dimensions. If the name is an element that is
an array, its dimensions consist of its inherited
dimensions plus its explicitly declared dimensions. A
structure with one or more inherited dimensions is
called a nonconnected aggregate. Contrast with
connected aggregate.
input/output. The transfer of data between auxiliary
medium and main storage.
insertion point character. A picture specification
character that is, on assignment of the associated data
to a character string, inserted in the indicated position.
When used in a P-format item for input, the insertion
character is used for checking purposes.
integer. An optionally signed sequence of digits or a
sequence of bits without a decimal or binary point. An
optionally signed whole number, commonly described
as FIXED BINARY (p,0) or FIXED DECIMAL (p,0).
integral boundary. A byte multiple address of any
8-bit unit on which data can be aligned. It usually is a
halfword, fullword, or doubleword (2-, 4-, or 8-byte
multiple respectively) boundary.
interleaved array. An array that refers to
nonconnected storage.
interleaved subscripts. Subscripts that exist in levels
other than the lowest level of a subscripted qualified
reference.
internal block. A block that is contained in another
block.
internal name. A name that is known only within the
block in which it is declared, and possibly within any
contained blocks.
internal procedure. A procedure that is contained in
another block. Contrast with external procedure.
interrupt. The redirection of the program's flow of
control as the result of raising a condition or attention.
invocation. The activation of a procedure.
invoke. To activate a procedure.
invoked procedure. A procedure that has been
activated.
invoking block. A block that activates a procedure.
iteration factor. In an INITIAL attribute specification,
an expression that specifies the number of consecutive
elements of an array that are to be initialized with the
given value. In a format list, an expression that
specifies the number of times a given format item or
list of format items is to be used in succession.
iterative do-group. A do-group whose DO statement
specifies a control variable and/or a WHILE or UNTIL
option.
K
key. Data that identifies a record within a direct access
data set. See source key and recorded key.
keyword. An identifier that has a specific meaning in
PL/I when used in a defined context.
keyword statement. A simple statement that begins
with a keyword, indicating the function of the
statement.
known (applied to a name). Recognized with its
declared meaning. A name is known throughout its
scope.
L
label. A name prefixed to a statement. A name on a
PROCEDURE statement is called an entry constant; a
name on a FORMAT statement is called a format
constant; a name on other kinds of statements is called
a label constant. A data item that has the LABEL
attribute.
label constant. A name written as the label prefix of a
statement (other than PROCEDURE, ENTRY, FORMAT,
or PACKAGE) so that, during execution, program
control can be transferred to that statement through a
reference to its label prefix.
label data. A label constant or the value of a label
variable.
label prefix. A label prefixed to a statement.
label variable. A variable declared with the LABEL
attribute. Its value is a label constant in the program.
leading zeroes. Zeros that have no significance in an
arithmetic value. All zeros to the left of the first
nonzero in a number.
level number. A number that precedes a name in a
DECLARE statement and specifies its relative position
in the hierarchy of structure names.
level-one variable. A major structure or union name.
Any unsubscripted variable not contained within a
structure or union.
lexically. Relating to the left-to-right order of units.
library. An MVS partitioned data set or a CMS
MACLIB that can be used to store other data sets called
members.
list-directed. The type of stream-oriented transmission
in which data in the stream appears as constants
separated by blanks or commas and for which
formatting is provided automatically.
locator. A control block that holds the address of a
variable or its descriptor.
locator/descriptor. A locator followed by a descriptor.
The locator holds the address of the variable, not the
address of the descriptor.
locator qualification. In a reference to a based
variable, either a locator variable or function reference
connected by an arrow to the left of a based variable to
specify the generation of the based variable to which
the reference refers. It might be an implicit reference.
locator value. A value that identifies or can be used to
identify the storage address.
locator variable. A variable whose value identifies the
location in main storage of a variable or a buffer. It has
the POINTER or OFFSET attribute.
locked record. A record in an EXCLUSIVE DIRECT
UPDATE file that has been made available to one task
only and cannot be accessed by other tasks until the
task using it relinquishes it.
logical level (of a structure or union member). The
depth indicated by a level number when all level
numbers are in direct sequence (when the increment
between successive level numbers is one).
logical operators. The bit-string operators not and
exclusive-or (), and (&), and or (|).
loop. A sequence of instructions that is executed
iteratively.
lower bound. The lower limit of an array dimension.
M
main procedure. An external procedure whose
PROCEDURE statement has the OPTIONS (MAIN)
attribute. This procedure is invoked automatically as
the first step in the execution of a program.
major structure. A structure whose name is declared
with level number 1.
member. A structure, union, or element name in a
structure or union. Data sets in a library.
minor structure. A structure that is contained within
another structure or union. The name of a minor
structure is declared with a level number greater than
one and greater than its parent structure or union.
mode (of arithmetic data). An attribute of arithmetic
data. It is either real or complex.
multiple declaration. Two or more declarations of the
same identifier internal to the same block without
different qualifications. Two or more external
declarations of the same identifier.
multiprocessing. The use of a computing system with
two or more processing units to execute two or more
programs simultaneously.
multiprogramming. The use of a computing system to
execute more than one program concurrently, using a
single processing unit.
multitasking. A facility that allows a program to
execute more than one PL/I procedure simultaneously.
N
name. Any identifier that the user gives to a variable
or to a constant. An identifier appearing in a context
where it is not a keyword. Sometimes called a
user-defined name.
nesting. The occurrence of:
v A block within another block
v A group within another group
v An IF statement in a THEN clause or in an ELSE
clause
v A function reference as an argument of a function
reference
v A remote format item in the format list of a
FORMAT statement
v A parameter descriptor list in another parameter
descriptor list
v An attribute specification within a parenthesized
name list for which one or more attributes are being
factored
Glossary 609
nonconnected storage. Storage occupied by
nonconnected data items. For example, interleaved
arrays and structures with inherited dimensions are in
nonconnected storage.
null locator value. A special locator value that cannot
identify any location in internal storage. It gives a
positive indication that a locator variable does not
currently identify any generation of data.
null statement. A statement that contains only the
semicolon symbol (;). It indicates that no action is to be
taken.
null string. A character, graphic, or bit string with a
length of zero.
numeric-character data. See decimal picture data.
numeric picture data. Picture data that has an
arithmetic value as well as a character value. This type
of picture data cannot contain the characters 'A' or 'X.'
O
object. A collection of data referred to by a single
name.
offset variable. A locator variable with the OFFSET
attribute, whose value identifies a location in storage
relative to the beginning of an area.
ON-condition. An occurrence, within a PL/I program,
that could cause a program interrupt. It can be the
detection of an unexpected error or of an occurrence
that is expected, but at an unpredictable time.
ON-statement action. The action explicitly established
for a condition that is executed when the condition is
raised. When the ON-statement is encountered in the
flow of control for the program, it executes, establishing
the action for the condition. The action executes when
the condition is raised if the ON-unit is still established
or a RESIGNAL statement reestablishes it. Contrast
with implicit action.
ON-unit. The specified action to be executed when the
appropriate condition is raised.
opening (of a file). The association of a file with a
data set.
operand. The value of an identifier, constant, or an
expression to which an operator is applied, possibly in
conjunction with another operand.
operational expression. An expression that consists of
one or more operators.
operator. A symbol specifying an operation to be
performed.
option. A specification in a statement that can be used
to influence the execution or interpretation of the
statement.
P
package constant. The label prefix on a PACKAGE
statement.
packed decimal. The internal representation of a
fixed-point decimal data item.
padding. One or more characters, graphics, or bits
concatenated to the right of a string to extend the string
to a required length. One or more bytes or bits inserted
in a structure or union so that the following element
within the structure or union is aligned on the
appropriate integral boundary.
parameter. A name in the parameter list following the
PROCEDURE statement, specifying an argument that
will be passed when the procedure is invoked.
parameter descriptor. The set of attributes specified
for a parameter in an ENTRY attribute specification.
parameter descriptor list. The list of all parameter
descriptors in an ENTRY attribute specification.
parameter list. A parenthesized list of one or more
parameters, separated by commas and following either
the keyword PROCEDURE in a procedure statement or
the keyword ENTRY in an ENTRY statement. The list
corresponds to a list of arguments passed at invocation.
partially qualified name. A qualified name that is
incomplete. It includes one or more, but not all, of the
names in the hierarchical sequence above the structure
or union member to which the name refers, as well as
the name of the member itself.
picture data. Numeric data, character data, or a mix of
both types, represented in character form.
picture specification. A data item that is described
using the picture characters in a declaration with the
PICTURE attribute or in a P-format item.
picture specification character. Any of the characters
that can be used in a picture specification.
PL/I character set. A set of characters that has been
defined to represent program elements in PL/I.
PL/I prompter. Command processor program for the
PLI command that checks the operands and allocates
the data sets required by the compiler.
point of invocation. The point in the invoking block
at which the reference to the invoked procedure
appears.
pointer. A type of variable that identifies a location in
storage.
pointer value. A value that identifies the pointer type.
pointer variable. A locator variable with the
POINTER attribute that contains a pointer value.
precision. The number of digits or bits contained in a
fixed-point data item, or the minimum number of
significant digits (excluding the exponent) maintained
for a floating-point data item.
prefix. A label or a parenthesized list of one or more
condition names included at the beginning of a
statement.
prefix operator. An operator that precedes an operand
and applies only to that operand. The prefix operators
are plus (+), minus (), and not ().
preprocessor. A program that examines the source
program before the compilation takes place.
preprocessor statement. A special statement appearing
in the source program that specifies the actions to be
performed by the preprocessor. It is executed as it is
encountered by the preprocessor.
primary entry point. The entry point identified by any
of the names in the label list of the PROCEDURE
statement.
priority. A value associated with a task, that specifies
the precedence of the task relative to other tasks.
problem data. Coded arithmetic, bit, character,
graphic, and picture data.
problem-state program. A program that operates in
the problem state of the operating system. It does not
contain input/output instructions or other privileged
instructions.
procedure. A collection of statements, delimited by
PROCEDURE and END statements. A procedure is a
program or a part of a program, delimits the scope of
names, and is activated by a reference to the procedure
or one of its entry names. See also external procedure
and internal procedure.
procedure reference. An entry constant or variable. It
can be followed by an argument list. It can appear in a
CALL statement or the CALL option, or as a function
reference.
program. A set of one or more external procedures or
packages. One of the external procedures must have
the OPTIONS(MAIN) specification in its procedure
statement.
program control data. Area, locator, label, format,
entry, and file data that is used to control the
processing of a PL/I program.
prologue. The processes that occur automatically on
block activation.
pseudovariable. Any of the built-in function names
that can be used to specify a target variable. It is
usually on the left-hand side of an assignment
statement.
Q
qualified name. A hierarchical sequence of names of
structure or union members, connected by periods,
used to identify a name within a structure. Any of the
names can be subscripted.
R
range (of a default specification). A set of identifiers
and/or parameter descriptors to which the attributes in
a DEFAULT statement apply.
record. The logical unit of transmission in a
record-oriented input or output operation. A collection
of one or more related data items. The items usually
have different data attributes and usually are described
by a structure or union declaration.
recorded key. A character string identifying a record
in a direct access data set where the character string
itself is also recorded as part of the data.
record-oriented data transmission. The transmission
of data in the form of separate records. Contrast with
stream data transmission.
recursive procedure. A procedure that can be called
from within itself or from within another active
procedure.
reentrant procedure. A procedure that can be
activated by multiple tasks, threads, or processes
simultaneously without causing any interference
between these tasks, threads, and processes.
REFER expression. The expression preceding the
keyword REFER, which is used as the bound, length,
or size when the based variable containing a REFER
option is allocated, either by an ALLOCATE or
LOCATE statement.
REFER object. The variable in a REFER option that
holds or will hold the current bound, length, or size for
the member. The REFER object must be a member of
the same structure or union. It must not be
locator-qualified or subscripted, and it must precede
the member with the REFER option.
reference. The appearance of a name, except in a
context that causes explicit declaration.
relative virtual origin (RVO). The actual origin of an
array minus the virtual origin of an array.
Glossary 611
remote format item. The letter R followed by the label
(enclosed in parentheses) of a FORMAT statement. The
format statement is used by edit-directed data
transmission statements to control the format of data
being transmitted.
repetition factor. A parenthesized unsigned integer
constant that specifies:
1. The number of times the string constant that
follows is to be repeated.
2. The number of times the picture character that
follows is to be repeated.
repetitive specification. An element of a data list that
specifies controlled iteration to transmit one or more
data items, generally used in conjunction with arrays.
restricted expression. An expression that can be
evaluated by the compiler during compilation, resulting
in a constant. Operands of such an expression are
constants, named constants, and restricted expressions.
returned value. The value returned by a function
procedure.
RETURNS descriptor. A descriptor used in a
RETURNS attribute, and in the RETURNS option of the
PROCEDURE and ENTRY statements.
S
scalar variable. A variable that is not a structure,
union, or array.
scale. A system of mathematical notation whose
representation of an arithmetic value is either
fixed-point or floating-point.
scale factor. A specification of the number of fractional
digits in a fixed-point number.
scaling factor. See scale factor.
scope (of a condition prefix). The portion of a
program throughout which a particular condition prefix
applies.
scope (of a declaration or name). The portion of a
program throughout which a particular name is known.
secondary entry point. An entry point identified by
any of the names in the label list of an entry statement.
select-group. A sequence of statements delimited by
SELECT and END statements.
selection clause. A WHEN or OTHERWISE clause of a
select-group.
self-defining data. An aggregate that contains data
items whose bounds, lengths, and sizes are determined
at program execution time and are stored in a member
of the aggregate.
separator. See delimiter.
shift. Change of data in storage to the left or to the
right of original position.
shift-in. Symbol used to signal the compiler at the end
of a double-byte string.
shift-out. Symbol used to signal the compiler at the
beginning of a double-byte string.
sign and currency symbol characters. The picture
specification characters. S, +, , and $ (or other national
currency symbols enclosed in < and >).
simple parameter. A parameter for which no storage
class attribute is specified. It can represent an argument
of any storage class, but only the current generation of
a controlled argument.
simple statement. A statement other than IF, ON,
WHEN, and OTHERWISE.
source. Data item to be converted for problem data.
source key. A key referred to in a record-oriented
transmission statement that identifies a particular
record within a direct access data set.
source program. A program that serves as input to the
source program processors and the compiler.
source variable. A variable whose value participates
in some other operation, but is not modified by the
operation. Contrast with target variable.
spill file. Data set named SYSUT1 that is used as a
temporary workfile.
standard default. The alternative attribute or option
assumed when none has been specified and there is no
applicable DEFAULT statement.
standard file. A file assumed by PL/I in the absence
of a FILE or STRING option in a GET or PUT
statement. SYSIN is the standard input file and
SYSPRINT is the standard output file.
standard system action. Action specified by the
language to be taken for an enabled condition in the
absence of an ON-unit for that condition.
statement. A PL/I statement, composed of keywords,
delimiters, identifiers, operators, and constants, and
terminated by a semicolon (;). Optionally, it can have a
condition prefix list and a list of labels. See also keyword
statement, assignment statement, and null statement.
statement body. A statement body can be either a
simple or a compound statement.
statement label. See label constant.
static storage allocation. The allocation of storage for
static variables.
static variable. A variable that is allocated before
execution of the program begins and that remains
allocated for the duration of execution.
stream-oriented data transmission. The transmission
of data in which the data is treated as though it were a
continuous stream of individual data values in
character form. Contrast with record-oriented data
transmission.
string. A contiguous sequence of characters, graphics,
or bits that is treated as a single data item.
string variable. A variable declared with the BIT,
CHARACTER, or GRAPHIC attribute, whose values
can be either bit, character, or graphic strings.
structure. A collection of data items that need not
have identical attributes. Contrast with array.
structure expression. An expression whose evaluation
yields a structure set of values.
structure of arrays. A structure that has the dimension
attribute.
structure member. See member.
structuring. The hierarchy of a structure, in terms of
the number of members, the order in which they
appear, their attributes, and their logical level.
subroutine. A procedure that has no RETURNS option
in the PROCEDURE statement. Contrast with function.
subroutine call. An entry reference that must
represent a subroutine, followed by an optional
argument list that appears in a CALL statement.
Contrast with function reference.
subscript. An element expression that specifies a
position within a dimension of an array. If the subscript
is an asterisk, it specifies all of the elements of the
dimension.
subscript list. A parenthesized list of one or more
subscripts, one for each dimension of the array, which
together uniquely identify either a single element or
cross section of the array.
subtask. A task that is attached by the given task or
any of the tasks in a direct line from the given task to
the last attached task.
synchronous. A single flow of control for serial
execution of a program.
T
target. Attributes to which a data item (source) is
converted.
target reference. A reference that designates a
receiving variable (or a portion of a receiving variable).
target variable. A variable to which a value is
assigned.
task. The execution of one or more procedures by a
single flow of control.
task name. An identifier used to refer to a task
variable.
task variable. A variable with the TASK attribute
whose value gives the relative priority of a task.
termination (of a block). Cessation of execution of a
block, and the return of control to the activating block
by means of a RETURN or END statement, or the
transfer of control to the activating block or to some
other active block by means of a GO TO statement.
termination (of a task). Cessation of the flow of
control for a task.
truncation. The removal of one or more digits,
characters, graphics, or bits from one end of an item of
data when a string length or precision of a target
variable has been exceeded.
type. The set of data attributes and storage attributes
that apply to a generation, a value, or an item of data.
U
undefined. Indicates something that a user must not
do. Use of a undefined feature is likely to produce
different results on different implementations of a PL/I
product. In that case, the application program is in
error.
union. A collection of data elements that overlay each
other, occupying the same storage. The members can be
structures, unions, elementary variables, or arrays. They
need not have identical attributes.
union of arrays. A union that has the DIMENSION
attribute.
upper bound. The upper limit of an array dimension.
V
value reference. A reference used to obtain the value
of an item of data.
Glossary 613
variable. A named entity used to refer to data and to
which values can be assigned. Its attributes remain
constant, but it can refer to different values at different
times.
variable reference. A reference that designates all or
part of a variable.
virtual origin (VO). The location where the element of
the array whose subscripts are all zero are held. If such
an element does not appear in the array, the virtual
origin is where it would be held.
Z
zero-suppression characters. The picture specification
characters Z and *, which are used to suppress zeros in
the corresponding digit positions and replace them
with blanks or asterisks respectively.
Index
Special characters
/ (forward slash) 260
*PROCESS, specifying options in 153
% statements 154
%INCLUDE statement 154, 230
control statement 154
source statement library 230
%NOPRINT 154
%NOPRINT statement 154
%PAGE 154
%PAGE statement 154
%POP statement 154
%PRINT 154
%PRINT statement 154
%PROCESS, specifying options in 153
%PUSH statement 155
%SKIP 154
%SKIP statement 154
A
access
ESDS 346
REGIONAL(1) data set 327
relative-record data set 364
access method services
regional data set 329
REGIONAL(1) data set
direct access 327
sequential access 327
accessibility
assistive technologies xxxiii
keyboard navigation xxxiii
of Enterprise PL/I for z/OS xxxiii
of this information xxxiii
ACCT EXEC statement parameter 222
aggregate
length table 159
AGGREGATE compiler option 9
ALIGNED compiler suboption 35
ALL option
hooks location suboption 139
ALLOCATE statement 159
alternate ddname under z/OS UNIX, in
TITLE option 260
AMP parameter 333
ANS
compiler suboption 35
APPEND option under z/OS UNIX 262
ARCH compiler option 10, 369
argument
sort program 392
argument passing
by descriptor list 541
by descriptor-locator 542
array descriptor
array descriptor 543, 545
ASCII
compiler suboption
description 35
assembler routines
FETCHing 248
ASSIGNABLE compiler suboption 35
assistive technologies xxxiii
ATTENTION ON-units 522
attention processing
attention interrupt,effect of 66
debugging tool 522
main description 521
attribute table 158
ATTRIBUTES option 12
automatic
padding 237
prompting
overriding 236
using 235
restart
after system failure 525
checkpoint/restart facility 523
within a program 525
auxiliary storage for sort 392
avoiding calls to library routines 381
B
BACKREG compiler option 13
batch compile
OS/390 225, 227
BIFPREC compiler option 14
BIN1ARG compiler suboption 36
BKWD option 275, 340
BLANK compiler option 15
BLKOFF compiler option 16
BLKSIZE
consecutive data sets 315
ENVIRONMENT 275
comparison with DCB
subparameter 276
for record I/O 278
option of ENVIRONMENT
for stream I/O 292
subparameter 272
block
and record 267
size
maximum 279
object module 229
PRINT files 300
record length 279
regional data sets 330
specifying 268
BUFFERS option
for stream I/O 292
BUFND option 340
BUFNI option 341
BUFSIZE option under z/OS UNIX 263
BUFSP option 341
BYADDR
description 372
effect on performance 373
using with DEFAULT option 36
BYVALUE
description 372
using with DEFAULT option 36
C
C routines
FETCHing 247
capacity record
REGIONAL(1) 325
carriage control character 78, 299
carriage return-line feed (CR - LF) 266
cataloged procedure
compile and bind 211
compile only 210
compile, bind, and run 213
compile, input data for 213, 216
compile, prelink and link-edit 214
compile, prelink, link-edit, and
run 216
compile, prelink, load and run 218
description of 209
invoking 220
listing 220
modifying
DD statement 222
EXEC statement 222
multiple invocations 220
under OS/390
IBM-supplied 209
to invoke 220
to modify 221
CEESTART compiler option 17
character string attribute table 158
characters
carriage control 78, 299
print control 78, 299
CHECK compiler option 18
checkpoint data
for sort 396
checkpoint data, defining, PLICKPT
built-in suboption 524
checkpoint/restart
deferred restart 525
PLICANC statement 526
checkpoint/restart facility
CALL PLIREST statement 525
checkpoint data set 524
description of 523
modify activity 526
PLICKPT built-in subroutine 523
request checkpoint record 523
request restart 525
checkpoint/restart facility (continued)
RESTART parameter 525
return codes 523
CHKPT sort option 390
CICS
preprocessor options 206
support 205
CKPT sort option 390
CMPAT compiler option 19
COBOL
map structure 159
CODE subparameter 272
CODEPAGE compiler option 21
coding
CICS statements 206
improving performance 375
SQL statements 195
comments
within options 153
communications area, SQL 195
compilation
user exit
activating 528
customizing 529
example 533
IBMUEXIT 528
procedures 527
compile and bind, input data for 211
COMPILE compiler option 23
compile-time options
under z/OS UNIX 226
compile, prelink, and link-edit, input data
for 214
compiler
% statements 154
DBCS identifier 55
descriptions of options 5
general description of 225
graphic string constant 55
invoking 225
JCL statements, using 227
listing
aggregate length table 159
attribute table 158
block level 157
cross-reference table 159
DO-level 157
file reference table 162
heading information 156
include source program 65
input to compiler 157
input to preprocessor 157
messages 163
printing options 229
return codes 163
SOURCE option program 157
source program 128
stack storage used 133
statement offset addresses 159
SYSPRINT 229
using 156
mixed string constant 55
PROCESS statement 153
reduce storage requirement 96
severity of error condition 23
temporary workfile (SYSUT1) 229
under OS/390 batch 227
compiler options
abbreviations 5
AGGREGATE 9
ARCH 10, 369
ATTRIBUTES 12
BACKREG 13
BIFPREC 14
BLANK 15
BLKOFF 16
CEESTART 17
CHECK 18
CMPAT 19
CODEPAGE 21
COMPILE 23
COPYRIGHT 24
CSECT 25
CSECTCUT 26
CURRENCY 27
DBCS 28
DD 29
DDSQL 30
default 5
DEFAULT 33, 372
DEPRECATE 42
description of 5
DISPLAY 44
DLLINIT 45
EXIT 46
EXTRN 47
FLAG 48
FLOAT 49
FLOATINMATH 52
GOFF 53
GONUMBER 54, 369
GRAPHIC 55
HGPR 56
IGNORE 57
INCAFTER 58
INCDIR 59
INSOURCE 65
INTERRUPT 66
LANGLVL 67
LIMITS 68
LINECOUNT 70
LINEDIR 71
LIST 72
LISTVIEW 73
MACRO 75
MAP 76
MARGINI 77
MARGINS 78
MAXGEN 80
MAXMEM 81
MAXMSG 82
MAXNEST 83
MAXSTMT 84
MAXTEMP 85
MDECK 86
NAME 87
NAMES 88
NATLANG 89
NEST 90
NOMARGINS 78
NOT 91
NUMBER 92
OBJECT 93
OFFSET 94
compiler options (continued)
ONSNAP 95
OPTIMIZE 96, 369
OPTIONS 98
OR 99
PP 100
PPCICS 101
PPINCLUDE 102
PPLIST 103
PPMACRO 104
PPSQL 105
PPTRACE 106
PRECTYPE 107
PREFIX 108, 371
PROCEED 109
QUOTE 111
REDUCE 112, 370
RENT 113
RESEXP 115
RESPECT 116
RTCHECK 117
RULES 118, 370
SEMANTIC 126
SERVICE 127
SOURCE 128
SPILL 129
STATIC 130
STDSYS 131
STMT 132
STORAGE 133
STRINGOFGRAPHIC 134
SYNTAX 135
SYSPARM 136
SYSTEM 137
TERMINAL 138
TEST 139
UNROLL 142
USAGE 143
WIDECHAR 144
WINDOW 145
WRITABLE 146
XINFO 148
XML 151
XREF 152
compiling
under z/OS UNIX 225
concatenating
data sets 257
external references 255
COND EXEC statement parameter 222
conditional compilation 23
conditional subparameter 271
CONNECTED compiler suboption
description 36
CONSECUTIVE
option of ENVIRONMENT 292, 312
consecutive data sets
controlling input from the terminal
capital and lowercase letters 307
condition format 304
COPY option of GET
statement 307
defining QSAM files 306
end-of-file 307
format of data 305
stream and record files 306
consecutive data sets (continued)
controlling output to the terminal
conditions 309
format of PRINT files 309
output from the PUT EDIT
command 309
defining and using 291
input from the terminal 304
output to the terminal 309
record-oriented data transmission
accessing and updating a data
set 316
creating a data set 315
defining files 312
specifying ENVIRONMENT
options 312
statements and options
allowed 311
record-oriented I/O 311
stream-oriented data
transmission 291
accessing a data set 297
creating a data set 294
defining files 291, 292
options 292
using PRINT files 299
using SYSIN and SYSPRINT
files 303
control
area 334
CONTROL option
EXEC statement 230
interval 334
control blocks
function-specific 528
global control 529
control characters
carriage 78, 299
print 78, 299
COPY option 307
COPYRIGHT compiler option 24
COPYRIGHT option 514
counter records, SYSADATA 551
cross-reference table
compiler listing 159
using XREF option 158
CSECT compiler option 25
CSECTCUT compiler option 26
CTLASA and CTL360 options
ENVIRONMENT option
for consecutive data sets 313
SCALARVARYING 282
CURRENCY compiler option 27
customizing
user exit
modifying SYSUEXIT 529
structure of global control
blocks 529
writing your own compiler
exit 529
CYLOFL subparameter
DCB parameter 272
D
data
conversion under z/OS UNIX 259
files
creating under z/OS UNIX 261
sort program 396
PLISRT(x) command 401
sorting
description of 385
types
equivalent Java and PL/I 440
equivalent SQL and PL/I 200
data definition (DD) information under
z/OS UNIX 259
data set
associating PL/I files with
closing a file 274
opening a file 273
specifying characteristics in the
ENVIRONMENT attribute 275
associating several data sets with one
file 256
blocks and records 267
checkpoint 524
conditional subparameter
characteristics 272
consecutive stream-oriented data 291
data set control block (DSCB) 271
ddnames 228
defining for dump
DD statement 508
logical record length 508
defining relative-record 362
direct 270
dissociating from a file 274
system-determined block size 274
dissociating from PL/I file 257
establishing characteristics 267
indexed
sequential 270
information interchange codes 268
input in cataloged procedures 209
label modification 273
labels 271, 285
libraries
extracting information 289
SPACE parameter 286
types of 285
updating 287
use 285
organization
conditional subparameters 271
data definition (DD)
statement 271
types of 270
partitioned 285
record format defaults 277
record formats
fixed-length 268
undefined-length 270
variable-length 269
records 268
regional 321
REGIONAL(1) 324
accessing and updating 326
creating 325
sequential 270
data set (continued)
sort program
checkpoint data set 396
input data set 395
output data set 396
sort work data set 395
sorting 395
SORTWK 392
SPACE parameter 228
stream files 291
temporary 229
to establish characteristics 267
types of
comparison 282
organization 270
used by PL/I record I/O 282
unlabeled 271
using 253
VSAM
blocking 334
data set type 337
defining 343
defining files 339
dummy data set 338
file attribute 338
indexed data set 347
keys 336
mass sequential insert 353
organization 334
running a program 333
options 339
VSAM option 342
VSAM.
performance options 342
data set under OS/390
associating one data set with several
files 256
concatenating 257
HFS 257
data set under z/OS UNIX
associating a PL/I file with a data set
how PL/I finds data sets 261
using environment variables
under 259
using the TITLE option of the
OPEN statement 260
using unassociated files 261
DD_DDNAME environment
variable 258
default identification 258
establishing a path 261
establishing characteristics
variable 262
extending on output 262
maximum number of regions 265
number of regions 265
recreating output 262
data sets
allocating files 253
associating data sets with files 254
closing 274
defining data sets under OS/390 254
data-directed I/O 376
coding for performance 375
Index 617
DBCS compiler option 28
DBCS identifier compilation 55
DCB subparameter 274, 277
equivalent ENVIRONMENT
options 277
main discussion of 272
overriding in cataloged
procedure 223
DD (data definition) information under
z/OS UNIX 259
DD compiler option 29
DD information under z/OS UNIX
TITLE statement 260
DD statement 271
%INCLUDE 155
add to cataloged procedure 222
cataloged procedure, modifying 222
checkpoint/restart 523
create a library 286
input data set in cataloged
procedure 209
modifying cataloged procedure 221
OS/390 batch compile 228
standard data set 228
input (SYSIN) 228
output (SYSLIN,
SYSPUNCH) 229
DD Statement
modify cataloged procedure 222
DD_DDNAME environment variables
alternate ddname under z/OS
UNIX 260
APPEND 262
DELAY 263
DELIMIT 264
LRECL 264
LRMSKIP 264
PROMPT 264
PUTPAGE 264
RECCOUNT 265
RECSIZE 265
SAMELINE 265
SKIP0 266
specifying characteristics under z/OS
UNIX 261
TYPE 266
ddname
%INCLUDE 155
standard data sets 228
DDSQL compiler option 30
deblocking of records 268
declaration
of files under OS/390 253, 254
declaring
host variables, SQL preprocessor 197
DEFAULT compiler option
description and syntax 33
suboptions
ALIGNED 35
ASCII or EBCDIC 35
ASSIGNABLE or
NONASSIGNABLE 35
BIN1ARG or NOBIN1ARG 36
BYADDR or BYVALUE 36
DEFAULT compiler option (continued)
suboptions (continued)
CONNECTED or
NONCONNECTED 36
DESCLIST or DESCLOCATOR 36
DESCRIPTOR or
NODESCRIPTOR 36
DUMMY 37
E 37
EVENDEC or NOEVENDEC 38
HEXADEC 38
IBM or ANS 35
INITFILL or NOINITFILL 38
INLINE or NOINLINE 38
LINKAGE 38
LOWERINC | UPPERINC 39
NATIVE or NONNATIVE 39
NATIVEADDR or
NONNATIVEADDR 39
NULLSTRADDR or
NONULLSTRADDR 40
NULLSTRPTR 40
NULLSYS or NULL370 40
ORDER or REORDER 40
ORDINAL(MIN | MAX) 40
OVERLAP | NOOVERLAP 40
PSEUDODUMMY or
NOPSEUDODUMMY 40
RECURSIVE or
NONRECURSIVE 41
RETCODE 41
RETURNS 41
SHORT 41
define data set
file 256
associating several files with one data
set 256
closing a file 274
concatenating several data sets 257
ESDS 345
opening a file 273
specifying characteristics 275
define file
associating several files with one data
set 256
closing a file 274
concatenating several data sets 257
opening a file 273
ENV options 323
keys 324
specifying characteristics 275
VSAM data set 339
define file under OS/390
file 256
DEFINED
versus UNION 379
DELAY option under z/OS UNIX
description 263
DELIMIT option under z/OS UNIX
description 264
DEPRECATE compiler option 42
DESCLIST compiler suboption 36
DESCLOCATOR compiler suboption 36
descriptor 541
descriptor area, SQL 195
DESCRIPTOR compiler option
DESCRIPTOR compiler suboption
description 36
descriptor list, argument passing 541
descriptor-locator, argument passing 542
DFSORT 385
direct data sets 270
DIRECT file
indexed ESDS with VSAM
accessing data set 351
updating data set 353
RRDS
access data set 364
DISP parameter
to delete a data set 285
DISPLAY compiler option 44
DLL
DYNAM=DLL linker option 240
linking considerations and
side-decks 233
RENT compiler option and
fetching 239
DLLINIT compiler option 45
DSA
saved in PLIDUMP built-in subroutine
for each block 509
DSCB (data set control block) 271, 287
DSNAME parameter
for consecutive data sets 315, 317
DSORG subparameter 272
DUMMY compiler suboption 37
dummy records
VSAM 338
dump
calling PLIDUMP 507
defining data set for
DD statement 508
logical record length 508
identifying beginning of 508
PLIDUMP built-in subroutine 507
producing z/OS Language
Environment dump 507
SNAP 508
DYNALLOC sort option 390
E
E compiler message 163
E compiler suboption 37
E15 input handling routine 397
E35 output handling routine 400
EBCDIC
compiler suboption 35
EBCDIC (Extended Binary Coded
Decimal Interchange Code) 268
embedded
CICS statements 206
SQL statements 196
ENDFILE
under OS/390 237
accessibility xxxiii
Enterprise PL/I library
Enterprise PL/I for z/OS library xiv
Language Environment library xiv
entry point
sort program 392
entry-sequenced data set
defining 345
updating 346
VSAM 335
loading an ESDS 344
SEQUENTIAL file 345
statements and options 344
ENVIRONMENT attribute
list 275
specifying characteristics under z/OS
UNIX
BUFSIZE 263
ENVIRONMENT options
BUFFERS option
comparison with DCB
subparameter 276
CONSECUTIVE 292, 312
CTLASA and CTL360 313
comparison with DCB
subparameter 276
equivalent DCB subparameters 277
GRAPHIC option 294
KEYLENGTH option
comparison with DCB
subparameter 276
LEAVE and REREAD 314
organization options 276
record format options 293
RECSIZE option
comparison with DCB
subparameter 276
record format 293
usage 293
VSAM
BKWD option 340
BUFND option 340
BUFNI option 341
BUFSP option 341
GENKEY option 341
PASSWORD option 341
REUSE option 341
SKIP option 342
VSAM option 342
environment variables
setting under z/OS UNIX 283
EQUALS sort option 390
error
severity of error compilation 23
error devices
redirecting 284
ESDS (entry-sequenced data set)
defining 345
nonunique key alternate index
path 355
ESDS (entry-sequenced data set)
(continued)
unique key alternate index path 354
updating 346
VSAM 335
loading 344
statements and options 344
EVENDEC compiler suboption 38
examples
calling PLIDUMP 507
EXEC SQL statements 175
EXEC statement
cataloged procedure, modifying 222
compiler 227
introduction 227
maximum length of option list 230
minimum region size 227
modify cataloged procedure 222
OS/390 batch compile 225, 227
PARM parameter 230
to specify options 230
Exit (E15) input handling routine 397
Exit (E35) output handling routine 400
EXIT compiler option 46
export command 261
extended binary coded decimal
interchange code (EBCDIC) 268
EXTERNAL attribute 158
external references
concatenating names 255
EXTRN compiler option 47
F
F option of ENVIRONMENT
for record I/O 277
for stream I/O 292
F-format records 268
FB option of ENVIRONMENT
for record I/O 277
for stream I/O 292
FB-format records 268
FBS option of ENVIRONMENT
for record I/O 277
for stream I/O 292
FETCH
assembler routines 248
Enterprise PL/I routines 239
OS/390 C routines 247
PL/I MAIN Routines 247
field for sorting 389
file
allocating files 253
associating data sets with files 254
closing 274
defining data sets under OS/390 254
establishing characteristics 267
FILE attribute 158
file records, SYSADATA 552
filespec 262
FILLERS, for tab control table 302
FILSZ sort option 390
filtering messages 528
FIXED
TYPE option under z/OS UNIX 267
fixed-length records 268
FLAG compiler option 48
flags, specifying compile-time
options 226
FLOAT option 49
FLOATINMATH compiler option 52
flowchart for sort 397
format notation, rules for xv
forward slash (/) 260
FS option of ENVIRONMENT
for record I/O 277
for stream I/O 292
FUNC subparameter
usage 272
Functions
using C functions with ILC 411
G
GENKEY option
key classification 280
usage 275
VSAM 339
GET DATA statement 236
GET EDIT statement 237
GET LIST statement 236
global control blocks
data entry fields 530
writing the initialization
procedure 530
writing the message filtering
procedure 531
writing the termination
procedure 533
GOFF compiler option 53
GONUMBER compiler option 369
definition 54
GOTO statements 376
graphic data 291
GRAPHIC option
compiler 55
of ENVIRONMENT 275, 294
stream I/O 292
graphic string constant compilation 55
H
handling routines
data for sort
input (sort exit E15) 397
output (sort exit E35) 400
PLISRTB 402
PLISRTC 403
PLISRTD 404
to determine success 394
variable length records 406
header label 271
HEXADEC compiler suboption 38
HGPR compiler option 56
hook
location suboptions 139
host
structures 204
variables, using in SQL
statements 197
Index 619
I
I compiler message 163
IBM compiler suboption 35
IBMUEXIT compiler exit 528
IBMZC cataloged procedure 210
IBMZCB cataloged procedure 211
IBMZCBG cataloged procedure 213
IBMZCPG cataloged procedure 218
IBMZCPL cataloged procedure 214
IBMZCPLG cataloged procedure 216
identifiers
not referenced 12
source program 12
IEC225I 272, 330
IGNORE option 57
ILC
linkage considerations 417
with C 409
data types 409
Enum data type 410
File type 411
Functions returning ENTRYs 416
parameter matching 412
Redirecting C standard
streams 420
sharing input 420
sharing output 419
string parameter type
matching 415
structure data type 410
Using the ATTACH statement 420
improving application performance 369
INCAFTER compiler option 58
INCDIR compiler option 59
include preprocessor
syntax 166
INCLUDE statement
compiler 155
INDEXAREA option 275
indexed data sets
indexed sequential data set 270
indexed ESDS (entry-sequenced data set)
DIRECT file 351
loading 349
SEQUENTIAL file 350
indicator variables, SQL 204
information interchange codes 268
INITFILL compiler suboption 38
initial volume label 271
initialization procedure of compiler user
exit 530
INLINE compiler suboption 38
input
data for PLISRTA 401
data for sort 396
defining data sets for stream
files 291, 292
redirecting 284
routines for sort program 396
SEQUENTIAL 315
skeletal code for sort 400
sort data set 396
input/output
compiler
data sets 228
data for compile and bind 211
input/output (continued)
data for compile, prelink, and
link-edit 214
in cataloged procedures 210
OS/390, punctuating long lines 236
sort data set 395
INSOURCE option 65
Inter Language Communication
linkage considerations 417
Redirecting C standard streams 420
Using the ATTACH statement 420
with C 409
data types 409
Enum data type 410
File type 411
Functions returning ENTRYs 416
parameter matching 412
sharing input 420
sharing output 419
string parameter type
matching 415
structure data type 410
using C functions 411
interactive program
attention interrupt 66
interblock gap (IBG) 267
interchange codes 268
INTERNAL attribute 158
INTERRUPT compiler option 66
interrupts
attention interrupts under interactive
system 66
debugging tool 522
main description 521
invoking
cataloged procedure 220
link-editing multitasking
programs 222
multiple invocations 220
J
Java 424, 425, 427, 428, 429, 430, 431,
432, 434, 435, 436, 440
JAVA 423
Java code, compiling 425, 429, 434, 436
Java code, writing 424, 428, 432, 436
JCL (job control language)
improving efficiency 209
using during compilation 227
jni
JNI sample program 424, 427, 432,
436
K
key indexed VSAM data set 336
key-sequenced data sets
accessing with a DIRECT file 351
accessing with a SEQUENTIAL
file 350
loading 349
statements and options for 347
keyboard navigation xxxiii
KEYLEN subparameter 272
KEYLENGTH option 275, 282
KEYLOC option
usage 275
keys
alternate index
nonunique 354
unique 354
dummy records 325
VSAM
indexed data set 336
relative byte address 337
relative record number 337
KEYTO option
under VSAM 344
KSDS (key-sequenced data set)
define and load 349
unique key alternate index path 355
updating 351
VSAM
DIRECT file 351
loading 349
SEQUENTIAL file 350
KSDS with VSAM
access data set 350
accessing data set 351
updating data set 353
L
label
for data sets 271
LANGLVL compiler option 67
Language Environment library xiv
LEAVE and REREAD options
ENVIRONMENT option
length of record
specifying under z/OS UNIX 265
library
compiled object modules 288
creating a data set library 286
creating a member 289
creating and updating a library
member 287
creating, examples of 287
directory 286
extracting information from a library
directory 289
general description of 270
how to use 285
information required to create 286
placing a load module 288
source statement 230
SPACE parameter 286
structure 289
system procedure
(SYS1.PROCLIB) 285
types of 285
updating a library member 289
using 285
LIMCT subparameter 272, 330
LIMITS compiler option 68
line
length 300
numbers in messages 54
line feed (LF)
definition 266
LINE option 293, 300
LINECOUNT compiler option 70
LINEDIR compiler option 71
LINESIZE option
for tab control table 302
OPEN statement 293
link-editing
description of 233
LINKAGE compiler suboption
syntax 38
linkage considerations with ILC 417
LIST compiler option 72
listing
compiler
aggregate length table 159
ATTRIBUTE and cross-reference
table 158
ddname list 5
file reference table 162
heading information 156
messages 163
options 157
preprocessor input 157
return codes 163
SOURCE option program 157
statement nesting level 157
statement offset addresses 159
storage offset listing 161
OS/390 batch compile 225, 229
source program 128
statement offset address 159
storage offset listing 161
SYSPRINT 229
LISTVIEW compiler option 73
literal records, SYSADATA 551
loader program, using 218
logical not 91
logical or 99
loops
control variables 377
LOWERINC compiler suboption 39
LRECL option under z/OS UNIX 264
LRECL subparameter 268, 272
LRMSKIP option under z/OS UNIX 264
M
MACRO option 75
macro preprocessor
macro definition 167
main storage for sort 391
MAP compiler option 76
MARGINI compiler option 77
MARGINS compiler option 78
MAXGEN compiler option 80
MAXMEM compiler option 81
MAXMSG compiler option 82, 83
MAXSTMT compiler option 84
MAXTEMP compiler option 85
MDECK compiler option
description 86
message
compiler list 163
printed format 303
runtime message line numbers 54
message records, SYSADATA 552
messages
filter function 531
modifying in compiler user exit 529
mixed string constant compilation 55
MODE subparameter
usage 272
module
create and store object module 93
multiple
invocations
Multitasking
options in PLIDUMP 507
N
NAME compiler option 87
named constants
defining 379
versus static variables 379
NAMES compiler option 88
NATIVE compiler suboption
description 39
NATIVEADDR compiler suboption 39
NATLANG compiler option 89
negative value
block-size 279
record length 278
NEST option 90
NOBIN1ARG compiler suboption 36
NODESCRIPTOR compiler suboption 36
NOEQUALS sort option 390
NOEVENDEC compiler suboption 38
NOINITFILL compiler suboption 38
NOINLINE compiler suboption 38
NOINTERRUPT compiler option 66
NOMAP option 159
NOMARGINS compiler option 78
NONASSIGNABLE compiler
suboption 35
NONCONNECTED compiler
suboption 36
NONE, hooks location suboption 139
NONNATIVE compiler suboption 39
NONNATIVEADDR compiler
suboption 39
NONRECURSIVE compiler
suboption 41
NONULLSTRADDR compiler
suboption 40
NOOVERLAP compiler suboption 40
NOPSEUDODUMMY compiler
suboption 40
NOSYNTAX compiler option 135
NOT compiler option 91
note statement 163
NTM subparameter
usage 272
NULL370 compiler suboption 40
NULLSTRADDR compiler suboption 40
NULLSTRPTR compiler suboption 40
NULLSYS compiler suboption 40
NUMBER compiler option 92
O
object
module
create and store 93
record size 229
OBJECT compiler option
definition 93
offset
of tab count 302
table 159
OFFSET compiler option 94
ONSNAP compiler option 95
OPEN statement
subroutines of PL/I library 273
TITLE option 272
Operating system
data definition (DD) information
under z/OS UNIX 259
OPTCD subparameter 271, 272
optimal coding
coding style 375
compiler options 369
OPTIMIZE compiler option 369
OPTIMIZE option 96
options
for compiling 157
for creating regional data set 321
saved options string in
PLIDUMP 515
specifying comments within 153
specifying strings within 153
to specify for compilation 230
OPTIONS option 98
options under z/OS UNIX
DD_DDNAME environment variables
APPEND 262
DELAY 263
DELIMIT 264
LRECL 264
LRMSKIP 264
PROMPT 264
PUTPAGE 264
RECCOUNT 265
RECSIZE 265
SAMELINE 265
SKIP0 266
TYPE 266
PL/I ENVIRONMENT attribute
BUFSIZE 263
using
DD information 260
TITLE 260
OR compiler option 99
ORDER compiler suboption
description 40
ORDINAL compiler suboption 40
ordinal element records,
SYSADATA 554
ordinal type records, SYSADATA 553
ORGANIZATION option 282
Index 621
ORGANIZATION option (continued)
usage 275
OS/390
batch compilation
DD statement 228
EXEC statement 227, 230
listing (SYSPRINT) 229
source statement library
(SYSLIB) 230
specifying options 230
temporary workfile (SYSUT1) 229
general compilation 225
output
data for PLISRTA 401
data for sort 396
defining data sets for stream
files 291, 292
limit preprocessor output 86
redirecting 284
routines for sort program 396
SEQUENTIAL 315
sort data set 396
SYSLIN 229
SYSPUNCH 229
OVERLAP compiler suboption 40
P
PACKAGEs versus nested
PROCEDUREs 377
PAGE option 293
PAGELENGTH, for tab control table 302
PAGESIZE, for tab control table 302
parameter passing
argument passing 541
CMPAT(LE) descriptors 544
CMPAT(V*) descriptors 542
PARM parameter
for cataloged procedure 222
specify options 230
passing an argument 541
PASSWORD option 341
performance
VSAM options 342
performance improvement
coding for performance
avoiding calls to library
routines 381
DATA-directed input and
output 375
DEFINED versus UNION 379
GOTO statements 376
input-only parameters 376
loop control variables 377
named constants versus static
variables 379
PACKAGEs versus nested
PROCEDUREs 377
preloading calls to library
routines 381
REDUCIBLE functions 378
string assignments 376
selecting compiler options
ARCH 369
DEFAULT 372
GONUMBER 369
performance improvement (continued)
selecting compiler options (continued)
OPTIMIZE 369
PREFIX 371
REDUCE 370
RULES 370
PL/I
compiler
user exit procedures 528
files
associating with a data set under
z/OS UNIX 258
PL/I code, compiling 427, 431, 435, 440
PL/I code, linking 427, 431, 435, 440
PL/I code, writing 425, 430, 434, 436
PL/I dynamic allocation
access HFS files under z/OS 257
allocate data sets 253
associating data sets with files under
z/OS UNIX 258
define files
QSAM files 306
REGIONAL(1) data sets 323
stream files 292
VSAM file 333
PL/I MAIN Routines
FETCHing 247
PLICANC statement, and
checkpoint/request 526
PLICKPT built-in subroutine 523
PLIDUMP built-in subroutine
calling to produce a z/OS Language
Environment dump 507
DSA saved for each block 509
H option 508
output
locating variables in 509
program unit name in dump
traceback table 509
saved load module timestamp 514
saved options string 515
syntax of 507
user-identifier 508
variables
locating AUTOMATIC variables
in 509
locating CONTROLLED variables
in 511
locating in PLIDUMP output 509
locating STATIC variables in 510
PLIREST statement 525
PLIRETC built-in subroutine
return codes for sort 395
PLISAXA 443, 444
PLISAXB 443, 444
PLISAXC 473, 474
PLISAXD 473, 474
PLISRTA interface 401
PLISRTB interface 402
PLISRTC interface 403
PLISRTD interface 404
PLITABS external structure
control section 303
declaration 234
PLIXOPT variable 234
PP compiler option 100
PPCICS compiler option 101
PPINCLUDE compiler option 102
PPLIST compiler option 103
PPMACRO compiler option 104
PPSQL compiler option 105
PPTRACE compiler option 106
PRECTYPE compiler option 107
PREFIX compiler option 108, 371
using default suboptions 371
preloading library routines 381
preprocessing
%INCLUDE statement 155
input 157
limit output to 80 bytes 86
source program 75
with MACRO 75
preprocessor options
APOSTSQL 178
ATTACH 179
CCSID0 180
CONNECT 181
DATE 182
DEC 183
FLOAT 184
GRAPHIC 185
INCONLY 186
LEVEL 187
NOFOR 188
SQL 189
SQLFLAG 190
STDSQL 191
TIME 192
VERSION 193
XREF 194
preprocessors
available with PL/I 165
CICS options 206
include 166
macro preprocessor 167
SQL options 177
SQL preprocessor 175
print
PRINT file
format 309
line length 300
stream I/O 299
print control character 78, 299
PRINT file
formatting conventions 234
punctuating output 234
record I/O 317
procedure
cataloged, using under OS/390 209
compile and bind (IBMZCB) 211
compile and link-edit
(IBMZCPL) 214
compile only (IBMZC) 210
compile, bind, and run
(IBMZCBG) 213
compile, prelink, link-edit, and run
(IBMZCPLG) 216
compile, prelink, load and run
(IBMZCPG) 218
PROCEED compiler option 109
PROCESS statement
description 153
override option defaults 230
program unit name in dump traceback
table 509
PROMPT option under z/OS UNIX 264
prompting
automatic, overriding 236
automatic, using 235
PRTSP subparameter
usage 272
PSEUDODUMMY compiler
suboption 40
punctuation
automatic prompting
overriding 236
using 235
OS/390
automatic padding for GET
EDIT 237
continuation character 236
entering ENDFILE at
terminal 237
GET DATA statement 236
GET LIST statement 236
long input lines 236
SKIP option 237
output from PRINT files 234
PUT EDIT command 309
PUTPAGE option under z/OS
UNIX 264
Q
QUOTE compiler option 111
R
REAL attribute 158
RECCOUNT option under z/OS
UNIX 265
RECFM subparameter 272
in organiza tion of data set 272
usage 272
record
checkpoint 523
data set 524
deblocking 268
maximum size for compiler
input 228
sort program 390
record format
fixed-length records 268
options 293
stream I/O 298
to specify 312
types 268
undefined-length records 270
variable-length records 269
record I/O
data set
access 316
create 315
types of 282
data transmission 311
ENVIRONMENT option 312
format 277
record format 312
record length
specify 268
value of 278
RECORD statement 390
recorded key
records
length under z/OS UNIX 265
RECSIZE option
consecutive data set 293
defaults 293
definition 277
description under z/OS UNIX 265
for stream I/O 292, 293
RECURSIVE compiler suboption 41
REDUCE compiler option 112
reduce storage requirement 96
REDUCIBLE functions 378
region
REGION parameter 222
size, EXEC statement 227
REGION size, minimum required 209
regional data sets
DD statement
accessing 331
creating 330
defining files for
options 323
using keys 324
operating system requirement 329
REGIONAL(1) data set
accessing and updating 326
creating 325
using 324
REGIONAL option of
ENVIRONMENT 324
REGIONAL(1) data sets
defining files for
regions under z/OS UNIX 265
relative byte address (RBA) 337
relative record number 337
relative-record data sets
accessing with a DIRECT file 364
accessing with a SEQUENTIAL
file 363
loading 362
statements and options for 360
RENT compiler option 113
REORDER compiler suboption
description 40
RESEXP compiler option 115
RESPECT compiler option 116
restarting
requesting 525
RESTART parameter 525
to request
automatic after system failure 525
automatic within a program 525
to cancel 525
to modify 526
RETCODE compiler suboption 41
return code
checkpoint/restart routine 523
PLIRETC 395
return codes
in compiler listing 163
RETURNS compiler suboption 41, 375
REUSE option 275, 341
RRDS (relative record data set)
define 362
load statements and options 360
load with VSAM 362
updating 364
VSAM
DIRECT file 364
loading 362
SEQUENTIAL file 363
RTCHECK compiler option 117
RULES compiler option 118
run time
message line numbers 54
run-time
OS/390 considerations
automatic prompting 235
formatting conventions 234
GET EDIT statement 237
GET LIST and GET DATA
statements 236
punctuating long lines 236
SKIP option 237
using PLIXOPT 234
S
S compiler message 163
SAMELINE option under z/OS
UNIX 265
sample program, running 427, 432, 436,
440
saved options string in PLIDUMP 515
SAX parser 443, 473
SCALARVARYING option 281
SEMANTIC compiler option 126
sequential access
sequential data set 270
SEQUENTIAL file
ESDS with VSAM
defining and loading 345
updating 346
access data set 350
RRDS, access data set 363
serial number volume label 271
SERVICE compiler option 127
shift code compilation 55
SHORT compiler suboption 41
SKIP option 342
in stream I/O 293
under OS/390 237
SKIP0 option under z/OS UNIX 266
SKIPREC sort option 390
sorting
assessing results 394
calling sort 392
CHKPT option 390
Index 623
sorting (continued)
choosing type of sort 386
CKPT option 390
data 385
data input and output 396
description of 385
DYNALLOC option 390
E15 input handling routine 397
EQUALS option 390
FILSZ option 390
maximum record length 391
PLISRT 385
PLISRTA(x) command 401, 406
preparation 385
RECORD statement 397
RETURN statement 397
SKIPREC option 390
SORTCKPT 396
SORTCNTL 396
SORTIN 395
sorting field 389
SORTLIB 395
SORTOUT 396
SORTWK 392, 395
storage
auxiliary 392
main 391
writing input/output routines 396
source
key
in REGIONAL(1) data sets 324
listing
location 78
program
compiler list 157
data set 228
identifiers 12
included in compiler list 65
list 128
preprocessor 75
shifting outside text 77
SOURCE compiler option 128
source records, SYSADATA 570
SPACE parameter
library 286
specifying compile-time options
using flags under 226
SPILL compiler option 129
spill file 229
SQL preprocessor
communications area 195
descriptor area 195
EXEC SQL statements 175
options 177
using host structures 204
using host variables 197
using IBMUEXIT with 205
using indicator variables 204
SQL preprocessor options 178, 179, 180,
181, 182, 183, 184, 185, 186, 187, 188,
189, 190, 191, 192, 193, 194
SQLCA 195
SQLDA 195
STACK subparameter
usage 272
standard data set 228
standard files (SYSPRINT and
SYSIN) 284
statement
nesting level 157
offset addresses 159
statements 154
STATIC compiler option 130
STDSYS compiler option 131
step abend 271
STMT compiler option 132
STMT suboption of test 139
storage
blocking print files 300
library data sets 286
report in listing 133
sort program 391
auxiliary storage 392
main storage 391
to reduce requirement 96
STORAGE compiler option 133
stream and record files 306, 309
STREAM attribute 291
stream I/O
data set
access 297
create 294
record format 298
DD statement 296, 298
ENVIRONMENT options 292
file
define 291
PRINT file 299
SYSIN and SYSPRINT files 303
record formats for data
transmission 277
string
graphic string constant
compilation 55
string assignments 376
string descriptors
string descriptors 542, 544
STRINGOFGRAPHIC compiler
option 134
structure of global control blocks
writing the initialization
procedure 530
writing the message filtering
procedure 531
writing the termination
procedure 533
SUB control character 268
summary record, SYSADATA 550
symbol records, SYSADATA 555
symbol table 139
SYNTAX option 135
syntax records, SYSADATA 572
syntax, diagrams, how to read xv
SYS1.PROCLIB (system procedure
library) 285
SYSADATA information, counter
records 551
SYSADATA information, file records 552
SYSADATA information,
introduction 549
SYSADATA information, literal
records 551
SYSADATA information, message
records 552
SYSADATA information, ordinal element
records 554
SYSADATA information, ordinal type
records 553
SYSADATA information, source
records 570
SYSADATA information, summary
record 550
SYSADATA information, symbol
information 553
SYSADATA information, symbol
records 555
SYSADATA information, syntax
information 570
SYSADATA information, syntax
records 572
SYSADATA information, token
records 571
SYSCHK default 523, 524
SYSIN 228, 284
sharing with C 420
SYSIN and SYSPRINT files 303
SYSLIB
%INCLUDE 155
preprocessing 230
SYSLIN 229
SYSOUT 395
SYSPARM compiler option 136
SYSPRINT 284
and z/OS UNIX 284
compiler listing written to 229
effect of STDSYS option on 131
required DD statement 228
shared with older PL/I 238
sharing between enclaves 237
sharing with C 419, 420
specifying on the DD option 29
using MSGFILE(SYSPRINT) 239
using with the PUT statement 303
SYSPUNCH 229
system
failure 525
restart after failure 525
SYSTEM compiler options
SYSTEM(CICS) 137
SYSTEM(IMS) 137
SYSTEM(MVS) 137
SYSTEM(OS) 137
SYSTEM(TSO) 137
type of parameter list 137
SYSUT1 compiler data set 229
T
tab control table 302
temporary workfile
SYSUT1 229
terminal
input 304
capital and lowercase letters 307
COPY option of GET
statement 307
defining QSAM files 306
terminal (continued)
input (continued)
end of file 307
format of data 305
output 309
format of PRINT file 309
output from PUT EDIT
command 309
TERMINAL compiler option 138
terminating
compilation 23
termination procedure
compiler user exit 533
example of procedure-specific control
block 533
syntax
global 529
specific 533
TEST compiler option
definition 139
TIME parameter 222
TIMESTAMP
saved load module timestamp in
PLIDUMP 514
TITLE option
associating standard SYSPRINT
file 238
description under z/OS UNIX 260
using 272
TITLE option under OS/390
specifying character string value 254
TITLE option under z/OS UNIX
using files not associated with data
sets 261
token records, SYSADATA 571
traceback table
program unit name in 509
trailer label 271
TYPE option under z/OS UNIX 266
U
U compiler message 163
U option of ENVIRONMENT
for record I/O 277
for stream I/O 292
U-format 270
undefined-length records 270
UNDEFINEDFILE condition
BLKSIZE error 279
line size conflict in OPEN 300
raising when opening a file under
z/OS UNIX 267
UNDEFINEDFILE condition under
OS/390
DD statement error 255
UNDEFINEDFILE condition under z/OS
UNIX
using files not associated with data
sets 267
UNIT parameter
unreferenced identifiers 12
UNROLL compiler option 142
updating
ESDS 346
UPPERINC compiler suboption 39
USAGE compiler option 143
user exit
compiler 527
customizing
modifying SYSUEXIT 529
structure of global control
blocks 529
writing your own compiler
exit 529
functions 528
sort 388
using
arrays as host variable, SQL
preprocessor 197
using host variables, SQL
preprocessor 197
V
V option of ENVIRONMENT
for record I/O 277
for stream I/O 292
variable-length records
format 269
sort program 406
VB option of ENVIRONMENT
for record I/O 277
for stream I/O 292
VB-format records 269
VBS option of ENVIRONMENT
for stream I/O 292
VOLUME parameter
consecutive data sets 315, 317
volume serial number
direct access volumes 271
VS option of ENVIRONMENT
for stream I/O 292
VSAM (virtual storage access method)
data sets
alternate index paths 342
alternate indexes 353
blocking 334
choosing a type 337
defining 333, 343
defining files for 339
dummy data set 338
entry-sequenced 344
file attribute 338
key-sequenced and indexed
entry-sequenced 347
keys for 336
organization 334
performance options 342
relative record 360
running a program with 333
options 339
using 333
defining files 339
ENV option 339
performance option 342
VSAM (virtual storage access method)
(continued)
indexed data set
load statement and options 347
VSAM option 342
VTOC 271
W
W compiler message 163
WIDECHAR compiler option 144
WINDOW compiler option 145
work data sets for sort 395
WRITABLE compiler option 146
X
XINFO compiler option 148
XML
support in the SAX parser 443, 473
XML compiler option 151
XREF compiler option 152
Z
z/OS UNIX
compile-time options
specifying 226
compiling under 225
variable 261
export command 261
setting environment variables 283
specifying compile-time options
command line 226
using flags 226
zero value 278
Index 625
Readers Comments We'd Like to Hear from You
Programming Guide
Version 4 Release 2
Publication No. GI11-9145-01
We appreciate your comments about this publication. Please comment on specific errors or omissions, accuracy,
organization, subject matter, or completeness of this book. The comments you send should pertain to only the
information in this manual or product and the way in which the information is presented.
For technical questions and information about products and prices, please contact your IBM branch office, your
IBM business partner, or your authorized remarketer.
When you send comments to IBM, you grant IBM a nonexclusive right to use or distribute your comments in any
way it believes appropriate without incurring any obligation to you. IBM or any other organizations will only use
the personal information that you supply to contact you about the issues that you state on this form.
Comments:
Thank you for your support.
Submit your comments using one of these channels:
v Send your comments to the address on the reverse side of this form.
v Send a fax to the following number: 1-800-426-7773
v Send your comments via email to: [email protected]
If you would like a response from IBM, please fill in the following information:
Name Address
Company or Organization
Phone No. Email address
Readers Comments We'd Like to Hear from You
GI11-9145-01
GI11-9145-01
Cut or Fold
Along Line
Cut or Fold
Along Line
Fold and Tape Please do not staple Fold and Tape
Fold and Tape Please do not staple Fold and Tape
NO POSTAGE
NECESSARY
IF MAILED IN THE
UNITED STATES
BUSINESS REPLY MAIL
FIRST-CLASS MAIL PERMIT NO. 40 ARMONK, NEW YORK
POSTAGE WILL BE PAID BY ADDRESSEE
IBM Corporation
Department H150/090
555 Bailey Ave.
San Jose, CA
95141-1099
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
Product Number: 5655-W67

Printed in USA
Enterprise PL/I for z/OS Library
GC14-7283
Licensed Program Specifications
GI11-9145
Programming Guide
GC14-7284
Compiler and Run-Time Migration Guide
SC14-7285
Language Reference
GC14-7286
Compile-Time Messages and Codes
GI11-9145-01

PLI Programming Guide

Uploaded by

Copyright:

Available Formats

PLI Programming Guide

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PLI Programming Guide

Uploaded by

Copyright:

Available Formats

Enterprise PL/I for z/OS

Enterprise PL/I for z/OS

Second Edition (September 2011)

as its run-time environment. It

platforms, and lowercase for UNIX

platforms. These items must be entered exactly as shown.

Multiple required or optional items

Enterprise Development Tools & Compilers Information Center

ABBREVIATIONS: AG, NAG

ABBREVIATIONS: STG, NSTG

Up to eight DD names may be specified. In order, they specify alternate DD names

Under INITAUTO(FULL), the compiler adds an INITIAL attribute to any

ABBREVIATION: IS, NIS

I Count all messages.

' upp_extraling_char '

For a PROCEDURE with either the OPTIONS(MAIN) or the

ABBREVIATIONS: OPT, NOPT

ABBREVIATIONS: OP, NOP

Chapter 1. Using compiler options and facilities 119

ABBREVIATIONS: AALL, ACICS, AMACRO, ASQL

The default is in the field DECIMAL ARITHMETIC on the Application

GRAPHIC and NOGRAPHIC are mutually exclusive options.

NO indicates conformance to DB2 rules.

The following constant declarations are generated by the SQL preprocessor.

1.4 in place of the prelinker

DD_DDNAME = " filespec "

Records can have one of the following formats:

The CTLASA option specifies American National Standard Vertical Carriage

or a program with the same interface. DFSORT is

Application Programming Guide, SC33-4035

Application Programming: Database Manager, SC26-8015

and character representation

Product Number: 5655-W67

You might also like