Kotlin Docs

Kotlin Language Documentation
Table of Contents
Overview 7
Using Kotlin for Server-side Development 7
Using Kotlin for Android Development 9
Kotlin JavaScript Overview 11
Kotlin/Native for Native 13
Coroutines for asynchronous programming and more 15
Multiplatform Programming 16
What's New in Kotlin 1.1 19
Standard library 45
Tooling 47
Getting Started 48
Basic Syntax 48
Idioms 54
Coding Conventions 59
Basics 76
Basic Types 76
Packages 85
Control Flow: if, when, for, while 87
Returns and Jumps 91
Classes and Objects 94
Classes and Inheritance 94
Properties and Fields 102
Interfaces 106
Visibility Modifiers 109
Extensions 112
2
Data Classes 118
Sealed Classes 120
Generics 121
Nested and Inner Classes 128
Enum Classes 129
Object Expressions and Declarations 131
Inline classes 136
Delegation 141
Delegated Properties 143
Functions and Lambdas 150
Functions 150
Higher-Order Functions and Lambdas 157
Inline Functions 165
Collections 169
Kotlin Collections Overview 169
Constructing Collections 174
Iterators 177
Ranges and Progressions 179
Sequences 181
Collection Operations Overview 184
Collection Transformations 186
Filtering 190
plus and minus Operators 193
Grouping 194
Retrieving Collection Parts 195
Retrieving Single Elements 198
Collection Ordering 201
Collection Aggregate Operations 204
Collection Write Operations 206
3
List Specific Operations 208
Set Specific Operations 213
Map Specific Operations 214
Multiplatform Programming 218
Platform-Specific Declarations 218
Building Multiplatform Projects with Gradle 220
Other 258
Destructuring Declarations 258
Type Checks and Casts: 'is' and 'as' 261
This Expression 265
Equality 266
Operator overloading 267
Null Safety 272
Exceptions 276
Annotations 279
Reflection 284
Scope Functions 289
Type-Safe Builders 297
Experimental API Markers 304
Reference 309
Keywords and Operators 309
Grammar 315
Java Interop 334
Calling Java code from Kotlin 334
Calling Kotlin from Java 348
JavaScript 359
Dynamic Type 359
Calling JavaScript from Kotlin 361
4
Calling Kotlin from JavaScript 364
JavaScript Modules 367
JavaScript Reflection 371
JavaScript DCE 372
Example 373
Native 374
Concurrency in Kotlin/Native 374
Immutability in Kotlin/Native 378
Kotlin/Native libraries 379
Advanced topics 380
Platform libraries 382
Kotlin/Native interoperability 383
Kotlin/Native interoperability with Swift/Objective-C 394
CocoaPods integration 399
Kotlin/Native Gradle plugin 402
Coroutines 417
Table of contents 417
Additional references 417
Coroutine basics 418
Cancellation and timeouts 424
Channels 429
Composing suspending functions 438
Coroutine context and dispatchers 443
Exception handling 453
Supervision 458
Select expression (experimental) 462
Shared mutable state and concurrency 469
Tools 475
Documenting Kotlin Code 475
5
Annotation Processing with Kotlin 479
Using Gradle 484
Using Maven 493
Using Ant 501
Kotlin and OSGi 505
Compiler Plugins 507
Code Style Migration Guide 514
Evolution 517
Kotlin Evolution 517
Stability of Different Components 523
Compatibility Guide for Kotlin 1.3 524
FAQ 537
FAQ 537
Comparison to Java Programming Language 541
6
Overview
Using Kotlin for Server-side Development
Kotlin is a great t for developing server-side applications, allowing you to write concise and
expressive code while maintaining full compatibility with existing Java-based technology stacks
and a smooth learning curve:
— Expressiveness: Kotlin's innovative language features, such as its support for type-safe
builders and delegated properties, help build powerful and easy-to-use abstractions.
— Scalability: Kotlin's support for coroutines helps build server-side applications that scale to
massive numbers of clients with modest hardware requirements.
— Interoperability: Kotlin is fully compatible with all Java-based frameworks, which lets you stay
on your familiar technology stack while reaping the bene ts of a more modern language.
— Migration: Kotlin supports gradual, step by step migration of large codebases from Java to
Kotlin. You can start writing new code in Kotlin while keeping older parts of your system in
Java.
— Tooling: In addition to great IDE support in general, Kotlin o ers framework-speci c tooling
(for example, for Spring) in the plugin for IntelliJ IDEA Ultimate.
— Learning Curve: For a Java developer, getting started with Kotlin is very easy. The automated
Java to Kotlin converter included in the Kotlin plugin helps with the rst steps. Kotlin Koans
o er a guide through the key features of the language with a series of interactive exercises.
Frameworks for Server-side Development with Kotlin

— Spring makes use of Kotlin's language features to o er more concise APIs, starting with
version 5.0. The online project generator allows you to quickly generate a new project in
Kotlin.
— Vert.x, a framework for building reactive Web applications on the JVM, o ers dedicated
support for Kotlin, including full documentation.
— Ktor is a framework built by JetBrains for creating Web applications in Kotlin, making use of
coroutines for high scalability and o ering an easy-to-use and idiomatic API.
— kotlinx.html is a DSL that can be used to build HTML in a Web application. It serves as an
alternative to traditional templating systems such as JSP and FreeMarker.
7
— The available options for persistence include direct JDBC access, JPA, as well as using NoSQL
databases through their Java drivers. For JPA, the kotlin-jpa compiler plugin adapts Kotlin-
compiled classes to the requirements of the framework.
Deploying Kotlin Server-side Applications

Kotlin applications can be deployed into any host that supports Java Web applications, including
Amazon Web Services, Google Cloud Platform and more.
To deploy Kotlin applications on Heroku, you can follow the o cial Heroku tutorial.
AWS Labs provides a sample project showing the use of Kotlin for writing AWS Lambda functions.
Google Cloud Platform o ers a series of tutorials for deploying Kotlin applications to GCP, both
for Ktor and App Engine and Spring and App engine. In addition there is an interactive code lab
for deploying a Kotlin Spring application.
Users of Kotlin on the Server Side

Corda is an open-source distributed ledger platform, supported by major banks, and built
entirely in Kotlin.
JetBrains Account, the system responsible for the entire license sales and validation process at
JetBrains, is written in 100% Kotlin and has been running in production since 2015 with no major
issues.
Next Steps
— The Creating Web Applications with Http Servlets and Creating a RESTful Web Service with
Spring Boot tutorials show you how you can build and run very small Web applications in
Kotlin.
— For a more in-depth introduction to the language, check out the reference documentation on
this site and Kotlin Koans.
8
Using Kotlin for Android Development
Kotlin is a great t for developing Android applications, bringing all of the advantages of a
modern language to the Android platform without introducing any new restrictions:
— Compatibility: Kotlin is fully compatible with JDK 6, ensuring that Kotlin applications can run
on older Android devices with no issues. The Kotlin tooling is fully supported in Android
Studio and compatible with the Android build system.
— Performance: A Kotlin application runs as fast as an equivalent Java one, thanks to very
similar bytecode structure. With Kotlin's support for inline functions, code using lambdas
often runs even faster than the same code written in Java.
— Interoperability: Kotlin is 100% interoperable with Java, allowing to use all existing Android
libraries in a Kotlin application. This includes annotation processing, so databinding and
Dagger work too.
— Footprint: Kotlin has a very compact runtime library, which can be further reduced through
the use of ProGuard. In a real application, the Kotlin runtime adds only a few hundred
methods and less than 100K to the size of the .apk le.
— Compilation Time: Kotlin supports e cient incremental compilation, so while there's some
additional overhead for clean builds, incremental builds are usually as fast or faster than with
Java.
— Learning Curve: For a Java developer, getting started with Kotlin is very easy. The automated
Java to Kotlin converter included in the Kotlin plugin helps with the rst steps. Kotlin Koans
o er a guide through the key features of the language with a series of interactive exercises.
Kotlin for Android Case Studies

Kotlin has been successfully adopted by major companies, and a few of them have shared their
experiences:
— Pinterest has successfully introduced Kotlin into their application, used by 150M people every
month.
— Basecamp's Android app is 100% Kotlin code, and they report a huge di erence in
programmer happiness and great improvements in work quality and speed.
— Keepsafe's App Lock app has also been converted to 100% Kotlin, leading to a 30% decrease
in source line count and 10% decrease in method count.
Tools for Android Development

The Kotlin team o ers a set of tools for Android development that goes beyond the standard
language features:
— Kotlin Android Extensions is a compiler extension that allows you to get rid of
findViewById() calls in your code and to replace them with synthetic compiler-generated
properties.
9
— Anko is a library providing a set of Kotlin-friendly wrappers around the Android APIs, as well
as a DSL that lets you replace your layout .xml les with Kotlin code.
Next Steps
— Download and install Android Studio which includes Kotlin support out of the box.
— Follow the Getting Started with Android and Kotlin tutorial to create your rst Kotlin
application.
— For a more in-depth introduction, check out the reference documentation on this site and
Kotlin Koans.
— Another great resource is Kotlin for Android Developers, a book that guides you step by step
through the process of creating a real Android application in Kotlin.
— Check out Google's sample projects written in Kotlin.
10
Kotlin JavaScript Overview
Kotlin provides the ability to target JavaScript. It does so by transpiling Kotlin to JavaScript. The
current implementation targets ECMAScript 5.1 but there are plans to eventually target
ECMAScript 2015 as well.
When you choose the JavaScript target, any Kotlin code that is part of the project as well as the
standard library that ships with Kotlin is transpiled to JavaScript. However, this excludes the JDK
and any JVM or Java framework or library used. Any le that is not Kotlin will be ignored during
compilation.
The Kotlin compiler tries to comply with the following goals:
— Provide output that is optimal in size

— Provide output that is readable JavaScript
— Provide interoperability with existing module systems
— Provide the same functionality in the standard library whether targeting JavaScript or the JVM
(to the largest possible degree).
How it can be used

You may want to compile Kotlin to JavaScript in the following scenarios:
— Creating Kotlin code that targets client-side JavaScript
— Interacting with DOM elements. Kotlin provides a series of statically typed interfaces to
interact with the Document Object Model, allowing creation and update of DOM elements.
— Interacting with graphics such as WebGL. You can use Kotlin to create graphical
elements on a web page using WebGL.
— Creating Kotlin code that targets server-side JavaScript
— Working with server-side technology. You can use Kotlin to interact with server-side
JavaScript such as Node.js
Kotlin can be used together with existing third-party libraries and frameworks, such as jQuery or
React. To access third-party frameworks with a strongly-typed API, you can convert TypeScript
de nitions from the De nitely Typed type de nitions repository to Kotlin using the ts2kt tool.
Alternatively, you can use the dynamic type to access any framework without strong typing.
JetBrains develops and maintains several tools speci cally for the React community: React
bindings as well as Create React Kotlin App. The latter helps you start building React apps with
Kotlin with no build con guration.
Kotlin is compatible with CommonJS, AMD and UMD, making interaction with di erent module
systems straightforward.
11
Getting Started with Kotlin to JavaScript
To nd out how to start using Kotlin for JavaScript, please refer to the tutorials.
12
Kotlin/Native for Native
Kotlin/Native is a technology for compiling Kotlin code to native binaries, which can run without a
virtual machine. It is an LLVM based backend for the Kotlin compiler and native implementation
of the Kotlin standard library.
Why Kotlin/Native?
Kotlin/Native is primarily designed to allow compilation for platforms where virtual machines are
not desirable or possible, for example, embedded devices or iOS. It solves the situations when a
developer needs to produce a self-contained program that does not require an additional
runtime or virtual machine.
Target Platforms
Kotlin/Native supports the following platforms:
— iOS (arm32, arm64, simulator x86_64)

— MacOS (x86_64)
— Android (arm32, arm64)
— Windows (mingw x86_64, x86)
— Linux (x86_64, arm32, MIPS, MIPS little endian, Raspberry Pi)
— WebAssembly (wasm32)
Interoperability
Kotlin/Native supports two-way interoperability with the Native world. On the one hand, the
compiler creates:
— an executable for many platforms

— a static library or dynamic library with C headers for C/C++ projects
— an Apple framework for Swift and Objective-C projects
On the other hand, Kotlin/Native supports interoperability to use existing libraries directly from
Kotlin/Native:
— static or dynamic C Libraries

— C, Swift, and Objective-C frameworks
It is easy to include a compiled Kotlin code into existing projects written in C, C++, Swift,
Objective-C, and other languages. It is also easy to use existing native code, static or dynamic C
libraries, Swift/Objective-C frameworks, graphical engines, and anything else directly from
Kotlin/Native.
13
Kotlin/Native libraries help to share Kotlin code between projects. POSIX, gzip, OpenGL, Metal,
Foundation, and many other popular libraries and Apple frameworks are pre-imported and
included as Kotlin/Native libraries into the compiler package.
Sharing Code between Platforms

Multiplatform projects are supported between di erent Kotlin and Kotlin/Native targets. This is
the way to share common Kotlin code between many platforms, including Android, iOS, server-
side, JVM, client-side, JavaScript, CSS, and native.
Multiplatform libraries provide the necessary APIs for the common Kotlin code and help to
develop shared parts of a project in Kotlin code once and share it with all of the target platforms.
How to Start
Tutorials and Documentation
New to Kotlin? Take a look at the Getting Started page.
Suggested documentation pages:
— C interop
— Swift/Objective-C interop
Recommended tutorials:
— A basic Kotlin/Native application

— Multiplatform Project: iOS and Android
— Types mapping between C and Kotlin/Native
— Kotlin/Native as a Dynamic Library
— Kotlin/Native as an Apple Framework
Example Projects
— Kotlin/Native sources and examples

— KotlinConf app
— KotlinConf Spinner app
— Kotlin/Native sources and examples (.tgz)
— Kotlin/Native sources and examples (.zip)
Even more examples are on GitHub.
14
Coroutines for asynchronous programming and more
Asynchronous or non-blocking programming is the new reality. Whether we're creating server-
side, desktop or mobile applications, it's important that we provide an experience that is not only
uid from the user's perspective, but scalable when needed.
There are many approaches to this problem, and in Kotlin we take a very exible one by
providing Coroutine support at the language level and delegating most of the functionality to
libraries, much in line with Kotlin's philosophy.
As a bonus, coroutines not only open the doors to asynchronous programming, but also provide
a wealth of other possibilities such as concurrency, actors, etc.
How to Start
Selected documentation pages:
— Coroutines Guide
— Basics
— Channels
— Coroutine Context and Dispatchers
— Shared Mutable State and Concurrency
— Your rst coroutine with Kotlin

— Asynchronous Programming
Example Projects
— kotlinx.coroutines Examples and Sources

— KotlinConf app
Even more examples are on GitHub
15
Multiplatform Programming
Multiplatform projects are an experimental feature in Kotlin 1.2 and 1.3. All of the
language and tooling features described in this document are subject to change in future
Kotlin versions.
Working on all platforms is an explicit goal for Kotlin, but we see it as a premise to a much more
important goal: sharing code between platforms. With support for JVM, Android, JavaScript, iOS,
Linux, Windows, Mac and even embedded systems like STM32, Kotlin can handle any and all
components of a modern application. And this brings the invaluable bene t of reuse for code
and expertise, saving the e ort for tasks more challenging than implementing everything twice or
multiple times.
How it works
Overall, multiplatform is not about compiling all code for all platforms. This model has its obvious
limitations, and we understand that modern applications need access to unique features of the
platforms they are running on. Kotlin doesn't limit you to the common subset of all APIs in the
world. Every component can share as much code as needed with others but can access platform
APIs at any time through the expect/actual mechanism provided by the language.
Here's an example of code sharing and interaction between the common and platform logic in a
minimalistic logging framework. The common code would look like this:
enum class LogLevel {

DEBUG, WARN, ERROR
}
internal expect fun writeLogMessage(message: String, logLevel: LogLevel)
fun logDebug(message: String) = writeLogMessage(message, LogLevel.DEBUG)

fun logWarn(message: String) = writeLogMessage(message, LogLevel.WARN)
fun logError(message: String) = writeLogMessage(message, LogLevel.ERROR)
├ compiled for all platforms
├ expected platform-specific API
├ expected API can be used in the common code
It expects the targets to provide platform-speci c implementations for writeLogMessage , and

the common code can now use this declaration without any consideration of how it is
implemented.
On the JVM, one could provide an implementation that writes the log to the standard output:
internal actual fun writeLogMessage(message: String, logLevel: LogLevel) {

println("[$logLevel]: $message")
}
16
In the JavaScript world, a completely di erent set of APIs is availiable, so one could instead
implement logging to the console:
internal actual fun writeLogMessage(message: String, logLevel: LogLevel) {

when (logLevel) {
LogLevel.DEBUG -> console.log(message)
LogLevel.WARN -> console.warn(message)
LogLevel.ERROR -> console.error(message)
}
}
In 1.3 we reworked the entire multiplatform model. The new DSL we have for describing
multiplatform Gradle projects is much more exible, and we'll keep working on it to make project
con guration straightforward.
Multiplatform Libraries
Common code can rely on a set of libraries that cover everyday tasks such as HTTP, serialization,
and managing coroutines. Also, an extensive standard library is available on all platforms.
You can always write your own library providing a common API and implementing it di erently
on every platform.
Use cases
Android — iOS
Sharing code between mobile platforms is one of the major Kotlin Multiplatform use cases, and it
is now possible to build mobile applications with parts of the code, such as business logic,
connectivity, and more, shared between Android and iOS.
See: Multiplatform Project: iOS and Android
Client — Server
Another scenario when code sharing may bring bene ts is a connected application where the
logic may be reused on both the server and the client side running in the browser. This is
covered by Kotlin Multiplatform as well.
The Ktor framework is suitable for building asynchronous servers and clients in connected
systems.
How to start
Suggested documentation pages:
17
— Setting up a Multiplatform Project
— Platform-Speci c Declarations
— Multiplatform Kotlin Library

— Multiplatform Project: iOS and Android
Example Projects
— KotlinConf app
— KotlinConf Spinner app
Even more examples are on GitHub
18
What's New in Kotlin 1.1
Table of Contents
— Coroutines
— Other language features
— Standard library
— JVM backend
— JavaScript backend
JavaScript
Starting with Kotlin 1.1, the JavaScript target is no longer considered experimental. All language
features are supported, and there are many new tools for integration with the front-end
development environment. See below for a more detailed list of changes.
Coroutines (experimental)
The key new feature in Kotlin 1.1 is coroutines, bringing the support of async / await , yield
and similar programming patterns. The key feature of Kotlin's design is that the implementation
of coroutine execution is part of the libraries, not the language, so you aren't bound to any
speci c programming paradigm or concurrency library.
A coroutine is e ectively a light-weight thread that can be suspended and resumed later.
Coroutines are supported through suspending functions: a call to such a function can potentially
suspend a coroutine, and to start a new coroutine we usually use an anonymous suspending
functions (i.e. suspending lambdas).
Let's look at async / await which is implemented in an external library, kotlinx.coroutines:
// runs the code in the background thread pool

fun asyncOverlay() = async(CommonPool) {
// start two async operations
val original = asyncLoadImage("original")
val overlay = asyncLoadImage("overlay")
// and then apply overlay to both results
applyOverlay(original.await(), overlay.await())
}
// launches new coroutine in UI context

launch(UI) {
// wait for async overlay to complete
val image = asyncOverlay().await()
// and then show it in UI
showImage(image)
}
Here, async { ... } starts a coroutine and, when we use await() , the execution of the
coroutine is suspended while the operation being awaited is executed, and is resumed (possibly
on a di erent thread) when the operation being awaited completes.
19
The standard library uses coroutines to support lazily generated sequences with yield and
yieldAll functions. In such a sequence, the block of code that returns sequence elements is
suspended after each element has been retrieved, and resumed when the next element is
requested. Here's an example:
val seq = buildSequence {

for (i in 1..5) {
// yield a square of i
yield(i * i)
}
// yield a range
yieldAll(26..28)
}
// print the sequence

intln(seq.toList())
Run the code above to see the result. Feel free to edit it and run again!
For more information, please refer to the coroutine documentation and tutorial.
Note that coroutines are currently considered an experimental feature, meaning that the Kotlin
team is not committing to supporting the backwards compatibility of this feature after the nal
1.1 release.
Other Language Features
Type aliases
A type alias allows you to de ne an alternative name for an existing type. This is most useful for
generic types such as collections, as well as for function types. Here is an example:
typealias OscarWinners = Map<String, String>
fun countLaLaLand(oscarWinners: OscarWinners) =

oscarWinners.count { it.value.contains("La La Land") }
// Note that the type names (initial and the type alias) are interchangeable:
fun checkLaLaLandIsTheBestMovie(oscarWinners: Map<String, String>) =
oscarWinners["Best picture"] == "La La Land"
See the documentation and KEEP for more details.
Bound callable references
You can now use the :: operator to get a member reference pointing to a method or property
of a speci c object instance. Previously this could only be expressed with a lambda. Here's an
example:
val numberRegex = "\\d+".toRegex()

val numbers = listOf("abc", "123", "456").filter(numberRegex::matches)
20
Read the documentation and KEEP for more details.
Sealed and data classes
Kotlin 1.1 removes some of the restrictions on sealed and data classes that were present in Kotlin
1.0. Now you can de ne subclasses of a top-level sealed class on the top level in the same le,
and not just as nested classes of the sealed class. Data classes can now extend other classes. This
can be used to de ne a hierarchy of expression classes nicely and cleanly:
sealed class Expr
data class Const(val number: Double) : Expr()

data class Sum(val e1: Expr, val e2: Expr) : Expr()
object NotANumber : Expr()
fun eval(expr: Expr): Double = when (expr) {

is Const -> expr.number
is Sum -> eval(expr.e1) + eval(expr.e2)
NotANumber -> Double.NaN
}
val e = eval(Sum(Const(1.0), Const(2.0)))
Read the documentation or sealed class and data class KEEPs for more detail.
Destructuring in lambdas
You can now use the destructuring declaration syntax to unpack the arguments passed to a
lambda. Here's an example:
val map = mapOf(1 to "one", 2 to "two")

// before
println(map.mapValues { entry ->
val (key, value) = entry
"$key -> $value!"
})
// now
println(map.mapValues { (key, value) -> "$key -> $value!" })
Underscores for unused parameters
For a lambda with multiple parameters, you can use the _ character to replace the names of the
parameters you don't use:
map.forEach { _, value -> println("$value!") }
This also works in destructuring declarations:
val (_, status) = getResult()
Read the KEEP for more details.
21
Underscores in numeric literals
Just as in Java 8, Kotlin now allows to use underscores in numeric literals to separate groups of
digits:
val oneMillion = 1_000_000

val hexBytes = 0xFF_EC_DE_5E
val bytes = 0b11010010_01101001_10010100_10010010
Shorter syntax for properties
For properties with the getter de ned as an expression body, the property type can now be
omitted:
data class Person(val name: String, val age: Int) {

val isAdult get() = age >= 20 // Property type inferred to be 'Boolean'
Inline property accessors
You can now mark property accessors with the inline modi er if the properties don't have a
backing eld. Such accessors are compiled in the same way as inline functions.
public val <T> List<T>.lastIndex: Int

inline get() = this.size - 1
You can also mark the entire property as inline - then the modi er is applied to both
accessors.
Local delegated properties
You can now use the delegated property syntax with local variables. One possible use is de ning
a lazily evaluated local variable:
val answer by lazy {

println("Calculating the answer...")
42
}
if (needAnswer()) { // returns the random value
println("The answer is $answer.") // answer is calculated at this point
}
else {
println("Sometimes no answer is the answer...")
}
22
Interception of delegated property binding
For delegated properties, it is now possible to intercept delegate to property binding using the
provideDelegate operator. For example, if we want to check the property name before
binding, we can write something like this:
class ResourceLoader<T>(id: ResourceID<T>) {

operator fun provideDelegate(thisRef: MyUI, prop: KProperty<*>):
ReadOnlyProperty<MyUI, T> {
checkProperty(thisRef, prop.name)
... // property creation
}
private fun checkProperty(thisRef: MyUI, name: String) { ... }

}
fun <T> bindResource(id: ResourceID<T>): ResourceLoader<T> { ... }
class MyUI {
val image by bindResource(ResourceID.image_id)
val text by bindResource(ResourceID.text_id)
}
The provideDelegate method will be called for each property during the creation of a MyUI
instance, and it can perform the necessary validation right away.
Read the documentation for more details.
Generic enum value access
It is now possible to enumerate the values of an enum class in a generic way.
enum class RGB { RED, GREEN, BLUE }
inline fun <reified T : Enum<T>> printAllValues() {

print(enumValues<T>().joinToString { it.name })
}
Scope control for implicit receivers in DSLs
The @DslMarker annotation allows to restrict the use of receivers from outer scopes in a DSL
context. Consider the canonical HTML builder example:
table {
tr {
td { + "Text" }
}
}
In Kotlin 1.0, code in the lambda passed to td has access to three implicit receivers: the one
passed to table , to tr and to td . This allows you to call methods that make no sense in the
context - for example to call tr inside td and thus to put a <tr> tag in a <td> .
23
In Kotlin 1.1, you can restrict that, so that only methods de ned on the implicit receiver of td
will be available inside the lambda passed to td . You do that by de ning your annotation
marked with the @DslMarker meta-annotation and applying it to the base class of the tag
classes.
rem operator
The mod operator is now deprecated, and rem is used instead. See this issue for motivation.
Standard library
String to number conversions
There is a bunch of new extensions on the String class to convert it to a number without throwing
an exception on invalid number: String.toIntOrNull(): Int? ,
String.toDoubleOrNull(): Double? etc.
val port = System.getenv("PORT")?.toIntOrNull() ?: 80
Also integer conversion functions, like Int.toString() , String.toInt() ,

String.toIntOrNull() , each got an overload with radix parameter, which allows to specify
the base of conversion (2 to 36).
onEach()
onEach is a small, but useful extension function for collections and sequences, which allows to
perform some action, possibly with side-e ects, on each element of the collection/sequence in a
chain of operations. On iterables it behaves like forEach but also returns the iterable instance
further. And on sequences it returns a wrapping sequence, which applies the given action lazily
as the elements are being iterated.
inputDir.walk()
.filter { it.isFile && it.name.endsWith(".txt") }
.onEach { println("Moving $it to $outputDir") }
.forEach { moveFile(it, File(outputDir, it.toRelativeString(inputDir))) }
also(), takeIf() and takeUnless()
These are three general-purpose extension functions applicable to any receiver.
also is like apply : it takes the receiver, does some action on it, and returns that receiver. The
di erence is that in the block inside apply the receiver is available as this , while in the block
inside also it's available as it (and you can give it another name if you want). This comes
handy when you do not want to shadow this from the outer scope:
24
fun Block.copy() = Block().also {
it.content = this.content
}
takeIf is like filter for a single value. It checks whether the receiver meets the predicate,
and returns the receiver, if it does or null if it doesn't. Combined with an elvis-operator and
early returns it allows to write constructs like:
val outDirFile = File(outputDir.path).takeIf { it.exists() } ?: return false

// do something with existing outDirFile
val index = input.indexOf(keyword).takeIf { it >= 0 } ?: error("keyword not found")

// do something with index of keyword in input string, given that it's found
takeUnless is the same as takeIf , but it takes the inverted predicate. It returns the receiver
when it doesn't meet the predicate and null otherwise. So one of the examples above could be
rewritten with takeUnless as following:
val index = input.indexOf(keyword).takeUnless { it < 0 } ?: error("keyword not found")
It is also convenient to use when you have a callable reference instead of the lambda:
val result = string.takeUnless(String::isEmpty)
groupingBy()
This API can be used to group a collection by key and fold each group simultaneously. For
example, it can be used to count the number of words starting with each letter:
val frequencies = words.groupingBy { it.first() }.eachCount()
Map.toMap() and Map.toMutableMap()
These functions can be used for easy copying of maps:
class ImmutablePropertyBag(map: Map<String, Any>) {

private val mapCopy = map.toMap()
}
Map.minus(key)
The operator plus provides a way to add key-value pair(s) to a read-only map producing a new
map, however there was not a simple way to do the opposite: to remove a key from the map you
have to resort to less straightforward ways to like Map.filter() or Map.filterKeys() .
Now the operator minus lls this gap. There are 4 overloads available: for removing a single
key, a collection of keys, a sequence of keys and an array of keys.
25
val map = mapOf("key" to 42)
val emptyMap = map - "key"
minOf() and maxOf()
These functions can be used to nd the lowest and greatest of two or three given values, where
values are primitive numbers or Comparable objects. There is also an overload of each function
that take an additional Comparator instance, if you want to compare objects that are not
comparable themselves.
val list1 = listOf("a", "b")

val list2 = listOf("x", "y", "z")
val minSize = minOf(list1.size, list2.size)
val longestList = maxOf(list1, list2, compareBy { it.size })
Array-like List instantiation functions
Similar to the Array constructor, there are now functions that create List and MutableList
instances and initialize each element by calling a lambda:
val squares = List(10) { index -> index * index }

val mutable = MutableList(10) { 0 }
Map.getValue()
This extension on Map returns an existing value corresponding to the given key or throws an
exception, mentioning which key was not found. If the map was produced with withDefault ,
this function will return the default value instead of throwing an exception.
val map = mapOf("key" to 42)

// returns non-nullable Int value 42
val value: Int = map.getValue("key")
val mapWithDefault = map.withDefault { k -> k.length }

// returns 4
val value2 = mapWithDefault.getValue("key2")
// map.getValue("anotherKey") // <- this will throw NoSuchElementException
Abstract collections
These abstract classes can be used as base classes when implementing Kotlin collection classes.
For implementing read-only collections there are AbstractCollection , AbstractList ,
AbstractSet and AbstractMap , and for mutable collections there are
AbstractMutableCollection , AbstractMutableList , AbstractMutableSet and
AbstractMutableMap . On JVM these abstract mutable collections inherit most of their
functionality from JDK's abstract collections.
26
Array manipulation functions
The standard library now provides a set of functions for element-by-element operations on
arrays: comparison ( contentEquals and contentDeepEquals ), hash code calculation
( contentHashCode and contentDeepHashCode ), and conversion to a string
( contentToString and contentDeepToString ). They're supported both for the JVM (where
they act as aliases for the corresponding functions in java.util.Arrays ) and for JS (where the
implementation is provided in the Kotlin standard library).
val array = arrayOf("a", "b", "c")

println(array.toString()) // JVM implementation: type-and-hash gibberish
println(array.contentToString()) // nicely formatted as list
JVM Backend
Java 8 bytecode support
Kotlin has now the option of generating Java 8 bytecode ( -jvm-target 1.8 command line
option or the corresponding options in Ant/Maven/Gradle). For now this doesn't change the
semantics of the bytecode (in particular, default methods in interfaces and lambdas are
generated exactly as in Kotlin 1.0), but we plan to make further use of this later.
Java 8 standard library support
There are now separate versions of the standard library supporting the new JDK APIs added in
Java 7 and 8. If you need access to the new APIs, use kotlin-stdlib-jre7 and kotlin-
stdlib-jre8 maven artifacts instead of the standard kotlin-stdlib . These artifacts are tiny
extensions on top of kotlin-stdlib and they bring it to your project as a transitive
dependency.
Parameter names in the bytecode
Kotlin now supports storing parameter names in the bytecode. This can be enabled using the -
java-parameters command line option.
Constant inlining
The compiler now inlines values of const val properties into the locations where they are
used.
Mutable closure variables
27
The box classes used for capturing mutable closure variables in lambdas no longer have volatile
elds. This change improves performance, but can lead to new race conditions in some rare
usage scenarios. If you're a ected by this, you need to provide your own synchronization for
accessing the variables.
javax.script support
Kotlin now integrates with the javax.script API (JSR-223). The API allows to evaluate snippets of
code at runtime:
val engine = ScriptEngineManager().getEngineByExtension("kts")!!

engine.eval("val x = 3")
println(engine.eval("x + 2")) // Prints out 5
See here for a larger example project using the API.
kotlin.re ect.full
To prepare for Java 9 support, the extension functions and properties in the kotlin-
reflect.jar library have been moved to the package kotlin.reflect.full . The names in
the old package ( kotlin.reflect ) are deprecated and will be removed in Kotlin 1.2. Note that
the core re ection interfaces (such as KClass ) are part of the Kotlin standard library, not
kotlin-reflect , and are not a ected by the move.
JavaScript Backend
Uni ed standard library
A much larger part of the Kotlin standard library can now be used from code compiled to
JavaScript. In particular, key classes such as collections ( ArrayList , HashMap etc.), exceptions
( IllegalArgumentException etc.) and a few others ( StringBuilder , Comparator ) are
now de ned under the kotlin package. On the JVM, the names are type aliases for the
corresponding JDK classes, and on the JS, the classes are implemented in the Kotlin standard
library.
Better code generation
JavaScript backend now generates more statically checkable code, which is friendlier to JS code
processing tools, like mini ers, optimisers, linters, etc.
The external modi er
28
If you need to access a class implemented in JavaScript from Kotlin in a typesafe way, you can
write a Kotlin declaration using the external modi er. (In Kotlin 1.0, the @native annotation
was used instead.) Unlike the JVM target, the JS one permits to use external modi er with classes
and properties. For example, here's how you can declare the DOM Node class:
external class Node {

val firstChild: Node
fun appendChild(child: Node): Node
fun removeChild(child: Node): Node
// etc
}
Improved import handling
You can now describe declarations which should be imported from JavaScript modules more
precisely. If you add the @JsModule("<module-name>") annotation on an external
declaration it will be properly imported to a module system (either CommonJS or AMD) during
the compilation. For example, with CommonJS the declaration will be imported via
require(...) function. Additionally, if you want to import a declaration either as a module or
as a global JavaScript object, you can use the @JsNonModule annotation.
For example, here's how you can import JQuery into a Kotlin module:
external interface JQuery {

fun toggle(duration: Int = definedExternally): JQuery
fun click(handler: (Event) -> Unit): JQuery
}
@JsModule("jquery")
@JsNonModule
@JsName("$")
external fun jquery(selector: String): JQuery
In this case, JQuery will be imported as a module named jquery . Alternatively, it can be used as
a $-object, depending on what module system Kotlin compiler is con gured to use.
You can use these declarations in your application like this:
fun main(args: Array<String>) {

jquery(".toggle-button").click {
jquery(".toggle-panel").toggle(300)
}
}
29
Table of Contents
— Multiplatform projects
— Other language features
— Standard library
— JVM backend
— JavaScript backend
Multiplatform Projects (experimental)

Multiplatform projects are a new experimental feature in Kotlin 1.2, allowing you to reuse code
between target platforms supported by Kotlin – JVM, JavaScript and (in the future) Native. In a
multiplatform project, you have three kinds of modules:
— A common module contains code that is not speci c to any platform, as well as declarations
without implementation of platform-dependent APIs.
— A platform module contains implementations of platform-dependent declarations in the
common module for a speci c platform, as well as other platform-dependent code.
— A regular module targets a speci c platform and can either be a dependency of platform
modules or depend on platform modules.
When you compile a multiplatform project for a speci c platform, the code for both the common
and platform-speci c parts is generated.
A key feature of the multiplatform project support is the possibility to express dependencies of
common code on platform-speci c parts through expected and actual declarations. An
expected declaration speci es an API (class, interface, annotation, top-level declaration etc.). An
actual declaration is either a platform-dependent implementation of the API or a typealias
referring to an existing implementation of the API in an external library. Here's an example:
In common code:
// expected platform-specific API:

expect fun hello(world: String): String
fun greet() {
// usage of the expected API:
val greeting = hello("multi-platform world")
println(greeting)
}
expect class URL(spec: String) {

open fun getHost(): String
open fun getPath(): String
}
In JVM platform code:
30
actual fun hello(world: String): String =
"Hello, $world, on the JVM platform!"
// using existing platform-specific implementation:

actual typealias URL = java.net.URL
See the documentation for details and steps to build a multiplatform project.
Other Language Features
Array literals in annotations
Starting with Kotlin 1.2, array arguments for annotations can be passed with the new array literal
syntax instead of the arrayOf function:
@CacheConfig(cacheNames = ["books", "default"])

public class BookRepositoryImpl {
// ...
}
The array literal syntax is constrained to annotation arguments.
Lateinit top-level properties and local variables
The lateinit modi er can now be used on top-level properties and local variables. The latter
can be used, for example, when a lambda passed as a constructor argument to one object refers
to another object which has to be de ned later:
class Node<T>(val value: T, val next: () -> Node<T>)

// A cycle of three nodes:
lateinit var third: Node<Int>
val second = Node(2, next = { third })

val first = Node(1, next = { second })
third = Node(3, next = { first })
val nodes = generateSequence(first) { it.next() }

println("Values in the cycle: ${nodes.take(7).joinToString { it.value.toString() }},
...")
}
Checking whether a lateinit var is initialized
You can now check whether a lateinit var has been initialized using isInitialized on the
property reference:
println("isInitialized before assignment: " + this::lateinitVar.isInitialized)

lateinitVar = "value"
println("isInitialized after assignment: " + this::lateinitVar.isInitialized)
31
Inline functions with default functional parameters
Inline functions are now allowed to have default values for their inlined functional parameters:
inline fun <E> Iterable<E>.strings(transform: (E) -> String = { it.toString() }) =

map { transform(it) }
val defaultStrings = listOf(1, 2, 3).strings()

val customStrings = listOf(1, 2, 3).strings { "($it)" }
Information from explicit casts is used for type inference
The Kotlin compiler can now use information from type casts in type inference. If you’re calling a
generic method that returns a type parameter T and casting the return value to a speci c type
Foo , the compiler now understands that T for this call needs to be bound to the type Foo .
This is particularly important for Android developers, since the compiler can now correctly
analyze generic findViewById calls in Android API level 26:
val button = findViewById(R.id.button) as Button
Smart cast improvements
When a variable is assigned from a safe call expression and checked for null, the smart cast is
now applied to the safe call receiver as well:
val firstChar = (s as? CharSequence)?.firstOrNull()

if (firstChar != null)
return s.count { it == firstChar } // s: Any is smart cast to CharSequence
val firstItem = (s as? Iterable<*>)?.firstOrNull()

if (firstItem != null)
return s.count { it == firstItem } // s: Any is smart cast to Iterable<*>
Also, smart casts in a lambda are now allowed for local variables that are only modi ed before
the lambda:
val flag = args.size == 0

var x: String? = null
if (flag) x = "Yahoo!"
run {
if (x != null) {
println(x.length) // x is smart cast to String
}
}
Support for ::foo as a shorthand for this::foo
32
A bound callable reference to a member of this can now be written without explicit receiver,
::foo instead of this::foo . This also makes callable references more convenient to use in
lambdas where you refer to a member of the outer receiver.
Breaking change: sound smart casts after try blocks
Earlier, Kotlin used assignments made inside a try block for smart casts after the block, which
could break type- and null-safety and lead to runtime failures. This release xes this issue,
making the smart casts more strict, but breaking some code that relied on such smart casts.
To switch to the old smart casts behavior, pass the fallback ag -Xlegacy-smart-cast-
after-try as the compiler argument. It will become deprecated in Kotlin 1.3.
Deprecation: data classes overriding copy
When a data class derived from a type that already had the copy function with the same
signature, the copy implementation generated for the data class used the defaults from the
supertype, leading to counter-intuitive behavior, or failed at runtime if there were no default
parameters in the supertype.
Inheritance that leads to a copy con ict has become deprecated with a warning in Kotlin 1.2 and
will be an error in Kotlin 1.3.
Deprecation: nested types in enum entries
Inside enum entries, de ning a nested type that is not an inner class has been deprecated
due to issues in the initialization logic. This causes a warning in Kotlin 1.2 and will become an
error in Kotlin 1.3.
Deprecation: single named argument for vararg
For consistency with array literals in annotations, passing a single item for a vararg parameter in
the named form ( foo(items = i) ) has been deprecated. Please use the spread operator with
the corresponding array factory functions:
foo(items = *intArrayOf(1))
There is an optimization that removes redundant arrays creation in such cases, which prevents
performance degradation. The single-argument form produces warnings in Kotlin 1.2 and is to be
dropped in Kotlin 1.3.
Deprecation: inner classes of generic classes extending Throwable
33
Inner classes of generic types that inherit from Throwable could violate type-safety in a throw-
catch scenario and thus have been deprecated, with a warning in Kotlin 1.2 and an error in Kotlin
1.3.
Deprecation: mutating backing eld of a read-only property
Mutating the backing eld of a read-only property by assigning field = ... in the custom
getter has been deprecated, with a warning in Kotlin 1.2 and an error in Kotlin 1.3.
Standard Library
Kotlin standard library artifacts and split packages
The Kotlin standard library is now fully compatible with the Java 9 module system, which forbids
split packages (multiple jar les declaring classes in the same package). In order to support that,
new artifacts kotlin-stdlib-jdk7 and kotlin-stdlib-jdk8 are introduced, which
replace the old kotlin-stdlib-jre7 and kotlin-stdlib-jre8 .
The declarations in the new artifacts are visible under the same package names from the Kotlin
point of view, but have di erent package names for Java. Therefore, switching to the new artifacts
will not require any changes to your source code.
Another change made to ensure compatibility with the new module system is removing the
deprecated declarations in the kotlin.reflect package from the kotlin-reflect library.
If you were using them, you need to switch to using the declarations in the
kotlin.reflect.full package, which is supported since Kotlin 1.1.
windowed, chunked, zipWithNext
New extensions for Iterable<T> , Sequence<T> , and CharSequence cover such use cases
as bu ering or batch processing ( chunked ), sliding window and computing sliding average
( windowed ) , and processing pairs of subsequent items ( zipWithNext ):
val items = (1..9).map { it * it }
val chunkedIntoLists = items.chunked(4)

val points3d = items.chunked(3) { (x, y, z) -> Triple(x, y, z) }
val windowed = items.windowed(4)
val slidingAverage = items.windowed(4) { it.average() }
val pairwiseDifferences = items.zipWithNext { a, b -> b - a }
ll, replaceAll, shu e/shu ed
A set of extension functions was added for manipulating lists: fill , replaceAll and
shuffle for MutableList , and shuffled for read-only List :
34
val items = (1..5).toMutableList()
items.shuffle()
println("Shuffled items: $items")
items.replaceAll { it * 2 }
println("Items doubled: $items")
items.fill(5)
println("Items filled with 5: $items")
Math operations in kotlin-stdlib
Satisfying the longstanding request, Kotlin 1.2 adds the kotlin.math API for math operations
that is common for JVM and JS and contains the following:
— Constants: PI and E ;
— Trigonometric: cos , sin , tan and inverse of them: acos , asin , atan , atan2 ;
— Hyperbolic: cosh , sinh , tanh and their inverse: acosh , asinh , atanh
— Exponentation: pow (an extension function), sqrt , hypot , exp , expm1 ;
— Logarithms: log , log2 , log10 , ln , ln1p ;
— Rounding:
— ceil , floor , truncate , round (half to even) functions;
— roundToInt , roundToLong (half to integer) extension functions;
— Sign and absolute value:

— abs and sign functions;
— absoluteValue and sign extension properties;
— withSign extension function;
— max and min of two values;
— Binary representation:
— ulp extension property;
— nextUp , nextDown , nextTowards extension functions;
— toBits , toRawBits , Double.fromBits (these are in the kotlin package).
The same set of functions (but without constants) is also available for Float arguments.
Operators and conversions for BigInteger and BigDecimal
Kotlin 1.2 introduces a set of functions for operating with BigInteger and BigDecimal and
creating them from other numeric types. These are:
35
— toBigInteger for Int and Long ;
— toBigDecimal for Int , Long , Float , Double , and BigInteger ;
— Arithmetic and bitwise operator functions:

— Binary operators + , - , * , / , % and in x functions and , or , xor , shl , shr ;
— Unary operators - , ++ , -- , and a function inv .
Floating point to bits conversions
New functions were added for converting Double and Float to and from their bit
representations:
— toBits and toRawBits returning Long for Double and Int for Float ;
— Double.fromBits and Float.fromBits for creating oating point numbers from the bit
representation.
Regex is now serializable
The kotlin.text.Regex class has become Serializable and can now be used in
serializable hierarchies.
Closeable.use calls Throwable.addSuppressed if available
The Closeable.use function calls Throwable.addSuppressed when an exception is thrown

during closing the resource after some other exception.
To enable this behavior you need to have kotlin-stdlib-jdk7 in your dependencies.
JVM Backend
Constructor calls normalization
Ever since version 1.0, Kotlin supported expressions with complex control ow, such as try-catch
expressions and inline function calls. Such code is valid according to the Java Virtual Machine
speci cation. Unfortunately, some bytecode processing tools do not handle such code quite well
when such expressions are present in the arguments of constructor calls.
To mitigate this problem for the users of such bytecode processing tools, we’ve added a
command-line option ( -Xnormalize-constructor-calls=MODE ) that tells the compiler to
generate more Java-like bytecode for such constructs. Here MODE is one of:
— disable (default) – generate bytecode in the same way as in Kotlin 1.0 and 1.1;
— enable – generate Java-like bytecode for constructor calls. This can change the order in
which the classes are loaded and initialized;
36
— preserve-class-initialization – generate Java-like bytecode for constructor calls,
ensuring that the class initialization order is preserved. This can a ect overall performance of
your application; use it only if you have some complex state shared between multiple classes
and updated on class initialization.
The “manual” workaround is to store the values of sub-expressions with control ow in variables,
instead of evaluating them directly inside the call arguments. It’s similar to -Xnormalize-
constructor-calls=enable .
Java-default method calls
Before Kotlin 1.2, interface members overriding Java-default methods while targeting JVM 1.6
produced a warning on super calls: Super calls to Java default methods are
deprecated in JVM target 1.6. Recompile with '-jvm-target 1.8' . In Kotlin 1.2,
there's an error instead, thus requiring any such code to be compiled with JVM target 1.8.
Breaking change: consistent behavior of x.equals(null) for platform types
Calling x.equals(null) on a platform type that is mapped to a Java primitive ( Int! ,

Boolean! , Short !, Long! , Float! , Double! , Char! ) incorrectly returned true when x
was null. Starting with Kotlin 1.2, calling x.equals(...) on a null value of a platform type
throws an NPE (but x == ... does not).
To return to the pre-1.2 behavior, pass the ag -Xno-exception-on-explicit-equals-

for-boxed-null to the compiler.
Breaking change: x for platform null escaping through an inlined extension

receiver
Inline extension functions that were called on a null value of a platform type did not check the
receiver for null and would thus allow null to escape into the other code. Kotlin 1.2 forces this
check at the call sites, throwing an exception if the receiver is null.
To switch to the old behavior, pass the fallback ag -Xno-receiver-assertions to the

compiler.
JavaScript Backend
TypedArrays support enabled by default
The JS typed arrays support that translates Kotlin primitive arrays, such as IntArray ,
DoubleArray , into JavaScript typed arrays, that was previously an opt-in feature, has been
enabled by default.
37
Tools
Warnings as errors
The compiler now provides an option to treat all warnings as errors. Use -Werror on the
command line, or the following Gradle snippet:
compileKotlin {
kotlinOptions.allWarningsAsErrors = true
}
38
Coroutines release
After some long and extensive battle testing, coroutines are now released! It means that from
Kotlin 1.3 the language support and the API are fully stable. Check out the new coroutines
overview page.
Kotlin 1.3 introduces callable references on suspend-functions and support of Coroutines in the
Re ection API.
Kotlin/Native
Kotlin 1.3 continues to improve and polish the Native target. See the Kotlin/Native overview for
details.
Multiplatform Projects
In 1.3, we've completely reworked the model of multiplatform projects in order to improve
expressiveness and exibility, and to make sharing common code easier. Also, Kotlin/Native is
now supported as one of the targets!
The key di erences to the old model are:
— In the old model, common and platform-speci c code needed to be placed in separate
modules, linked by expectedBy dependencies. Now, common and platform-speci c code is
placed in di erent source roots of the same module, making projects easier to con gure.
— There is now a large number of preset platform con gurations for di erent supported
platforms.
— The dependencies con guration has been changed; dependencies are now speci ed
separately for each source root.
— Source sets can now be shared between an arbitrary subset of platforms (for example, in a
module that targets JS, Android and iOS, you can have a source set that is shared only
between Android and iOS).
— Publishing multiplatform libraries is now supported.
For more information, please refer to the Multiplatform Programming documentation.
Contracts
The Kotlin compiler does extensive static analysis to provide warnings and reduce boilerplate.
One of the most notable features is smartcasts — with the ability to perform a cast automatically
based on the performed type checks:
fun foo(s: String?) {

if (s != null) s.length // Compiler automatically casts 's' to 'String'
}
39
However, as soon as these checks are extracted in a separate function, all the smartcasts
immediately disappear:
fun String?.isNotNull(): Boolean = this != null

if (s.isNotNull()) s.length // No smartcast :(
}
To improve the behavior in such cases, Kotlin 1.3 introduces experimental mechanism called
contracts.
Contracts allow a function to explicitly describe its behavior in a way which is understood by the
compiler. Currently, two wide classes of cases are supported:
— Improving smartcasts analysis by declaring the relation between a function's call outcome and
the passed arguments values:
fun require(condition: Boolean) {

// This is a syntax form, which tells compiler:
// "if this function returns successfully, then passed 'condition' is true"
contract { returns() implies condition }
if (!condition) throw IllegalArgumentException(...)
}

require(s is String)
// s is smartcasted to 'String' here, because otherwise
// 'require' would have throw an exception
}
— Improving the variable initialization analysis in the presence of high-order functions:
fun synchronize(lock: Any?, block: () -> Unit) {

// It tells compiler:
// "This function will invoke 'block' here and now, and exactly one time"
contract { callsInPlace(block, EXACTLY_ONCE) }
}
fun foo() {
val x: Int
synchronize(lock) {
x = 42 // Compiler knows that lambda passed to 'synchronize' is called
// exactly once, so no reassignment is reported
}
println(x) // Compiler knows that lambda will be definitely called, performing
// initialization, so 'x' is considered to be initialized here
}
Contracts in stdlib
stdlib already makes use of contracts, which leads to improvements in the analyses described
above. This part of contracts is stable, meaning that you can bene t from the improved analysis
right now without any additional opt-ins:
40
fun bar(x: String?) {
if (!x.isNullOrEmpty()) {
println("length of '$x' is ${x.length}") // Yay, smartcasted to not-null!
}
}
Custom Contracts
It is possible to declare contracts for your own functions, but this feature is experimental, as the
current syntax is in a state of early prototype and will most probably be changed. Also, please
note, that currently the Kotlin compiler does not verify contracts, so it's a programmer's
responsibility to write correct and sound contracts.
Custom contracts are introduced by the call to contract stdlib function, which provides DSL
scope:
fun String?.isNullOrEmpty(): Boolean {

contract {
returns(false) implies (this@isNullOrEmpty != null)
}
return this == null || isEmpty()
}
See the details on the syntax as well as the compatibility notice in the KEEP.
Capturing when subject in a variable

In Kotlin 1.3, it is now possible to capture the when subject into variable:
fun Request.getBody() =
when (val response = executeRequest()) {
is Success -> response.body
is HttpError -> throw HttpException(response.status)
}
While it was already possible to extract this variable just before when , val in when has its
scope properly restricted to the body of when , and so preventing namespace pollution. See the
full documentation on when here.
@JvmStatic and @JvmField in companion of interfaces

With Kotlin 1.3, it is possible to mark members of a companion object of interfaces with
annotations @JvmStatic and @JvmField . In the class le, such members will be lifted to the
corresponding interface and marked as static .
For example, the following Kotlin code:
41
interface Foo {
companion object {
@JvmField
val answer: Int = 42
@JvmStatic
fun sayHello() {
println("Hello, world!")
}
}
}
It is equivalent to this Java code:
interface Foo {
public static int answer = 42;
public static void sayHello() {
// ...
}
}
Nested declarations in annotation classes

In Kotlin 1.3 it is possible for annotations to have nested classes, interfaces, objects, and
companions:
annotation class Foo {

enum class Direction { UP, DOWN, LEFT, RIGHT }
annotation class Bar
companion object {
fun foo(): Int = 42
val bar: Int = 42
}
}
Parameterless main
By convention, the entry point of a Kotlin program is a function with a signature like
main(args: Array<String>) , where args represent the command-line arguments passed
to the program. However, not every application supports command-line arguments, so this
parameter often ends up not being used.
Kotlin 1.3 introduced a simpler form of main which takes no parameters. Now “Hello, World” in
Kotlin is 19 characters shorter!
fun main() {
println("Hello, world!")
}
Functions with big arity
42
In Kotlin, functional types are represented as generic classes taking a di erent number of
parameters: Function0<R> , Function1<P0, R> , Function2<P0, P1, R> , … This
approach has a problem in that this list is nite, and it currently ends with Function22 .
Kotlin 1.3 relaxes this limitation and adds support for functions with bigger arity:
fun trueEnterpriseComesToKotlin(block: (Any, Any, ... /* 42 more */, Any) -> Any) {
block(Any(), Any(), ..., Any())
}
Progressive mode
Kotlin cares a lot about stability and backward compatibility of code: Kotlin compatibility policy
says that "breaking changes" (e.g., a change which makes the code that used to compile ne, not
compile anymore) can be introduced only in the major releases (1.2, 1.3, etc.).
We believe that a lot of users could use a much faster cycle, where critical compiler bug xes
arrive immediately, making the code more safe and correct. So, Kotlin 1.3 introduces progressive
compiler mode, which can be enabled by passing the argument -progressive to the compiler.
In progressive mode, some xes in language semantics can arrive immediately. All these xes
have two important properties:
— they preserve backward-compatibility of source code with older compilers, meaning that all
the code which is compilable by the progressive compiler will be compiled ne by non-
progressive one.
— they only make code safer in some sense — e.g., some unsound smartcast can be forbidden,
behavior of the generated code may be changed to be more predictable/stable, and so on.
Enabling the progressive mode can require you to rewrite some of your code, but it shouldn't be
too much — all the xes which are enabled under progressive are carefully handpicked,
reviewed, and provided with tooling migration assistance. We expect that the progressive mode
will be a nice choice for any actively maintained codebases which are updated to the latest
language versions quickly.
Inline classes
Inline classes are available only since Kotlin 1.3 and currently are experimental. See
details in the reference.
Kotlin 1.3 introduces a new kind of declaration — inline class . Inline classes can be viewed
as a restricted version of the usual classes, in particular, inline classes must have exactly one
property:
inline class Name(val s: String)
43
The Kotlin compiler will use this restriction to aggressively optimize runtime representation of
inline classes and substitute their instances with the value of the underlying property where
possible removing constructor calls, GC pressure, and enabling other optimizations:
fun main() {
// In the next line no constructor call happens, and
// at the runtime 'name' contains just string "Kotlin"
val name = Name("Kotlin")
println(name.s)
}
See reference for inline classes for details.
Unsigned integers
Unsigned integers are available only since Kotlin 1.3 and currently are experimental. See
details in the reference.
Kotlin 1.3 introduces unsigned integer types:
— kotlin.UByte : an unsigned 8-bit integer, ranges from 0 to 255
— kotlin.UShort : an unsigned 16-bit integer, ranges from 0 to 65535
— kotlin.UInt : an unsigned 32-bit integer, ranges from 0 to 2^32 - 1
— kotlin.ULong : an unsigned 64-bit integer, ranges from 0 to 2^64 - 1
Most of the functionality of signed types are supported for unsigned counterparts too:
// You can define unsigned types using literal suffixes

val uint = 42u
val ulong = 42uL
val ubyte: UByte = 255u
// You can convert signed types to unsigned and vice versa via stdlib extensions:
val int = uint.toInt()
val byte = ubyte.toByte()
val ulong2 = byte.toULong()
// Unsigned types support similar operators:

val x = 20u + 22u
val y = 1u shl 8
val z = "128".toUByte()
val range = 1u..5u
See reference for details.
@JvmDefault
@JvmDefault is only available since Kotlin 1.3 and currently is experimental. See details
in the reference page.
44
Kotlin targets a wide range of the Java versions, including Java 6 and Java 7, where default
methods in the interfaces are not allowed. For your convenience, the Kotlin compiler works
around that limitation, but this workaround isn't compatible with the default methods,
introduced in Java 8.
This could be an issue for Java-interoperability, so Kotlin 1.3 introduces the @JvmDefault
annotation. Methods, annotated with this annotation will be generated as default methods for
JVM:
interface Foo {
// Will be generated as 'default' method
@JvmDefault
fun foo(): Int = 42
}
Warning! Annotating your API with @JvmDefault has serious implications on binary
compatibility. Make sure to carefully read the reference page before using
@JvmDefault in production.
Standard library
Multiplatform Random
Prior to Kotlin 1.3, there was no uniform way to generate random numbers on all platforms —
we had to resort to platform speci c solutions, like java.util.Random on JVM. This release
xes this issue by introducing the class kotlin.random.Random , which is available on all
platforms:
val number = Random.nextInt(42) // number is in range [0, limit)

println(number)
isNullOrEmpty/orEmpty extensions
isNullOrEmpty and orEmpty extensions for some types are already present in stdlib . The
rst one returns true if the receiver is null or empty, and the second one falls back to an
empty instance if the receiver is null . Kotlin 1.3 provides similar extensions on collections,
maps, and arrays of objects.
Copying elements between two existing arrays

The array.copyInto(targetArray, targetOffset, startIndex, endIndex)
functions for the existing array types, including the unsigned arrays, make it easier to implement
array-based containers in pure Kotlin.
45
val sourceArr = arrayOf("k", "o", "t", "l", "i", "n")
val targetArr = sourceArr.copyInto(arrayOfNulls<String>(6), 3, startIndex = 3, endIndex =
6)
println(targetArr.contentToString())
sourceArr.copyInto(targetArr, startIndex = 0, endIndex = 3)

println(targetArr.contentToString())
associateWith
It is quite a common situation to have a list of keys and want to build a map by associating each
of these keys with some value. It was possible to do it before with the associate { it to
getValue(it) } function, but now we’re introducing a more e cient and easy to explore
alternative: keys.associateWith { getValue(it) } .
val keys = 'a'..'f'

val map = keys.associateWith { it.toString().repeat(5).capitalize() }
map.forEach { println(it) }
ifEmpty and ifBlank functions

Collections, maps, object arrays, char sequences, and sequences now have an ifEmpty function,
which allows specifying a fallback value that will be used instead of the receiver if it is empty:
fun printAllUppercase(data: List<String>) {

val result = data
.filter { it.all { c -> c.isUpperCase() } }
.ifEmpty { listOf("<no uppercase>") }
result.forEach { println(it) }
}
printAllUppercase(listOf("foo", "Bar"))
printAllUppercase(listOf("FOO", "BAR"))
Char sequences and strings in addition have an ifBlank extension that does the same thing as
ifEmpty , but checks for a string being all whitespace instead of empty.
val s = " \n"

println(s.ifBlank { "<blank>" })
println(s.ifBlank { null })
Sealed classes in re ection

We’ve added a new API to kotlin-reflect that can be used to enumerate all the direct
subtypes of a sealed class, namely KClass.sealedSubclasses .
Smaller changes
— Boolean type now has companion.
— Any?.hashCode() extension, which returns 0 for null .
— Char now provides MIN_VALUE / MAX_VALUE constants.
—
46
— SIZE_BYTES and SIZE_BITS constants in primitive type companions.
Tooling
Code Style Support in IDE

Kotlin 1.3 introduces support for the recommended code style in the IDE. Check out this page for
the migration guidelines.
kotlinx.serialization
kotlinx.serialization is a library which provides multiplatform support for (de)serializing objects in
Kotlin. Previously, it was a separate project, but since Kotlin 1.3, it ships with the Kotlin compiler
distribution on par with the other compiler plugins. The main di erence is that you don't need to
manually watch out for the Serialization IDE Plugin being compatible with the Kotlin IDE Plugin
version you're using: now the Kotlin IDE Plugin already includes serialization!
See here for details.
Please, note, that even though kotlinx.serialization now ships with the Kotlin Compiler
distribution, it is still considered to be an experimental feature.
Scripting update
Please note, that scripting is an experimental feature, meaning that no compatibility

guarantees on the API are given.
Kotlin 1.3 continues to evolve and improve scripting API, introducing some experimental support
for scripts customization, such as adding external properties, providing static or dynamic
dependencies, and so on.
For additional details, please consult the KEEP-75.
Scratches support
Kotlin 1.3 introduces support for runnable Kotlin scratch les. Scratch le is a kotlin script le with
a .kts extension which you can run and get evaluation results directly in the editor.
Consult the general Scratches documentation for details.
47
Getting Started
Basic Syntax
De ning packages
Package speci cation should be at the top of the source le:
package my.demo
import java.util.*
// ...
It is not required to match directories and packages: source les can be placed arbitrarily in the
le system.
See Packages.
De ning functions
Function having two Int parameters with Int return type:
fun sum(a: Int, b: Int): Int {

return a + b
}
Function with an expression body and inferred return type:
fun sum(a: Int, b: Int) = a + b
Function returning no meaningful value:
fun printSum(a: Int, b: Int): Unit {

println("sum of $a and $b is ${a + b}")
}
Unit return type can be omitted:
fun printSum(a: Int, b: Int) {

println("sum of $a and $b is ${a + b}")
}
See Functions.
De ning variables
48
Read-only local variables are de ned using the keyword val . They can be assigned a value only
once.
val a: Int = 1 // immediate assignment

val b = 2 // Ìnt` type is inferred
val c: Int // Type required when no initializer is provided
c = 3 // deferred assignment
Variables that can be reassigned use the var keyword:
var x = 5 // Ìnt` type is inferred

x += 1
Top-level variables:
val PI = 3.14
var x = 0
fun incrementX() {
x += 1
}
See also Properties And Fields.
Comments
Just like Java and JavaScript, Kotlin supports end-of-line and block comments.
// This is an end-of-line comment
/* This is a block comment

on multiple lines. */
Unlike Java, block comments in Kotlin can be nested.
See Documenting Kotlin Code for information on the documentation comment syntax.
Using string templates

var a = 1
// simple name in template:
val s1 = "a is $a"
a = 2
// arbitrary expression in template:
val s2 = "${s1.replace("is", "was")}, but now is $a"
See String templates.
Using conditional expressions
49
fun maxOf(a: Int, b: Int): Int {
if (a > b) {
return a
} else {
return b
}
}
Using if as an expression:
fun maxOf(a: Int, b: Int) = if (a > b) a else b
See if-expressions.
Using nullable values and checking for null

A reference must be explicitly marked as nullable when null value is possible.
Return null if str does not hold an integer:
fun parseInt(str: String): Int? {

// ...
}
Use a function returning nullable value:
fun printProduct(arg1: String, arg2: String) {

val x = parseInt(arg1)
val y = parseInt(arg2)
// Using `x * y` yields error because they may hold nulls.

if (x != null && y != null) {
// x and y are automatically cast to non-nullable after null check
println(x * y)
}
else {
println("either '$arg1' or '$arg2' is not a number")
}
}
or
// ...
if (x == null) {
println("Wrong number format in arg1: '$arg1'")
return
}
if (y == null) {
println("Wrong number format in arg2: '$arg2'")
return
}
// x and y are automatically cast to non-nullable after null check

println(x * y)
See Null-safety.
50
Using type checks and automatic casts
The is operator checks if an expression is an instance of a type. If an immutable local variable or
property is checked for a speci c type, there's no need to cast it explicitly:
fun getStringLength(obj: Any): Int? {

if (obj is String) {
// òbj` is automatically cast to `String` in this branch
return obj.length
}
// òbj` is still of type Àny` outside of the type-checked branch

return null
}
or

if (obj !is String) return null
// òbj` is automatically cast to `String` in this branch

return obj.length
}
or even

// òbj` is automatically cast to `String` on the right-hand side of `&&`
if (obj is String && obj.length > 0) {
return obj.length
}
return null
}
See Classes and Type casts.
Using a for loop

val items = listOf("apple", "banana", "kiwifruit")
for (item in items) {
println(item)
}
or

for (index in items.indices) {
println("item at $index is ${items[index]}")
}
See for loop.
Using a while loop
51
var index = 0
while (index < items.size) {
println("item at $index is ${items[index]}")
index++
}
See while loop.
Using when expression

fun describe(obj: Any): String =
when (obj) {
1 -> "One"
"Hello" -> "Greeting"
is Long -> "Long"
!is String -> "Not a string"
else -> "Unknown"
}
See when expression.
Using ranges
Check if a number is within a range using in operator:
val x = 10
val y = 9
if (x in 1..y+1) {
println("fits in range")
}
Check if a number is out of range:
val list = listOf("a", "b", "c")
if (-1 !in 0..list.lastIndex) {

println("-1 is out of range")
}
if (list.size !in list.indices) {
println("list size is out of valid list indices range, too")
}
Iterating over a range:
for (x in 1..5) {
print(x)
}
or over a progression:
52
for (x in 1..10 step 2) {
print(x)
}
println()
for (x in 9 downTo 0 step 3) {
print(x)
}
See Ranges.
Using collections
Iterating over a collection:
for (item in items) {

println(item)
}
Checking if a collection contains an object using in operator:
when {
"orange" in items -> println("juicy")
"apple" in items -> println("apple is fine too")
}
Using lambda expressions to lter and map collections:
val fruits = listOf("banana", "avocado", "apple", "kiwifruit")

fruits
.filter { it.startsWith("a") }
.sortedBy { it }
.map { it.toUpperCase() }
.forEach { println(it) }
See Higher-order functions and Lambdas.
Creating basic classes and their instances:

val rectangle = Rectangle(5.0, 2.0) //no 'new' keyword required
val triangle = Triangle(3.0, 4.0, 5.0)
See classes and objects and instances.
53
Idioms
A collection of random and frequently used idioms in Kotlin. If you have a favorite idiom,
contribute it by sending a pull request.
Creating DTOs (POJOs/POCOs)
data class Customer(val name: String, val email: String)
provides a Customer class with the following functionality:
— getters (and setters in case of vars) for all properties
— equals()
— hashCode()
— toString()
— copy()
— component1() , component2() , …, for all properties (see Data classes)
Default values for function parameters
fun foo(a: Int = 0, b: String = "") { ... }
Filtering a list
val positives = list.filter { x -> x > 0 }
Or alternatively, even shorter:
val positives = list.filter { it > 0 }
String Interpolation
println("Name $name")
Instance Checks
when (x) {
is Foo -> ...
is Bar -> ...
else -> ...
}
Traversing a map/list of pairs
54
for ((k, v) in map) {
println("$k -> $v")
}
k , v can be called anything.
Using ranges
for (i in 1..100) { ... } // closed range: includes 100

for (i in 1 until 100) { ... } // half-open range: does not include 100
for (x in 2..10 step 2) { ... }
for (x in 10 downTo 1) { ... }
if (x in 1..10) { ... }
Read-only list
val list = listOf("a", "b", "c")
Read-only map
val map = mapOf("a" to 1, "b" to 2, "c" to 3)
Accessing a map
println(map["key"])
map["key"] = value
Lazy property
val p: String by lazy {

// compute the string
}
Extension Functions
fun String.spaceToCamelCase() { ... }
"Convert this to camelcase".spaceToCamelCase()
Creating a singleton
object Resource {
val name = "Name"
}
If not null shorthand
55
val files = File("Test").listFiles()
println(files?.size)
If not null and else shorthand
val files = File("Test").listFiles()
println(files?.size ?: "empty")
Executing a statement if null
val values = ...

val email = values["email"] ?: throw IllegalStateException("Email is missing!")
Get rst item of a possibly empty collection
val emails = ... // might be empty

val mainEmail = emails.firstOrNull() ?: ""
Execute if not null
val value = ...
value?.let {
... // execute this block if not null
}
Map nullable value if not null
val value = ...
val mapped = value?.let { transformValue(it) } ?: defaultValueIfValueIsNull
Return on when statement
fun transform(color: String): Int {

return when (color) {
"Red" -> 0
"Green" -> 1
"Blue" -> 2
else -> throw IllegalArgumentException("Invalid color param value")
}
}
'try/catch' expression
56
fun test() {
val result = try {
count()
} catch (e: ArithmeticException) {
throw IllegalStateException(e)
}
// Working with result

}
'if' expression
fun foo(param: Int) {

val result = if (param == 1) {
"one"
} else if (param == 2) {
"two"
} else {
"three"
}
}
Builder-style usage of methods that return Unit
fun arrayOfMinusOnes(size: Int): IntArray {

return IntArray(size).apply { fill(-1) }
}
Single-expression functions
fun theAnswer() = 42
This is equivalent to
fun theAnswer(): Int {

return 42
}
This can be e ectively combined with other idioms, leading to shorter code. E.g. with the when-
expression:
fun transform(color: String): Int = when (color) {

"Red" -> 0
"Green" -> 1
"Blue" -> 2
else -> throw IllegalArgumentException("Invalid color param value")
}
Calling multiple methods on an object instance (with)
57
class Turtle {
fun penDown()
fun penUp()
fun turn(degrees: Double)
fun forward(pixels: Double)
}
val myTurtle = Turtle()

with(myTurtle) { //draw a 100 pix square
penDown()
for(i in 1..4) {
forward(100.0)
turn(90.0)
}
penUp()
}
Java 7's try with resources
val stream = Files.newInputStream(Paths.get("/some/file.txt"))

stream.buffered().reader().use { reader ->
println(reader.readText())
}
Convenient form for a generic function that requires the generic type
information
// public final class Gson {

// ...
// public <T> T fromJson(JsonElement json, Class<T> classOfT) throws
JsonSyntaxException {
// ...
inline fun <reified T: Any> Gson.fromJson(json: JsonElement): T = this.fromJson(json,

T::class.java)
Consuming a nullable Boolean
val b: Boolean? = ...

if (b == true) {
...
} else {
// `b` is false or null
}
Swapping two variables
var a = 1
var b = 2
a = b.also { b = a }
58
Coding Conventions
This page contains the current coding style for the Kotlin language.
— Source code organization

— Naming rules
— Formatting
— Documentation comments
— Avoiding redundant constructs
— Idiomatic use of language features
— Coding conventions for libraries
Applying the style guide
To con gure the IntelliJ formatter according to this style guide, please install Kotlin plugin version
1.2.20 or newer, go to Settings | Editor | Code Style | Kotlin, click on "Set from…" link in the
upper right corner, and select "Prede ned style / Kotlin style guide" from the menu.
To verify that your code is formatted according to the style guide, go to the inspection settings
and enable the "Kotlin | Style issues | File is not formatted according to project settings"
inspection. Additional inspections that verify other issues described in the style guide (such as
naming conventions) are enabled by default.
Source code organization
Directory structure
In mixed-language projects, Kotlin source les should reside in the same source root as the Java
source les, and follow the same directory structure (each le should be stored in the directory
corresponding to each package statement).
In pure Kotlin projects, the recommended directory structure is to follow the package structure
with the common root package omitted (e.g. if all the code in the project is in the
"org.example.kotlin" package and its subpackages, les with the "org.example.kotlin" package
should be placed directly under the source root, and les in "org.example.kotlin.foo.bar" should
be in the "foo/bar" subdirectory of the source root).
Source le names
If a Kotlin le contains a single class (potentially with related top-level declarations), its name
should be the same as the name of the class, with the .kt extension appended. If a le contains
multiple classes, or only top-level declarations, choose a name describing what the le contains,
and name the le accordingly. Use camel humps with an uppercase rst letter (e.g.
ProcessDeclarations.kt ).
59
The name of the le should describe what the code in the le does. Therefore, you should avoid
using meaningless words such as "Util" in le names.
Source le organization
Placing multiple declarations (classes, top-level functions or properties) in the same Kotlin source
le is encouraged as long as these declarations are closely related to each other semantically and
the le size remains reasonable (not exceeding a few hundred lines).
In particular, when de ning extension functions for a class which are relevant for all clients of this
class, put them in the same le where the class itself is de ned. When de ning extension
functions that make sense only for a speci c client, put them next to the code of that client. Do
not create les just to hold "all extensions of Foo".
Class layout
Generally, the contents of a class is sorted in the following order:
— Property declarations and initializer blocks

— Secondary constructors
— Method declarations
— Companion object
Do not sort the method declarations alphabetically or by visibility, and do not separate regular
methods from extension methods. Instead, put related stu together, so that someone reading
the class from top to bottom would be able to follow the logic of what's happening. Choose an
order (either higher-level stu rst, or vice versa) and stick to it.
Put nested classes next to the code that uses those classes. If the classes are intended to be used
externally and aren't referenced inside the class, put them in the end, after the companion
object.
Interface implementation layout
When implementing an interface, keep the implementing members in the same order as
members of the interface (if necessary, interspersed with additional private methods used for
the implementation)
Overload layout
Always put overloads next to each other in a class.
Naming rules
Kotlin follows the Java naming conventions. In particular:
60
Names of packages are always lower case and do not use underscores
( org.example.myproject ). Using multi-word names is generally discouraged, but if you do
need to use multiple words, you can either simply concatenate them together or use camel
humps ( org.example.myProject ).
Names of classes and objects start with an upper case letter and use camel humps:
open class DeclarationProcessor { ... }
object EmptyDeclarationProcessor : DeclarationProcessor() { ... }
Function names
Names of functions, properties and local variables start with a lower case letter and use camel
humps and no underscores:
fun processDeclarations() { ... }

var declarationCount = ...
Exception: factory functions used to create instances of classes can have the same name as the
class being created:
abstract class Foo { ... }
class FooImpl : Foo { ... }
fun Foo(): Foo { return FooImpl(...) }
Names for test methods
In tests (and only in tests), it's acceptable to use method names with spaces enclosed in backticks.
(Note that such method names are currently not supported by the Android runtime.)
Underscores in method names are also allowed in test code.
class MyTestCase {
@Test fun ènsure everything works`() { ... }
@Test fun ensureEverythingWorks_onAndroid() { ... }

}
Property names
Names of constants (properties marked with const , or top-level or object val properties with
no custom get function that hold deeply immutable data) should use uppercase underscore-
separated names:
const val MAX_COUNT = 8

val USER_NAME_FIELD = "UserName"
61
Names of top-level or object properties which hold objects with behavior or mutable data should
use regular camel-hump names:
val mutableCollection: MutableSet<String> = HashSet()
Names of properties holding references to singleton objects can use the same naming style as
object declarations:
val PersonComparator: Comparator<Person> = ...
For enum constants, it's OK to use either uppercase underscore-separated names ( enum class
Color { RED, GREEN } ) or regular camel-humps names starting with an uppercase letter,
depending on the usage.
Names for backing properties
If a class has two properties which are conceptually the same but one is part of a public API and
another is an implementation detail, use an underscore as the pre x for the name of the private
property:
class C {
private val _elementList = mutableListOf<Element>()
val elementList: List<Element>

get() = _elementList
}
Choosing good names
The name of a class is usually a noun or a noun phrase explaining what the class is: List ,
PersonReader .
The name of a method is usually a verb or a verb phrase saying what the method does: close ,
readPersons . The name should also suggest if the method is mutating the object or returning
a new one. For instance sort is sorting a collection in place, while sorted is returning a sorted
copy of the collection.
The names should make it clear what the purpose of the entity is, so it's best to avoid using
meaningless words ( Manager , Wrapper etc.) in names.
When using an acronym as part of a declaration name, capitalize it if it consists of two letters
( IOStream ); capitalize only the rst letter if it is longer ( XmlFormatter , HttpInputStream ).
Formatting
In most cases, Kotlin follows the Java coding conventions.
Use 4 spaces for indentation. Do not use tabs.
62
For curly braces, put the opening brace in the end of the line where the construct begins, and the
closing brace on a separate line aligned horizontally with the opening construct.
if (elements != null) {
for (element in elements) {
// ...
}
}
(Note: In Kotlin, semicolons are optional, and therefore line breaks are signi cant. The language
design assumes Java-style braces, and you may encounter surprising behavior if you try to use a
di erent formatting style.)
Horizontal whitespace
Put spaces around binary operators ( a + b ). Exception: don't put spaces around the "range to"
operator ( 0..i ).
Do not put spaces around unary operators ( a++ )
Put spaces between control ow keywords ( if , when , for and while ) and the corresponding
opening parenthesis.
Do not put a space before an opening parenthesis in a primary constructor declaration, method
declaration or method call.
class A(val x: Int)
fun foo(x: Int) { ... }
fun bar() {
foo(1)
}
Never put a space after ( , [ , or before ] , ) .
Never put a space around . or ?. : foo.bar().filter { it > 2 }.joinToString() ,

foo?.bar()
Put a space after // : // This is a comment
Do not put spaces around angle brackets used to specify type parameters: class Map<K, V> {
... }
Do not put spaces around :: : Foo::class , String::length
Do not put a space before ? used to mark a nullable type: String?
As a general rule, avoid horizontal alignment of any kind. Renaming an identi er to a name with a
di erent length should not a ect the formatting of either the declaration or any of the usages.
63
Colon
Put a space before : in the following cases:
— when it's used to separate a type and a supertype;

— when delegating to a superclass constructor or a di erent constructor of the same class;
— after the object keyword.
Don't put a space before : when it separates a declaration and its type.
Always put a space after : .
abstract class Foo<out T : Any> : IFoo {

abstract fun foo(a: Int): T
}
class FooImpl : Foo() {

constructor(x: String) : this(x) { ... }
val x = object : IFoo { ... }

}
Class header formatting
Classes with a few primary constructor parameters can be written in a single line:
class Person(id: Int, name: String)
Classes with longer headers should be formatted so that each primary constructor parameter is
in a separate line with indentation. Also, the closing parenthesis should be on a new line. If we
use inheritance, then the superclass constructor call or list of implemented interfaces should be
located on the same line as the parenthesis:
class Person(
id: Int,
name: String,
surname: String
) : Human(id, name) { ... }
For multiple interfaces, the superclass constructor call should be located rst and then each
interface should be located in a di erent line:
class Person(
id: Int,
name: String,
surname: String
) : Human(id, name),
KotlinMaker { ... }
For classes with a long supertype list, put a line break after the colon and align all supertype
names horizontally:
64
class MyFavouriteVeryLongClassHolder :
MyLongHolder<MyFavouriteVeryLongClass>(),
SomeOtherInterface,
AndAnotherOne {
fun foo() { ... }

}
To clearly separate the class header and body when the class header is long, either put a blank
line following the class header (as in the example above), or put the opening curly brace on a
separate line:
class MyFavouriteVeryLongClassHolder :
MyLongHolder<MyFavouriteVeryLongClass>(),
SomeOtherInterface,
AndAnotherOne
{
fun foo() { ... }
}
Use regular indent (4 spaces) for constructor parameters.
Rationale: This ensures that properties declared in the primary constructor have the same
indentation as properties declared in the body of a class.
Modi ers
If a declaration has multiple modi ers, always put them in the following order:
public / protected / private / internal

expect / actual
final / open / abstract / sealed / const
external
override
lateinit
tailrec
vararg
suspend
inner
enum / annotation
companion
inline
infix
operator
data
Place all annotations before modi ers:
@Named("Foo")
private val foo: Foo
Unless you're working on a library, omit redundant modi ers (e.g. public ).
Annotation formatting
65
Annotations are typically placed on separate lines, before the declaration to which they are
attached, and with the same indentation:
@Target(AnnotationTarget.PROPERTY)
annotation class JsonExclude
Annotations without arguments may be placed on the same line:
@JsonExclude @JvmField
var x: String
A single annotation without arguments may be placed on the same line as the corresponding
declaration:
@Test fun foo() { ... }
File annotations
File annotations are placed after the le comment (if any), before the package statement, and
are separated from package with a blank line (to emphasize the fact that they target the le and
not the package).
/** License, copyright and whatever */

@file:JvmName("FooBar")
package foo.bar
Function formatting
If the function signature doesn't t on a single line, use the following syntax:
fun longMethodName(
argument: ArgumentType = defaultValue,
argument2: AnotherArgumentType
): ReturnType {
// body
}
Use regular indent (4 spaces) for function parameters.
Rationale: Consistency with constructor parameters
Prefer using an expression body for functions with the body consisting of a single expression.
fun foo(): Int { // bad

return 1
}
fun foo() = 1 // good
66
Expression body formatting
If the function has an expression body that doesn't t in the same line as the declaration, put the
= sign on the rst line. Indent the expression body by 4 spaces.
fun f(x: String) =

x.length
Property formatting
For very simple read-only properties, consider one-line formatting:
val isEmpty: Boolean get() = size == 0
For more complex properties, always put get and set keywords on separate lines:
val foo: String

get() { ... }
For properties with an initializer, if the initializer is long, add a line break after the equals sign
and indent the initializer by four spaces:
private val defaultCharset: Charset? =

EncodingRegistry.getInstance().getDefaultCharsetForPropertiesFiles(file)
Formatting control ow statements
If the condition of an if or when statement is multiline, always use curly braces around the
body of the statement. Indent each subsequent line of the condition by 4 spaces relative to
statement begin. Put the closing parentheses of the condition together with the opening curly
brace on a separate line:
if (!component.isSyncing &&
!hasAnyKotlinRuntimeInScope(module)
) {
return createKotlinNotConfiguredPanel(module)
}
Rationale: Tidy alignment and clear separation of condition and statement body
Put the else , catch , finally keywords, as well as the while keyword of a do/while loop,
on the same line as the preceding curly brace:
67
if (condition) {
// body
} else {
// else part
}
try {
// body
} finally {
// cleanup
}
In a when statement, if a branch is more than a single line, consider separating it from adjacent
case blocks with a blank line:
private fun parsePropertyValue(propName: String, token: Token) {

when (token) {
is Token.ValueToken ->
callback.visitValue(propName, token.value)
Token.LBRACE -> { // ...

}
}
}
Put short branches on the same line as the condition, without braces.
when (foo) {
true -> bar() // good
false -> { baz() } // bad
}
Method call formatting
In long argument lists, put a line break after the opening parenthesis. Indent arguments by 4
spaces. Group multiple closely related arguments on the same line.
drawSquare(
x = 10, y = 10,
width = 100, height = 100,
fill = true
)
Put spaces around the = sign separating the argument name and value.
Chained call wrapping
When wrapping chained calls, put the . character or the ?. operator on the next line, with a
single indent:
val anchor = owner

?.firstChild!!
.siblings(forward = true)
.dropWhile { it is PsiComment || it is PsiWhiteSpace }
68
The rst call in the chain usually should have a line break before it, but it's OK to omit it if the
code makes more sense that way.
Lambda formatting
In lambda expressions, spaces should be used around the curly braces, as well as around the
arrow which separates the parameters from the body. If a call takes a single lambda, it should be
passed outside of parentheses whenever possible.
list.filter { it > 10 }
If assigning a label for a lambda, do not put a space between the label and the opening curly
brace:
fun foo() {
ints.forEach lit@{
// ...
}
}
When declaring parameter names in a multiline lambda, put the names on the rst line, followed
by the arrow and the newline:
appendCommaSeparated(properties) { prop ->

val propertyValue = prop.get(obj) // ...
}
If the parameter list is too long to t on a line, put the arrow on a separate line:
foo {
context: Context,
environment: Env
->
context.configureEnv(environment)
}
Documentation comments
For longer documentation comments, place the opening /** on a separate line and begin each
subsequent line with an asterisk:
/**
* This is a documentation comment
* on multiple lines.
*/
Short comments can be placed on a single line:
/** This is a short documentation comment. */
69
Generally, avoid using @param and @return tags. Instead, incorporate the description of
parameters and return values directly into the documentation comment, and add links to
parameters wherever they are mentioned. Use @param and @return only when a lengthy
description is required which doesn't t into the ow of the main text.
// Avoid doing this:
/**
* Returns the absolute value of the given number.
* @param number The number to return the absolute value for.
* @return The absolute value.
*/
fun abs(number: Int) = ...
// Do this instead:
/**
* Returns the absolute value of the given [number].
*/
fun abs(number: Int) = ...
Avoiding redundant constructs

In general, if a certain syntactic construction in Kotlin is optional and highlighted by the IDE as
redundant, you should omit it in your code. Do not leave unnecessary syntactic elements in code
just "for clarity".
Unit
If a function returns Unit, the return type should be omitted:
fun foo() { // ": Unit" is omitted here
Semicolons
Omit semicolons whenever possible.
String templates
Don't use curly braces when inserting a simple variable into a string template. Use curly braces
only for longer expressions.
println("$name has ${children.size} children")
Idiomatic use of language features
Immutability
70
Prefer using immutable data to mutable. Always declare local variables and properties as val
rather than var if they are not modi ed after initialization.
Always use immutable collection interfaces ( Collection , List , Set , Map ) to declare
collections which are not mutated. When using factory functions to create collection instances,
always use functions that return immutable collection types when possible:
// Bad: use of mutable collection type for value which will not be mutated
fun validateValue(actualValue: String, allowedValues: HashSet<String>) { ... }
// Good: immutable collection type used instead

fun validateValue(actualValue: String, allowedValues: Set<String>) { ... }
// Bad: arrayListOf() returns ArrayList<T>, which is a mutable collection type

val allowedValues = arrayListOf("a", "b", "c")
// Good: listOf() returns List<T>

val allowedValues = listOf("a", "b", "c")
Default parameter values
Prefer declaring functions with default parameter values to declaring overloaded functions.
// Bad
fun foo() = foo("a")
fun foo(a: String) { ... }
// Good
fun foo(a: String = "a") { ... }
Type aliases
If you have a functional type or a type with type parameters which is used multiple times in a
codebase, prefer de ning a type alias for it:
typealias MouseClickHandler = (Any, MouseEvent) -> Unit

typealias PersonIndex = Map<String, Person>
Lambda parameters
In lambdas which are short and not nested, it's recommended to use the it convention instead
of declaring the parameter explicitly. In nested lambdas with parameters, parameters should be
always declared explicitly.
Returns in a lambda
Avoid using multiple labeled returns in a lambda. Consider restructuring the lambda so that it will
have a single exit point. If that's not possible or not clear enough, consider converting the lambda
into an anonymous function.
71
Do not use a labeled return for the last statement in a lambda.
Named arguments
Use the named argument syntax when a method takes multiple parameters of the same
primitive type, or for parameters of Boolean type, unless the meaning of all parameters is
absolutely clear from context.
drawSquare(x = 10, y = 10, width = 100, height = 100, fill = true)
Using conditional statements
Prefer using the expression form of try , if and when . Examples:
return if (x) foo() else bar()
return when(x) {
0 -> "zero"
else -> "nonzero"
}
The above is preferable to:
if (x)
return foo()
else
return bar()
when(x) {
0 -> return "zero"
else -> return "nonzero"
}
if versus when
Prefer using if for binary conditions instead of when . Instead of
when (x) {
null -> ...
else -> ...
}
use if (x == null) ... else ...
Prefer using when if there are three or more options.
Using nullable Boolean values in conditions
If you need to use a nullable Boolean in a conditional statement, use if (value == true)
or if (value == false) checks.
72
Using loops
Prefer using higher-order functions ( filter , map etc.) to loops. Exception: forEach (prefer
using a regular for loop instead, unless the receiver of forEach is nullable or forEach is
used as part of a longer call chain).
When making a choice between a complex expression using multiple higher-order functions and
a loop, understand the cost of the operations being performed in each case and keep
performance considerations in mind.
Loops on ranges
Use the until function to loop over an open range:
for (i in 0..n - 1) { ... } // bad

for (i in 0 until n) { ... } // good
Using strings
Prefer using string templates to string concatenation.
Prefer to use multiline strings instead of embedding \n escape sequences into regular string
literals.
To maintain indentation in multiline strings, use trimIndent when the resulting string does not
require any internal indentation, or trimMargin when internal indentation is required:
assertEquals(
"""
Foo
Bar
""".trimIndent(),
value
)
val a = """if(a > 1) {

| return a
|}""".trimMargin()
Functions vs Properties
In some cases functions with no arguments might be interchangeable with read-only properties.
Although the semantics are similar, there are some stylistic conventions on when to prefer one to
another.
Prefer a property over a function when the underlying algorithm:
— does not throw

— is cheap to calculate (or caсhed on the rst run)
73
— returns the same result over invocations if the object state hasn't changed
Using extension functions
Use extension functions liberally. Every time you have a function that works primarily on an
object, consider making it an extension function accepting that object as a receiver. To minimize
API pollution, restrict the visibility of extension functions as much as it makes sense. As
necessary, use local extension functions, member extension functions, or top-level extension
functions with private visibility.
Using in x functions
Declare a function as in x only when it works on two objects which play a similar role. Good
examples: and , to , zip . Bad example: add .
Don't declare a method as in x if it mutates the receiver object.
Factory functions
If you declare a factory function for a class, avoid giving it the same name as the class itself.
Prefer using a distinct name making it clear why the behavior of the factory function is special.
Only if there is really no special semantics, you can use the same name as the class.
Example:
class Point(val x: Double, val y: Double) {

companion object {
fun fromPolar(angle: Double, radius: Double) = Point(...)
}
}
If you have an object with multiple overloaded constructors that don't call di erent superclass
constructors and can't be reduced to a single constructor with default argument values, prefer to
replace the overloaded constructors with factory functions.
Platform types
A public function/method returning an expression of a platform type must declare its Kotlin type
explicitly:
fun apiCall(): String = MyJavaApi.getProperty("name")
Any property (package-level or class-level) initialised with an expression of a platform type must
declare its Kotlin type explicitly:
class Person {
val name: String = MyJavaApi.getProperty("name")
}
74
A local value initialised with an expression of a platform type may or may not have a type
declaration:
fun main() {
val name = MyJavaApi.getProperty("name")
println(name)
}
Using scope functions apply/with/run/also/let
Kotlin provides a variety of functions to execute a block of code in the context of a given object:
let , run , with , apply , and also . For the guidance on choosing the right scope function for
your case, refer to Scope Functions.
Coding conventions for libraries

When writing libraries, it's recommended to follow an additional set of rules to ensure API
stability:
— Always explicitly specify member visibility (to avoid accidentally exposing declarations as
public API)
— Always explicitly specify function return types and property types (to avoid accidentally
changing the return type when the implementation changes)
— Provide KDoc comments for all public members, with the exception of overrides that do not
require any new documentation (to support generating documentation for the library)
75
Basics
Basic Types
In Kotlin, everything is an object in the sense that we can call member functions and properties
on any variable. Some of the types can have a special internal representation - for example,
numbers, characters and booleans can be represented as primitive values at runtime - but to the
user they look like ordinary classes. In this section we describe the basic types used in Kotlin:
numbers, characters, booleans, arrays, and strings.
Numbers
Kotlin handles numbers in a way close to Java, but not exactly the same. For example, there are
no implicit widening conversions for numbers, and literals are slightly di erent in some cases.
Kotlin provides the following built-in types representing numbers (this is close to Java):
Type Bit width

Double 64
Float 32
Long 64
Int 32
Short 16
Byte 8
Note that characters are not numbers in Kotlin.
Literal Constants
There are the following kinds of literal constants for integral values:
Decimals: 123
—
— Longs are tagged by a capital L : 123L
— Hexadecimals: 0x0F
— Binaries: 0b00001011
NOTE: Octal literals are not supported.
Kotlin also supports a conventional notation for oating-point numbers:
— Doubles by default: 123.5 , 123.5e10
76
— Floats are tagged by f or F : 123.5f
Underscores in numeric literals (since 1.1)
You can use underscores to make number constants more readable:
val oneMillion = 1_000_000

val creditCardNumber = 1234_5678_9012_3456L
val socialSecurityNumber = 999_99_9999L
val hexBytes = 0xFF_EC_DE_5E
val bytes = 0b11010010_01101001_10010100_10010010
Representation
On the Java platform, numbers are physically stored as JVM primitive types, unless we need a
nullable number reference (e.g. Int? ) or generics are involved. In the latter cases numbers are
boxed.
Note that boxing of numbers does not necessarily preserve identity:
val a: Int = 10000

println(a === a) // Prints 'true'
val boxedA: Int? = a
val anotherBoxedA: Int? = a
println(boxedA === anotherBoxedA) // !!!Prints 'false'!!!
On the other hand, it preserves equality:
val a: Int = 10000

println(a == a) // Prints 'true'
val boxedA: Int? = a
val anotherBoxedA: Int? = a
println(boxedA == anotherBoxedA) // Prints 'true'
Explicit Conversions
Due to di erent representations, smaller types are not subtypes of bigger ones. If they were, we
would have troubles of the following sort:
// Hypothetical code, does not actually compile:

val a: Int? = 1 // A boxed Int (java.lang.Integer)
val b: Long? = a // implicit conversion yields a boxed Long (java.lang.Long)
print(b == a) // Surprise! This prints "false" as Long's equals() checks whether the
other is Long as well
So equality would have been lost silently all over the place, not to mention identity.
As a consequence, smaller types are NOT implicitly converted to bigger types. This means that we
cannot assign a value of type Byte to an Int variable without an explicit conversion
val b: Byte = 1 // OK, literals are checked statically

val i: Int = b // ERROR
77
We can use explicit conversions to widen numbers
val i: Int = b.toInt() // OK: explicitly widened

print(i)
Every number type supports the following conversions:
— toByte(): Byte
— toShort(): Short
— toInt(): Int
— toLong(): Long
— toFloat(): Float
— toDouble(): Double
— toChar(): Char
Absence of implicit conversions is rarely noticeable because the type is inferred from the context,
and arithmetical operations are overloaded for appropriate conversions, for example
val l = 1L + 3 // Long + Int => Long
Operations
Kotlin supports the standard set of arithmetical operations over numbers, which are declared as
members of appropriate classes (but the compiler optimizes the calls down to the corresponding
instructions). See Operator overloading.
As of bitwise operations, there're no special characters for them, but just named functions that
can be called in in x form, for example:
val x = (1 shl 2) and 0x000FF000
Here is the complete list of bitwise operations (available for Int and Long only):
— shl(bits) – signed shift left (Java's << )
— shr(bits) – signed shift right (Java's >> )
— ushr(bits) – unsigned shift right (Java's >>> )
— and(bits) – bitwise and
— or(bits) – bitwise or
— xor(bits) – bitwise xor
— inv() – bitwise inversion
Floating Point Numbers Comparison
78
The operations on oating point numbers discussed in this section are:
— Equality checks: a == b and a != b
— Comparison operators: a b , a <= b , a >= b
— Range instantiation and range checks: a..b , x in a..b , x !in a..b
When the operands a and b are statically known to be Float or Double or their nullable
counterparts (the type is declared or inferred or is a result of a smart cast), the operations on the
numbers and the range that they form follow the IEEE 754 Standard for Floating-Point Arithmetic.
However, to support generic use cases and provide total ordering, when the operands are not
statically typed as oating point numbers (e.g. Any , Comparable<...> , a type parameter), the
operations use the equals and compareTo implementations for Float and Double , which
disagree with the standard, so that:
— NaN is considered equal to itself
— NaN is considered greater than any other element including POSITIVE_INFINITY

— -0.0 is considered less than 0.0
Characters
Characters are represented by the type Char . They can not be treated directly as numbers
fun check(c: Char) {

if (c == 1) { // ERROR: incompatible types
// ...
}
}
Character literals go in single quotes: '1' . Special characters can be escaped using a backslash.
The following escape sequences are supported: \t , \b , \n , \r , \' , \" , \\ and \$ . To
encode any other character, use the Unicode escape sequence syntax: '\uFF00' .
We can explicitly convert a character to an Int number:
fun decimalDigitValue(c: Char): Int {

if (c !in '0'..'9')
throw IllegalArgumentException("Out of range")
return c.toInt() - '0'.toInt() // Explicit conversions to numbers
}
Like numbers, characters are boxed when a nullable reference is needed. Identity is not
preserved by the boxing operation.
Booleans
The type Boolean represents booleans, and has two values: true and false.
Booleans are boxed if a nullable reference is needed.
79
Built-in operations on booleans include
— || – lazy disjunction
— && – lazy conjunction
— ! - negation
Arrays
Arrays in Kotlin are represented by the Array class, that has get and set functions (that turn
into [] by operator overloading conventions), and size property, along with a few other
useful member functions:
class Array<T> private constructor() {

val size: Int
operator fun get(index: Int): T
operator fun set(index: Int, value: T): Unit
operator fun iterator(): Iterator<T>

// ...
}
To create an array, we can use a library function arrayOf() and pass the item values to it, so
that arrayOf(1, 2, 3) creates an array [1, 2, 3] . Alternatively, the arrayOfNulls()
library function can be used to create an array of a given size lled with null elements.
Another option is to use the Array constructor that takes the array size and the function that
can return the initial value of each array element given its index:
// Creates an Array<String> with values ["0", "1", "4", "9", "16"]

val asc = Array(5) { i -> (i * i).toString() }
asc.forEach { println(it) }
As we said above, the [] operation stands for calls to member functions get() and set() .
Note: unlike Java, arrays in Kotlin are invariant. This means that Kotlin does not let us assign an
Array<String> to an Array<Any> , which prevents a possible runtime failure (but you can
use Array<out Any> , see Type Projections).
Kotlin also has specialized classes to represent arrays of primitive types without boxing
overhead: ByteArray , ShortArray , IntArray and so on. These classes have no inheritance
relation to the Array class, but they have the same set of methods and properties. Each of them
also has a corresponding factory function:
val x: IntArray = intArrayOf(1, 2, 3)

x[0] = x[1] + x[2]
Unsigned integers
80
Unsigned types are available only since Kotlin 1.3 and currently are experimental. See
details below
Kotlin introduces following types for unsigned integers:
— kotlin.UByte : an unsigned 8-bit integer, ranges from 0 to 255
— kotlin.UShort : an unsigned 16-bit integer, ranges from 0 to 65535
— kotlin.UInt : an unsigned 32-bit integer, ranges from 0 to 2^32 - 1
— kotlin.ULong : an unsigned 64-bit integer, ranges from 0 to 2^64 - 1
Unsigned types support most of the operations of their signed counterparts.
Note that changing type from unsigned type to signed counterpart (and vice versa) is a
binary incompatible change
Unsigned types are implemented using another experimental feature, namely inline classes.
Specialized classes
Same as for primitives, each of unsigned type has corresponding type that represents array,
specialized for that unsigned type:
— kotlin.UByteArray : an array of unsigned bytes
— kotlin.UShortArray : an array of unsigned shorts
— kotlin.UIntArray : an array of unsigned ints
— kotlin.ULongArray : an array of unsigned longs
Same as for signed integer arrays, they provide similar API to Array class without boxing
overhead.
Also, ranges and progressions supported for UInt and ULong by classes
kotlin.ranges.UIntRange , kotlin.ranges.UIntProgression ,
kotlin.ranges.ULongRange , kotlin.ranges.ULongProgression
Literals
To make unsigned integers easier to use, Kotlin provides an ability to tag an integer literal with a
su x indicating a speci c unsigned type (similarly to Float/Long):
— su xes u and U tag literal as unsigned. Exact type will be determined based on the expected
type. If no expected type is provided, UInt or ULong will be chosen based on the size of
literal
81
val b: UByte = 1u // UByte, expected type provided
val s: UShort = 1u // UShort, expected type provided
val l: ULong = 1u // ULong, expected type provided
val a1 = 42u // UInt: no expected type provided, constant fits in UInt

val a2 = 0xFFFF_FFFF_FFFFu // ULong: no expected type provided, constant doesn't fit in
UInt
— su xes uL and UL explicitly tag literal as unsigned long.
val a = 1UL // ULong, even though no expected type provided and constant fits into UInt
Experimental status of unsigned integers
The design of unsigned types is experimental, meaning that this feature is moving fast and no
compatibility guarantees are given. When using unsigned arithmetics in Kotlin 1.3+, warning will
be reported, indicating that this feature is experimental. To remove warning, you have to opt-in
for experimental usage of unsigned types.
There are two possible ways to opt-in for unsigned types: with marking your API as experimental
too, or without doing that.
— to propagate experimentality, either annotate declarations which use unsigned integers with
@ExperimentalUnsignedTypes or pass -
Xexperimental=kotlin.ExperimentalUnsignedTypes to the compiler (note that the
latter will make all declaration in compiled module experimental)
— to opt-in without propagating experimentality, either annotate declarations with

@UseExperimental(ExperimentalUnsignedTypes::class) or pass -Xuse-
experimental=kotlin.ExperimentalUnsignedTypes
It's up to you to decide if your clients have to explicitly opt-in into usage of your API, but bear in
mind that unsigned types are an experimental feature, so API which uses them can be suddenly
broken due to changes in language.
See also or Experimental API KEEP for technical details.
Further discussion
See language proposal for unsigned types for technical details and further discussion.
Strings
Strings are represented by the type String . Strings are immutable. Elements of a string are
characters that can be accessed by the indexing operation: s[i] . A string can be iterated over
with a for-loop:
82
for (c in str) {
println(c)
}
You can concatenate strings using the + operator. This also works for concatenating strings with
values of other types, as long as the rst element in the expression is a string:
val s = "abc" + 1
println(s + "def")
Note that in most cases using string templates or raw strings is preferable to string
concatenation.
String Literals
Kotlin has two types of string literals: escaped strings that may have escaped characters in them
and raw strings that can contain newlines and arbitrary text. An escaped string is very much like a
Java string:
val s = "Hello, world!\n"
Escaping is done in the conventional way, with a backslash. See Characters above for the list of
supported escape sequences.
A raw string is delimited by a triple quote ( """ ), contains no escaping and can contain newlines
and any other characters:
val text = """

for (c in "foo")
print(c)
"""
You can remove leading whitespace with trimMargin() function:
val text = """

|Tell me and I forget.
|Teach me and I remember.
|Involve me and I learn.
|(Benjamin Franklin)
""".trimMargin()
By default | is used as margin pre x, but you can choose another character and pass it as a
parameter, like trimMargin(">") .
String Templates
String literals may contain template expressions, i.e. pieces of code that are evaluated and whose
results are concatenated into the string. A template expression starts with a dollar sign ($) and
consists of either a simple name:
83
val i = 10
println("i = $i") // prints "i = 10"
or an arbitrary expression in curly braces:
val s = "abc"
println("$s.length is ${s.length}") // prints "abc.length is 3"
Templates are supported both inside raw strings and inside escaped strings. If you need to
represent a literal $ character in a raw string (which doesn't support backslash escaping), you
can use the following syntax:
val price = """

${'$'}9.99
"""
84
Packages
A source le may start with a package declaration:
package foo.bar
fun baz() { ... }

class Goo { ... }
// ...
All the contents (such as classes and functions) of the source le are contained by the package
declared. So, in the example above, the full name of baz() is foo.bar.baz , and the full name
of Goo is foo.bar.Goo .
If the package is not speci ed, the contents of such a le belong to "default" package that has no
name.
Default Imports
A number of packages are imported into every Kotlin le by default:
— kotlin.*
— kotlin.annotation.*
— kotlin.collections.*
— kotlin.comparisons.* (since 1.1)
— kotlin.io.*
— kotlin.ranges.*
— kotlin.sequences.*
— kotlin.text.*
Additional packages are imported depending on the target platform:
— JVM:
— java.lang.*
— kotlin.jvm.*
— JS:
— kotlin.js.*
Imports
Apart from the default imports, each le may contain its own import directives. Syntax for
imports is described in the grammar.
We can import either a single name, e.g.
85
import foo.Bar // Bar is now accessible without qualification
or all the accessible contents of a scope (package, class, object etc):
import foo.* // everything in 'foo' becomes accessible
If there is a name clash, we can disambiguate by using as keyword to locally rename the clashing
entity:
import foo.Bar // Bar is accessible

import bar.Bar as bBar // bBar stands for 'bar.Bar'
The import keyword is not restricted to importing classes; you can also use it to import other
declarations:
— top-level functions and properties;

— functions and properties declared in object declarations;
— enum constants.
Unlike Java, Kotlin does not have a separate "import static" syntax; all of these declarations are
imported using the regular import keyword.
Visibility of Top-level Declarations

If a top-level declaration is marked private, it is private to the le it's declared in (see Visibility
Modi ers).
86
Control Flow: if, when, for, while
If Expression
In Kotlin, if is an expression, i.e. it returns a value. Therefore there is no ternary operator
(condition ? then : else), because ordinary if works ne in this role.
// Traditional usage
var max = a
if (a < b) max = b
// With else
var max: Int
if (a > b) {
max = a
} else {
max = b
}
// As expression
val max = if (a > b) a else b
if branches can be blocks, and the last expression is the value of a block:
val max = if (a > b) {

print("Choose a")
a
} else {
print("Choose b")
b
}
If you're using if as an expression rather than a statement (for example, returning its value or
assigning it to a variable), the expression is required to have an else branch.
See the grammar for if.
When Expression
when replaces the switch operator of C-like languages. In the simplest form it looks like this
when (x) {
1 -> print("x == 1")
2 -> print("x == 2")
else -> { // Note the block
print("x is neither 1 nor 2")
}
}
when matches its argument against all branches sequentially until some branch condition is
satis ed. when can be used either as an expression or as a statement. If it is used as an
expression, the value of the satis ed branch becomes the value of the overall expression. If it is
used as a statement, the values of individual branches are ignored. (Just like with if, each branch
can be a block, and its value is the value of the last expression in the block.)
87
The else branch is evaluated if none of the other branch conditions are satis ed. If when is used
as an expression, the else branch is mandatory, unless the compiler can prove that all possible
cases are covered with branch conditions (as, for example, with enum class entries and sealed
class subtypes).
If many cases should be handled in the same way, the branch conditions may be combined with a
comma:
when (x) {
0, 1 -> print("x == 0 or x == 1")
else -> print("otherwise")
}
We can use arbitrary expressions (not only constants) as branch conditions
when (x) {
parseInt(s) -> print("s encodes x")
else -> print("s does not encode x")
}
We can also check a value for being in or !in a range or a collection:
when (x) {
in 1..10 -> print("x is in the range")
in validNumbers -> print("x is valid")
!in 10..20 -> print("x is outside the range")
else -> print("none of the above")
}
Another possibility is to check that a value is or !is of a particular type. Note that, due to smart
casts, you can access the methods and properties of the type without any extra checks.
fun hasPrefix(x: Any) = when(x) {

is String -> x.startsWith("prefix")
else -> false
}
when can also be used as a replacement for an if-else if chain. If no argument is supplied, the
branch conditions are simply boolean expressions, and a branch is executed when its condition is
true:
when {
x.isOdd() -> print("x is odd")
x.isEven() -> print("x is even")
else -> print("x is funny")
}
Since Kotlin 1.3, it is possible to capture when subject in a variable using following syntax:
88
fun Request.getBody() =
when (val response = executeRequest()) {
is Success -> response.body
is HttpError -> throw HttpException(response.status)
}
Scope of variable, introduced in when subject, is restricted to when body.
See the grammar for when.
For Loops
for loop iterates through anything that provides an iterator. This is equivalent to the foreach
loop in languages like C#. The syntax is as follows:
for (item in collection) print(item)
The body can be a block.
for (item: Int in ints) {

// ...
}
As mentioned before, for iterates through anything that provides an iterator, i.e.
— has a member- or extension-function iterator() , whose return type
— has a member- or extension-function next() , and
— has a member- or extension-function hasNext() that returns Boolean .
All of these three functions need to be marked as operator .
To iterate over a range of numbers, use a range expression:
for (i in 1..3) {
println(i)
}
for (i in 6 downTo 0 step 2) {
println(i)
}
A for loop over a range or an array is compiled to an index-based loop that does not create an
iterator object.
If you want to iterate through an array or a list with an index, you can do it this way:
for (i in array.indices) {
println(array[i])
}
Alternatively, you can use the withIndex library function:
89
for ((index, value) in array.withIndex()) {
println("the element at $index is $value")
}
See the grammar for for.
While Loops
while and do..while work as usual
while (x > 0) {
x--
}
do {
val y = retrieveData()
} while (y != null) // y is visible here!
See the grammar for while.
Break and continue in loops

Kotlin supports traditional break and continue operators in loops. See Returns and jumps.
90
Returns and Jumps
Kotlin has three structural jump expressions:
— return. By default returns from the nearest enclosing function or anonymous function.
— break. Terminates the nearest enclosing loop.
— continue. Proceeds to the next step of the nearest enclosing loop.
All of these expressions can be used as part of larger expressions:
val s = person.name ?: return
The type of these expressions is the Nothing type.
Break and Continue Labels

Any expression in Kotlin may be marked with a label. Labels have the form of an identi er
followed by the @ sign, for example: abc@ , fooBar@ are valid labels (see the grammar). To
label an expression, we just put a label in front of it
loop@ for (i in 1..100) {

// ...
}
Now, we can qualify a break or a continue with a label:
loop@ for (i in 1..100) {

for (j in 1..100) {
if (...) break@loop
}
}
A break quali ed with a label jumps to the execution point right after the loop marked with that
label. A continue proceeds to the next iteration of that loop.
Return at Labels
With function literals, local functions and object expression, functions can be nested in Kotlin.
Quali ed returns allow us to return from an outer function. The most important use case is
returning from a lambda expression. Recall that when we write this:
fun foo() {
listOf(1, 2, 3, 4, 5).forEach {
if (it == 3) return // non-local return directly to the caller of foo()
print(it)
}
println("this point is unreachable")
}
91
The return-expression returns from the nearest enclosing function, i.e. foo . (Note that such
non-local returns are supported only for lambda expressions passed to inline functions.) If we
need to return from a lambda expression, we have to label it and qualify the return:
fun foo() {
listOf(1, 2, 3, 4, 5).forEach lit@{
if (it == 3) return@lit // local return to the caller of the lambda, i.e. the
forEach loop
print(it)
}
print(" done with explicit label")
}
Now, it returns only from the lambda expression. Oftentimes it is more convenient to use implicit
labels: such a label has the same name as the function to which the lambda is passed.
fun foo() {
if (it == 3) return@forEach // local return to the caller of the lambda, i.e. the
forEach loop
print(it)
}
print(" done with implicit label")
}
Alternatively, we can replace the lambda expression with an anonymous function. A return
statement in an anonymous function will return from the anonymous function itself.
fun foo() {
listOf(1, 2, 3, 4, 5).forEach(fun(value: Int) {
if (value == 3) return // local return to the caller of the anonymous fun, i.e.
the forEach loop
print(value)
})
print(" done with anonymous function")
}
Note that the use of local returns in previous three examples is similar to the use of continue in
regular loops. There is no direct equivalent for break, but it can be simulated by adding another
nesting lambda and non-locally returning from it:
fun foo() {
run loop@{
if (it == 3) return@loop // non-local return from the lambda passed to run
print(it)
}
}
print(" done with nested loop")
}
When returning a value, the parser gives preference to the quali ed return, i.e.
return@a 1
92
means "return 1 at label @a " and not "return a labeled expression (@a 1) ".
93
Classes and Objects
Classes and Inheritance
Classes
Classes in Kotlin are declared using the keyword class:
class Invoice { ... }
The class declaration consists of the class name, the class header (specifying its type parameters,
the primary constructor etc.) and the class body, surrounded by curly braces. Both the header
and the body are optional; if the class has no body, curly braces can be omitted.
class Empty
Constructors
A class in Kotlin can have a primary constructor and one or more secondary constructors. The
primary constructor is part of the class header: it goes after the class name (and optional type
parameters).
class Person constructor(firstName: String) { ... }
If the primary constructor does not have any annotations or visibility modi ers, the
constructor keyword can be omitted:
class Person(firstName: String) { ... }
The primary constructor cannot contain any code. Initialization code can be placed in initializer
blocks, which are pre xed with the init keyword.
During an instance initialization, the initializer blocks are executed in the same order as they
appear in the class body, interleaved with the property initializers:
94
class InitOrderDemo(name: String) {
val firstProperty = "First property: $name".also(::println)
init {
println("First initializer block that prints ${name}")
}
val secondProperty = "Second property: ${name.length}".also(::println)
init {
println("Second initializer block that prints ${name.length}")
}
}
Note that parameters of the primary constructor can be used in the initializer blocks. They can
also be used in property initializers declared in the class body:
class Customer(name: String) {

val customerKey = name.toUpperCase()
}
In fact, for declaring properties and initializing them from the primary constructor, Kotlin has a
concise syntax:
class Person(val firstName: String, val lastName: String, var age: Int) { ... }
Much the same way as regular properties, the properties declared in the primary constructor can
be mutable (var) or read-only (val).
If the constructor has annotations or visibility modi ers, the constructor keyword is required,
and the modi ers go before it:
class Customer public @Inject constructor(name: String) { ... }
For more details, see Visibility Modi ers.
Secondary Constructors
The class can also declare secondary constructors, which are pre xed with constructor:
class Person {
constructor(parent: Person) {
parent.children.add(this)
}
}
If the class has a primary constructor, each secondary constructor needs to delegate to the
primary constructor, either directly or indirectly through another secondary constructor(s).
Delegation to another constructor of the same class is done using the this keyword:
95
class Person(val name: String) {
constructor(name: String, parent: Person) : this(name) {
parent.children.add(this)
}
}
Note that code in initializer blocks e ectively becomes part of the primary constructor.
Delegation to the primary constructor happens as the rst statement of a secondary constructor,
so the code in all initializer blocks is executed before the secondary constructor body. Even if the
class has no primary constructor, the delegation still happens implicitly, and the initializer blocks
are still executed:
class Constructors {
init {
println("Init block")
}
constructor(i: Int) {
println("Constructor")
}
}
If a non-abstract class does not declare any constructors (primary or secondary), it will have a
generated primary constructor with no arguments. The visibility of the constructor will be public.
If you do not want your class to have a public constructor, you need to declare an empty primary
constructor with non-default visibility:
class DontCreateMe private constructor () { ... }
NOTE: On the JVM, if all of the parameters of the primary constructor have default values,
the compiler will generate an additional parameterless constructor which will use the default
values. This makes it easier to use Kotlin with libraries such as Jackson or JPA that create class
instances through parameterless constructors.
class Customer(val customerName: String = "")
Creating instances of classes
To create an instance of a class, we call the constructor as if it were a regular function:
val invoice = Invoice()
val customer = Customer("Joe Smith")
Note that Kotlin does not have a new keyword.
Creating instances of nested, inner and anonymous inner classes is described in Nested classes.
Class Members
96
Classes can contain:
— Constructors and initializer blocks

— Functions
— Properties
— Nested and Inner Classes
— Object Declarations
Inheritance
All classes in Kotlin have a common superclass Any , that is the default superclass for a class with
no supertypes declared:
class Example // Implicitly inherits from Any
Note: Any is not java.lang.Object; in particular, it does not have any members other
than equals(), hashCode() and toString(). Please consult the Java interoperability
section for more details.
To declare an explicit supertype, we place the type after a colon in the class header:
open class Base(p: Int)
class Derived(p: Int) : Base(p)
If the derived class has a primary constructor, the base class can (and must) be initialized right
there, using the parameters of the primary constructor.
If the derived class has no primary constructor, then each secondary constructor has to initialize
the base type using the super keyword, or to delegate to another constructor which does that.
Note that in this case di erent secondary constructors can call di erent constructors of the base
type:
class MyView : View {

constructor(ctx: Context) : super(ctx)
constructor(ctx: Context, attrs: AttributeSet) : super(ctx, attrs)

}
Overriding Methods
As we mentioned before, we stick to making things explicit in Kotlin. And unlike Java, Kotlin
requires explicit modi ers for overridable members (we call them open) and for overrides:
97
open class Base {
open fun v() { ... }
fun nv() { ... }
}
class Derived() : Base() {
override fun v() { ... }
}
The override modi er is required for Derived.v() . If it were missing, the compiler would
complain. If there is no open modi er on a function, like Base.nv() , declaring a method with
the same signature in a subclass is illegal, either with override or without it. The open modi er
has no e ect when added on members of a nal class (i.e.. a class with no open modi er).
A member marked override is itself open, i.e. it may be overridden in subclasses. If you want to
prohibit re-overriding, use final:
open class AnotherDerived() : Base() {

final override fun v() { ... }
}
Overriding Properties
Overriding properties works in a similar way to overriding methods; properties declared on a

superclass that are then redeclared on a derived class must be prefaced with override, and
they must have a compatible type. Each declared property can be overridden by a property with
an initializer or by a property with a getter method.
open class Foo {

open val x: Int get() { ... }
}
class Bar1 : Foo() {

override val x: Int = ...
}
You can also override a val property with a var property, but not vice versa. This is allowed
because a val property essentially declares a getter method, and overriding it as a var
additionally declares a setter method in the derived class.
Note that you can use the override keyword as part of the property declaration in a primary
constructor.
interface Foo {
val count: Int
}
class Bar1(override val count: Int) : Foo
class Bar2 : Foo {

override var count: Int = 0
}
98
Derived class initialization order
During construction of a new instance of a derived class, the base class initialization is done as
the rst step (preceded only by evaluation of the arguments for the base class constructor) and
thus happens before the initialization logic of the derived class is run.
open class Base(val name: String) {
init { println("Initializing Base") }
open val size: Int =

name.length.also { println("Initializing size in Base: $it") }
}
class Derived(
name: String,
val lastName: String
) : Base(name.capitalize().also { println("Argument for Base: $it") }) {
init { println("Initializing Derived") }
override val size: Int =

(super.size + lastName.length).also { println("Initializing size in Derived:
$it") }
}
It means that, by the time of the base class constructor execution, the properties declared or
overridden in the derived class are not yet initialized. If any of those properties are used in the
base class initialization logic (either directly or indirectly, through another overridden open
member implementation), it may lead to incorrect behavior or a runtime failure. When designing
a base class, you should therefore avoid using open members in the constructors, property
initializers, and init blocks.
Calling the superclass implementation
Code in a derived class can call its superclass functions and property accessors implementations
using the super keyword:
open class Foo {

open fun f() { println("Foo.f()") }
open val x: Int get() = 1
}
class Bar : Foo() {

override fun f() {
super.f()
println("Bar.f()")
}
override val x: Int get() = super.x + 1

}
99
Inside an inner class, accessing the superclass of the outer class is done with the super keyword
quali ed with the outer class name: super@Outer :
class Bar : Foo() {

override fun f() { /* ... */ }
override val x: Int get() = 0
inner class Baz {

fun g() {
[email protected]() // Calls Foo's implementation of f()
println([email protected]) // Uses Foo's implementation of x's getter
}
}
}
Overriding Rules
In Kotlin, implementation inheritance is regulated by the following rule: if a class inherits many
implementations of the same member from its immediate superclasses, it must override this
member and provide its own implementation (perhaps, using one of the inherited ones). To
denote the supertype from which the inherited implementation is taken, we use super quali ed
by the supertype name in angle brackets, e.g. super<Base> :
open class A {
open fun f() { print("A") }
fun a() { print("a") }
}
interface B {
fun f() { print("B") } // interface members are 'open' by default
fun b() { print("b") }
}
class C() : A(), B {

// The compiler requires f() to be overridden:
override fun f() {
super<A>.f() // call to A.f()
super.f() // call to B.f()
}
}
It's ne to inherit from both A and B , and we have no problems with a() and b() since C
inherits only one implementation of each of these functions. But for f() we have two
implementations inherited by C , and thus we have to override f() in C and provide our own
implementation that eliminates the ambiguity.
Abstract Classes
A class and some of its members may be declared abstract. An abstract member does not have
an implementation in its class. Note that we do not need to annotate an abstract class or function
with open – it goes without saying.
100
We can override a non-abstract open member with an abstract one
open class Base {

open fun f() {}
}
abstract class Derived : Base() {

override abstract fun f()
}
Companion Objects
In Kotlin, unlike Java or C#, classes do not have static methods. In most cases, it's recommended
to simply use package-level functions instead.
If you need to write a function that can be called without having a class instance but needs access
to the internals of a class (for example, a factory method), you can write it as a member of an
object declaration inside that class.
Even more speci cally, if you declare a companion object inside your class, you'll be able to call
its members with the same syntax as calling static methods in Java/C#, using only the class name
as a quali er.
101
Properties and Fields
Declaring Properties
Properties in Kotlin classes can be declared either as mutable using the var keyword, or as read-
only using the val keyword.
class Address {
var name: String = ...
var street: String = ...
var city: String = ...
var state: String? = ...
var zip: String = ...
}
To use a property, we simply refer to it by name, as if it were a eld in Java:
fun copyAddress(address: Address): Address {

val result = Address() // there's no 'new' keyword in Kotlin
result.name = address.name // accessors are called
result.street = address.street
// ...
return result
}
Getters and Setters

The full syntax for declaring a property is
var <propertyName>[: <PropertyType>] [= <property_initializer>]

[<getter>]
[<setter>]
The initializer, getter and setter are optional. Property type is optional if it can be inferred from
the initializer (or from the getter return type, as shown below).
Examples:
var allByDefault: Int? // error: explicit initializer required, default getter and setter
implied
var initialized = 1 // has type Int, default getter and setter
The full syntax of a read-only property declaration di ers from a mutable one in two ways: it
starts with val instead of var and does not allow a setter:
val simple: Int? // has type Int, default getter, must be initialized in constructor
val inferredType = 1 // has type Int and a default getter
We can de ne custom accessors for a property. If we de ne a custom getter, it will be called

every time we access the property (this allows us to implement a computed property). Here's an
example of a custom getter:
102
val isEmpty: Boolean
get() = this.size == 0
If we de ne a custom setter, it will be called every time we assign a value to the property. A
custom setter looks like this:
var stringRepresentation: String

get() = this.toString()
set(value) {
setDataFromString(value) // parses the string and assigns values to other
properties
}
By convention, the name of the setter parameter is value , but you can choose a di erent name
if you prefer.
Since Kotlin 1.1, you can omit the property type if it can be inferred from the getter:
val isEmpty get() = this.size == 0 // has type Boolean
If you need to change the visibility of an accessor or to annotate it, but don't need to change the
default implementation, you can de ne the accessor without de ning its body:
var setterVisibility: String = "abc"

private set // the setter is private and has the default implementation
var setterWithAnnotation: Any? = null

@Inject set // annotate the setter with Inject
Backing Fields
Fields cannot be declared directly in Kotlin classes. However, when a property needs a backing
eld, Kotlin provides it automatically. This backing eld can be referenced in the accessors using
the field identi er:
var counter = 0 // Note: the initializer assigns the backing field directly
set(value) {
if (value >= 0) field = value
}
The field identi er can only be used in the accessors of the property.
A backing eld will be generated for a property if it uses the default implementation of at least
one of the accessors, or if a custom accessor references it through the field identi er.
For example, in the following case there will be no backing eld:
val isEmpty: Boolean

get() = this.size == 0
Backing Properties
103
If you want to do something that does not t into this "implicit backing eld" scheme, you can
always fall back to having a backing property:
private var _table: Map<String, Int>? = null

public val table: Map<String, Int>
get() {
if (_table == null) {
_table = HashMap() // Type parameters are inferred
}
return _table ?: throw AssertionError("Set to null by another thread")
}
In all respects, this is just the same as in Java since access to private properties with default
getters and setters is optimized so that no function call overhead is introduced.
Compile-Time Constants
Properties the value of which is known at compile time can be marked as compile time constants
using the const modi er. Such properties need to ful l the following requirements:
— Top-level, or member of an object declaration or a companion object.
— Initialized with a value of type String or a primitive type
— No custom getter
Such properties can be used in annotations:
const val SUBSYSTEM_DEPRECATED: String = "This subsystem is deprecated"
@Deprecated(SUBSYSTEM_DEPRECATED) fun foo() { ... }
Late-Initialized Properties and Variables

Normally, properties declared as having a non-null type must be initialized in the constructor.
However, fairly often this is not convenient. For example, properties can be initialized through
dependency injection, or in the setup method of a unit test. In this case, you cannot supply a non-
null initializer in the constructor, but you still want to avoid null checks when referencing the
property inside the body of a class.
To handle this case, you can mark the property with the lateinit modi er:
public class MyTest {

lateinit var subject: TestSubject
@SetUp fun setup() {

subject = TestSubject()
}
@Test fun test() {

subject.method() // dereference directly
}
}
104
The modi er can be used on var properties declared inside the body of a class (not in the
primary constructor, and only when the property does not have a custom getter or setter) and,
since Kotlin 1.2, for top-level properties and local variables. The type of the property or variable
must be non-null, and it must not be a primitive type.
Accessing a lateinit property before it has been initialized throws a special exception that
clearly identi es the property being accessed and the fact that it hasn't been initialized.
Checking whether a lateinit var is initialized (since 1.2)
To check whether a lateinit var has already been initialized, use .isInitialized on the
reference to that property:
if (foo::bar.isInitialized) {
println(foo.bar)
}
This check is only available for the properties that are lexically accessible, i.e. declared in the
same type or in one of the outer types, or at top level in the same le.
Overriding Properties
See Overriding Properties
Delegated Properties
The most common kind of properties simply reads from (and maybe writes to) a backing eld. On
the other hand, with custom getters and setters one can implement any behaviour of a property.
Somewhere in between, there are certain common patterns of how a property may work. A few
examples: lazy values, reading from a map by a given key, accessing a database, notifying listener
on access, etc.
Such common behaviours can be implemented as libraries using delegated properties.
105
Interfaces
Interfaces in Kotlin are very similar to Java 8. They can contain declarations of abstract methods,
as well as method implementations. What makes them di erent from abstract classes is that
interfaces cannot store state. They can have properties but these need to be abstract or to
provide accessor implementations.
An interface is de ned using the keyword interface
interface MyInterface {
fun bar()
fun foo() {
// optional body
}
}
Implementing Interfaces
A class or object can implement one or more interfaces
class Child : MyInterface {

override fun bar() {
// body
}
}
Properties in Interfaces
You can declare properties in interfaces. A property declared in an interface can either be
abstract, or it can provide implementations for accessors. Properties declared in interfaces can't
have backing elds, and therefore accessors declared in interfaces can't reference them.
interface MyInterface {
val prop: Int // abstract
val propertyWithImplementation: String

get() = "foo"
fun foo() {
print(prop)
}
}
class Child : MyInterface {

override val prop: Int = 29
}
Interfaces Inheritance
An interface can derive from other interfaces and thus both provide implementations for their
members and declare new functions and properties. Quite naturally, classes implementing such
an interface are only required to de ne the missing implementations:
106
interface Named {
val name: String
}
interface Person : Named {

val firstName: String
val lastName: String
override val name: String get() = "$firstName $lastName"

}
data class Employee(

// implementing 'name' is not required
override val firstName: String,
override val lastName: String,
val position: Position
) : Person
Resolving overriding con icts

When we declare many types in our supertype list, it may appear that we inherit more than one
implementation of the same method. For example
interface A {
fun foo() { print("A") }
fun bar()
}
interface B {
fun foo() { print("B") }
fun bar() { print("bar") }
}
class C : A {
override fun bar() { print("bar") }
}
class D : A, B {
override fun foo() {
super<A>.foo()
super.foo()
}
override fun bar() {

super.bar()
}
}
Interfaces A and B both declare functions foo() and bar(). Both of them implement foo(), but only B
implements bar() (bar() is not marked abstract in A, because this is the default for interfaces, if the
function has no body). Now, if we derive a concrete class C from A, we, obviously, have to
override bar() and provide an implementation.
107
However, if we derive D from A and B, we need to implement all the methods which we have
inherited from multiple interfaces, and to specify how exactly D should implement them. This
rule applies both to methods for which we've inherited a single implementation (bar()) and
multiple implementations (foo()).
108
Visibility Modi ers
Classes, objects, interfaces, constructors, functions, properties and their setters can have visibility
modi ers. (Getters always have the same visibility as the property.) There are four visibility
modi ers in Kotlin: private , protected , internal and public . The default visibility, used
if there is no explicit modi er, is public .
Below please nd explanations of how the modi ers apply to di erent types of declaring scopes.
Packages
Functions, properties and classes, objects and interfaces can be declared on the "top-level", i.e.
directly inside a package:
// file name: example.kt

package foo
fun baz() { ... }

class Bar { ... }
— If you do not specify any visibility modi er, public is used by default, which means that your
declarations will be visible everywhere;
— If you mark a declaration private , it will only be visible inside the le containing the
declaration;
— If you mark it internal , it is visible everywhere in the same module;
— protected is not available for top-level declarations.
Note: to use a visible top-level declaration from another package, you should still import it.
Examples:
// file name: example.kt

package foo
private fun foo() { ... } // visible inside example.kt
public var bar: Int = 5 // property is visible everywhere

private set // setter is visible only in example.kt
internal val baz = 6 // visible inside the same module
Classes and Interfaces

For members declared inside a class:
— private means visible inside this class only (including all its members);
— protected — same as private + visible in subclasses too;
— internal — any client inside this module who sees the declaring class sees its internal
members;
109
— public — any client who sees the declaring class sees its public members.
NOTE for Java users: outer class does not see private members of its inner classes in Kotlin.
If you override a protected member and do not specify the visibility explicitly, the overriding
member will also have protected visibility.
Examples:
open class Outer {

private val a = 1
protected open val b = 2
internal val c = 3
val d = 4 // public by default
protected class Nested {

public val e: Int = 5
}
}
class Subclass : Outer() {

// a is not visible
// b, c and d are visible
// Nested and e are visible
override val b = 5 // 'b' is protected

}
class Unrelated(o: Outer) {

// o.a, o.b are not visible
// o.c and o.d are visible (same module)
// Outer.Nested is not visible, and Nested::e is not visible either
}
Constructors
To specify a visibility of the primary constructor of a class, use the following syntax (note that you
need to add an explicit constructor keyword):
class C private constructor(a: Int) { ... }
Here the constructor is private. By default, all constructors are public , which e ectively
amounts to them being visible everywhere where the class is visible (i.e. a constructor of an
internal class is only visible within the same module).
Local declarations
Local variables, functions and classes can not have visibility modi ers.
Modules
The internal visibility modi er means that the member is visible within the same module.
More speci cally, a module is a set of Kotlin les compiled together:
110
— an IntelliJ IDEA module;
— a Maven project;
— a Gradle source set (with the exception that the test source set can access the internal
declarations of main );
— a set of les compiled with one invocation of the <kotlinc> Ant task.
111
Extensions
Kotlin, similar to C# and Gosu, provides the ability to extend a class with new functionality
without having to inherit from the class or use any type of design pattern such as Decorator. This
is done via special declarations called extensions. Kotlin supports extension functions and extension
properties.
Extension Functions
To declare an extension function, we need to pre x its name with a receiver type, i.e. the type
being extended. The following adds a swap function to MutableList<Int> :
fun MutableList<Int>.swap(index1: Int, index2: Int) {

val tmp = this[index1] // 'this' corresponds to the list
this[index1] = this[index2]
this[index2] = tmp
}
The this keyword inside an extension function corresponds to the receiver object (the one that
is passed before the dot). Now, we can call such a function on any MutableList<Int> :
val l = mutableListOf(1, 2, 3)
l.swap(0, 2) // 'this' inside 'swap()' will hold the value of 'l'
Of course, this function makes sense for any MutableList<T> , and we can make it generic:
fun <T> MutableList<T>.swap(index1: Int, index2: Int) {

val tmp = this[index1] // 'this' corresponds to the list
this[index1] = this[index2]
this[index2] = tmp
}
We declare the generic type parameter before the function name for it to be available in the
receiver type expression. See Generic functions.
Extensions are resolved statically

Extensions do not actually modify classes they extend. By de ning an extension, you do not
insert new members into a class, but merely make new functions callable with the dot-notation
on variables of this type.
We would like to emphasize that extension functions are dispatched statically, i.e. they are not
virtual by receiver type. This means that the extension function being called is determined by the
type of the expression on which the function is invoked, not by the type of the result of
evaluating that expression at runtime. For example:
112
open class C
class D: C()
fun C.foo() = "c"
fun D.foo() = "d"
fun printFoo(c: C) {
println(c.foo())
}
printFoo(D())
This example will print "c", because the extension function being called depends only on the
declared type of the parameter c , which is the C class.
If a class has a member function, and an extension function is de ned which has the same
receiver type, the same name is applicable to given arguments, the member always wins. For
example:
class C {
fun foo() { println("member") }
}
fun C.foo() { println("extension") }
If we call c.foo() of any c of type C , it will print "member", not "extension".
However, it's perfectly OK for extension functions to overload member functions which have the
same name but a di erent signature:
class C {
fun foo() { println("member") }
}
fun C.foo(i: Int) { println("extension") }
The call to C().foo(1) will print "extension".
Nullable Receiver
Note that extensions can be de ned with a nullable receiver type. Such extensions can be called
on an object variable even if its value is null, and can check for this == null inside the body.
This is what allows you to call toString() in Kotlin without checking for null: the check happens
inside the extension function.
fun Any?.toString(): String {

if (this == null) return "null"
// after the null check, 'this' is autocast to a non-null type, so the toString()
below
// resolves to the member function of the Any class
return toString()
}
113
Extension Properties
Similarly to functions, Kotlin supports extension properties:
val <T> List<T>.lastIndex: Int

get() = size - 1
Note that, since extensions do not actually insert members into classes, there's no e cient way
for an extension property to have a backing eld. This is why initializers are not allowed for
extension properties. Their behavior can only be de ned by explicitly providing getters/setters.
Example:
val Foo.bar = 1 // error: initializers are not allowed for extension properties
Companion Object Extensions

If a class has a companion object de ned, you can also de ne extension functions and properties
for the companion object:
class MyClass {
companion object { } // will be called "Companion"
}
fun MyClass.Companion.foo() { ... }
Just like regular members of the companion object, they can be called using only the class name
as the quali er:
MyClass.foo()
Scope of Extensions
Most of the time we de ne extensions on the top level, i.e. directly under packages:
package foo.bar
fun Baz.goo() { ... }
To use such an extension outside its declaring package, we need to import it at the call site:
package com.example.usage
import foo.bar.goo // importing all extensions by name "goo"

// or
import foo.bar.* // importing everything from "foo.bar"
fun usage(baz: Baz) {

baz.goo()
}
See Imports for more information.
114
Declaring Extensions as Members
Inside a class, you can declare extensions for another class. Inside such an extension, there are
multiple implicit receivers - objects members of which can be accessed without a quali er. The
instance of the class in which the extension is declared is called dispatch receiver, and the instance
of the receiver type of the extension method is called extension receiver.
class D {
fun bar() { ... }
}
class C {
fun baz() { ... }
fun D.foo() {
bar() // calls D.bar
baz() // calls C.baz
}
fun caller(d: D) {
d.foo() // call the extension function
}
}
In case of a name con ict between the members of the dispatch receiver and the extension
receiver, the extension receiver takes precedence. To refer to the member of the dispatch
receiver you can use the quali ed this syntax.
class C {
fun D.foo() {
toString() // calls D.toString()
[email protected]() // calls C.toString()
}
}
Extensions declared as members can be declared as open and overridden in subclasses. This
means that the dispatch of such functions is virtual with regard to the dispatch receiver type, but
static with regard to the extension receiver type.
115
open class D { }
class D1 : D() { }
open class C {
open fun D.foo() {
println("D.foo in C")
}
open fun D1.foo() {

println("D1.foo in C")
}
fun caller(d: D) {
d.foo() // call the extension function
}
}
class C1 : C() {
override fun D.foo() {
println("D.foo in C1")
}
override fun D1.foo() {

println("D1.foo in C1")
}
}
fun main() {
C().caller(D()) // prints "D.foo in C"
C1().caller(D()) // prints "D.foo in C1" - dispatch receiver is resolved virtually
C().caller(D1()) // prints "D.foo in C" - extension receiver is resolved statically
}
Note on visibility
Extensions utilize the same visibility of other entities as regular functions declared in the same
scope would. For example:
— An extension declared on top level of a le has access to the other private top-level
declarations in the same le;
— If an extension is declared outside its receiver type, such an extension cannot access the
receiver's private members.
Motivation
In Java, we are used to classes named "*Utils": FileUtils , StringUtils and so on. The
famous java.util.Collections belongs to the same breed. And the unpleasant part about
these Utils-classes is that the code that uses them looks like this:
// Java
Collections.swap(list, Collections.binarySearch(list,
Collections.max(otherList)),
Collections.max(list));
116
Those class names are always getting in the way. We can use static imports and get this:
// Java
swap(list, binarySearch(list, max(otherList)), max(list));
This is a little better, but we have no or little help from the powerful code completion of the IDE.
It would be so much better if we could say:
// Java
list.swap(list.binarySearch(otherList.max()), list.max());
But we don't want to implement all the possible methods inside the class List , right? This is
where extensions help us.
117
Data Classes
We frequently create classes whose main purpose is to hold data. In such a class some standard
functionality and utility functions are often mechanically derivable from the data. In Kotlin, this is
called a data class and is marked as data :
data class User(val name: String, val age: Int)
The compiler automatically derives the following members from all properties declared in the
primary constructor:
— equals() / hashCode() pair;
— toString() of the form "User(name=John, age=42)" ;
— componentN() functions corresponding to the properties in their order of declaration;
— copy() function (see below).
To ensure consistency and meaningful behavior of the generated code, data classes have to ful ll
the following requirements:
— The primary constructor needs to have at least one parameter;

— All primary constructor parameters need to be marked as val or var ;
— Data classes cannot be abstract, open, sealed or inner;

— (before 1.1) Data classes may only implement interfaces.
Additionally, the members generation follows these rules with regard to the members
inheritance:
— If there are explicit implementations of equals() , hashCode() or toString() in the

data class body or final implementations in a superclass, then these functions are not
generated, and the existing implementations are used;
— If a supertype has the componentN() functions that are open and return compatible types,
the corresponding functions are generated for the data class and override those of the
supertype. If the functions of the supertype cannot be overridden due to incompatible
signatures or being nal, an error is reported;
— Deriving a data class from a type that already has a copy(...) function with a matching
signature is deprecated in Kotlin 1.2 and is prohibited in Kotlin 1.3.
— Providing explicit implementations for the componentN() and copy() functions is not
allowed.
Since 1.1, data classes may extend other classes (see Sealed classes for examples).
On the JVM, if the generated class needs to have a parameterless constructor, default values for
all properties have to be speci ed (see Constructors).
118
data class User(val name: String = "", val age: Int = 0)
Properties Declared in the Class Body

Note that the compiler only uses the properties de ned inside the primary constructor for the
automatically generated functions. To exclude a property from the generated implementations,
declare it inside the class body:
data class Person(val name: String) {

var age: Int = 0
}
Only the property name will be used inside the toString() , equals() , hashCode() , and
copy() implementations, and there will only be one component function component1() .
While two Person objects can have di erent ages, they will be treated as equal.
val person1 = Person("John")

val person2 = Person("John")
person1.age = 10
person2.age = 20
Copying
It's often the case that we need to copy an object altering some of its properties, but keeping the
rest unchanged. This is what copy() function is generated for. For the User class above, its
implementation would be as follows:
fun copy(name: String = this.name, age: Int = this.age) = User(name, age)
This allows us to write:
val jack = User(name = "Jack", age = 1)

val olderJack = jack.copy(age = 2)
Data Classes and Destructuring Declarations

Component functions generated for data classes enable their use in destructuring declarations:
val jane = User("Jane", 35)

val (name, age) = jane
println("$name, $age years of age") // prints "Jane, 35 years of age"
Standard Data Classes

The standard library provides Pair and Triple . In most cases, though, named data classes are
a better design choice, because they make the code more readable by providing meaningful
names for properties.
119
Sealed Classes
Sealed classes are used for representing restricted class hierarchies, when a value can have one
of the types from a limited set, but cannot have any other type. They are, in a sense, an extension
of enum classes: the set of values for an enum type is also restricted, but each enum constant
exists only as a single instance, whereas a subclass of a sealed class can have multiple instances
which can contain state.
To declare a sealed class, you put the sealed modi er before the name of the class. A sealed
class can have subclasses, but all of them must be declared in the same le as the sealed class
itself. (Before Kotlin 1.1, the rules were even more strict: classes had to be nested inside the
declaration of the sealed class).
sealed class Expr

data class Const(val number: Double) : Expr()
data class Sum(val e1: Expr, val e2: Expr) : Expr()
object NotANumber : Expr()
(The example above uses one additional new feature of Kotlin 1.1: the possibility for data classes
to extend other classes, including sealed classes.)
A sealed class is abstract by itself, it cannot be instantiated directly and can have abstract
members.
Sealed classes are not allowed to have non-private constructors (their constructors are
private by default).
Note that classes which extend subclasses of a sealed class (indirect inheritors) can be placed
anywhere, not necessarily in the same le.
The key bene t of using sealed classes comes into play when you use them in a when expression.
If it's possible to verify that the statement covers all cases, you don't need to add an else clause
to the statement. However, this works only if you use when as an expression (using the result)
and not as a statement.
fun eval(expr: Expr): Double = when(expr) {

is Const -> expr.number
is Sum -> eval(expr.e1) + eval(expr.e2)
NotANumber -> Double.NaN
// the èlse` clause is not required because we've covered all the cases
}
120
Generics
As in Java, classes in Kotlin may have type parameters:
class Box<T>(t: T) {
var value = t
}
In general, to create an instance of such a class, we need to provide the type arguments:
val box: Box<Int> = Box<Int>(1)
But if the parameters may be inferred, e.g. from the constructor arguments or by some other
means, one is allowed to omit the type arguments:
val box = Box(1) // 1 has type Int, so the compiler figures out that we are talking about
Box<Int>
Variance
One of the most tricky parts of Java's type system is wildcard types (see Java Generics FAQ). And
Kotlin doesn't have any. Instead, it has two other things: declaration-site variance and type
projections.
First, let's think about why Java needs those mysterious wildcards. The problem is explained in
E ective Java, 3rd Edition, Item 31: Use bounded wildcards to increase API exibility. First, generic
types in Java are invariant, meaning that List<String> is not a subtype of List<Object> .
Why so? If List was not invariant, it would have been no better than Java's arrays, since the
following code would have compiled and caused an exception at runtime:
// Java
List<String> strs = new ArrayList<String>();
List<Object> objs = strs; // !!! The cause of the upcoming problem sits here. Java
prohibits this!
objs.add(1); // Here we put an Integer into a list of Strings
String s = strs.get(0); // !!! ClassCastException: Cannot cast Integer to String
So, Java prohibits such things in order to guarantee run-time safety. But this has some
implications. For example, consider the addAll() method from Collection interface. What's
the signature of this method? Intuitively, we'd put it this way:
// Java
interface Collection<E> ... {
void addAll(Collection<E> items);
}
But then, we would not be able to do the following simple thing (which is perfectly safe):
121
// Java
void copyAll(Collection<Object> to, Collection<String> from) {
to.addAll(from);
// !!! Would not compile with the naive declaration of addAll:
// Collection<String> is not a subtype of Collection<Object>
}
(In Java, we learned this lesson the hard way, see E ective Java, 3rd Edition, Item 28: Prefer lists to
arrays)
That's why the actual signature of addAll() is the following:
// Java
interface Collection<E> ... {
void addAll(Collection<? extends E> items);
}
The wildcard type argument ? extends E indicates that this method accepts a collection of
objects of E or some subtype of E , not just E itself. This means that we can safely read E 's from
items (elements of this collection are instances of a subclass of E), but cannot write to it since we
do not know what objects comply to that unknown subtype of E . In return for this limitation, we
have the desired behaviour: Collection<String> is a subtype of Collection<? extends
Object> . In "clever words", the wildcard with an extends-bound (upper bound) makes the type
covariant.
The key to understanding why this trick works is rather simple: if you can only take items from a
collection, then using a collection of String s and reading Object s from it is ne. Conversely,
if you can only put items into the collection, it's OK to take a collection of Object s and put
String s into it: in Java we have List<? super String> a supertype of List<Object> .
The latter is called contravariance, and you can only call methods that take String as an
argument on List<? super String> (e.g., you can call add(String) or set(int,
String) ), while if you call something that returns T in List<T> , you don't get a String , but
an Object .
Joshua Bloch calls those objects you only read from Producers, and those you only write to
Consumers. He recommends: "For maximum exibility, use wildcard types on input parameters that
represent producers or consumers", and proposes the following mnemonic:
PECS stands for Producer-Extends, Consumer-Super.
NOTE: if you use a producer-object, say, List<? extends Foo> , you are not allowed to call
add() or set() on this object, but this does not mean that this object is immutable: for
example, nothing prevents you from calling clear() to remove all items from the list, since
clear() does not take any parameters at all. The only thing guaranteed by wildcards (or other
types of variance) is type safety. Immutability is a completely di erent story.
122
Declaration-site variance
Suppose we have a generic interface Source<T> that does not have any methods that take T
as a parameter, only methods that return T :
// Java
interface Source<T> {
T nextT();
}
Then, it would be perfectly safe to store a reference to an instance of Source<String> in a

variable of type Source<Object> – there are no consumer-methods to call. But Java does not
know this, and still prohibits it:
// Java
void demo(Source<String> strs) {
Source<Object> objects = strs; // !!! Not allowed in Java
// ...
}
To x this, we have to declare objects of type Source<? extends Object> , which is sort of
meaningless, because we can call all the same methods on such a variable as before, so there's
no value added by the more complex type. But the compiler does not know that.
In Kotlin, there is a way to explain this sort of thing to the compiler. This is called declaration-site
variance: we can annotate the type parameter T of Source to make sure that it is only
returned (produced) from members of Source<T> , and never consumed. To do this we
provide the out modi er:
interface Source<out T> {

fun nextT(): T
}
fun demo(strs: Source<String>) {

val objects: Source<Any> = strs // This is OK, since T is an out-parameter
// ...
}
The general rule is: when a type parameter T of a class C is declared out, it may occur only in
out-position in the members of C , but in return C<Base> can safely be a supertype of
C<Derived> .
In "clever words" they say that the class C is covariant in the parameter T , or that T is a
covariant type parameter. You can think of C as being a producer of T 's, and NOT a consumer
of T 's.
The out modi er is called a variance annotation, and since it is provided at the type parameter
declaration site, we talk about declaration-site variance. This is in contrast with Java's use-site
variance where wildcards in the type usages make the types covariant.
123
In addition to out, Kotlin provides a complementary variance annotation: in. It makes a type
parameter contravariant: it can only be consumed and never produced. A good example of a
contravariant type is Comparable :
interface Comparable<in T> {

operator fun compareTo(other: T): Int
}
fun demo(x: Comparable<Number>) {

x.compareTo(1.0) // 1.0 has type Double, which is a subtype of Number
// Thus, we can assign x to a variable of type Comparable<Double>
val y: Comparable<Double> = x // OK!
}
We believe that the words in and out are self-explaining (as they were successfully used in C# for
quite some time already), thus the mnemonic mentioned above is not really needed, and one can
rephrase it for a higher purpose:
The Existential Transformation: Consumer in, Producer out! :-)
Type projections
Use-site variance: Type projections
It is very convenient to declare a type parameter T as out and avoid trouble with subtyping on the
use site, but some classes can't actually be restricted to only return T 's! A good example of this
is Array:
class Array<T>(val size: Int) {

fun get(index: Int): T { ... }
fun set(index: Int, value: T) { ... }
}
This class cannot be either co- or contravariant in T . And this imposes certain in exibilities.
Consider the following function:
fun copy(from: Array<Any>, to: Array<Any>) {

assert(from.size == to.size)
for (i in from.indices)
to[i] = from[i]
}
This function is supposed to copy items from one array to another. Let's try to apply it in practice:
val ints: Array<Int> = arrayOf(1, 2, 3)

val any = Array<Any>(3) { "" }
copy(ints, any)
// ^ type is Array<Int> but Array<Any> was expected
124
Here we run into the same familiar problem: Array<T> is invariant in T , thus neither of
Array<Int> and Array<Any> is a subtype of the other. Why? Again, because copy might be
doing bad things, i.e. it might attempt to write, say, a String to from , and if we actually passed an
array of Int there, a ClassCastException would have been thrown sometime later.
Then, the only thing we want to ensure is that copy() does not do any bad things. We want to
prohibit it from writing to from , and we can:
fun copy(from: Array<out Any>, to: Array<Any>) { ... }
What has happened here is called type projection: we said that from is not simply an array, but
a restricted (projected) one: we can only call those methods that return the type parameter T , in
this case it means that we can only call get() . This is our approach to use-site variance, and
corresponds to Java's Array<? extends Object> , but in a slightly simpler way.
You can project a type with in as well:
fun fill(dest: Array<in String>, value: String) { ... }
Array<in String> corresponds to Java's Array<? super String> , i.e. you can pass an
array of CharSequence or an array of Object to the fill() function.
Star-projections
Sometimes you want to say that you know nothing about the type argument, but still want to use
it in a safe way. The safe way here is to de ne such a projection of the generic type, that every
concrete instantiation of that generic type would be a subtype of that projection.
Kotlin provides so called star-projection syntax for this:
— For Foo<out T : TUpper> , where T is a covariant type parameter with the upper bound
TUpper , Foo<*> is equivalent to Foo<out TUpper> . It means that when the T is
unknown you can safely read values of TUpper from Foo<*> .
— For Foo<in T> , where T is a contravariant type parameter, Foo<*> is equivalent to

Foo<in Nothing> . It means there is nothing you can write to Foo<*> in a safe way when
T is unknown.
— For Foo<T : TUpper> , where T is an invariant type parameter with the upper bound
TUpper , Foo<*> is equivalent to Foo<out TUpper> for reading values and to Foo<in
Nothing> for writing values.
If a generic type has several type parameters each of them can be projected independently. For
example, if the type is declared as interface Function<in T, out U> we can imagine the
following star-projections:
— Function<*, String> means Function<in Nothing, String> ;
125
— Function<Int, *> means Function<Int, out Any?> ;
— Function<*, *> means Function<in Nothing, out Any?> .
Note: star-projections are very much like Java's raw types, but safe.
Generic functions
Not only classes can have type parameters. Functions can, too. Type parameters are placed
before the name of the function:
fun <T> singletonList(item: T): List<T> {

// ...
}
fun <T> T.basicToString(): String { // extension function

// ...
}
To call a generic function, specify the type arguments at the call site after the name of the
function:
val l = singletonList<Int>(1)
Type arguments can be omitted if they can be inferred from the context, so the following
example works as well:
val l = singletonList(1)
Generic constraints
The set of all possible types that can be substituted for a given type parameter may be restricted
by generic constraints.
Upper bounds
The most common type of constraint is an upper bound that corresponds to Java's extends
keyword:
fun <T : Comparable<T>> sort(list: List<T>) { ... }
The type speci ed after a colon is the upper bound: only a subtype of Comparable<T> may be
substituted for T . For example:
sort(listOf(1, 2, 3)) // OK. Int is a subtype of Comparable<Int>

sort(listOf(HashMap<Int, String>())) // Error: HashMap<Int, String> is not a subtype of
Comparable<HashMap<Int, String>>
The default upper bound (if none speci ed) is Any? . Only one upper bound can be speci ed
inside the angle brackets. If the same type parameter needs more than one upper bound, we
need a separate where-clause:
126
fun <T> copyWhenGreater(list: List<T>, threshold: T): List<String>
where T : CharSequence,
T : Comparable<T> {
return list.filter { it > threshold }.map { it.toString() }
}
The passed type must satisfy all conditions of the where clause simultaneously. In the above
example, the T type must implement both CharSequence and Comparable .
Type erasure
The type safety checks that Kotlin performs for generic declaration usages are only done at
compile time. At runtime, the instances of generic types do not hold any information about their
actual type arguments. The type information is said to be erased. For example, the instances of
Foo<Bar> and Foo<Baz?> are erased to just Foo<*> .
Therefore, there is no general way to check whether an instance of a generic type was created
with certain type arguments at runtime, and the compiler prohibits such is-checks.
Type casts to generic types with concrete type arguments, e.g. foo as List<String> , cannot
be checked at runtime.
These unchecked casts can be used when type safety is implied by the high-level program logic
but cannot be inferred directly by the compiler. The compiler issues a warning on unchecked
casts, and at runtime, only the non-generic part is checked (equivalent to foo as List<*> ).
The type arguments of generic function calls are also only checked at compile time. Inside the
function bodies, the type parameters cannot be used for type checks, and type casts to type
parameters ( foo as T ) are unchecked. However, rei ed type parameters of inline functions
are substituted by the actual type arguments in the inlined function body at the call sites and thus
can be used for type checks and casts, with the same restrictions for instances of generic types as
described above.
127
Nested and Inner Classes
Classes can be nested in other classes:
class Outer {
private val bar: Int = 1
class Nested {
fun foo() = 2
}
}
val demo = Outer.Nested().foo() // == 2
Inner classes
A class may be marked as inner to be able to access members of outer class. Inner classes carry
a reference to an object of an outer class:
class Outer {
private val bar: Int = 1
inner class Inner {
fun foo() = bar
}
}
val demo = Outer().Inner().foo() // == 1
See Quali ed this expressions to learn about disambiguation of this in inner classes.
Anonymous inner classes

Anonymous inner class instances are created using an object expression:
window.addMouseListener(object : MouseAdapter() {
override fun mouseClicked(e: MouseEvent) { ... }
override fun mouseEntered(e: MouseEvent) { ... }

})
If the object is an instance of a functional Java interface (i.e. a Java interface with a single abstract
method), you can create it using a lambda expression pre xed with the type of the interface:
val listener = ActionListener { println("clicked") }
128
Enum Classes
The most basic usage of enum classes is implementing type-safe enums:
enum class Direction {

NORTH, SOUTH, WEST, EAST
}
Each enum constant is an object. Enum constants are separated with commas.
Initialization
Since each enum is an instance of the enum class, they can be initialized as:
enum class Color(val rgb: Int) {

RED(0xFF0000),
GREEN(0x00FF00),
BLUE(0x0000FF)
}
Anonymous Classes
Enum constants can also declare their own anonymous classes:
enum class ProtocolState {

WAITING {
override fun signal() = TALKING
},
TALKING {
override fun signal() = WAITING
};
abstract fun signal(): ProtocolState

}
with their corresponding methods, as well as overriding base methods. Note that if the enum
class de nes any members, you need to separate the enum constant de nitions from the
member de nitions with a semicolon, just like in Java.
Enum entries cannot contain nested types other than inner classes (deprecated in Kotlin 1.2).
Implementing Interfaces in Enum Classes

An enum class may implement an interface (but not derive from a class), providing either a single
interface members implementation for all of the entries, or separate ones for each entry within
its anonymous class. This is done by adding the interfaces to the enum class declaration as
follows:
129
enum class IntArithmetics : BinaryOperator<Int>, IntBinaryOperator {
PLUS {
override fun apply(t: Int, u: Int): Int = t + u
},
TIMES {
override fun apply(t: Int, u: Int): Int = t * u
};
override fun applyAsInt(t: Int, u: Int) = apply(t, u)

}
Working with Enum Constants

Just like in Java, enum classes in Kotlin have synthetic methods allowing to list the de ned enum
constants and to get an enum constant by its name. The signatures of these methods are as
follows (assuming the name of the enum class is EnumClass ):
EnumClass.valueOf(value: String): EnumClass

EnumClass.values(): Array<EnumClass>
The valueOf() method throws an IllegalArgumentException if the speci ed name does

not match any of the enum constants de ned in the class.
Since Kotlin 1.1, it's possible to access the constants in an enum class in a generic way, using the
enumValues<T>() and enumValueOf<T>() functions:
enum class RGB { RED, GREEN, BLUE }
inline fun <reified T : Enum<T>> printAllValues() {

print(enumValues<T>().joinToString { it.name })
}
printAllValues<RGB>() // prints RED, GREEN, BLUE
Every enum constant has properties to obtain its name and position in the enum class
declaration:
val name: String

val ordinal: Int
The enum constants also implement the Comparable interface, with the natural order being the
order in which they are de ned in the enum class.
130
Object Expressions and Declarations
Sometimes we need to create an object of a slight modi cation of some class, without explicitly
declaring a new subclass for it. Java handles this case with anonymous inner classes. Kotlin slightly
generalizes this concept with object expressions and object declarations.
Object expressions
To create an object of an anonymous class that inherits from some type (or types), we write:

})
If a supertype has a constructor, appropriate constructor parameters must be passed to it. Many
supertypes may be speci ed as a comma-separated list after the colon:
open class A(x: Int) {

public open val y: Int = x
}
interface B { ... }
val ab: A = object : A(1), B {

override val y = 15
}
If, by any chance, we need "just an object", with no nontrivial supertypes, we can simply say:
fun foo() {
val adHoc = object {
var x: Int = 0
var y: Int = 0
}
print(adHoc.x + adHoc.y)
}
Note that anonymous objects can be used as types only in local and private declarations. If you
use an anonymous object as a return type of a public function or the type of a public property,
the actual type of that function or property will be the declared supertype of the anonymous
object, or Any if you didn't declare any supertype. Members added in the anonymous object will
not be accessible.
131
class C {
// Private function, so the return type is the anonymous object type
private fun foo() = object {
val x: String = "x"
}
// Public function, so the return type is Any

fun publicFoo() = object {
val x: String = "x"
}
fun bar() {
val x1 = foo().x // Works
val x2 = publicFoo().x // ERROR: Unresolved reference 'x'
}
}
Just like Java's anonymous inner classes, code in object expressions can access variables from the
enclosing scope. (Unlike Java, this is not restricted to nal or e ectively nal variables.)
fun countClicks(window: JComponent) {

var clickCount = 0
var enterCount = 0
override fun mouseClicked(e: MouseEvent) {
clickCount++
}
override fun mouseEntered(e: MouseEvent) {

enterCount++
}
})
// ...
}
Object declarations
Singleton may be useful in several cases, and Kotlin (after Scala) makes it easy to declare
singletons:
object DataProviderManager {
fun registerDataProvider(provider: DataProvider) {
// ...
}
val allDataProviders: Collection<DataProvider>

get() = // ...
}
This is called an object declaration, and it always has a name following the object keyword. Just
like a variable declaration, an object declaration is not an expression, and cannot be used on the
right hand side of an assignment statement.
Object declaration's initialization is thread-safe.
132
To refer to the object, we use its name directly:
DataProviderManager.registerDataProvider(...)
Such objects can have supertypes:
object DefaultListener : MouseAdapter() {


}
NOTE: object declarations can't be local (i.e. be nested directly inside a function), but they can be
nested into other object declarations or non-inner classes.
Companion Objects
An object declaration inside a class can be marked with the companion keyword:
class MyClass {
companion object Factory {
fun create(): MyClass = MyClass()
}
}
Members of the companion object can be called by using simply the class name as the quali er:
val instance = MyClass.create()
The name of the companion object can be omitted, in which case the name Companion will be
used:
class MyClass {
companion object { }
}
val x = MyClass.Companion
The name of a class used by itself (not as a quali er to another name) acts as a reference to the
companion object of the class (whether named or not):
class MyClass1 {
companion object Named { }
}
val x = MyClass1
class MyClass2 {
companion object { }
}
val y = MyClass2
133
Note that, even though the members of companion objects look like static members in other
languages, at runtime those are still instance members of real objects, and can, for example,
implement interfaces:
interface Factory<T> {
fun create(): T
}
class MyClass {
companion object : Factory<MyClass> {
override fun create(): MyClass = MyClass()
}
}
val f: Factory<MyClass> = MyClass
However, on the JVM you can have members of companion objects generated as real static
methods and elds, if you use the @JvmStatic annotation. See the Java interoperability section
for more details.
Semantic di erence between object expressions and declarations
There is one important semantic di erence between object expressions and object declarations:
— object expressions are executed (and initialized) immediately, where they are used;
— object declarations are initialized lazily, when accessed for the rst time;
— a companion object is initialized when the corresponding class is loaded (resolved), matching
the semantics of a Java static initializer.
134
Type aliases
Type aliases provide alternative names for existing types. If the type name is too long you can
introduce a di erent shorter name and use the new one instead.
It's useful to shorten long generic types. For instance, it's often tempting to shrink collection
types:
typealias NodeSet = Set<Network.Node>
typealias FileTable<K> = MutableMap<K, MutableList<File>>
You can provide di erent aliases for function types:
typealias MyHandler = (Int, String, Any) -> Unit
typealias Predicate<T> = (T) -> Boolean
You can have new names for inner and nested classes:
class A {
inner class Inner
}
class B {
inner class Inner
}
typealias AInner = A.Inner

typealias BInner = B.Inner
Type aliases do not introduce new types. They are equivalent to the corresponding underlying
types. When you add typealias Predicate<T> and use Predicate<Int> in your code, the
Kotlin compiler always expands it to (Int) -> Boolean . Thus you can pass a variable of your
type whenever a general function type is required and vice versa:
typealias Predicate<T> = (T) -> Boolean
fun foo(p: Predicate<Int>) = p(42)
fun main() {
val f: (Int) -> Boolean = { it > 0 }
println(foo(f)) // prints "true"
val p: Predicate<Int> = { it > 0 }

println(listOf(1, -2).filter(p)) // prints "[1]"
}
135
Inline classes
Inline classes are available only since Kotlin 1.3 and currently are experimental. See
details below
Sometimes it is necessary for business logic to create a wrapper around some type. However, it
introduces runtime overhead due to additional heap allocations. Moreover, if the wrapped type
is primitive, the performance hit is terrible, because primitive types are usually heavily optimized
by the runtime, while their wrappers don't get any special treatment.
To solve such issues, Kotlin introduces a special kind of class called an inline class , which is
declared by placing an inline modi er before the name of the class:
inline class Password(val value: String)
An inline class must have a single property initialized in the primary constructor. At runtime,
instances of the inline class will be represented using this single property (see details about
runtime representation below):
// No actual instantiation of class 'Password' happens

// At runtime 'securePassword' contains just 'String'
val securePassword = Password("Don't try this in production")
This is the main feature of inline classes, which inspired the name "inline": data of the class is
"inlined" into its usages (similar to how content of inline functions is inlined to call sites).
Members
Inline classes support some functionality of regular classes. In particular, they are allowed to
declare properties and functions:
inline class Name(val s: String) {

val length: Int
get() = s.length
fun greet() {
println("Hello, $s")
}
}
fun main() {
name.greet() // method `greet` is called as a static method
println(name.length) // property getter is called as a static method
}
However, there are some restrictions for inline class members:
— inline classes cannot have init blocks
— inline class properties cannot have backing elds
136
— it follows that inline classes can only have simple computable properties (no
lateinit/delegated properties)
Inheritance
Inline classes are allowed to inherit from interfaces:
interface Printable {
fun prettyPrint(): String
}
inline class Name(val s: String) : Printable {

override fun prettyPrint(): String = "Let's $s!"
}
fun main() {
println(name.prettyPrint()) // Still called as a static method
}
It is forbidden for inline classes to participate in a class hierarchy. This means that inline classes
cannot extend other classes and must be final.
Representation
In generated code, the Kotlin compiler keeps a wrapper for each inline class. Inline class instances
can be represented at runtime either as wrappers or as the underlying type. This is similar to
how Int can be represented either as a primitive int or as the wrapper Integer .
The Kotlin compiler will prefer using underlying types instead of wrappers to produce the most
performant and optimized code. However, sometimes it is necessary to keep wrappers around.
As a rule of thumb, inline classes are boxed whenever they are used as another type.
137
interface I
inline class Foo(val i: Int) : I
fun asInline(f: Foo) {}

fun <T> asGeneric(x: T) {}
fun asInterface(i: I) {}
fun asNullable(i: Foo?) {}
fun <T> id(x: T): T = x
fun main() {
val f = Foo(42)
asInline(f) // unboxed: used as Foo itself

asGeneric(f) // boxed: used as generic type T
asInterface(f) // boxed: used as type I
asNullable(f) // boxed: used as Foo?, which is different from Foo
// below, 'f' first is boxed (while being passed to 'id') and then unboxed (when
returned from 'id')
// In the end, 'c' contains unboxed representation (just '42'), as 'f'
val c = id(f)
}
Because inline classes may be represented both as the underlying value and as a wrapper,
referential equality is pointless for them and is therefore prohibited.
Mangling
Since inline classes are compiled to their underlying type, it may lead to various obscure errors,
for example unexpected platform signature clashes:
inline class UInt(val x: Int)
// Represented as 'public final void compute(int x)' on the JVM

fun compute(x: Int) { }
// Also represented as 'public final void compute(int x)' on the JVM!

fun compute(x: UInt) { }
To mitigate such issues, functions using inline classes are mangled by adding some stable
hashcode to the function name. Therefore, fun compute(x: UInt) will be represented as
public final void compute-<hashcode>(int x) , which solves the clash problem.
Note that - is an invalid symbol in Java, meaning that it's impossible to call functions
which accept inline classes from Java.
Inline classes vs type aliases

At rst sight, inline classes may appear to be very similar to type aliases. Indeed, both seem to
introduce a new type and both will be represented as the underlying type at runtime.
138
However, the crucial di erence is that type aliases are assignment-compatible with their
underlying type (and with other type aliases with the same underlying type), while inline classes
are not.
In other words, inline classes introduce a truly new type, contrary to type aliases which only
introduce an alternative name (alias) for an existing type:
typealias NameTypeAlias = String

inline class NameInlineClass(val s: String)
fun acceptString(s: String) {}

fun acceptNameTypeAlias(n: NameTypeAlias) {}
fun acceptNameInlineClass(p: NameInlineClass) {}
fun main() {
val nameAlias: NameTypeAlias = ""
val nameInlineClass: NameInlineClass = NameInlineClass("")
val string: String = ""
acceptString(nameAlias) // OK: pass alias instead of underlying type

acceptString(nameInlineClass) // Not OK: can't pass inline class instead of
underlying type
// And vice versa:

acceptNameTypeAlias(string) // OK: pass underlying type instead of alias
acceptNameInlineClass(string) // Not OK: can't pass underlying type instead of inline
class
}
Experimental status of inline classes

The design of inline classes is experimental, meaning that this feature is moving fast and no
compatibility guarantees are given. When using inline classes in Kotlin 1.3+, a warning will be
reported, indicating that this feature is experimental.
To remove the warning you have to opt in to the usage of experimental features by passing the
argument -XXLanguage:+InlineClasses to kotlinc .
Enabling inline classes in Gradle:
compileKotlin {
kotlinOptions.freeCompilerArgs += ["-XXLanguage:+InlineClasses"]
}
See Compiler options in Gradle for details. For Multiplatform Projects settings, see building
Multiplatform Projects with Gradle section.
Enabling inline classes in Maven
139
<configuration>
<args>
<arg>-XXLanguage:+InlineClasses</arg>
</args>
</configuration>
See Compiler options in Maven for details.
Further discussion
See this language proposal for inline classes for other technical details and discussion.
140
Delegation
Property Delegation
Delegated properties are described on a separate page: Delegated Properties.
Implementation by Delegation
The Delegation pattern has proven to be a good alternative to implementation inheritance, and
Kotlin supports it natively requiring zero boilerplate code. A class Derived can implement an
interface Base by delegating all of its public members to a speci ed object:
interface Base {
fun print()
}
class BaseImpl(val x: Int) : Base {

override fun print() { print(x) }
}
class Derived(b: Base) : Base by b
fun main() {
val b = BaseImpl(10)
Derived(b).print()
}
The by-clause in the supertype list for Derived indicates that b will be stored internally in
objects of Derived and the compiler will generate all the methods of Base that forward to b .
Overriding a member of an interface implemented by delegation
Overrides work as you might expect: the compiler will use your override implementations
instead of those in the delegate object. If we were to add override fun printMessage() {
print("abc") } to Derived , the program would print "abc" instead of "10" when
printMessage is called:
141
interface Base {
fun printMessage()
fun printMessageLine()
}

override fun printMessage() { print(x) }
override fun printMessageLine() { println(x) }
}
class Derived(b: Base) : Base by b {

override fun printMessage() { print("abc") }
}
fun main() {
Derived(b).printMessage()
Derived(b).printMessageLine()
}
Note, however, that members overridden in this way do not get called from the members of the
delegate object, which can only access its own implementations of the interface members:
interface Base {
val message: String
fun print()
}

override val message = "BaseImpl: x = $x"
override fun print() { println(message) }
}
class Derived(b: Base) : Base by b {

// This property is not accessed from b's implementation of `print`
override val message = "Message of Derived"
}
fun main() {
val derived = Derived(b)
derived.print()
println(derived.message)
}
142
Delegated Properties
There are certain common kinds of properties, that, though we can implement them manually
every time we need them, would be very nice to implement once and for all, and put into a
library. Examples include:
— lazy properties: the value gets computed only upon rst access;
— observable properties: listeners get noti ed about changes to this property;
— storing properties in a map, instead of a separate eld for each property.
To cover these (and other) cases, Kotlin supports delegated properties:
class Example {
var p: String by Delegate()
}
The syntax is: val/var <property name>: <Type> by <expression> . The expression
after by is the delegate, because get() (and set() ) corresponding to the property will be
delegated to its getValue() and setValue() methods. Property delegates don’t have to
implement any interface, but they have to provide a getValue() function (and setValue()
— for vars). For example:
class Delegate {
operator fun getValue(thisRef: Any?, property: KProperty<*>): String {
return "$thisRef, thank you for delegating '${property.name}' to me!"
}
operator fun setValue(thisRef: Any?, property: KProperty<*>, value: String) {

println("$value has been assigned to '${property.name}' in $thisRef.")
}
}
When we read from p that delegates to an instance of Delegate , the getValue() function
from Delegate is called, so that its rst parameter is the object we read p from and the
second parameter holds a description of p itself (e.g. you can take its name). For example:
val e = Example()
println(e.p)
This prints:
Example@33a17727, thank you for delegating ‘p’ to me!
Similarly, when we assign to p , the setValue() function is called. The rst two parameters are
the same, and the third holds the value being assigned:
e.p = "NEW"
This prints
NEW has been assigned to ‘p’ in Example@33a17727.
143
The speci cation of the requirements to the delegated object can be found below.
Note that since Kotlin 1.1 you can declare a delegated property inside a function or code block, it
shouldn't necessarily be a member of a class. Below you can nd the example.
Standard Delegates
The Kotlin standard library provides factory methods for several useful kinds of delegates.
Lazy
lazy() is a function that takes a lambda and returns an instance of Lazy<T> which can serve
as a delegate for implementing a lazy property: the rst call to get() executes the lambda
passed to lazy() and remembers the result, subsequent calls to get() simply return the
remembered result.
val lazyValue: String by lazy {

println("computed!")
"Hello"
}
fun main() {
println(lazyValue)
println(lazyValue)
}
By default, the evaluation of lazy properties is synchronized: the value is computed only in one
thread, and all threads will see the same value. If the synchronization of initialization delegate is
not required, so that multiple threads can execute it simultaneously, pass
LazyThreadSafetyMode.PUBLICATION as a parameter to the lazy() function. And if you're
sure that the initialization will always happen on a single thread, you can use
LazyThreadSafetyMode.NONE mode, which doesn't incur any thread-safety guarantees and
the related overhead.
Observable
Delegates.observable() takes two arguments: the initial value and a handler for
modi cations. The handler gets called every time we assign to the property (after the assignment
has been performed). It has three parameters: a property being assigned to, the old value and
the new one:
144
import kotlin.properties.Delegates
class User {
var name: String by Delegates.observable("<no name>") {
prop, old, new ->
println("$old -> $new")
}
}
fun main() {
val user = User()
user.name = "first"
user.name = "second"
}
If you want to be able to intercept an assignment and "veto" it, use vetoable() instead of
observable() . The handler passed to the vetoable is called before the assignment of a new
property value has been performed.
Storing Properties in a Map

One common use case is storing the values of properties in a map. This comes up often in
applications like parsing JSON or doing other “dynamic” things. In this case, you can use the map
instance itself as the delegate for a delegated property.
class User(val map: Map<String, Any?>) {

val name: String by map
val age: Int by map
}
In this example, the constructor takes a map:
val user = User(mapOf(

"name" to "John Doe",
"age" to 25
))
Delegated properties take values from this map (by the string keys –– names of properties):
println(user.name) // Prints "John Doe"

println(user.age) // Prints 25
This works also for var’s properties if you use a MutableMap instead of read-only Map :
class MutableUser(val map: MutableMap<String, Any?>) {

var name: String by map
var age: Int by map
}
Local Delegated Properties (since 1.1)

You can declare local variables as delegated properties. For instance, you can make a local
variable lazy:
145
fun example(computeFoo: () -> Foo) {
val memoizedFoo by lazy(computeFoo)
if (someCondition && memoizedFoo.isValid()) {

memoizedFoo.doSomething()
}
}
The memoizedFoo variable will be computed on the rst access only. If someCondition fails,
the variable won't be computed at all.
Property Delegate Requirements

Here we summarize requirements to delegate objects.
For a read-only property (i.e. a val), a delegate has to provide a function named getValue that
takes the following parameters:
— thisRef — must be the same or a supertype of the property owner (for extension properties
— the type being extended);
— property — must be of type KProperty<*> or its supertype.
this function must return the same type as property (or its subtype).
For a mutable property (a var), a delegate has to additionally provide a function named
setValue that takes the following parameters:
— thisRef — same as for getValue() ;
— property — same as for getValue() ;
— new value — must be of the same type as a property or its supertype.
getValue() and/or setValue() functions may be provided either as member functions of

the delegate class or extension functions. The latter is handy when you need to delegate
property to an object which doesn't originally provide these functions. Both of the functions need
to be marked with the operator keyword.
The delegate class may implement one of the interfaces ReadOnlyProperty and
ReadWriteProperty containing the required operator methods. These interfaces are
declared in the Kotlin standard library:
interface ReadOnlyProperty<in R, out T> {

operator fun getValue(thisRef: R, property: KProperty<*>): T
}
interface ReadWriteProperty<in R, T> {

operator fun getValue(thisRef: R, property: KProperty<*>): T
operator fun setValue(thisRef: R, property: KProperty<*>, value: T)
}
146
Translation Rules
Under the hood for every delegated property the Kotlin compiler generates an auxiliary property
and delegates to it. For instance, for the property prop the hidden property prop$delegate is
generated, and the code of the accessors simply delegates to this additional property:
class C {
var prop: Type by MyDelegate()
}
// this code is generated by the compiler instead:

class C {
private val prop$delegate = MyDelegate()
var prop: Type
get() = prop$delegate.getValue(this, this::prop)
set(value: Type) = prop$delegate.setValue(this, this::prop, value)
}
The Kotlin compiler provides all the necessary information about prop in the arguments: the
rst argument this refers to an instance of the outer class C and this::prop is a re ection
object of the KProperty type describing prop itself.
Note that the syntax this::prop to refer a bound callable reference in the code directly is
available only since Kotlin 1.1.
Providing a delegate (since 1.1)
By de ning the provideDelegate operator you can extend the logic of creating the object to
which the property implementation is delegated. If the object used on the right hand side of by
de nes provideDelegate as a member or extension function, that function will be called to
create the property delegate instance.
One of the possible use cases of provideDelegate is to check property consistency when the
property is created, not only in its getter or setter.
For example, if you want to check the property name before binding, you can write something
like this:
147
class ResourceDelegate<T> : ReadOnlyProperty<MyUI, T> {
override fun getValue(thisRef: MyUI, property: KProperty<*>): T { ... }
}
class ResourceLoader<T>(id: ResourceID<T>) {

operator fun provideDelegate(
thisRef: MyUI,
prop: KProperty<*>
): ReadOnlyProperty<MyUI, T> {
checkProperty(thisRef, prop.name)
// create delegate
return ResourceDelegate()
}
private fun checkProperty(thisRef: MyUI, name: String) { ... }

}
class MyUI {
fun <T> bindResource(id: ResourceID<T>): ResourceLoader<T> { ... }
val image by bindResource(ResourceID.image_id)

val text by bindResource(ResourceID.text_id)
}
The parameters of provideDelegate are the same as for getValue :
— thisRef — must be the same or a supertype of the property owner (for extension properties
— the type being extended);
— property — must be of type KProperty<*> or its supertype.
The provideDelegate method is called for each property during the creation of the MyUI
instance, and it performs the necessary validation right away.
Without this ability to intercept the binding between the property and its delegate, to achieve the
same functionality you'd have to pass the property name explicitly, which isn't very convenient:
// Checking the property name without "provideDelegate" functionality

class MyUI {
val image by bindResource(ResourceID.image_id, "image")
val text by bindResource(ResourceID.text_id, "text")
}
fun <T> MyUI.bindResource(

id: ResourceID<T>,
propertyName: String
): ReadOnlyProperty<MyUI, T> {
checkProperty(this, propertyName)
// create delegate
}
In the generated code, the provideDelegate method is called to initialize the auxiliary
prop$delegate property. Compare the generated code for the property declaration val
prop: Type by MyDelegate() with the generated code above (when the
provideDelegate method is not present):
148
class C {
var prop: Type by MyDelegate()
}
// this code is generated by the compiler

// when the 'provideDelegate' function is available:
class C {
// calling "provideDelegate" to create the additional "delegate" property
private val prop$delegate = MyDelegate().provideDelegate(this, this::prop)
var prop: Type
get() = prop$delegate.getValue(this, this::prop)
set(value: Type) = prop$delegate.setValue(this, this::prop, value)
}
Note that the provideDelegate method a ects only the creation of the auxiliary property and
doesn't a ect the code generated for getter or setter.
149
Functions and Lambdas
Functions
Function Declarations
Functions in Kotlin are declared using the fun keyword:
fun double(x: Int): Int {

return 2 * x
}
Function Usage
Calling functions uses the traditional approach:
val result = double(2)
Calling member functions uses the dot notation:
Sample().foo() // create instance of class Sample and call foo
Parameters
Function parameters are de ned using Pascal notation, i.e. name: type. Parameters are separated
using commas. Each parameter must be explicitly typed:
fun powerOf(number: Int, exponent: Int) { ... }
Default Arguments
Function parameters can have default values, which are used when a corresponding argument is
omitted. This allows for a reduced number of overloads compared to other languages:
fun read(b: Array<Byte>, off: Int = 0, len: Int = b.size) { ... }
Default values are de ned using the = after type along with the value.
Overriding methods always use the same default parameter values as the base method. When
overriding a method with default parameter values, the default parameter values must be
omitted from the signature:
150
open class A {
open fun foo(i: Int = 10) { ... }
}
class B : A() {
override fun foo(i: Int) { ... } // no default value allowed
}
If a default parameter precedes a parameter with no default value, the default value can only be
used by calling the function with named arguments:
fun foo(bar: Int = 0, baz: Int) { ... }
foo(baz = 1) // The default value bar = 0 is used
If the last argument after default parameters is a lambda, it can be passed in either as a named
argument or outside the parentheses:
fun foo(bar: Int = 0, baz: Int = 1, qux: () -> Unit) { ... }
foo(1) { println("hello") } // Uses the default value baz = 1

foo(qux = { println("hello") }) // Uses both default values bar = 0 and baz = 1
foo { println("hello") } // Uses both default values bar = 0 and baz = 1
Named Arguments
Function parameters can be named when calling functions. This is very convenient when a
function has a high number of parameters or default ones.
Given the following function:
fun reformat(str: String,

normalizeCase: Boolean = true,
upperCaseFirstLetter: Boolean = true,
divideByCamelHumps: Boolean = false,
wordSeparator: Char = ' ') {
...
}
We could call this using default arguments:
reformat(str)
However, when calling it with non-default, the call would look something like:
reformat(str, true, true, false, '_')
With named arguments we can make the code much more readable:
151
reformat(str,
normalizeCase = true,
upperCaseFirstLetter = true,
divideByCamelHumps = false,
wordSeparator = '_'
)
and if we do not need all arguments:
reformat(str, wordSeparator = '_')
When a function is called with both positional and named arguments, all the positional
arguments should be placed before the rst named one. For example, the call f(1, y = 2) is
allowed, but f(x = 1, 2) is not.
Variable number of arguments (vararg) can be passed in the named form by using the spread
operator:
fun foo(vararg strings: String) { ... }
foo(strings = *arrayOf("a", "b", "c"))
Note that the named argument syntax cannot be used when calling Java functions, because Java
bytecode does not always preserve names of function parameters.
Unit-returning functions
If a function does not return any useful value, its return type is Unit . Unit is a type with only
one value - Unit . This value does not have to be returned explicitly:
fun printHello(name: String?): Unit {

if (name != null)
println("Hello ${name}")
else
println("Hi there!")
// `return Unit` or `return` is optional
}
The Unit return type declaration is also optional. The above code is equivalent to:
fun printHello(name: String?) { ... }
Single-Expression functions
When a function returns a single expression, the curly braces can be omitted and the body is
speci ed after a = symbol:
fun double(x: Int): Int = x * 2
Explicitly declaring the return type is optional when this can be inferred by the compiler:
152
fun double(x: Int) = x * 2
Explicit return types
Functions with block body must always specify return types explicitly, unless it's intended for
them to return Unit , in which case it is optional. Kotlin does not infer return types for functions
with block bodies because such functions may have complex control ow in the body, and the
return type will be non-obvious to the reader (and sometimes even for the compiler).
Variable number of arguments (Varargs)
A parameter of a function (normally the last one) may be marked with vararg modi er:
fun <T> asList(vararg ts: T): List<T> {

val result = ArrayList<T>()
for (t in ts) // ts is an Array
result.add(t)
return result
}
allowing a variable number of arguments to be passed to the function:
val list = asList(1, 2, 3)
Inside a function a vararg -parameter of type T is visible as an array of T , i.e. the ts variable
in the example above has type Array<out T> .
Only one parameter may be marked as vararg . If a vararg parameter is not the last one in
the list, values for the following parameters can be passed using the named argument syntax, or,
if the parameter has a function type, by passing a lambda outside parentheses.
When we call a vararg -function, we can pass arguments one-by-one, e.g. asList(1, 2, 3) ,
or, if we already have an array and want to pass its contents to the function, we use the spread
operator (pre x the array with * ):
val a = arrayOf(1, 2, 3)
val list = asList(-1, 0, *a, 4)
In x notation
Functions marked with the infix keyword can also be called using the in x notation (omitting
the dot and the parentheses for the call). In x functions must satisfy the following requirements:
— They must be member functions or extension functions;

— They must have a single parameter;
— The parameter must not accept variable number of arguments and must have no default
value.
153
infix fun Int.shl(x: Int): Int { ... }
// calling the function using the infix notation

1 shl 2
// is the same as
1.shl(2)
In x function calls have lower precedence than the arithmetic operators, type casts, and
the rangeTo operator. The following expressions are equivalent:
— 1 shl 2 + 3 and 1 shl (2 + 3)
— 0 until n * 2 and 0 until (n * 2)
— xs union ys as Set<*> and xs union (ys as Set<*>)
On the other hand, in x function call's precedence is higher than that of the boolean
operators && and ||, is- and in-checks, and some other operators. These expressions
are equivalent as well:
— a && b xor c and a && (b xor c)
— a xor b in c and (a xor b) in c
See the Grammar reference for the complete operators precedence hierarchy.
Note that in x functions always require both the receiver and the parameter to be speci ed.
When you're calling a method on the current receiver using the in x notation, you need to use
this explicitly; unlike regular method calls, it cannot be omitted. This is required to ensure
unambiguous parsing.
class MyStringCollection {
infix fun add(s: String) { ... }
fun build() {
this add "abc" // Correct
add("abc") // Correct
add "abc" // Incorrect: the receiver must be specified
}
}
Function Scope
In Kotlin functions can be declared at top level in a le, meaning you do not need to create a class
to hold a function, which you are required to do in languages such as Java, C# or Scala. In addition
to top level functions, Kotlin functions can also be declared local, as member functions and
extension functions.
Local Functions
154
Kotlin supports local functions, i.e. a function inside another function:
fun dfs(graph: Graph) {

fun dfs(current: Vertex, visited: Set<Vertex>) {
if (!visited.add(current)) return
for (v in current.neighbors)
dfs(v, visited)
}
dfs(graph.vertices[0], HashSet())
}
Local function can access local variables of outer functions (i.e. the closure), so in the case above,
the visited can be a local variable:
fun dfs(graph: Graph) {

val visited = HashSet<Vertex>()
fun dfs(current: Vertex) {
if (!visited.add(current)) return
for (v in current.neighbors)
dfs(v)
}
dfs(graph.vertices[0])
}
Member Functions
A member function is a function that is de ned inside a class or object:
class Sample() {
fun foo() { print("Foo") }
}
Member functions are called with dot notation:
Sample().foo() // creates instance of class Sample and calls foo
For more information on classes and overriding members see Classes and Inheritance.
Generic Functions
Functions can have generic parameters which are speci ed using angle brackets before the
function name:
fun <T> singletonList(item: T): List<T> { ... }
For more information on generic functions see Generics.
Inline Functions
Inline functions are explained here.
Extension Functions
155
Extension functions are explained in their own section.
Higher-Order Functions and Lambdas

Higher-Order functions and Lambdas are explained in their own section.
Tail recursive functions

Kotlin supports a style of functional programming known as tail recursion. This allows some
algorithms that would normally be written using loops to instead be written using a recursive
function, but without the risk of stack over ow. When a function is marked with the tailrec
modi er and meets the required form, the compiler optimises out the recursion, leaving behind
a fast and e cient loop based version instead:
val eps = 1E-10 // "good enough", could be 10^-15
tailrec fun findFixPoint(x: Double = 1.0): Double

= if (Math.abs(x - Math.cos(x)) < eps) x else findFixPoint(Math.cos(x))
This code calculates the xpoint of cosine, which is a mathematical constant. It simply calls
Math.cos repeatedly starting at 1.0 until the result doesn't change any more, yielding a result of
0.7390851332151611 for the speci ed eps precision. The resulting code is equivalent to this
more traditional style:
val eps = 1E-10 // "good enough", could be 10^-15
private fun findFixPoint(): Double {

var x = 1.0
while (true) {
val y = Math.cos(x)
if (Math.abs(x - y) < eps) return x
x = Math.cos(x)
}
}
To be eligible for the tailrec modi er, a function must call itself as the last operation it
performs. You cannot use tail recursion when there is more code after the recursive call, and you
cannot use it within try/catch/ nally blocks. Currently tail recursion is only supported in the JVM
backend.
156
Higher-Order Functions and Lambdas
Kotlin functions are rst-class, which means that they can be stored in variables and data
structures, passed as arguments to and returned from other higher-order functions. You can
operate with functions in any way that is possible for other non-function values.
To facilitate this, Kotlin, as a statically typed programming language, uses a family of function
types to represent functions and provides a set of specialized language constructs, such as
lambda expressions.
Higher-Order Functions
A higher-order function is a function that takes functions as parameters, or returns a function.
A good example is the functional programming idiom fold for collections, which takes an initial
accumulator value and a combining function and builds its return value by consecutively
combining current accumulator value with each collection element, replacing the accumulator:
fun <T, R> Collection<T>.fold(

initial: R,
combine: (acc: R, nextElement: T) -> R
): R {
var accumulator: R = initial
for (element: T in this) {
accumulator = combine(accumulator, element)
}
return accumulator
}
In the code above, the parameter combine has a function type (R, T) -> R , so it accepts a
function that takes two arguments of types R and T and returns a value of type R . It is invoked
inside the for-loop, and the return value is then assigned to accumulator .
To call fold , we need to pass it an instance of the function type as an argument, and lambda
expressions (described in more detail below) are widely used for this purpose at higher-order
function call sites:
157
val items = listOf(1, 2, 3, 4, 5)
// Lambdas are code blocks enclosed in curly braces.

items.fold(0, {
// When a lambda has parameters, they go first, followed by '->'
acc: Int, i: Int ->
print("acc = $acc, i = $i, ")
val result = acc + i
println("result = $result")
// The last expression in a lambda is considered the return value:
result
})
// Parameter types in a lambda are optional if they can be inferred:

val joinedToString = items.fold("Elements:", { acc, i -> acc + " " + i })
// Function references can also be used for higher-order function calls:

val product = items.fold(1, Int::times)
The following sections explain in more detail the concepts mentioned so far.
Function types
Kotlin uses a family of function types like (Int) -> String for declarations that deal with
functions: val onClick: () -> Unit = ... .
These types have a special notation that corresponds to the signatures of the functions, i.e. their
parameters and return values:
— All function types have a parenthesized parameter types list and a return type: (A, B) -> C
denotes a type that represents functions taking two arguments of types A and B and
returning a value of type C . The parameter types list may be empty, as in () -> A . The
Unit return type cannot be omitted.
— Function types can optionally have an additional receiver type, which is speci ed before a dot
in the notation: the type A.(B) -> C represents functions that can be called on a receiver
object of A with a parameter of B and return a value of C . Function literals with receiver are
often used along with these types.
— Suspending functions belong to function types of a special kind, which have a suspend
modi er in the notation, such as suspend () -> Unit or suspend A.(B) -> C .
The function type notation can optionally include names for the function parameters: (x: Int,
y: Int) -> Point . These names can be used for documenting the meaning of the
parameters.
158
To specify that a function type is nullable, use parentheses: ((Int, Int) -> Int)?.
Function types can be combined using parentheses: (Int) -> ((Int) -> Unit)
The arrow notation is right-associative, (Int) -> (Int) -> Unit is equivalent to the
previous example, but not to ((Int) -> (Int)) -> Unit.
You can also give a function type an alternative name by using a type alias:
typealias ClickHandler = (Button, ClickEvent) -> Unit
Instantiating a function type
There are several ways to obtain an instance of a function type:
— Using a code block within a function literal, in one of the forms:

— a lambda expression: { a, b -> a + b } ,
— an anonymous function: fun(s: String): Int { return s.toIntOrNull() ?: 0

}
Function literals with receiver can be used as values of function types with receiver.
— Using a callable reference to an existing declaration:

— a top-level, local, member, or extension function: ::isOdd , String::toInt ,
— a top-level, member, or extension property: List<Int>::size ,
— a constructor: ::Regex
These include bound callable references that point to a member of a particular instance:
foo::toString .
— Using instances of a custom class that implements a function type as an interface:
class IntTransformer: (Int) -> Int {

override operator fun invoke(x: Int): Int = TODO()
}
val intFunction: (Int) -> Int = IntTransformer()
The compiler can infer the function types for variables if there is enough information:
val a = { i: Int -> i + 1 } // The inferred type is (Int) -> Int
Non-literal values of function types with and without receiver are interchangeable, so that the
receiver can stand in for the rst parameter, and vice versa. For instance, a value of type (A, B)
-> C can be passed or assigned where a A.(B) -> C is expected and the other way around:
159
val repeatFun: String.(Int) -> String = { times -> this.repeat(times) }
val twoParameters: (String, Int) -> String = repeatFun // OK
fun runTransformation(f: (String, Int) -> String): String {

return f("hello", 3)
}
val result = runTransformation(repeatFun) // OK
Note that a function type with no receiver is inferred by default, even if a variable is
initialized with a reference to an extension function. To alter that, specify the variable type
explicitly.
Invoking a function type instance
A value of a function type can be invoked by using its invoke(...) operator: f.invoke(x) or
just f(x) .
If the value has a receiver type, the receiver object should be passed as the rst argument.
Another way to invoke a value of a function type with receiver is to prepend it with the receiver
object, as if the value were an extension function: 1.foo(2) ,
Example:
val stringPlus: (String, String) -> String = String::plus

val intPlus: Int.(Int) -> Int = Int::plus
println(stringPlus.invoke("<-", "->"))
println(stringPlus("Hello, ", "world!"))
println(intPlus.invoke(1, 1))
println(intPlus(1, 2))
println(2.intPlus(3)) // extension-like call
Inline functions
Sometimes it is bene cial to use inline functions, which provide exible control ow, for higher-
order functions.
Lambda Expressions and Anonymous Functions

Lambda expressions and anonymous functions are 'function literals', i.e. functions that are not
declared, but passed immediately as an expression. Consider the following example:
max(strings, { a, b -> a.length < b.length })
Function max is a higher-order function, it takes a function value as the second argument. This
second argument is an expression that is itself a function, i.e. a function literal, which is
equivalent to the following named function:
fun compare(a: String, b: String): Boolean = a.length < b.length
160
Lambda expression syntax
The full syntactic form of lambda expressions is as follows:
val sum = { x: Int, y: Int -> x + y }
A lambda expression is always surrounded by curly braces, parameter declarations in the full
syntactic form go inside curly braces and have optional type annotations, the body goes after an
-> sign. If the inferred return type of the lambda is not Unit , the last (or possibly single)
expression inside the lambda body is treated as the return value.
If we leave all the optional annotations out, what's left looks like this:
val sum: (Int, Int) -> Int = { x, y -> x + y }
Passing a lambda to the last parameter
In Kotlin, there is a convention that if the last parameter of a function accepts a function, a
lambda expression that is passed as the corresponding argument can be placed outside the
parentheses:
val product = items.fold(1) { acc, e -> acc * e }
If the lambda is the only argument to that call, the parentheses can be omitted entirely:
run { println("...") }
it: implicit name of a single parameter
It's very common that a lambda expression has only one parameter.
If the compiler can gure the signature out itself, it is allowed not to declare the only parameter
and omit -> . The parameter will be implicitly declared under the name it :
ints.filter { it > 0 } // this literal is of type '(it: Int) -> Boolean'
Returning a value from a lambda expression
We can explicitly return a value from the lambda using the quali ed return syntax. Otherwise, the
value of the last expression is implicitly returned.
Therefore, the two following snippets are equivalent:
161
ints.filter {
val shouldFilter = it > 0
shouldFilter
}
ints.filter {
val shouldFilter = it > 0
return@filter shouldFilter
}
This convention, along with passing a lambda expression outside parentheses, allows for LINQ-
style code:
strings.filter { it.length == 5 }.sortedBy { it }.map { it.toUpperCase() }
Underscore for unused variables (since 1.1)
If the lambda parameter is unused, you can place an underscore instead of its name:
map.forEach { _, value -> println("$value!") }
Destructuring in lambdas (since 1.1)
Destructuring in lambdas is described as a part of destructuring declarations.
Anonymous functions
One thing missing from the lambda expression syntax presented above is the ability to specify
the return type of the function. In most cases, this is unnecessary because the return type can be
inferred automatically. However, if you do need to specify it explicitly, you can use an alternative
syntax: an anonymous function.
fun(x: Int, y: Int): Int = x + y
An anonymous function looks very much like a regular function declaration, except that its name
is omitted. Its body can be either an expression (as shown above) or a block:
fun(x: Int, y: Int): Int {

return x + y
}
The parameters and the return type are speci ed in the same way as for regular functions,
except that the parameter types can be omitted if they can be inferred from context:
ints.filter(fun(item) = item > 0)
The return type inference for anonymous functions works just like for normal functions: the
return type is inferred automatically for anonymous functions with an expression body and has
to be speci ed explicitly (or is assumed to be Unit ) for anonymous functions with a block body.
162
Note that anonymous function parameters are always passed inside the parentheses. The
shorthand syntax allowing to leave the function outside the parentheses works only for lambda
expressions.
One other di erence between lambda expressions and anonymous functions is the behavior of
non-local returns. A return statement without a label always returns from the function declared
with the fun keyword. This means that a return inside a lambda expression will return from the
enclosing function, whereas a return inside an anonymous function will return from the
anonymous function itself.
Closures
A lambda expression or anonymous function (as well as a local function and an object
expression) can access its closure, i.e. the variables declared in the outer scope. Unlike Java, the
variables captured in the closure can be modi ed:
var sum = 0
ints.filter { it > 0 }.forEach {
sum += it
}
print(sum)
Function literals with receiver
Function types with receiver, such as A.(B) -> C , can be instantiated with a special form of
function literals – function literals with receiver.
As said above, Kotlin provides the ability to call an instance of a function type with receiver
providing the receiver object.
Inside the body of the function literal, the receiver object passed to a call becomes an implicit
this, so that you can access the members of that receiver object without any additional
quali ers, or access the receiver object using a this expression.
This behavior is similar to extension functions, which also allow you to access the members of
the receiver object inside the body of the function.
Here is an example of a function literal with receiver along with its type, where plus is called on
the receiver object:
val sum: Int.(Int) -> Int = { other -> plus(other) }
The anonymous function syntax allows you to specify the receiver type of a function literal
directly. This can be useful if you need to declare a variable of a function type with receiver, and
to use it later.
val sum = fun Int.(other: Int): Int = this + other
163
Lambda expressions can be used as function literals with receiver when the receiver type can be
inferred from context. One of the most important examples of their usage is type-safe builders:
class HTML {
fun body() { ... }
}
fun html(init: HTML.() -> Unit): HTML {

val html = HTML() // create the receiver object
html.init() // pass the receiver object to the lambda
return html
}
html { // lambda with receiver begins here

body() // calling a method on the receiver object
}
164
Inline Functions
Using higher-order functions imposes certain runtime penalties: each function is an object, and it
captures a closure, i.e. those variables that are accessed in the body of the function. Memory
allocations (both for function objects and classes) and virtual calls introduce runtime overhead.
But it appears that in many cases this kind of overhead can be eliminated by inlining the lambda
expressions. The functions shown below are good examples of this situation. I.e., the lock()
function could be easily inlined at call-sites. Consider the following case:
lock(l) { foo() }
Instead of creating a function object for the parameter and generating a call, the compiler could
emit the following code:
l.lock()
try {
foo()
}
finally {
l.unlock()
}
Isn't it what we wanted from the very beginning?
To make the compiler do this, we need to mark the lock() function with the inline modi er:
inline fun <T> lock(lock: Lock, body: () -> T): T { ... }
The inline modi er a ects both the function itself and the lambdas passed to it: all of those
will be inlined into the call site.
Inlining may cause the generated code to grow; however, if we do it in a reasonable way (i.e.
avoiding inlining large functions), it will pay o in performance, especially at "megamorphic" call-
sites inside loops.
noinline
In case you want only some of the lambdas passed to an inline function to be inlined, you can
mark some of your function parameters with the noinline modi er:
inline fun foo(inlined: () -> Unit, noinline notInlined: () -> Unit) { ... }
Inlinable lambdas can only be called inside the inline functions or passed as inlinable arguments,
but noinline ones can be manipulated in any way we like: stored in elds, passed around etc.
Note that if an inline function has no inlinable function parameters and no rei ed type
parameters, the compiler will issue a warning, since inlining such functions is very unlikely to be
bene cial (you can suppress the warning if you are sure the inlining is needed using the
annotation @Suppress("NOTHING_TO_INLINE") ).
165
Non-local returns
In Kotlin, we can only use a normal, unquali ed return to exit a named function or an
anonymous function. This means that to exit a lambda, we have to use a label, and a bare
return is forbidden inside a lambda, because a lambda cannot make the enclosing function
return:
fun foo() {
ordinaryFunction {
return // ERROR: cannot make `foo` return here
}
}
But if the function the lambda is passed to is inlined, the return can be inlined as well, so it is
allowed:
inline fun inlined(block: () -> Unit) { println("hi!") }
fun foo() {
inlined {
return // OK: the lambda is inlined
}
}
Such returns (located in a lambda, but exiting the enclosing function) are called non-local returns.
We are used to this sort of construct in loops, which inline functions often enclose:
fun hasZeros(ints: List<Int>): Boolean {

ints.forEach {
if (it == 0) return true // returns from hasZeros
}
return false
}
Note that some inline functions may call the lambdas passed to them as parameters not directly
from the function body, but from another execution context, such as a local object or a nested
function. In such cases, non-local control ow is also not allowed in the lambdas. To indicate that,
the lambda parameter needs to be marked with the crossinline modi er:
inline fun f(crossinline body: () -> Unit) {

val f = object: Runnable {
override fun run() = body()
}
// ...
}
break and continue are not yet available in inlined lambdas, but we are planning to
support them too.
Rei ed type parameters

Sometimes we need to access a type passed to us as a parameter:
166
fun <T> TreeNode.findParentOfType(clazz: Class<T>): T? {
var p = parent
while (p != null && !clazz.isInstance(p)) {
p = p.parent
}
@Suppress("UNCHECKED_CAST")
return p as T?
}
Here, we walk up a tree and use re ection to check if a node has a certain type. It’s all ne, but
the call site is not very pretty:
treeNode.findParentOfType(MyTreeNode::class.java)
What we actually want is simply pass a type to this function, i.e. call it like this:
treeNode.findParentOfType<MyTreeNode>()
To enable this, inline functions support rei ed type parameters, so we can write something like
this:
inline fun <reified T> TreeNode.findParentOfType(): T? {

var p = parent
while (p != null && p !is T) {
p = p.parent
}
return p as T?
}
We quali ed the type parameter with the reified modi er, now it’s accessible inside the
function, almost as if it were a normal class. Since the function is inlined, no re ection is needed,
normal operators like !is and as are working now. Also, we can call it as mentioned above:
myTree.findParentOfType<MyTreeNodeType>() .
Though re ection may not be needed in many cases, we can still use it with a rei ed type
parameter:
inline fun <reified T> membersOf() = T::class.members
fun main(s: Array<String>) {

println(membersOf<StringBuilder>().joinToString("\n"))
}
Normal functions (not marked as inline) cannot have rei ed parameters. A type that does not
have a run-time representation (e.g. a non-rei ed type parameter or a ctitious type like
Nothing ) cannot be used as an argument for a rei ed type parameter.
For a low-level description, see the spec document.
Inline properties (since 1.1)
167
The inline modi er can be used on accessors of properties that don't have a backing eld. You
can annotate individual property accessors:
val foo: Foo

inline get() = Foo()
var bar: Bar

get() = ...
inline set(v) { ... }
You can also annotate an entire property, which marks both of its accessors as inline:
inline var bar: Bar

get() = ...
set(v) { ... }
At the call site, inline accessors are inlined as regular inline functions.
Restrictions for public API inline functions

When an inline function is public or protected and is not a part of a private or
internal declaration, it is considered a module's public API. It can be called in other modules
and is inlined at such call sites as well.
This imposes certain risks of binary incompatibility caused by changes in the module that
declares an inline function in case the calling module is not re-compiled after the change.
To eliminate the risk of such incompatibility being introduced by a change in non-public API of a
module, the public API inline functions are not allowed to use non-public-API declarations, i.e.
private and internal declarations and their parts, in their bodies.
An internal declaration can be annotated with @PublishedApi , which allows its use in
public API inline functions. When an internal inline function is marked as @PublishedApi ,
its body is checked too, as if it were public.
168
Collections
Kotlin Collections Overview
The Kotlin Standard Library provides a comprehensive set of tools for managing collections –
groups of a variable number of items (possibly zero) that share signi cance to the problem being
solved and are operated upon commonly.
Collections are a common concept for most programming languages, so if you're familiar with,
for example, Java or Python collections, you can skip this introduction and proceed to the
detailed sections.
A collection usually contains a number of objects (this number may also be zero) of the same
type. Objects in a collection are called elements or items. For example, all the students in a
department form a collection that can be used to calculate their average age. The following
collection types are relevant for Kotlin:
— List is an ordered collection with access to elements by indices – integer numbers that re ect
their position. Elements can occur more than once in a list. An example of a list is a sentence:
it's a group of words, their order is important, and they can repeat.
— Set is a collection of unique elements. It re ects the mathematical abstraction of set: a group
of objects without repetitions. Generally, the order of set elements has no signi cance. For
example, an alphabet is a set of letters.
— Map (or dictionary) is a set of key-value pairs. Keys are unique, and each of them maps to
exactly one value. The values can be duplicates. Maps are useful for storing logical
connections between objects, for example, an employee's ID and their position.
Kotlin lets you manipulate collections independently of the exact type of objects stored in them.
In other words, you add a String to a list of String s the same way as you would do with
Int s or a user-de ned class. So, the Kotlin Standard Library o ers generic interfaces, classes,
and functions for creating, populating, and managing collections of any type.
The collection interfaces and related functions are located in the kotlin.collections package. Let's
get an overview of its contents.
Collection types
The Kotlin Standard Library provides implementations for basic collection types: sets, lists, and
maps. A pair of interfaces represent each collection type:
169
— A read-only interface that provides operations for accessing collection elements.
— A mutable interface that extends the corresponding read-only interface with write operations:
adding, removing, and updating its elements.
Note that altering a mutable collection doesn't require it to be a var: write operations modify the
same mutable collection object, so the reference doesn't change. Although, if you try to reassign
a val collection, you'll get a compilation error.
val numbers = mutableListOf("one", "two", "three", "four")

numbers.add("five") // this is OK
//numbers = mutableListOf("six", "seven") // compilation error
The read-only collection types are covariant. This means that, if a Rectangle class inherits from
Shape , you can use a List<Rectangle> anywhere the List<Shape> is required. In other
words, the collection types have the same subtyping relationship as the element types. Maps are
covariant on the value type, but not on the key type.
In turn, mutable collections aren't covariant; otherwise, this would lead to runtime failures. If
MutableList<Rectangle> was a subtype of MutableList<Shape> , you could insert other
Shape inheritors (for example, Circle ) into it, thus violating its Rectangle type argument.
Below is a diagram of the Kotlin collection interfaces:
Let's walk through the interfaces and their implementations.
Collection
Collection<T> is the root of the collection hierarchy. This interface represents the common
behavior of a read-only collection: retrieving size, checking item membership, and so on.
Collection inherits from the Iterable<T> interface that de nes the operations for iterating
elements. You can use Collection as a parameter of a function that applies to di erent
collection types. For more speci c cases, use the Collection 's inheritors: List and Set.
fun printAll(strings: Collection<String>) {

for(s in strings) print("$s ")
println()
}
fun main() {
val stringList = listOf("one", "two", "one")
printAll(stringList)
val stringSet = setOf("one", "two", "three")

printAll(stringSet)
}
MutableCollection is a Collection with write operations, such as add and remove .
170
fun List<String>.getShortWordsTo(shortWords: MutableList<String>, maxLength: Int) {
this.filterTo(shortWords) { it.length <= maxLength}
// throwing away the articles
val articles = setOf("a", "A", "an", "An", "the", "The")
shortWords -= articles
}
fun main() {
val words = "A long time ago in a galaxy far far away".split(" ")
val shortWords = mutableListOf<String>()
words.getShortWordsTo(shortWords, 3)
println(shortWords)
}
List
List<T> stores elements in a speci ed order and provides indexed access to them. Indices start
from zero – the index of the rst element – and go to lastIndex which is the (list.size -
1) .
val numbers = listOf("one", "two", "three", "four")

println("Number of elements: ${numbers.size}")
println("Third element: ${numbers.get(2)}")
println("Fourth element: ${numbers[3]}")
println("Index of element \"two\" ${numbers.indexOf("two")}")
List elements (including nulls) can duplicate: a list can contain any number of equal objects or
occurrences of a single object. Two lists are considered equal if they have the same sizes and
structurally equal elements at the same positions.
val bob = Person("Bob", 31)

val people = listOf<Person>(Person("Adam", 20), bob, bob)
val people2 = listOf<Person>(Person("Adam", 20), Person("Bob", 31), bob)
println(people == people2)
bob.age = 32
println(people == people2)
MutableList is a List with list-speci c write operations, for example, to add or remove an
element at a speci c position.
val numbers = mutableListOf(1, 2, 3, 4)

numbers.add(5)
numbers.removeAt(1)
numbers[0] = 0
numbers.shuffle()
println(numbers)
As you see, in some aspects lists are very similar to arrays. However, there is one important
di erence: an array's size is de ned upon initialization and is never changed; in turn, a list
doesn't have a prede ned size; a list's size can be changed as a result of write operations: adding,
updating, or removing elements.
171
In Kotlin, the default implementation of List is ArrayList which you can think of as a
resizable array.
Set
Set<T> stores unique elements; their order is generally unde ned. null elements are unique
as well: a Set can contain only one null . Two sets are equal if they have the same size, and for
each element of a set there is an equal element in the other set.
val numbers = setOf(1, 2, 3, 4)

println("Number of elements: ${numbers.size}")
if (numbers.contains(1)) println("1 is in the set")
val numbersBackwards = setOf(4, 3, 2, 1)

println("The sets are equal: ${numbers == numbersBackwards}")
MutableSet is a Set with write operations from MutableCollection .
The default implementation of Set – LinkedHashSet – preserves the order of elements

insertion. Hence, the functions that rely on the order, such as first() or last() , return
predictable results on such sets.
val numbers = setOf(1, 2, 3, 4) // LinkedHashSet is the default implementation

val numbersBackwards = setOf(4, 3, 2, 1)
println(numbers.first() == numbersBackwards.first())
println(numbers.first() == numbersBackwards.last())
An alternative implementation – HashSet – says nothing about the elements order, so calling
such functions on it returns unpredictable results. However, HashSet requires less memory to
store the same number of elements.
Map
Map<K, V> is not an inheritor of the Collection interface; however, it's a Kotlin collection
type as well. A Map stores key-value pairs (or entries); keys are unique, but di erent keys can be
paired with equal values. The Map interface provides speci c functions, such as access to value
by key, searching keys and values, and so on.
val numbersMap = mapOf("key1" to 1, "key2" to 2, "key3" to 3, "key4" to 1)
println("All keys: ${numbersMap.keys}")

println("All values: ${numbersMap.values}")
if ("key2" in numbersMap) println("Value by key \"key2\": ${numbersMap["key2"]}")
if (1 in numbersMap.values) println("The value 1 is in the map")
if (numbersMap.containsValue(1)) println("The value 1 is in the map") // same as previous
Two maps containing the equal pairs are equal regardless of the pair order.
172
val anotherMap = mapOf("key2" to 2, "key1" to 1, "key4" to 1, "key3" to 3)
println("The maps are equal: ${numbersMap == anotherMap}")
MutableMap is a Map with map write operations, for example, you can add a new key-value pair
or update the value associated with the given key.
val numbersMap = mutableMapOf("one" to 1, "two" to 2)

numbersMap.put("three", 3)
numbersMap["one"] = 11
println(numbersMap)
The default implementation of Map – LinkedHashMap – preserves the order of elements

insertion when iterating the map. In turn, an alternative implementation – HashMap – says
nothing about the elements order.
173
Constructing Collections
Constructing from elements

The most common way to create a collection is with the standard library functions listOf<T>(),
setOf<T>(), mutableListOf<T>(), mutableSetOf<T>(). If you provide a comma-separated
list of collection elements as arguments, the compiler detects the element type automatically.
When creating empty collections, specify the type explicitly.
val numbersSet = setOf("one", "two", "three", "four")

val emptySet = mutableSetOf<String>()
The same is available for maps with the functions mapOf() and mutableMapOf(). The map's
keys and values are passed as Pair objects (usually created with to in x function).
Note that the to notation creates a short-living Pair object, so it's recommended that you use
it only if performance isn't critical. To avoid excessive memory usage, use alternative ways. For
example, you can create a mutable map and populate it using the write operations. The apply()
function can help to keep the initialization uent here.
val numbersMap = mutableMapOf<String, String>().apply { this["one"] = "1"; this["two"] =

"2" }
Empty collections
There are also functions for creating collections without any elements: emptyList(),
emptySet(), and emptyMap(). When creating empty collections, you should specify the type of
elements that the collection will hold.
val empty = emptyList<String>()
Initializer functions for lists

For lists, there is a constructor that takes the list size and the initializer function that de nes the
element value based on its index.
val doubled = List(3, { it * 2 }) // or MutableList if you want to change its content

later
println(doubled)
Concrete type constructors

To create a concrete type collection, such as an ArrayList or LinkedList , you can use the
available constructors for these types. Similar constructors are available for implementations of
Set and Map .
174
val linkedList = LinkedList<String>(listOf("one", "two", "three"))
val presizedSet = HashSet<Int>(32)
Copying
To create a collection with the same elements as an existing collection, you can use copying
operations. Collection copying operations from the standard library create shallow copy
collections with references to the same elements. Thus, a change made to a collection element
re ects in all its copies.
Collection copying functions, such as toList(), toMutableList(), toSet() and others,

create a snapshot of a collection at a speci c moment. Their result is a new collection of the same
elements. If you add or remove elements from the original collection, this won't a ect the copies.
Copies may be changed independently of the source as well.
val sourceList = mutableListOf(1, 2, 3)

val copyList = sourceList.toMutableList()
val readOnlyCopyList = sourceList.toList()
sourceList.add(4)
println("Copy size: ${copyList.size}")
//readOnlyCopyList.add(4) // compilation error

println("Read-only copy size: ${readOnlyCopyList.size}")
These functions can also be used for converting collections to other types, for example, build a
set from a list or vice versa.

val copySet = sourceList.toMutableSet()
copySet.add(3)
copySet.add(4)
println(copySet)
Alternatively, you can create new references to the same collection instance. New references are
created when you initialize a collection variable with an existing collection. So, when the
collection instance is altered through a reference, the changes are re ected in all its references.

val referenceList = sourceList
referenceList.add(4)
println("Source size: ${sourceList.size}")
Collection initialization can be used for restricting mutability. For example, if you create a List
reference to a MutableList , the compiler will produce errors if you try to modify the collection
through this reference.

val referenceList: List<Int> = sourceList
//referenceList.add(4) //compilation error
sourceList.add(4)
println(referenceList) // shows the current state of sourceList
175
Invoking functions on other collections
Collections can be created in result of various operations on other collections. For example,
ltering a list creates a new list of elements that match the lter:

val longerThan3 = numbers.filter { it.length > 3 }
println(longerThan3)
Mapping produces a list of a transformation results:
val numbers = setOf(1, 2, 3)

println(numbers.map { it * 3 })
println(numbers.mapIndexed { idx, value -> value * idx })
Association produces maps:

println(numbers.associateWith { it.length })
For more information about operations on collections in Kotlin, see Collection Operations
Overview.
176
Iterators
For traversing collection elements, the Kotlin standard library supports the commonly used
mechanism of iterators – objects that provide access to the elements sequentially without
exposing the underlying structure of the collection. Iterators are useful when you need to
process all the elements of a collection one-by-one, for example, print values or make similar
updates to them.
Iterators can be obtained for inheritors of the Iterable<T> interface, including Set and
List , by calling the iterator() function. Once you obtain an iterator, it points to the rst
element of a collection; calling the next() function returns this element and moves the iterator
position to the following element if it exists. Once the iterator passes through the last element, it
can no longer be used for retrieving elements; neither can it be reset to any previous position. To
iterate through the collection again, create a new iterator.

val numbersIterator = numbers.iterator()
while (numbersIterator.hasNext()) {
println(numbersIterator.next())
}
Another way to go through an Iterable collection is the well-known for loop. When using
for on a collection, you obtain the iterator implicitly. So, the following code is equivalent to the
example above:

for (item in numbers) {
println(item)
}
Finally, there is a useful forEach() function that lets you automatically iterate a collection and
execute the given code for each element. So, the same example would look like this:

numbers.forEach {
println(it)
}
List iterators
For lists, there is a special iterator implementation: ListIterator. It supports iterating lists in
both directions: forwards and backwards. Backward iteration is implemented by the functions
hasPrevious() and previous(). Additionally, the ListIterator provides information
about the element indices with the functions nextIndex() and previousIndex().
177
val listIterator = numbers.listIterator()
while (listIterator.hasNext()) listIterator.next()
println("Iterating backwards:")
while (listIterator.hasPrevious()) {
print("Index: ${listIterator.previousIndex()}")
println(", value: ${listIterator.previous()}")
}
Having the ability to iterate in both directions, means the ListIterator can still be used after
it reaches the last element.
Mutable iterators
For iterating mutable collections, there is MutableIterator that extends Iterator with the
element removal function remove(). So, you can remove elements from a collection while
iterating it.

val mutableIterator = numbers.iterator()
mutableIterator.next()
mutableIterator.remove()
println("After removal: $numbers")
In addition to removing elements, the MutableListIterator can also insert and replace
elements while iterating the list.
val numbers = mutableListOf("one", "four", "four")

val mutableListIterator = numbers.listIterator()
mutableListIterator.next()
mutableListIterator.add("two")
mutableListIterator.next()
mutableListIterator.set("three")
println(numbers)
178
Ranges and Progressions
Kotlin lets you easily create ranges of values using the rangeTo() function from the
kotlin.ranges package and its operator form .. . Usually, rangeTo() is complemented by
in or !in functions.
if (i in 1..4) { // equivalent of 1 <= i && i <= 4

print(i)
}
Integral type ranges (IntRange, LongRange, CharRange) have an extra feature: they can be
iterated over. These ranges are also progressions of the corresponding integral types. Such
ranges are generally used for iteration in the for loops.
for (i in 1..4) print(i)
To iterate numbers in reverse order, use the downTo function instead of .. .
for (i in 4 downTo 1) print(i)
It is also possible to iterate over numbers with an arbitrary step (not necessarily 1). This is done
via the step function.
for (i in 1..8 step 2) print(i)

println()
for (i in 8 downTo 1 step 2) print(i)
To iterate a number range which does not include its end element, use the until function:
for (i in 1 until 10) { // i in [1, 10), 10 is excluded

print(i)
}
Range
A range de nes a closed interval in the mathematical sense: it is de ned by its two endpoint
values which are both included in the range. Ranges are de ned for comparable types: having an
order, you can de ne whether an arbitrary instance is in the range between two given instances.
The main operation on ranges is contains , which is usually used in the form of in and !in
operators.
To create a range for your class, call the rangeTo() function on the range start value and
provide the end value as an argument. rangeTo() is often called in its operator form .. .
val versionRange = Version(1, 11)..Version(1, 30)

println(Version(0, 9) in versionRange)
println(Version(1, 20) in versionRange)
Progression
179
As shown in the examples above, the ranges of integral types, such as Int , Long , and Char ,
can be treated as arithmetic progressions of them. In Kotlin, these progressions are de ned by
special types: IntProgression, LongProgression, and CharProgression.
Progressions have three essential properties: the first element, the last element, and a
non-zero step . The rst element is first , subsequent elements are the previous element
plus a step . The last element is always hit by iteration unless the progression is empty. Iteration
over a progression with a positive step is equivalent to an indexed for loop in Java/JavaScript.
for (int i = first; i <= last; i += step) {

// ...
}
When you create a progression implicitly by iterating a range, this progression's first and
last elements are the range's endpoints, and the step is 1.
for (i in 1..10) print(i)
To de ne a custom progression step, use the step function on a range.
for (i in 1..8 step 2) print(i)
The last element of the progression is calculated to nd the maximum value not greater than the
end value for a positive step or the minimum value not less than the end value for a negative
step such that (last - first) % step == 0 .
To create a progression for iterating in reverse order, use downTo instead of .. when de ning
the range for it.
for (i in 4 downTo 1) print(i)
Progressions implement Iterable<N> , where N is Int , Long , or Char respectively, so you

can use them in various collection functions like map , filter , and other.
println((1..10).filter { it % 2 == 0 })
180
Sequences
Along with collections, the Kotlin standard library contains another container type – sequences
(Sequence<T>). Sequences o er the same functions as Iterable but implement another
approach to multi-step collection processing.
When the processing of an Iterable includes multiple steps, they are executed eagerly: each
processing step completes and returns its result – an intermediate collection. The following step
executes on this collection. In turn, multi-step processing of sequences is executed lazily when
possible: actual computing happens only when the result of the whole processing chain is
requested.
The order of operations execution is di erent as well: Sequence performs all the processing
steps one-by-one for every single element. In turn, Iterable completes each step for the
whole collection and then proceeds to the next step.
So, the sequences let you avoid building results of intermediate steps, therefore improving the
performance of the whole collection processing chain. However, the lazy nature of sequences
adds some overhead which may be signi cant when processing smaller collections or doing
simpler computations. Hence, you should consider both Sequence and Iterable and decide
which one is better for your case.
Constructing
From elements
To create a sequence, call the sequenceOf() function listing the elements as its arguments.
val numbersSequence = sequenceOf("four", "three", "two", "one")
From Iterable
If you already have an Iterable object (such as a List or a Set ), you can create a sequence
from it by calling asSequence().

val numbersSequence = numbers.asSequence()
From function
One more way to create a sequence is by building it with a function that calculates its elements.
To build a sequence based on a function, call generateSequence() with this function as an
argument. Optionally, you can specify the rst element as an explicit value or a result of a
function call. The sequence generation stops when the provided function returns null . So, the
sequence in the example below is in nite.
181
val oddNumbers = generateSequence(1) { it + 2 } // ìt` is the previous element
println(oddNumbers.take(5).toList())
//println(oddNumbers.count()) // error: the sequence is infinite
To create a nite sequence with generateSequence() , provide a function that returns null
after the last element you need.
val oddNumbersLessThan10 = generateSequence(1) { if (it < 10) it + 2 else null }

println(oddNumbersLessThan10.count())
From chunks
Finally, there is a function that lets you produce sequence elements one by one or by chunks of
arbitrary sizes – the sequence() function. This function takes a lambda expression containing
calls of yield() and yieldAll() functions. They return an element to the sequence consumer
and suspend the execution of sequence() until the next element is requested by the
consumer. yield() takes a single element as an argument; yieldAll() can take an
Iterable object, an Iterator , or another Sequence . A Sequence argument of
yieldAll() can be in nite. However, such a call must be the last: all subsequent calls will
never be executed.
val oddNumbers = sequence {

yield(1)
yieldAll(listOf(3, 5))
yieldAll(generateSequence(7) { it + 2 })
}
println(oddNumbers.take(5).toList())
Sequence operations
The sequence operations can be classi ed into the following groups regarding their state
requirements:
— Stateless operations require no state and process each element independently, for example,
map() or filter(). Stateless operations can also require a small constant amount of state
to process an element, for example, take() or drop().
— Stateful operations require a signi cant amount of state, usually proportional to the number
of elements in a sequence.
If a sequence operation returns another sequence, which is produced lazily, it's called
intermediate. Otherwise, the operation is terminal. Examples of terminal operations are
toList() or sum(). Sequence elements can be retrieved only with terminal operations.
Sequences can be iterated multiple times; however some sequence implementations might
constrain themselves to be iterated only once. That is mentioned speci cally in their
documentation.
182
Sequence processing example
Let's take a look at the di erence between Iterable and Sequence with an example.
Iterable
Assume that you have a list of words. The code below lters the words longer than three
characters and prints the lengths of rst four such words.
val words = "The quick brown fox jumps over the lazy dog".split(" ")
val lengthsList = words.filter { println("filter: $it"); it.length > 3 }
.map { println("length: ${it.length}"); it.length }
.take(4)
println("Lengths of first 4 words longer than 3 chars:")

println(lengthsList)
When you run this code, you'll see that the filter() and map() functions are executed in the
same order as they appear in the code. First, you see filter: for all elements, then length:
for the elements left after ltering, and then the output of the two last lines. This is how the list
processing goes:
Sequence
Now let's write the same with sequences:
val words = "The quick brown fox jumps over the lazy dog".split(" ")
//convert the List to a Sequence
val wordsSequence = words.asSequence()
val lengthsSequence = wordsSequence.filter { println("filter: $it"); it.length > 3 }

.map { println("length: ${it.length}"); it.length }
.take(4)
println("Lengths of first 4 words longer than 3 chars")

// terminal operation: obtaining the result as a List
println(lengthsSequence.toList())
The output of this code shows that the filter() and map() functions are called only when
building the result list. So, you rst see the line of text “Lengths of..” and then the sequence
processing starts. Note that for elements left after ltering, the map executes before ltering the
next element. When the result size reaches 4, the processing stops because it's the largest
possible size that take(4) can return.
The sequence processing goes like this:
In this example, the sequence processing takes 18 steps instead of 23 steps for doing the same
with lists.
183
Collection Operations Overview
The Kotlin standard library o ers a broad variety of functions for performing operations on
collections. This includes simple operations, such as getting or adding elements, as well as more
complex ones including search, sorting, ltering, transformations, and so on.
Extension and member functions

Collection operations are declared in the standard library in two ways: member functions of
collection interfaces and extension functions.
Member functions de ne operations that are essential for a collection type. For example,
Collection contains the function isEmpty() for checking its emptiness; List contains get()
for index access to elements, and so on.
When you create own implementations of collection interfaces, you must implement their
member functions. To make the creation of new implementations easier, use the skeletal
implementations of collection interfaces from the standard library: AbstractCollection,
AbstractList, AbstractSet, AbstractMap, and their mutable counterparts.
Other collection operations are declared as extension functions. These are ltering,
transformation, ordering, and other collection processing functions.
Common operations
Common operations are available for both read-only and mutable collections. Common
operations fall into these groups:
— Transformations
— Filtering
— plus and minus operators
— Grouping
— Retrieving collection parts
— Retrieving single elements
— Ordering
— Aggregate operations
Operations described on these pages return their results without a ecting the original collection.
For example, a ltering operation produces a new collection that contains all the elements
matching the ltering predicate. Results of such operations should be either stored in variables,
or used in some other way, for example, passed in other functions.
184
numbers.filter { it.length > 3 } // nothing happens with `numbers`, result is lost
println("numbers are still $numbers")
val longerThan3 = numbers.filter { it.length > 3 } // result is stored in `longerThan3`
println("numbers longer than 3 chars are $longerThan3")
For certain collection operations, there is an option to specify the destination object. Destination
is a mutable collection to which the function appends its resulting items instead of returning
them in a new object. For performing operations with destinations, there are separate functions
with the To post x in their names, for example, filterTo() instead of filter() or
associateTo() instead of associate(). These functions take the destination collection as an
additional parameter.

val filterResults = mutableListOf<String>() //destination object
numbers.filterTo(filterResults) { it.length > 3 }
numbers.filterIndexedTo(filterResults) { index, _ -> index == 0 }
println(filterResults) // contains results of both operations
For convenience, these functions return the destination collection back, so you can create it right
in the corresponding argument of the function call:
// filter numbers right into a new hash set,

// thus eliminating duplicates in the result
val result = numbers.mapTo(HashSet()) { it.length }
println("distinct item lengths are $result")
Functions with destination are available for ltering, association, grouping, attening, and other
operations. For the complete list of destination operations see the Kotlin collections reference.
Write operations
For mutable collections, there are also write operations that change the collection state. Such
operations include adding, removing, and updating elements. Write operations are listed in the
Write operations and corresponding sections of List speci c operations and Map speci c
operations.
For certain operations, there are pairs of functions for performing the same operation: one
applies the operation in-place and the other returns the result as a separate collection. For
example, sort() sorts a mutable collection in-place, so it's state changes; sorted() creates a
new collection that contains the same elements in the sorted order.

val sortedNumbers = numbers.sorted()
println(numbers == sortedNumbers) // false
numbers.sort()
println(numbers == sortedNumbers) // true
185
Collection Transformations
The Kotlin standard library provides a set of extension functions for collection transformations.
These functions build new collections from existing ones based on the transformation rules
provided. In this page, we'll give an overview of the available collection transformation functions.
Mapping
The mapping transformation creates a collection from the results of a function on the elements
of another collection. The basic mapping function is map(). It applies the given lambda function
to each subsequent element and returns the list of the lambda results. The order of results is the
same as the original order of elements. To apply a transformation that additionally uses the
element index as an argument, use mapIndexed().

println(numbers.map { it * 3 })
println(numbers.mapIndexed { idx, value -> value * idx })
If the transformation produces null on certain elements, you can lter out the null s from the
result collection by calling the mapNotNull() function instead of map() , or
mapIndexedNotNull() instead of mapIndexed() .

println(numbers.mapNotNull { if ( it == 2) null else it * 3 })
println(numbers.mapIndexedNotNull { idx, value -> if (idx == 0) null else value * idx })
When transforming maps, you have two options: transform keys leaving values unchanged and
vice versa. To apply a given transformation to keys, use mapKeys(); in turn, mapValues()
transforms values. Both functions use the transformations that take a map entry as an argument,
so you can operate both its key and value.

println(numbersMap.mapKeys { it.key.toUpperCase() })
println(numbersMap.mapValues { it.value + it.key.length })
Zipping
Zipping transformation is building pairs from elements with the same positions in both
collections. In the Kotlin standard library, this is done by the zip() extension function. When
called on a collection or an array with another collection (array) as an argument, zip() returns
the List of Pair objects. The elements of the receiver collection are the rst elements in
these pairs. If the collections have di erent sizes, the result of the zip() is the smaller size; the
last elements of the larger collection are not included in the result. zip() can also be called in
the in x form a zip b .
186
val colors = listOf("red", "brown", "grey")
val animals = listOf("fox", "bear", "wolf")
println(colors zip animals)
val twoAnimals = listOf("fox", "bear")

println(colors.zip(twoAnimals))
You can also call zip() with a transformation function that takes two parameters: the receiver
element and the argument element. In this case, the result List contains the return values of
the transformation function called on pairs of the receiver and the argument elements with the
same positions.
val colors = listOf("red", "brown", "grey")

val animals = listOf("fox", "bear", "wolf")
println(colors.zip(animals) { color, animal -> "The ${animal.capitalize()} is $color"})
When you have a List of Pair s, you can do the reverse transformation – unzipping – that
builds two lists from these pairs:
— The rst list contains the rst elements of each Pair in the original list.
— The second list contains the second elements.
To unzip a list of pairs, call unzip().
val numberPairs = listOf("one" to 1, "two" to 2, "three" to 3, "four" to 4)

println(numberPairs.unzip())
Association
Association transformations allow building maps from the collection elements and certain values
associated with them. In di erent association types, the elements can be either keys or values in
the association map.
The basic association function associateWith() creates a Map in which the elements of the
original collection are keys, and values are produced from them by the given transformation
function. If two elements are equal, only the last one remains in the map.

println(numbers.associateWith { it.length })
For building maps with collection elements as values, there is the function associateBy(). It
takes a function that returns a key based on an element's value. If two elements are equal, only
the last one remains in the map. associateBy() can also be called with a value transformation
function.
187
println(numbers.associateBy { it.first().toUpperCase() })
println(numbers.associateBy(keySelector = { it.first().toUpperCase() }, valueTransform =
{ it.length }))
Another way to build maps in which both keys and values are somehow produced from collection
elements is the function associate(). It takes a lambda function that returns a Pair : the key
and the value of the corresponding map entry.
Note that associate() produces short-living Pair objects which may a ect the performance.
Thus, associate() should be used when the performance isn't critical or it's more preferable
than other options.
An example of the latter is when a key and the corresponding value are produced from an
element together.
val names = listOf("Alice Adams", "Brian Brown", "Clara Campbell")

println(names.associate { name -> parseFullName(name).let { it.lastName to it.firstName }
})
Here we call a transform function on an element rst, and then build a pair from the properties
of that function's result.
Flattening
If you operate nested collections, you may nd the standard library functions that provide at
access to nested collection elements useful.
The rst function is flatten(). You can call it on a collection of collections, for example, a
List of Set s. The function returns a single List of all the elements of the nested collections.
val numberSets = listOf(setOf(1, 2, 3), setOf(4, 5, 6), setOf(1, 2))

println(numberSets.flatten())
Another function – flatMap() provides a exible way to process nested collections. It takes a
function that maps a collection element to another collection. As a result, flatMap() returns a
single list of its return values on all the elements. So, flatMap() behaves as a subsequent call
of map() (with a collection as a mapping result) and flatten() .
val containers = listOf(

StringContainer(listOf("one", "two", "three")),
StringContainer(listOf("four", "five", "six")),
StringContainer(listOf("seven", "eight"))
)
println(containers.flatMap { it.values })
String representation
188
If you need to retrieve the collection content in a readable format, use functions that transform
the collections to strings: joinToString() and joinTo().
joinToString() builds a single String from the collection elements based on the provided
arguments. joinTo() does the same but appends the result to the given Appendable object.
When called with the default arguments, the functions return the result similar to calling
toString() on the collection: a String of elements' string representations separated by
commas with spaces.
println(numbers)
println(numbers.joinToString())
val listString = StringBuffer("The list of numbers: ")

numbers.joinTo(listString)
println(listString)
To build a custom string representation, you can specify its parameters in function arguments
separator , prefix , and postfix . The resulting string will start with the prefix and end
with the postfix . The separator will come after each element except the last.

println(numbers.joinToString(separator = " | ", prefix = "start: ", postfix = ": end"))
For bigger collections, you may want to specify the limit – a number of elements that will be
included into result. If the collection size exceeds the limit , all the other elements will be
replaced with a single value of the truncated argument.
val numbers = (1..100).toList()

println(numbers.joinToString(limit = 10, truncated = "<...>"))
Finally, to customize the representation of elements themselves, provide the transform

function.

println(numbers.joinToString { "Element: ${it.toUpperCase()}"})
189
Filtering
Filtering is one of the most popular tasks in the collection processing. In Kotlin, ltering
conditions are de ned by predicates – lambda functions that take a collection element and return
a boolean value: true means that the given element matches the predicate, false means the
opposite.
The standard library contains a group of extension functions that let you lter collections in a
single call. These functions leave the original collection unchanged, so they are available for both
mutable and read-only collections. To operate the ltering result, you should assign it to a
variable or chain the functions after ltering.
Filtering by predicate
The basic ltering function is filter(). When called with a predicate, filter() returns the
collection elements that match it. For both List and Set , the resulting collection is a List ,
for Map it's a Map as well.

val longerThan3 = numbers.filter { it.length > 3 }
println(longerThan3)

val filteredMap = numbersMap.filter { (key, value) -> key.endsWith("1") && value > 10}
println(filteredMap)
The predicates in filter() can only check the values of the elements. If you want to use
element positions in the lter, use filterIndexed(). It takes a predicate with two arguments:
the index and the value of an element.
To lter collections by negative conditions, use filterNot(). It returns a list of elements for
which the predicate yields false .
val filteredIdx = numbers.filterIndexed { index, s -> (index != 0) && (s.length < 5) }

val filteredNot = numbers.filterNot { it.length <= 3 }
println(filteredIdx)
println(filteredNot)
There are also functions that narrow the element type by ltering elements of a given type:
— filterIsInstance() returns collection elements of a given type. Being called on a

List<Any> , filterIsInstance<T>() returns a List<T> , thus allowing you to call
functions of the T type on its items.
190
val numbers = listOf(null, 1, "two", 3.0, "four")
println("All String elements in upper case:")
numbers.filterIsInstance<String>().forEach {
println(it.toUpperCase())
}
— filterNotNull() returns all non-null elements. Being called on a List<T?> ,

filterNotNull() returns a List<T: Any> , thus allowing you to treat the elements as
non-null objects.
val numbers = listOf(null, "one", "two", null)

numbers.filterNotNull().forEach {
println(it.length) // length is unavailable for nullable Strings
}
Partitioning
Another ltering function – partition() – lters a collection by a predicate and keeps the
elements that don't match it in a separate list. So, you have a Pair of List s as a return value:
the rst list containing elements that match the predicate and the second one containing
everything else from the original collection.

val (match, rest) = numbers.partition { it.length > 3 }
println(match)
println(rest)
Testing predicates
Finally, there are functions that simply test a predicate against collection elements:
— any() returns true if at least one element matches the given predicate.
— none() returns true if none of the elements match the given predicate.
— all() returns true if all elements match the given predicate. Note that all() returns
true when called with any valid predicate on an empty collection. Such behavior is known in
logic as vacuous truth.
println(numbers.any { it.endsWith("e") })
println(numbers.none { it.endsWith("a") })
println(numbers.all { it.endsWith("e") })
println(emptyList<Int>().all { it > 5 }) // vacuous truth
any() and none() can also be used without a predicate: in this case they just check the
collection emptiness. any() returns true if there are elements and false if there aren't;
none() does the opposite.
191
println(numbers.any())
println(empty.any())
println(numbers.none())
println(empty.none())
192
plus and minus Operators
In Kotlin, plus ( + ) and minus ( - ) operators are de ned for collections. They take a collection as
the rst operand; the second operand can be either an element or another collection. The return
value is a new read-only collection:
— The result of plus contains the elements from the original collection and from the second
operand.
— The result of minus contains the elements of the original collection except the elements from
the second operand. If it's an element, minus removes its rst occurrence; if it's a collection,
all occurrences of its elements are removed.
val plusList = numbers + "five"

val minusList = numbers - listOf("three", "four")
println(plusList)
println(minusList)
For the details on plus and minus operators for maps, see Map Speci c Operations. The
augmented assignment operators plusAssign ( += ) and minusAssign ( -= ) are also de ned
for collections. However, for read-only collections, they actually use the plus or minus
operators and try to assign the result to the same variable. Thus, they are available only on var
read-only collections. For mutable collections, they modify the collection if it's a val . For more
details see Collection Write Operations.
193
Grouping
The Kotlin standard library provides extension functions for grouping collection elements. The
basic function groupBy() takes a lambda function and returns a Map . In this map, each key is
the lambda result and the corresponding value is the List of elements on which this result is
returned. This function can be used, for example, to group a list of String s by their rst letter.
You can also call groupBy() with a second lambda argument – a value transformation function.
In the result map of groupBy() with two lambdas, the keys produced by keySelector
function are mapped to the results of the value transformation function instead of the original
elements.
val numbers = listOf("one", "two", "three", "four", "five")
println(numbers.groupBy { it.first().toUpperCase() })
println(numbers.groupBy(keySelector = { it.first() }, valueTransform = { it.toUpperCase()
}))
If you want to group elements and then apply an operation to all groups at one time, use the
function groupingBy(). It returns an instance of the Grouping type. The Grouping instance
lets you apply operations to all groups in a lazy manner: the groups are actually built right before
the operation execution.
Namely, Grouping supports the following operations:
— eachCount() counts the elements in each group.
— fold() and reduce() perform fold and reduce operations on each group as a separate
collection and return the results.
— aggregate() applies a given operation subsequently to all the elements in each group and
returns the result. This is the generic way to perform any operations on a Grouping . Use it
to implement custom operations when fold or reduce are not enough.
val numbers = listOf("one", "two", "three", "four", "five", "six")

println(numbers.groupingBy { it.first() }.eachCount())
194
Retrieving Collection Parts
The Kotlin standard library contains extension functions for retrieving parts of a collection. These
functions provide a variety of ways to select elements for the result collection: listing their
positions explicitly, specifying the result size, and others.
Slice
slice() returns a list of the collection elements with given indices. The indices may be passed
either as a range or as a collection of integer values.

println(numbers.slice(1..3))
println(numbers.slice(0..4 step 2))
println(numbers.slice(setOf(3, 5, 0)))
Take and drop

To get the speci ed number of elements starting from the rst, use the take() function. For
getting the last elements, use takeLast(). When called with a number larger than the collection
size, both functions return the whole collection.
To take all the elements except a given number of rst or last elements, call the drop() and
dropLast() functions respectively.

println(numbers.take(3))
println(numbers.takeLast(3))
println(numbers.drop(1))
println(numbers.dropLast(5))
You can also use predicates to de ne the number of elements for taking or dropping. There are
four functions similar to the ones described above:
— takeWhile() is take() with a predicate: it takes the elements up to but excluding the rst
one not matching the predicate. If the rst collection element doesn't match the predicate, the
result is empty.
— takeLastWhile() is similar to takeLast() : it takes the range of elements matching the
predicate from the end of the collection. The rst element of the range is the element next to
the last element not matching the predicate. If the last collection element doesn't match the
predicate, the result is empty;
— dropWhile() is the opposite to takeWhile() with the same predicate: it returns the
elements from the rst one not matching the predicate to the end.
— dropLastWhile() is the opposite to takeLastWhile() with the same predicate: it returns
the elements from the beginning to the last one not matching the predicate.
195
println(numbers.takeWhile { !it.startsWith('f') })
println(numbers.takeLastWhile { it != "three" })
println(numbers.dropWhile { it.length == 3 })
println(numbers.dropLastWhile { it.contains('i') })
Chunked
To break a collection onto parts of a given size, use the chunked() function. chunked() takes
a single argument – the size of the chunk – and returns a List of List s of the given size. The
rst chunk starts from the rst element and contains the size elements, the second chunk
holds the next size elements, and so on. The last chunk may have a smaller size.

println(numbers.chunked(3))
You can also apply a transformation for the returned chunks right away. To do this, provide the
transformation as a lambda function when calling chunked() . The lambda argument is a chunk
of the collection. When chunked() is called with a transformation, the chunks are short-living
List s that should be consumed right in that lambda.

println(numbers.chunked(3) { it.sum() }) // ìt` is a chunk of the original collection
Windowed
You can retrieve all possible ranges of the collection elements of a given size. The function for
getting them is called windowed(): it returns a list of element ranges that you would see if you
were looking at the collection through a sliding window of the given size. Unlike chunked() ,
windowed() returns element ranges (windows) starting from each collection element. All the
windows are returned as elements of a single List .

println(numbers.windowed(3))
windowed() provides more exibility with optional parameters:
— step de nes a distance between rst elements of two adjacent windows. By default the
value is 1, so the result contains windows starting from all elements. If you increase the step
to 2, you will receive only windows starting from odd elements: rst, third, an so on.
— partialWindows includes windows of smaller sizes that start from the elements at the end
of the collection. For example, if you request windows of three elements, you can't build them
for the last two elements. Enabling partialWindows in this case includes two more lists of
sizes 2 and 1.
Finally, you can apply a transformation to the returned ranges right away. To do this, provide the
transformation as a lambda function when calling windowed() .
196
println(numbers.windowed(3, step = 2, partialWindows = true))
println(numbers.windowed(3) { it.sum() })
To build two-element windows, there is a separate function - zipWithNext(). It creates pairs of

adjacent elements of the receiver collection. Note that zipWithNext() doesn't break the
collection into pairs; it creates a Pair for each element except the last one, so its result on [1,
2, 3, 4] is [[1, 2], [2, 3], [3, 4]] , not [[1, 2 ], [3, 4]] . zipWithNext() can
be called with a transformation function as well; it should take two elements of the receiver
collection as arguments.

println(numbers.zipWithNext())
println(numbers.zipWithNext() { s1, s2 -> s1.length > s2.length})
197
Retrieving Single Elements
Kotlin collections provide a set of functions for retrieving single elements from collections.
Functions described on this page apply to both lists and sets.
As the de nition of list says, a list is an ordered collection. Hence, every element of a list has its
position that you can use for referring. In addition to functions described on this page, lists o er
a wider set of ways to retrieve and search for elements by indices. For more details, see List
Speci c Operations.
In turn, set is not an ordered collection by de nition. However, the Kotlin Set stores elements in
certain orders. These can be the order of insertion (in LinkedHashSet ), natural sorting order
(in SortedSet ), or another order. The order of a set of elements can also be unknown. In such
cases, the elements are still ordered somehow, so the functions that rely on the element
positions still return their results. However, such results are unpredictable to the caller unless
they know the speci c implementation of Set used.
Retrieving by position
For retrieving an element at a speci c position, there is the function elementAt(). Call it with
the integer number as an argument, and you'll receive the collection element at the given
position. The rst element has the position 0 , and the last one is (size - 1) .
elementAt() is useful for collections that do not provide indexed access, or are not statically
known to provide one. In case of List , it's more idiomatic to use indexed access operator
( get() or [] ).
val numbers = linkedSetOf("one", "two", "three", "four", "five")

println(numbers.elementAt(3))
val numbersSortedSet = sortedSetOf("one", "two", "three", "four")

println(numbersSortedSet.elementAt(0)) // elements are stored in the ascending order
There are also useful aliases for retrieving the rst and the last element of the collection:
first() and last().

println(numbers.first())
println(numbers.last())
To avoid exceptions when retrieving element with non-existing positions, use safe variations of
elementAt() :
— elementAtOrNull() returns null when the speci ed position is out of the collection bounds.
— elementAtOrElse() additionally takes a lambda function that maps an Int argument to an

instance of the collection element type. When called with an out-of-bounds position, the
elementAtOrElse() returns the result of the lambda on the given value.
198
println(numbers.elementAtOrNull(5))
println(numbers.elementAtOrElse(5) { index -> "The value for index $index is undefined"})
Retrieving by condition
Functions first() and last() also let you search a collection for elements matching a given
predicate. When you call first() with a predicate that tests a collection element, you'll receive
the rst element on which the predicate yields true . In turn, last() with a predicate returns
the last element matching it.

println(numbers.first { it.length > 3 })
println(numbers.last { it.startsWith("f") })
If no elements match the predicate, both functions throw exceptions. To avoid them, use
firstOrNull() and lastOrNull() instead: they return null if no matching elements are
found.

println(numbers.firstOrNull { it.length > 6 })
Alternatively, you can use the aliases if their names suit your situation better:
— find() instead of firstOrNull()
— findLast() instead of lastOrNull()
val numbers = listOf(1, 2, 3, 4)

println(numbers.find { it % 2 == 0 })
println(numbers.findLast { it % 2 == 0 })
Random element
If you need to retrieve an arbitrary element of a collection, call the random() function. You can
call it without arguments or with a Random object as a source of the randomness.

println(numbers.random())
Checking existence
To check the presence of an element in a collection, use the contains() function. It returns
true if there is a collection element that equals() the function argument. You can call
contains() in the operator form with the in keyword.
To check the presence of multiple instances together at once, call containsAll() with a
collection of these instances as an argument.
199
println(numbers.contains("four"))
println("zero" in numbers)
println(numbers.containsAll(listOf("four", "two")))
println(numbers.containsAll(listOf("one", "zero")))
Additionally, you can check if the collection contains any elements by calling isEmpty() or
isNotEmpty().

println(numbers.isEmpty())
println(numbers.isNotEmpty())

println(empty.isEmpty())
println(empty.isNotEmpty())
200
Collection Ordering
The order of elements is an important aspect of certain collection types. For example, two lists of
the same elements are not equal if their elements are ordered di erently.
In Kotlin, the orders of objects can be de ned in several ways.
First, there is natural order. It is de ned for inheritors of the Comparable interface. Natural
order is used for sorting them when no other order is speci ed.
Most built-in types are comparable:
— Numeric types use the traditional numerical order: 1 is greater than 0 ; -3.4f is greater
than -5f , an so on.
— Char and String use the lexicographical order: b is greater than a ; world is greater
than hello .
To de ne a natural order for a user-de ned type, make the type an inheritor of Comparable .
This requires implementing the compareTo() function. compareTo() must take another
object of the same type as an argument and return an integer value showing which object is
greater:
— Positive values show that the receiver object is greater.

— Negative values show that it's less than the argument.
— Zero shows that the objects are equal.
Below is a class that can be used for ordering versions that consist of the major and the minor
part.
class Version(val major: Int, val minor: Int): Comparable<Version> {

override fun compareTo(other: Version): Int {
if (this.major != other.major) {
return this.major - other.major
} else if (this.minor != other.minor) {
return this.minor - other.minor
} else return 0
}
}
fun main() {
println(Version(1, 2) > Version(1, 3))
println(Version(2, 0) > Version(1, 5))
}
201
Custom orders let you sort instances of any type in a way you like. Particularly, you can de ne an
order for non-comparable objects or de ne an order other than natural for a comparable type.
To de ne a custom order for a type, create a Comparator for it. Comparator contains the
compare() function: it takes two instances of a class and returns the integer result of the
comparison between them. The result is interpreted in the same way as the result of a
compareTo() as is described above.
val lengthComparator = Comparator { str1: String, str2: String -> str1.length -

str2.length }
println(listOf("aaa", "bb", "c").sortedWith(lengthComparator))
Having the lengthComparator , you are able to arrange strings by their length instead of the
default lexicographical order.
A shorter way to de ne a Comparator is the compareBy() function from the standard library.
compareBy() takes a lambda function that produces a Comparable value from an instance
and de nes the custom order as the natural order of the produced values. With compareBy() ,
the length comparator from the example above looks like this:
tln(listOf("aaa", "bb", "c").sortedWith(compareBy { it.length }))
The Kotlin collections package provides functions for sorting collections in natural, custom, and
even random orders. On this page, we'll describe sorting functions that apply to read-only
collections. These functions return their result as a new collection containing the elements of the
original collection in the requested order. To learn about functions for sorting mutable
collections in place, see the List Speci c Operations.
Natural order
The basic functions sorted() and sortedDescending() return elements of a collection sorted
into ascending and descending sequence according to their natural order. These functions apply
to collections of Comparable elements.
println("Sorted ascending: ${numbers.sorted()}")

println("Sorted descending: ${numbers.sortedDescending()}")
Custom orders
For sorting in custom orders or sorting non-comparable objects, there are the functions
sortedBy() and sortedByDescending(). They take a selector function that maps collection
elements to Comparable values and sort the collection in natural order of that values.
202
val sortedNumbers = numbers.sortedBy { it.length }

println("Sorted by length ascending: $sortedNumbers")
val sortedByLast = numbers.sortedByDescending { it.last() }
println("Sorted by the last letter descending: $sortedByLast")
To de ne a custom order for the collection sorting, you can provide your own Comparator . To
do this, call the sortedWith() function passing in your Comparator . With this function,
sorting strings by their length looks like this:

println("Sorted by length ascending: ${numbers.sortedWith(compareBy { it.length })}")
Reverse order
You can retrieve the collection in the reversed order using the reversed() function.

println(numbers.reversed())
reversed() returns a new collection with the copies of the elements. So, if you change the
original collection later, this won't a ect the previously obtained results of reversed() .
Another reversing function - asReversed() - returns a reversed view of the same collection
instance, so it may be more lightweight and preferable than reversed() if the original list is
not going to change.

val reversedNumbers = numbers.asReversed()
println(reversedNumbers)
If the original list is mutable, all its changes re ect in its reversed views and vice versa.

val reversedNumbers = numbers.asReversed()
numbers.add("five")
However, if the mutability of the list is unknown or the source is not a list at all, reversed() is
more preferable since its result is a copy that won't change in the future.
Random order
Finally, there is a function that returns a new List containing the collection elements in a
random order - shuffled(). You can call it without arguments or with a Random object.

println(numbers.shuffled())
203
Collection Aggregate Operations
Kotlin collections contain functions for commonly used aggregate operations – operations that
return a single value based on the collection content. Most of them are well known and work the
same way as they do in other languages:
— min() and max() return the smallest and the largest element respectively;
— average() returns the average value of elements in the collection of numbers;
— sum() returns the sum of elements in the collection of numbers;
— count() returns the number of elements in a collection;
println("Count: ${numbers.count()}")
println("Max: ${numbers.max()}")
println("Min: ${numbers.min()}")
println("Average: ${numbers.average()}")
println("Sum: ${numbers.sum()}")
There are also functions for retrieving the smallest and the largest elements by certain selector
function or custom Comparator:
— maxBy()/minBy() take a selector function and return the element for which it returns the
largest or the smallest value.
— maxWith()/minWith() take a Comparator object and return the largest or smallest
element according to that Comparator .

val min3Remainder = numbers.minBy { it % 3 }
println(min3Remainder)
val strings = listOf("one", "two", "three", "four")

val longestString = strings.maxWith(compareBy { it.length })
println(longestString)
Additionally, there are advanced summation functions that take a function and return the sum of
its return values on all elements:
— sumBy() applies functions that return Int values on collection elements.
— sumByDouble() works with functions that return Double .

println(numbers.sumBy { it * 2 })
println(numbers.sumByDouble { it.toDouble() / 2 })
Fold and reduce
204
For more speci c cases, there are the functions reduce() and fold() that apply the provided
operation to the collection elements sequentially and return the accumulated result. The
operation takes two arguments: the previously accumulated value and the collection element.
The di erence between the two functions is that fold() takes an initial value and uses it as the
accumulated value on the rst step, whereas the rst step of reduce() uses the rst and the
second elements as operation arguments on the rst step.
val sum = numbers.reduce { sum, element -> sum + element }

println(sum)
val sumDoubled = numbers.fold(0) { sum, element -> sum + element * 2 }
println(sumDoubled)
//val sumDoubledReduce = numbers.reduce { sum, element -> sum + element * 2 }

//incorrect: the first element isn't doubled in the result
//println(sumDoubledReduce)
The example above shows the di erence: fold() is used for calculating the sum of doubled
elements. If you pass the same function to reduce() , it will return another result because it
uses the list's rst and second elements as arguments on the rst step, so the rst element won't
be doubled.
To apply a function to elements in the reverse order, use functions reduceRight() and
foldRight(). They work in a way similar to fold() and reduce() but start from the last
element and then continue to previous. Note that when folding or reducing right, the operation
arguments change their order: rst goes the element, and then the accumulated value.

val sumDoubledRight = numbers.foldRight(0) { element, sum -> sum + element * 2 }
println(sumDoubledRight)
You can also apply operations that take element indices as parameters. For this purpose, use
functions reduceIndexed() and foldIndexed() passing element index as the rst argument
of the operation.
Finally, there are functions that apply such operations to collection elements from right to left -
reduceRightIndexed() and foldRightIndexed().

val sumEven = numbers.foldIndexed(0) { idx, sum, element -> if (idx % 2 == 0) sum +
element else sum }
println(sumEven)
val sumEvenRight = numbers.foldRightIndexed(0) { idx, element, sum -> if (idx % 2 == 0)

sum + element else sum }
println(sumEvenRight)
205
Collection Write Operations
Mutable collections support operations for changing the collection contents, for example, adding
or removing elements. On this page, we'll describe write operations available for all
implementations of MutableCollection . For more speci c operations available for List and
Map , see List Speci c Operations and Map Speci c Operations respectively.
Adding elements
To add a single element to a list or a set, use the add() function. The speci ed object is
appended to the end of the collection.

numbers.add(5)
println(numbers)
addAll() adds every element of the argument object to a list or a set. The argument can be an
Iterable , a Sequence , or an Array . The types of the receiver and the argument may be
di erent, for example, you can add all items from a Set to a List .
When called on lists, addAll() adds new elements in the same order as they go in the
argument. You can also call addAll() specifying an element position as the rst argument. The
rst element of the argument collection will be inserted at this position. Other elements of the
argument collection will follow it, shifting the receiver elements to the end.

numbers.addAll(arrayOf(7, 8))
println(numbers)
numbers.addAll(2, setOf(3, 4))
println(numbers)
You can also add elements using the in-place version of the plus operator - plusAssign ( += )
When applied to a mutable collection, += appends the second operand (an element or another
collection) to the end of the collection.
val numbers = mutableListOf("one", "two")

numbers += "three"
println(numbers)
numbers += listOf("four", "five")
println(numbers)
Removing elements
To remove an element from a mutable collection, use the remove() function. remove()
accepts the element value and removes one occurrence of this value.
206
val numbers = mutableListOf(1, 2, 3, 4, 3)
numbers.remove(3) // removes the first `3`
println(numbers)
numbers.remove(5) // removes nothing
println(numbers)
For removing multiple elements at once, there are the following functions :
— removeAll() removes all elements that are present in the argument collection. Alternatively,
you can call it with a predicate as an argument; in this case the function removes all elements
for which the predicate yields true .
— retainAll() is the opposite of removeAll() : it removes all elements except the ones
from the argument collection. When used with a predicate, it leaves only elements that match
it.
— clear() removes all elements from a list and leaves it empty.

println(numbers)
numbers.retainAll { it >= 3 }
println(numbers)
numbers.clear()
println(numbers)
val numbersSet = mutableSetOf("one", "two", "three", "four")

numbersSet.removeAll(setOf("one", "two"))
println(numbersSet)
Another way to remove elements from a collection is with the minusAssign ( -= ) operator – the
in-place version of minus. The second argument can be a single instance of the element type or
another collection. With a single element on the right-hand side, -= removes the rst occurrence
of it. In turn, if it's a collection, all occurrences of its elements are removed. For example, if a list
contains duplicate elements, they are removed at once. The second operand can contain
elements that are not present in the collection. Such elements don't a ect the operation
execution.
val numbers = mutableListOf("one", "two", "three", "three", "four")

numbers -= "three"
println(numbers)
numbers -= listOf("four", "five")
//numbers -= listOf("four") // does the same as above
println(numbers)
Updating elements
Lists and sets also provide operations for updating elements. They are described in List Speci c
Operations and Map Speci c Operations. For sets, updating doesn't make sense since it's actually
removing an element and adding another one.
207
List Speci c Operations
List is the most popular type of built-in collection in Kotlin. Index access to the elements of lists
provides a powerful set of operations for lists.
Retrieving elements by index

Lists support all common operations for element retrieval: elementAt() , first() , last() ,
and others listed in Retrieving Single Elements. What is speci c for lists is index access to the
elements, so the simplest way to read an element is retrieving it by index. That is done with the
get() function with the index passed in the argument or the shorthand [index] syntax.
If the list size is less than the speci ed index, an exception is thrown. There are two other
functions that help you avoid such exceptions:
— getOrElse() lets you provide the function for calculating the default value to return if the
index isn't present in the collection.
— getOrNull() returns null as the default value.

println(numbers.get(0))
println(numbers[0])
//numbers.get(5) // exception!
println(numbers.getOrNull(5)) // null
println(numbers.getOrElse(5, {it})) // 5
Retrieving list parts

In addition to common operations for Retrieving Collection Parts, lists provide the subList()
function that returns a view of the speci ed elements range as a list. Thus, if an element of the
original collection changes, it also changes in the previously created sublists and vice versa.

println(numbers.subList(3, 6))
Finding element positions
Linear search
In any lists, you can nd the position of an element using the functions indexOf() and
lastIndexOf(). They return the rst and the last position of an element equal to the given
argument in the list. If there are no such elements, both functions return -1 .
val numbers = listOf(1, 2, 3, 4, 2, 5)

println(numbers.indexOf(2))
println(numbers.lastIndexOf(2))
There is also a pair of functions that take a predicate and search for elements matching it:
208
indexOfFirst() returns the index of the rst element matching the predicate or -1 if there
— are no such elements.
— indexOfLast() returns the index of the last element matching the predicate or -1 if there
are no such elements.

println(numbers.indexOfFirst { it > 2})
println(numbers.indexOfLast { it % 2 == 1})
Binary search in sorted lists
There is one more way to search elements in lists – binary search. It works signi cantly faster
than other built-in search functions but requires the list to be sorted in ascending order according
to a certain ordering: natural or another one provided in the function parameter. Otherwise, the
result is unde ned.
To search an element in a sorted list, call the binarySearch() function passing the value as an
argument. If such an element exists, the function returns its index; otherwise, it returns (-
insertionPoint - 1) where insertionPoint is the index where this element should be
inserted so that the list remains sorted. If there is more than one element with the given value,
the search can return any of their indices.
You can also specify an index range to search in: in this case, the function searches only between
two provided indices.

numbers.sort()
println(numbers)
println(numbers.binarySearch("two")) // 3
println(numbers.binarySearch("z")) // -5
println(numbers.binarySearch("two", 0, 2)) // -3
Comparator binary search
When list elements aren't Comparable , you should provide a Comparator to use in the binary
search. The list must be sorted in ascending order according to this Comparator . Let's have a
look at an example:
val productList = listOf(

Product("WebStorm", 49.0),
Product("AppCode", 99.0),
Product("DotTrace", 129.0),
Product("ReSharper", 149.0))
println(productList.binarySearch(Product("AppCode", 99.0), compareBy<Product> { it.price

}.thenBy { it.name }))
209
Here's a list of Product instances that aren't Comparable and a Comparator that de nes the
order: product p1 precedes product p2 if p1 's price is less than p2 's price. So, having a list
sorted ascending according to this order, we use binarySearch() to nd the index of the
speci ed Product .
Custom comparators are also handy when a list uses an order di erent from natural one, for
example, a case-insensitive order for String elements.
val colors = listOf("Blue", "green", "ORANGE", "Red", "yellow")

println(colors.binarySearch("RED", String.CASE_INSENSITIVE_ORDER)) // 3
Comparison binary search
Binary search with comparison function lets you nd elements without providing explicit search
values. Instead, it takes a comparison function mapping elements to Int values and searches for
the element where the function returns zero. The list must be sorted in the ascending order
according to the provided function; in other words, the return values of comparison must grow
from one list element to the next one.
data class Product(val name: String, val price: Double)
fun priceComparison(product: Product, price: Double) = sign(product.price -

price).toInt()
fun main() {
val productList = listOf(
Product("WebStorm", 49.0),
Product("AppCode", 99.0),
Product("DotTrace", 129.0),
Product("ReSharper", 149.0))
println(productList.binarySearch { priceComparison(it, 99.0) })

}
Both comparator and comparison binary search can be performed for list ranges as well.
List write operations

In addition to the collection modi cation operations described in Collection Write Operations,
mutable lists support speci c write operations. Such operations use the index to access elements
to broaden the list modi cation capabilities.
Adding
To add elements to a speci c position in a list, use add() and addAll() providing the position
for element insertion as an additional argument. All elements that come after the position shift
to the right.
210
val numbers = mutableListOf("one", "five", "six")
numbers.add(1, "two")
numbers.addAll(2, listOf("three", "four"))
println(numbers)
Updating
Lists also o er a function to replace an element at a given position - set() and its operator form
[] . set() doesn't change the indexes of other elements.
val numbers = mutableListOf("one", "five", "three")

numbers[1] = "two"
println(numbers)
fill() simply replaces all the collection elements with the speci ed value.

numbers.fill(3)
println(numbers)
Removing
To remove an element at a speci c position from a list, use the removeAt() function providing
the position as an argument. All indices of elements that come after the element being removed
will decrease by one.
val numbers = mutableListOf(1, 2, 3, 4, 3)

numbers.removeAt(1)
println(numbers)
Sorting
In Collection Ordering, we describe operations that retrieve collection elements in speci c

orders. For mutable lists, the standard library o ers similar extension functions that perform the
same ordering operations in place. When you apply such an operation to a list instance, it
changes the order of elements in that exact instance.
The in-place sorting functions have similar names to the functions that apply to read-only lists,
but without the ed/d su x:
— sort* instead of sorted* in the names of all sorting functions: sort(),

sortDescending(), sortBy(), and so on.
— shuffle() instead of shuffled() .
— reverse() instead of reversed() .
211
asReversed() called on a mutable list returns another mutable list which is a reversed view of
the original list. Changes in that view are re ected in the original list. The following example
shows sorting functions for mutable lists:
numbers.sort()
println("Sort into ascending: $numbers")
numbers.sortDescending()
println("Sort into descending: $numbers")
numbers.sortBy { it.length }
println("Sort into ascending by length: $numbers")
numbers.sortByDescending { it.last() }
println("Sort into descending by the last letter: $numbers")
numbers.sortWith(compareBy<String> { it.length }.thenBy { it })

println("Sort by Comparator: $numbers")
numbers.shuffle()
println("Shuffle: $numbers")
numbers.reverse()
println("Reverse: $numbers")
212
Set Speci c Operations
The Kotlin collections package contains extension functions for popular operations on sets:
nding intersections, merging, or subtracting collections from each other.
To merge two collections into one, use the union() function. It can be used in the in x form a
union b . Note that for ordered collections the order of the operands is important: in the
resulting collection, the elements of the rst operand go before the elements of the second.
To nd an intersection between two collections (elements present in both of them), use

intersect(). To nd collection elements not present in another collection, use subtract().
Both these functions can be called in the in x form as well, for example, a intersect b .
val numbers = setOf("one", "two", "three")
println(numbers union setOf("four", "five"))

println(setOf("four", "five") union numbers)
println(numbers intersect setOf("two", "one"))

println(numbers subtract setOf("three", "four"))
println(numbers subtract setOf("four", "three")) // same output
Note that set operations are supported by List as well. However, the result of set operations
on lists is still a Set , so all the duplicate elements are removed.
213
Map Speci c Operations
In maps, types of both keys and values are user-de ned. Key-based access to map entries
enables various map-speci c processing capabilities from getting a value by key to separate
ltering of keys and values. On this page, we provide descriptions of the map processing
functions from the standard library.
Retrieving keys and values

For retrieving a value from a map, you must provide its key as an argument of the get()
function. The shorthand [key] syntax is also supported. If the given key is not found, it returns
null . There is also the function getValue() which has slightly di erent behavior: it throws an
exception if the key is not found in the map. Additionally, you have two more options to handle
the key absence:
— getOrElse() works the same way as for lists: the values for non-existent keys are returned
from the given lambda function.
— getOrDefault() returns the speci ed default value if the key is not found.
val numbersMap = mapOf("one" to 1, "two" to 2, "three" to 3)

println(numbersMap.get("one"))
println(numbersMap["one"])
println(numbersMap.getOrDefault("four", 10))
println(numbersMap["five"]) // null
//numbersMap.getValue("six") // exception!
To perform operations on all keys or all values of a map, you can retrieve them from the
properties keys and values accordingly. keys is a set of all map keys and values is a
collection of all map values.

println(numbersMap.keys)
println(numbersMap.values)
Filtering
You can lter maps with the filter() function as well as other collections. When calling
filter() on a map, pass to it a predicate with a Pair as an argument. This enables you to use
both the key and the value in the ltering predicate.

val filteredMap = numbersMap.filter { (key, value) -> key.endsWith("1") && value > 10}
println(filteredMap)
There are also two speci c ways for ltering maps: by keys and by values. For each way, there is a
function: filterKeys() and filterValues(). Both return a new map of entries which match
the given predicate. The predicate for filterKeys() checks only the element keys, the one for
filterValues() checks only values.
214
val filteredKeysMap = numbersMap.filterKeys { it.endsWith("1") }
val filteredValuesMap = numbersMap.filterValues { it < 10 }
println(filteredKeysMap)
println(filteredValuesMap)
plus and minus operators

Due to the key access to elements, plus ( + ) and minus ( - ) operators work for maps di erently
than for other collections. plus returns a Map that contains elements of its both operands: a
Map on the left and a Pair or another Map on the right. When the right-hand side operand
contains entries with keys present in the left-hand side Map , the result map contains the entries
from the right side.

println(numbersMap + Pair("four", 4))
println(numbersMap + Pair("one", 10))
println(numbersMap + mapOf("five" to 5, "one" to 11))
minus creates a Map from entries of a Map on the left except those with keys from the right-
hand side operand. So, the right-hand side operand can be either a single key or a collection of
keys: list, set, and so on.

println(numbersMap - "one")
println(numbersMap - listOf("two", "four"))
For details on using plusAssign ( += ) and minusAssign ( -= ) operators on mutable maps, see
Map write operations below.
Map write operations

Mutable maps o er map-speci c write operations. These operations let you change the map
content using the key-based access to the values.
There are certain rules that de ne write operations on maps:
— Values can be updated. In turn, keys never change: once you add an entry, its key is constant.
— For each key, there is always a single value associated with it. You can add and remove whole
entries.
Below are descriptions of the standard library functions for write operations available on
mutable maps.
Adding and updating entries
215
To add a new key-value pair to a mutable map, use put(). When a new entry is put into a
LinkedHashMap (the default map implementation), it is added so that it comes last when
iterating the map. In sorted maps, the positions of new elements are de ned by the order of
their keys.

numbersMap.put("three", 3)
println(numbersMap)
To add multiple entries at a time, use putAll(). Its argument can be a Map or a group of
Pair s: Iterable , Sequence , or Array .
val numbersMap = mutableMapOf("one" to 1, "two" to 2, "three" to 3)

numbersMap.putAll(setOf("four" to 4, "five" to 5))
println(numbersMap)
Both put() and putAll() overwrite the values if the given keys already exist in the map.
Thus, you can use them to update values of map entries.

val previousValue = numbersMap.put("one", 11)
println("value associated with 'one', before: $previousValue, after:
${numbersMap["one"]}")
println(numbersMap)
You can also add new entries to maps using the shorthand operator form. There are two ways:
— plusAssign ( += ) operator.
— the [] operator alias for put() .

numbersMap["three"] = 3 // calls numbersMap.put("three", 3)
numbersMap += mapOf("four" to 4, "five" to 5)
println(numbersMap)
When called with the key present in the map, operators overwrite the values of the
corresponding entries.
Removing entries
To remove an entry from a mutable map, use the remove() function. When calling remove() ,
you can pass either a key or a whole key-value-pair. If you specify both the key and value, the
element with this key will be removed only if its value matches the second argument.

numbersMap.remove("one")
println(numbersMap)
numbersMap.remove("three", 4) //doesn't remove anything
println(numbersMap)
216
You can also remove entries from a mutable map by their keys or values. To do this, call
remove() on the map's keys or values providing the key or the value of an entry. When called
on values, remove() removes only the rst entry with the given value.
val numbersMap = mutableMapOf("one" to 1, "two" to 2, "three" to 3, "threeAgain" to 3)

numbersMap.keys.remove("one")
println(numbersMap)
numbersMap.values.remove(3)
println(numbersMap)
The minusAssign ( -= ) operator is also available for mutable maps.

numbersMap -= "two"
println(numbersMap)
numbersMap -= "five" //doesn't remove anything
println(numbersMap)
217
Multiplatform Programming
Platform-Speci c Declarations
Kotlin versions.
One of the key capabilities of Kotlin's multiplatform code is a way for common code to depend
on platform-speci c declarations. In other languages, this can often be accomplished by building
a set of interfaces in the common code and implementing these interfaces in platform-speci c
modules. However, this approach is not ideal in cases when you have a library on one of the
platforms that implements the functionality you need, and you'd like to use the API of this library
directly without extra wrappers. Also, it requires common declarations to be expressed as
interfaces, which doesn't cover all possible cases.
As an alternative, Kotlin provides a mechanism of expected and actual declarations. With this
mechanism, a common module can de ne expected declarations, and a platform module can
provide actual declarations corresponding to the expected ones. To see how this works, let's look
at an example rst. This code is part of a common module:
package org.jetbrains.foo
expect class Foo(bar: String) {

fun frob()
}
fun main() {
Foo("Hello").frob()
}
And this is the corresponding JVM module:
package org.jetbrains.foo
actual class Foo actual constructor(val bar: String) {

actual fun frob() {
println("Frobbing the $bar")
}
}
This illustrates several important points:
218
— An expected declaration in the common module and its actual counterparts always have
exactly the same fully quali ed name.
— An expected declaration is marked with the expect keyword; the actual declaration is
marked with the actual keyword.
— All actual declarations that match any part of an expected declaration need to be marked as
actual .
— Expected declarations never contain any implementation code.
Note that expected declarations are not restricted to interfaces and interface members. In this
example, the expected class has a constructor and can be created directly from common code.
You can apply the expect modi er to other declarations as well, including top-level declarations
and annotations:
// Common
expect fun formatString(source: String, vararg args: Any): String
expect annotation class Test
// JVM
actual fun formatString(source: String, vararg args: Any) =
String.format(source, *args)
actual typealias Test = org.junit.Test
The compiler ensures that every expected declaration has actual declarations in all platform
modules that implement the corresponding common module, and reports an error if any actual
declarations are missing. The IDE provides tools that help you create the missing actual
declarations.
If you have a platform-speci c library that you want to use in common code while providing your
own implementation for another platform, you can provide a typealias to an existing class as the
actual declaration:
expect class AtomicRef<V>(value: V) {

fun get(): V
fun set(value: V)
fun getAndSet(value: V): V
fun compareAndSet(expect: V, update: V): Boolean
}
actual typealias AtomicRef<V> = java.util.concurrent.atomic.AtomicReference<V>
219
Building Multiplatform Projects with Gradle
Kotlin versions.
This document explains the structure of Kotlin multiplatform projects and describes how those
are con gured and built using Gradle.
Table of Contents
— Project Structure
— Setting up a Multiplatform Project
— Gradle Plugin
— Setting up Targets
— Supported platforms
— Con guring compilations
— Con guring Source Sets

— Connecting source sets
— Adding dependencies
— Language settings
— Default Project Layout

— Running Tests
— Publishing a Multiplatform Library
— Android Support
— Publishing Android libraries
— Using Kotlin/Native Targets

— Building nal native binaries
Project Structure
The layout of a Kotlin multiplatform project is constructed out of the following building blocks:
— A target is a part of the build that is responsible for building, testing, and packaging a
complete piece of software for one of the platforms. Therefore, a multiplatform project
usually contains multiple targets.
— Building each target involves compiling Kotlin sources once or multiple times. In other words,
a target may have one or more compilations. For example, one compilation for production
sources, the other for tests.
220
— The Kotlin sources are arranged into source sets. In addition to Kotlin source les and
resources, each source set may have its own dependencies. Source sets form a hierarchy that
is built with the "depends on" relation. A source set by itself is platform agnostic, but it may
contain platform-speci c code and dependencies if it's only compiled for a single platform.
Each compilation has a default source set, which is the place for sources and dependencies that
are speci c to that compilation. The default source set is also used for directing other source sets
to the compilation by the means of the "depends on" relation.
Here's an illustration of what a project targeting the JVM and JS looks like:
Here, the two targets, jvm and js , each compile the production and test sources, and some of
the sources are shared. This layout is achieved by just creating the two targets, with no additional
con guration for the compilations and source sets: those are created by default for these targets.
In the example above, the production sources for the JVM target are compiled by its main
compilation and therefore include the sources and dependencies from the source sets jvmMain
and commonMain (due to the depends on relation):
Here, the jvmMain source set provides plaform-speci c implementations for the expected API
in the shared commonMain sources. This is how the code is shared between the platforms in a
exible way with platform-speci c implementations where needed.
In further sections, these concepts are described in more detail along with the DSL to con gure
them in a project.
Setting up a Multiplatform Project

You can create a new multiplatform project in the IDE by selecting one of the multiplatform
project templates in the New Project dialog under the "Kotlin" section.
For example, if you choose "Kotlin (Multiplatform Library)", a library project is created that has
three targets, one for the JVM, one for JS, and one for the Native platform that you are using.
These are con gured in the build.gradle script in the following way:
221
plugins {
id 'org.jetbrains.kotlin.multiplatform' version '1.3.31'
}
repositories {
mavenCentral()
}
kotlin {
jvm() // Creates a JVM target with the default name 'jvm'
js() // JS target named 'js'
mingwX64("mingw") // Windows (MinGW X64) target named 'mingw'
sourceSets { /* ... */ }
}
plugins {
kotlin("multiplatform") version "1.3.31"
}
repositories {
mavenCentral()
}
kotlin {
jvm() // Creates a JVM target with the default name 'jvm'
js() // JS target named 'js'
mingwX64("mingw") // Windows (MinGW X64) target named 'mingw'
sourceSets { /* ... */ }
}
The three targets are created with the preset functions jvm() , js() , and mingwX64() that
provide some default con guration. There are presets for each of the supported platforms.
The source sets and their dependencies are then con gured as follows:
plugins { /* ... */ }
kotlin {
/* Targets declarations omitted */
sourceSets {
commonMain {
dependencies {
implementation kotlin('stdlib-common')
}
}
commonTest {
dependencies {
implementation kotlin('test-common')
implementation kotlin('test-annotations-common')
}
}
// Default source set for JVM-specific sources and dependencies.

// Alternatively, jvmMain { ... } would work as well:
jvm().compilations.main.defaultSourceSet {
dependencies {
implementation kotlin('stdlib-jdk8')
222
implementation kotlin('stdlib-jdk8')
}
}
// JVM-specific tests and their dependencies:
jvm().compilations.test.defaultSourceSet {
dependencies {
implementation kotlin('test-junit')
}
}
js().compilations.main.defaultSourceSet { /* ... */ }
js().compilations.test.defaultSourceSet { /* ... */ }
mingwX64('mingw').compilations.main.defaultSourceSet { /* ... */ }
mingwX64('mingw').compilations.test.defaultSourceSet { /* ... */ }
}
}
plugins { /* ... */ }
kotlin {
/* Targets declarations omitted */
sourceSets {
val commonMain by getting {
dependencies {
implementation(kotlin("stdlib-common"))
}
}
val commonTest by getting {
dependencies {
implementation(kotlin("test-common"))
implementation(kotlin("test-annotations-common"))
}
}
// Default source set for JVM-specific sources and dependencies:

jvm().compilations["main"].defaultSourceSet {
dependencies {
implementation(kotlin("stdlib-jdk8"))
}
}
// JVM-specific tests and their dependencies:
jvm().compilations["test"].defaultSourceSet {
dependencies {
implementation(kotlin("test-junit"))
}
}
js().compilations["main"].defaultSourceSet { /* ... */ }
js().compilations["test"].defaultSourceSet { /* ... */ }
mingwX64("mingw").compilations["main"].defaultSourceSet { /* ... */ }
mingwX64("mingw").compilations["test"].defaultSourceSet { /* ... */ }
}
}
223
These are the default source set names for the production and test sources for the targets
con gured above. The source sets commonMain and commonTest are included into production
and test compilations, respectively, of all targets. Note that the dependencies for common source
sets commonMain and commonTest are the common artifacts, and the platform libraries go to
the source sets of the speci c targets.
Gradle Plugin
Kotlin Multiplatform projects require Gradle version 4.7 and above, older Gradle versions are not
supported.
To setup a multiplatform project from scratch in a Gradle project, rst apply the kotlin-
multiplatform plugin to the project by adding the following to the beginning of the
build.gradle le:
plugins {
id 'org.jetbrains.kotlin.multiplatform' version '1.3.31'
}
plugins {
kotlin("multiplatform") version "1.3.31"
}
This creates the kotlin extension at the top level. You can then access it in the build script for:
— setting up the targets for multiple platforms (no targets are created by default);
— con guring the source sets and their dependencies;
Setting up Targets
A target is a part of the build responsible for compiling, testing, and packaging a piece of software
aimed for one of the supported platforms.
All of the targets may share some of the sources and may have platform-speci c sources as well.
As the platforms are di erent, targets are built in di erent ways as well and have various
platform-speci c settings. The Gradle plugin bundles a number of presets for the supported
platforms.
To create a target, use one of the preset functions, which are named according to the target
platforms and optionally accept the target name and a con guring code block:
kotlin {
jvm() // Create a JVM target with the default name 'jvm'
js("nodeJs") // Create a JS target with a custom name 'nodeJs'
linuxX64("linux") {
/* Specify additional settings for the 'linux' target here */
}
}
224
The preset functions return an existing target if there is one. This can be used to con gure an
existing target:
kotlin {
/* ... */
// Configure the attributes of the 'jvm6' target:

jvm("jvm6").attributes { /* ... */ }
}
Note that both the target platform and the name matter: if a target was created as
jvm('jvm6') , using jvm() will create a separate target (with the default name jvm ). If the
preset function used to create the target under that name was di erent, an error is reported.
The targets created from presets are added to the kotlin.targets domain object collection,
which can be used to access them by their names or con gure all targets:
kotlin {
jvm()
js("nodeJs")
println(targets.names) // Prints: [jvm, metadata, nodeJs]
// Configure all targets, including those which will be added later:

targets.all {
compilations["main"].defaultSourceSet { /* ... */ }
}
}
To create or access several targets from multiple presets dynamically, you can use the
targetFromPreset function which accepts a preset (those are contained in the
kotlin.presets domain object collection) and, optionally, a target name and a con guration
code block.
For example, to create a target for each of the Kotlin/Native supported platforms (see below), use
this code:
kotlin {
presets.withType(org.jetbrains.kotlin.gradle.plugin.mpp.KotlinNativeTargetPreset).each {
targetFromPreset(it) {
/* Configure each of the created targets */
}
}
}
225
import org.jetbrains.kotlin.gradle.plugin.mpp.KotlinNativeTargetPreset
/* ... */
kotlin {
presets.withType<KotlinNativeTargetPreset>().forEach {
targetFromPreset(it) {
/* Configure each of the created targets */
}
}
}
Supported platforms
There are target presets that one can apply using the preset functions, as shown above, for the
following target platforms:
— jvm for Kotlin/JVM. Note: jvm targets do not compile Java;
— js for Kotlin/JS;
— android for Android applications and libraries. Note that one of the Android Gradle plugins
should be applied before the target is created;
— Kotlin/Native target presets (see the notes below):
— androidNativeArm32 and androidNativeArm64 for Android NDK;
— iosArm32 , iosArm64 , iosX64 for iOS;
— linuxArm32Hfp , linuxMips32 , linuxMipsel32 , linuxX64 for Linux;
— macosX64 for MacOS;
— mingwX64 for Windows;
— wasm32 for WebAssembly.
Note that some of the Kotlin/Native targets require an appropriate host machine to build on.
Some targets may require additional con guration. For Android and iOS examples, see the
Multiplatform Project: iOS and Android tutorial.
Con guring compilations
Building a target requires compiling Kotlin once or multiple times. Each Kotlin compilation of a
target may serve a di erent purpose (e.g. production code, tests) and incorporate di erent
source sets. The compilations of a target may be accessed in the DSL, for example, to get the
tasks, con gure the Kotlin compiler options or get the dependency les and compilation outputs:
kotlin {
jvm {
compilations.main.kotlinOptions {
// Setup the Kotlin compiler options for the 'main' compilation:
jvmTarget = "1.8"
226
}
compilations.main.compileKotlinTask // get the Kotlin task 'compileKotlinJvm'

compilations.main.output // get the main compilation output
compilations.test.runtimeDependencyFiles // get the test runtime classpath
}
// Configure all compilations of all targets:

targets.all {
compilations.all {
kotlinOptions {
allWarningsAsErrors = true
}
}
}
}
kotlin {
jvm {
val main by compilations.getting {
kotlinOptions {
// Setup the Kotlin compiler options for the 'main' compilation:
jvmTarget = "1.8"
}
compileKotlinTask // get the Kotlin task 'compileKotlinJvm'

output // get the main compilation output
}
compilations["test"].runtimeDependencyFiles // get the test runtime classpath

}
// Configure all compilations of all targets:

targets.all {
compilations.all {
kotlinOptions {
allWarningsAsErrors = true
}
}
}
}
Each compilation is accompanied by a default source set, which is created automatically and
should be used for sources and dependencies that are speci c to that compilation. The default
source set for a compilation foo of a target bar has the name barFoo . It can also be accessed
from a compilation using defaultSourceSet :
227
kotlin {
sourceSets {
// The default source set for the 'main` compilation of the 'jvm' target:
jvmMain {
/* ... */
}
}
// Alternatively, access it from the target's compilation:

jvm().compilations.main.defaultSourceSet {
/* ... */
}
}
kotlin {
sourceSets {
// The default source set for the 'main` compilation of the 'jvm' target:
val jvmMain by getting {
/* ... */
}
}
// Alternatively, access it from the target's compilation:

jvm().compilations["main"].defaultSourceSet {
/* ... */
}
}
To collect all source sets participating in a compilation, including those added via the depends-on
relation, one can use the property allKotlinSourceSets .
For some speci c use cases, creating a custom compilation may be required. This can be done
within the target's compilations domain object collection. Note that the dependencies need
to be set up manually for all custom compilations, and the usage of a custom compilation's
outputs is up to the build author. For example, consider a custom compilation for integration
tests of a jvm() target:
228
kotlin {
jvm() {
compilations.create('integrationTest') {
defaultSourceSet {
dependencies {
def main = compilations.main
// Compile against the main compilation's compile classpath and
outputs:
implementation(main.compileDependencyFiles + main.output.classesDirs)
implementation kotlin('test-junit')
/* ... */
}
}
// Create a test task to run the tests produced by this compilation:

tasks.create('jvmIntegrationTest', Test) {
// Run the tests with the classpath containing the compile dependencies
(including 'main'),
// runtime dependencies, and the outputs of this compilation:
classpath = compileDependencyFiles + runtimeDependencyFiles +
output.allOutputs
// Run only the tests from this compilation's outputs:

testClassesDirs = output.classesDirs
}
}
}
}
kotlin {
jvm() {
compilations {
val main by getting
229
val integrationTest by compilations.creating {
defaultSourceSet {
dependencies {
// Compile against the main compilation's compile classpath and
outputs:
implementation(main.compileDependencyFiles +
main.output.classesDirs)
implementation(kotlin("test-junit"))
/* ... */
}
}
// Create a test task to run the tests produced by this compilation:

tasks.create<Test>("integrationTest") {
// Run the tests with the classpath containing the compile
dependencies (including 'main'),
// runtime dependencies, and the outputs of this compilation:
classpath = compileDependencyFiles + runtimeDependencyFiles +
output.allOutputs
// Run only the tests from this compilation's outputs:

testClassesDirs = output.classesDirs
}
}
}
}
}
Also note that the default source set of a custom compilation depends on neither commonMain
nor commonTest by default.
Con guring Source Sets

A Kotlin source set is a collection of Kotlin sources, along with their resources, dependencies, and
language settings, which may take part in Kotlin compilations of one or more targets.
A source set is not bound to be platform-speci c or "shared"; what it's allowed to contain
depends on its usage: a source set added to multiple compilations is limited to the common
language features and dependencies, while a source set that is only used by a single target can
have platform-speci c dependencies, and its code may use language features speci c to that
target's platform.
Some source sets are created and con gured by default: commonMain , commonTest , and the
default source sets for the compilations. See Default Project Layout.
The source sets are con gured within a sourceSets { ... } block of the kotlin { ... }
extension:
kotlin {
sourceSets {
foo { /* ... */ } // create or configure a source set by the name 'foo'
bar { /* ... */ }
}
}
230
kotlin {
sourceSets {
val foo by creating { /* ... */ } // create a new source set by the name 'foo'
val bar by getting { /* ... */ } // configure an existing source set by the name
'bar'
}
}
Note: creating a source set does not link it to any target. Some source sets are
prede ned and thus compiled by default. However, custom source sets always need to
be explicitly directed to the compilations. See: Connecting source sets.
The source set names are case-sensitive. When referring to a default source set by its name,
make sure the name pre x matches a target's name, for example, a source set iosX64Main for
a target iosX64 .
A source set by itself is platform-agnostic, but it can be considered platform-speci c if it is only

compiled for a single platform. A source set can, therefore, contain either common code shared
between the platforms or platform-speci c code.
Each source set has a default source directory for Kotlin sources: src/<source set
name>/kotlin . To add Kotlin source directories and resources to a source set, use its kotlin
and resources SourceDirectorySet s:
kotlin {
sourceSets {
commonMain {
kotlin.srcDir('src')
resources.srcDir('res')
}
}
}
kotlin {
sourceSets {
kotlin.srcDir("src")
resources.srcDir("res")
}
}
}
Connecting source sets
Kotlin source sets may be connected with the 'depends on' relation, so that if a source set foo
depends on a source set bar then:
— whenever foo is compiled for a certain target, bar takes part in that compilation as well and
is also compiled into the same target binary form, such as JVM class les or JS code;
231
— sources of foo 'see' the declarations of bar , including the internal ones, and the
dependencies of bar , even those speci ed as implementation dependencies;
— foo may contain platform-speci c implementations for the expected declarations of bar ;
— the resources of bar are always processed and copied along with the resources of foo ;
— the language settings of foo and bar should be consistent;
Circular source set dependencies are prohibited.
The source sets DSL can be used to de ne these connections between the source sets:
kotlin {
sourceSets {
commonMain { /* ... */ }
allJvm {
dependsOn commonMain
/* ... */
}
}
}
kotlin {
sourceSets {
val commonMain by getting { /* ... */ }
val allJvm by creating {
dependsOn(commonMain)
/* ... */
}
}
}
Custom source sets created in addition to the default ones should be explicitly included into the
dependencies hierarchy to be able to use declarations from other source sets and, most
importantly, to take part in compilations. Most often, they need a dependsOn(commonMain) or
dependsOn(commonTest) statement, and some of the default platform-speci c source sets
should depend on the custom ones, directly or indirectly:
232
kotlin {
mingwX64()
linuxX64()
sourceSets {
// custom source set with tests for the two targets
desktopTest {
dependsOn commonTest
/* ... */
}
// Make the 'windows' default test source set for depend on 'desktopTest'
mingwX64().compilations.test.defaultSourceSet {
dependsOn desktopTest
/* ... */
}
// And do the same for the other target:
linuxX64().compilations.test.defaultSourceSet {
dependsOn desktopTest
/* ... */
}
}
}
kotlin {
mingwX64()
linuxX64()
sourceSets {
// custom source set with tests for the two targets
val desktopTest by creating {
dependsOn(getByName("commonTest"))
/* ... */
}
// Make the 'windows' default test source set for depend on 'desktopTest'
mingwX64().compilations["test"].defaultSourceSet {
dependsOn(desktopTest)
/* ... */
}
// And do the same for the other target:
linuxX64().compilations["test"].defaultSourceSet {
dependsOn(desktopTest)
/* ... */
}
}
}
Adding Dependencies
To add a dependency to a source set, use a dependencies { ... } block of the source sets
DSL. Four kinds of dependencies are supported:
— api dependencies are used both during compilation and at runtime and are exported to
library consumers. If any types from a dependency are used in the public API of the current
module, then it should be an api dependency;
233
— implementation dependencies are used during compilation and at runtime for the current
module, but are not exposed for compilation of other modules depending on the one with
the implementation dependency. The implementation dependency kind should be used
for dependencies needed for the internal logic of a module. If a module is an endpoint
application which is not published, it may use implementation dependencies instead of
api ones.
— compileOnly dependencies are only used for compilation of the current module and are
available neither at runtime nor during compilation of other modules. These dependencies
should be used for APIs which have a third-party implementation available at runtime.
— runtimeOnly dependencies are available at runtime but are not visible during compilation
of any module.
Dependencies are speci ed per source set as follows:
kotlin {
sourceSets {
commonMain {
dependencies {
api 'com.example:foo-metadata:1.0'
}
}
jvm6Main {
dependencies {
api 'com.example:foo-jvm6:1.0'
}
}
}
}
kotlin {
sourceSets {
dependencies {
api("com.example:foo-metadata:1.0")
}
}
val jvm6Main by getting {
dependencies {
api("com.example:foo-jvm6:1.0")
}
}
}
}
Note that for the IDE to correctly analyze the dependencies of the common sources, the common
source sets need to have corresponding dependencies on the Kotlin metadata packages in
addition to the platform-speci c artifact dependencies of the platform-speci c source sets.
Usually, an artifact with a su x -common (as in kotlin-stdlib-common ) or -metadata is
required when using a published library (unless it is published with Gradle metadata, as
described below).
234
However, a project('...') dependency on another multiplatform project is resolved to an
appropriate target automatically. It is enough to specify a single project('...') dependency
in a source set's dependencies, and the compilations that include the source set will receive a
corresponding platform-speci c artifact of that project, given that it has a compatible target:
kotlin {
sourceSets {
commonMain {
dependencies {
// All of the compilations that include source set 'commonMain'
// will get this dependency resolved to a compatible target, if any:
api project(':foo-lib')
}
}
}
}
kotlin {
sourceSets {
dependencies {
// All of the compilations that include source set 'commonMain'
// will get this dependency resolved to a compatible target, if any:
api(project(":foo-lib"))
}
}
}
}
Likewise, if a multiplatform library is published in the experimental Gradle metadata publishing

mode and the project is set up to consume the metadata as well, then it is enough to specify a
dependency only once, for the common source set. Otherwise, each platform-speci c source set
should be provided with a corresponding platform module of the library, in addition to the
common module, as shown above.
An alternative way to specify the dependencies is to use the Gradle built-in DSL at the top level
with the con guration names following the pattern <sourceSetName><DependencyKind> :
dependencies {
commonMainApi 'com.example:foo-common:1.0'
jvm6MainApi 'com.example:foo-jvm6:1.0'
}
dependencies {
"commonMainApi"("com.example:foo-common:1.0")
"jvm6MainApi"("com.example:foo-jvm6:1.0")
}
Some of the Gradle built-in dependencies, like gradleApi() , localGroovy() , or

gradleTestKit() are not available in the source sets dependency DSL. You can, however, add
them within the top-level dependency block, as shown above.
235
A dependency on a Kotlin module like kotlin-stdlib or kotlin-reflect may be added
with the notation kotlin("stdlib") , which is a shorthand for
"org.jetbrains.kotlin:kotlin-stdlib" .
Language settings
The language settings for a source set can be speci ed as follows:
kotlin {
sourceSets {
commonMain {
languageSettings {
languageVersion = '1.3' // possible values: '1.0', '1.1', '1.2', '1.3'
apiVersion = '1.3' // possible values: '1.0', '1.1', '1.2', '1.3'
enableLanguageFeature('InlineClasses') // language feature name
useExperimentalAnnotation('kotlin.ExperimentalUnsignedTypes') //
annotation FQ-name
progressiveMode = true // false by default
}
}
}
}
kotlin {
sourceSets {
languageSettings.apply {
languageVersion = "1.3" // possible values: '1.0', '1.1', '1.2', '1.3'
apiVersion = "1.3" // possible values: '1.0', '1.1', '1.2', '1.3'
enableLanguageFeature("InlineClasses") // language feature name
useExperimentalAnnotation("kotlin.ExperimentalUnsignedTypes") //
annotation FQ-name
progressiveMode = true // false by default
}
}
}
}
It is possible to con gure the language settings of all source sets at once:
kotlin.sourceSets.all {
languageSettings.progressiveMode = true
}
Language settings of a source set a ect how the sources are analyzed in the IDE. Due to the
current limitations, in a Gradle build, only the language settings of the compilation's default
source set are used and are applied to all of the sources participating in the compilation.
The language settings are checked for consistency between source sets depending on each other.
Namely, if foo depends on bar :
— foo should set languageVersion that is greater than or equal to that of bar ;
— foo should enable all unstable language features that bar enables (there's no such
236
requirement for bug x features);
— foo should use all experimental annotations that bar uses;
— apiVersion , bug x language features, and progressiveMode can be set arbitrarily;
Default Project Layout

By default, each project contains two source sets, commonMain and commonTest , where one
can place all the code that should be shared between all of the target platforms. These source
sets are added to each production and test compilation, respectively.
Then, once a target is added, default compilations are created for it:
— main and test compilations for JVM, JS, and Native targets;
— a compilation per Android build variant, for Android targets;
For each compilation, there is a default source set under the name composed as <targetName>
<CompilationName> . This default source set participates in the compilation, and thus it should
be used for the platform-speci c code and dependencies, and for adding other source sets to the
compilation by the means of 'depends on'. For example, a project with targets jvm6 (JVM) and
nodeJs (JS) will have source sets: commonMain , commonTest , jvm6Main , jvm6Test ,
nodeJsMain , nodeJsTest .
Numerous use cases are covered by just the default source sets and don't require custom source
sets.
Each source set by default has its Kotlin sources under src/<sourceSetName>/kotlin
directory and the resources under src/<sourceSetName>/resources .
In Android projects, additional Kotlin source sets are created for each Android source set. If the
Android target has a name foo , the Android source set bar gets a Kotlin source set counterpart
fooBar . The Kotlin compilations, however, are able to consume Kotlin sources from all of the
directories src/bar/java , src/bar/kotlin , and src/fooBar/kotlin . Java sources are
only read from the rst of these directories.
Running Tests
Running tests in a Gradle build is currently supported by default for JVM, Android, Linux,
Windows and macOS; JS and other Kotlin/Native targets need to be manually con gured to run
the tests with an appropriate environment, an emulator or a test framework.
A test task is created under the name <targetName>Test for each target that is suitable for
testing. Run the check task to run the tests for all targets.
As the commonTest default source set is added to all test compilations, tests and test tools that
are needed on all target platforms may be placed there.
237
The kotlin.test API is availble for multiplatform tests. Add the kotlin-test-common and
kotlin-test-annotations-common dependencies to commonTest to use the assertion
functions like kotlin.test.assertTrue(...)
and @Test / @Ignore / @BeforeTest / @AfterTest annotations in the common tests.
For JVM targets, use kotlin-test-junit or kotlin-test-testng for the corresponding

asserter implementation and annotations mapping.
For Kotlin/JS targets, add kotlin-test-js as a test dependency. At this point, test tasks for
Kotlin/JS are created but do not run tests by default; they should be manually con gured to run
the tests with a JavaScript test framework.
Kotlin/Native targets do not require additional test dependencies, and the kotlin.test API
implementations are built-in.
Publishing a Multiplatform Library
The set of target platforms is de ned by a multiplatform library author, and they should
provide all of the platform-speci c implementations for the library. Adding new targets
for a multiplatform library at the consumer's side is not supported.
A library built from a multiplatform project may be published to a Maven repository with the
maven-publish Gradle plugin, which can be applied as follows:
plugins {
/* ... */
id("maven-publish")
}
Once this plugin is applied, default publications are created for each of the targets that can be
built on the current host. This requires group and version to be set in the project:
plugins { /* ... */ }
group = "com.example.my.library"
version = "0.0.1"
Android library targets, however, don't have any artifacts published by default and need an
additional step to con gure publishing, see Publishing Android libraries.
The default artifact IDs follow the pattern <projectName>-<targetNameToLowerCase> , for

example sample-lib-nodejs for a target named nodeJs in a project sample-lib .
By default, a sources JAR is added to each publication in addition to its main artifact. The sources
JAR contains the sources used by the main compilation of the target. If you also need to publish
a documentation artifact (like a Javadoc JAR), you need to con gure its build manually and add it
as an artifact to the relevant publications, as shown below.
238
Also, an additional publication under the name metadata is added by default which contains
serialized Kotlin declarations and is used by the IDE to analyze multiplatform libraries.
The Maven coordinates can be altered and additional artifact les may be added to the
publication within the targets { ... } block or using the publishing { ... } DSL:
kotlin {
jvm('jvm6') {
mavenPublication { // Setup the publication for the target 'jvm6'
// The default artifactId was 'foo-jvm6', change it:
artifactId = 'foo-jvm'
// Add a docs JAR artifact (it should be a custom task):
artifact(jvmDocsJar)
}
}
}
// Alternatively, configure the publications with the `publishing { ... }` DSL:

publishing {
publications {
jvm6 { /* Setup the publication for target 'jvm6' */ }
metadata { /* Setup the publication for Kotlin metadata */ }
}
}
kotlin {
jvm("jvm6") {
mavenPublication { // Setup the publication for the target 'jvm6'
// The default artifactId was 'foo-jvm6', change it:
artifactId = "foo-jvm"
// Add a docs JAR artifact (it should be a custom task):
artifact(jvmDocsJar)
}
}
}
// Alternatively, configure the publications with the `publishing { ... }` DSL:

publishing {
publications.withType<MavenPublication>().apply {
val jvm6 by getting { /* Setup the publication for target 'jvm6' */ }
val metadata by getting { /* Setup the publication for Kotlin metadata */ }
}
}
As assembling Kotlin/Native artifacts requires several builds to run on di erent host platforms,
publishing a multiplatform library that includes Kotlin/Native targets needs to be done with that
same set of host machines. To avoid duplicate publications of modules that can be built on more
than one of the platforms (like JVM, JS, Kotlin metadata, WebAssembly), the publishing tasks for
these modules may be con gured to run conditionally, for example:
239
kotlin {
jvm()
js()
mingwX64()
linuxX64()
// Given that `-PisLinux=true` command line argument is passed when running on Linux,
// these targets get published only from a Linux machine.
// Note that the Kotlin metadata is here, too.
// The mingwx64() target is automatically skipped as incompatible in Linux builds.
configure([targets["metadata"], jvm(), js()]) {
mavenPublication { linuxOnlyPublication ->
tasks.withType(AbstractPublishToMaven).all {
onlyIf {
publication != linuxOnlyPublication || findProperty("isLinux") ==
"true"
}
}
}
}
}
kotlin {
jvm()
js()
mingwX64()
linuxX64()
// Given that `-PisLinux=true` command line argument is passed when running on Linux,
// these targets get published only from a Linux machine.
// Note that the Kotlin metadata is here, too.
// The mingwx64() target is automatically skipped as incompatible in Linux builds.
configure(listOf(metadata(), jvm(), js())) {
mavenPublication {
val linuxOnlyPublication = this@mavenPublication
tasks.withType<AbstractPublishToMaven>().all {
onlyIf {
publication != linuxOnlyPublication || findProperty("isLinux") ==
"true"
}
240
}
}
}
}
}
Experimental metadata publishing mode
An experimental publishing and dependency consumption mode can be enabled by adding

enableFeaturePreview("GRADLE_METADATA") to the root project's settings.gradle
le.
With Gradle metadata enabled, an additional publication kotlinMultiplatform is added

which references the target publications as its variants. The default artifact ID of this publication
matches the project name.
Gradle metadata publishing is an experimental Gradle feature which is not guaranteed

to be backward-compatible. Future Gradle versions may fail to resolve a dependency to
a library published with current versions of Gradle metadata. Library authors are
recommended to use it to publish experimental versions of the library alongside with
the stable publishing mechanism until the feature is considered stable.
If a library is published with Gradle metadata enabled and a consumer enables the metadata as
well, the consumer may specify a single dependency on the library in a common source set, and a
corresponding platform-speci c variant will be chosen, if available, for each of the compilations.
Consider a sample-lib library built for the JVM and JS and published with experimental Gradle
metadata. Then it is enough for the consumers to add
enableFeaturePreview('GRADLE_METADATA') and specify a single dependency:
kotlin {
jvm('jvm6')
js('nodeJs')
sourceSets {
commonMain {
dependencies {
// Resolved to the appropriate target modules,
// for example, `sample-lib-jvm6` for JVM, `sample-lib-js` for JS:
api 'com.example:sample-lib:1.0'
}
}
}
}
241
kotlin {
jvm("jvm6")
js("nodeJs")
sourceSets {
dependencies {
// Resolved to the appropriate target modules,
// for example, `sample-lib-jvm6` for JVM, `sample-lib-js` for JS:
api("com.example:sample-lib:1.0")
}
}
}
}
Disambiguating targets
It is possible to have more than one target for a single platform in a multiplatform library. For
example, these targets may provide the same API and di er in the libraries they cooperate with
at runtime, like testing frameworks or logging solutions.
However, dependencies on such a multiplatform library may be ambiguous and may thus fail to
resolve because there is not enough information to decide which of the targets to choose.
The solution is to mark the targets with a custom attribute, which is taken into account by Gradle
during dependency resolution. This, however, must be done on both the library author and the
consumer sides, and it's the library author's responsibility to communicate the attribute and its
possible values to the consumers.
Adding attributes is done symmetrically, to both the library and the consumer projects. For
example, consider a testing library that supports both JUnit and TestNG in the two targets. The
library author needs to add an attribute to both targets as follows:
def testFrameworkAttribute = Attribute.of('com.example.testFramework', String)
kotlin {
jvm('junit') {
attributes.attribute(testFrameworkAttribute, 'junit')
}
jvm('testng') {
attributes.attribute(testFrameworkAttribute, 'testng')
}
}
242
val testFrameworkAttribute = Attribute.of("com.example.testFramework",
String::class.java)
kotlin {
jvm("junit") {
attributes.attribute(testFrameworkAttribute, "junit")
}
jvm("testng") {
attributes.attribute(testFrameworkAttribute, "testng")
}
}
The consumer may only need to add the attribute to a single target where the ambiguity arises.
If the same kind of ambiguity arises when a dependency is added to a custom con guration
rather than one of the con gurations created by the plugin, you can add the attributes to the
con guration in the same way:
def testFrameworkAttribute = Attribute.of('com.example.testFramework', String)
configurations {
myConfiguration {
attributes.attribute(testFrameworkAttribute, 'junit')
}
}
val testFrameworkAttribute = Attribute.of("com.example.testFramework",

String::class.java)
configurations {
val myConfiguration by creating {
attributes.attribute(testFrameworkAttribute, "junit")
}
}
Alternatively, if a dependency library is published with experimental Gradle metadata, one can
still replace the single dependency with unambiguous dependencies on its separate target
modules, as if it had no experimental Gradle metadata. This way, the dependency will also be
published in the speci ed way, so the consumers won't encounter any ambiguity.
Android Support
Kotlin Multiplatform projects support the Android platform by providing the android preset.
Creating an Android target requires that one of the Android Gradle plugins, like
com.android.application or com.android.library is manually applied to the project.
Only one Android target may be created per Gradle subproject:
243
plugins {
id("com.android.library")
id("org.jetbrains.kotlin.multiplatform").version("1.3.31")
}
android { /* ... */ }
kotlin {
android { // Create the Android target
// Provide additional configuration if necessary
}
}
plugins {
id("com.android.library")
kotlin("multiplatform").version("1.3.31")
}
android { /* ... */ }
kotlin {
android { // Create the Android target
// Provide additional configuration if necessary
}
}
An Android target's compilations created by default are tied to Android build variants: for each
build variant, a Kotlin compilation is created under the same name.
Then, for each Android source set compiled by the variants, a Kotlin source set is created under
that source set name prepended by the target name, like Kotlin source set androidDebug for
an Android source set debug and the Kotlin target named android . These Kotlin source sets
are added to the variants compilations accordingly.
The default source set commonMain is added to each production (application or library) variant's
compilation. The commonTest source set is, similarly, added to the compilations of unit test and
instrumented test variants.
Annotation processing with kapt is also supported but, due to the current limitations, it requires
that the Android target is created before the kapt dependencies are con gured, which needs to
be done in a top-level dependencies { ... } block rather than within Kotlin source sets
dependencies.
// ...
kotlin {
android { /* ... */ }
}
dependencies {
kapt("com.my.annotation:processor:1.0.0")
}
244
Publishing Android libraries
To publish an Android library as a part of a multiplatform library, one needs to setup publishing
for the library and provide additional con guration for the Android library target.
By default, no artifacts of an Android library are published. To publish artifacts produced by a set
of Android variants, specify the variant names in the Android target block as follows:
kotlin {
android {
publishLibraryVariants("release", "debug")
}
}
The example above will work for Android libraries with no product avors. For a library with
product avors, the variant names also contain the avors, like fooBarDebug or
fooBazRelease .
Note that if a library consumer de nes variants that are missing in the library, they need to
provide matching fallbacks. For example, if a library does not have or does not publish a
staging build type, it will be necessary to provide a fallback for the consumers who have such
a build type, specifying at least one of the build types that the library publishes:
android {
buildTypes {
staging {
// ...
matchingFallbacks = ['release', 'debug']
}
}
}
android {
buildTypes {
val staging by creating {
// ...
matchingFallbacks = listOf("release", "debug")
}
}
}
Similarly, a library consumer may need to provide matching fallbacks for custom product avors
if some are missing in the library publications.
There is an option to publish variants grouped by the product avor, so that the outputs of the
di erent build types are placed in a single module, with the build type becoming a classi er for
the artifacts (the release build type is still published with no classi er). This mode is disabled by
default and can be enabled as follows:
245
kotlin {
android {
publishLibraryVariantsGroupedByFlavor = true
}
}
It is not recommended to publish variants grouped by the product avor in case they have
di erent dependencies, as those will be merged into one dependencies list.
Using Kotlin/Native Targets

It is important to note that some of the Kotlin/Native targets may only be built with an
appropriate host machine:
— Linux MIPS targets ( linuxMips32 and linuxMipsel32 ) require a Linux host. Other Linux
targets can be built on any supported host;
— Windows targets require a Windows host;
— macOS and iOS targets can only be built on a macOS host;
— The 64-bit Android Native target require a Linux or macOS host. The 32-bit Android Native
target can be built on any supported host.
A target that is not supported by the current host is ignored during build and therefore not
published. A library author may want to set up builds and publishing from di erent hosts as
required by the library target platforms.
Building nal native binaries
By default, a Kotlin/Native target is compiled down to a *.klib library artifact, which can be
consumed by Kotlin/Native itself as a dependency but cannot be executed or used as a native
library. To declare nal native binaries like executables or shared libraries a binaries property
of a native target is used. This property represents a collection of native binaries built for this
target in addition to the default *.klib artifact and provides a set of methods for declaring and
con guring them.
Note that the kotlin-multiplaform plugin doesn't create any production binaries by default.
The only binary available by default is a debug executable allowing one to run tests from the
test compilation.
Declaring binaries
A set of factory methods is used for declaring elements of the binaries collection. These
methods allow one to specify what kinds of binaries are to be created and con gure them. The
following binary kinds are supported (note that not all the kinds are available for all native
platforms):
246
Factory method Binary kind Available for
executable an executable program all native targets
sharedLib a shared native library all native targets except wasm32
staticLib a static native library all native targets except wasm32
framework an Objective-C framework macOS and iOS targets only
Each factory method exists in several versions. Consider them by example of the executable
method. All the same versions are available for all other factory methods.
The simplest version doesn't require any additional parameters and creates one binary for each
build type. Currently there a two build types available: DEBUG (produces a not optimized binary
with a debug information) and RELEASE (produces an optimized binary without debug
information). Consequently the following snippet creates two executable binaries: debug and
release.
kotlin {
linuxX64 { // Use your target instead.
binaries {
executable {
// Binary configuration.
}
}
}
}
A lambda expression accepted by the executable method in the example above is applied to
each binary created and allows one to con gure the binary (see the corresponding section). Note
that this lambda can be dropped if there is no need for additional con guration:
binaries {
executable()
}
It is possible to specify which build types will be used to create binaries and which won't. In the
following example only debug executable is created.
binaries {
executable([DEBUG]) {
}
}
binaries {
executable(listOf(DEBUG)) {
}
}
Finally the last factory method version allows customizing the binary name.
247
binaries {
executable('foo', [DEBUG]) {
}
// It's possible to drop the list of build types (all the available build types will
be used in this case).
executable('bar') {
}
}
binaries {
executable("foo", listOf(DEBUG)) {
}
// It's possible to drop the list of build types (all the available build types will
be used in this case).
executable("bar") {
}
}
The rst argument in this example allows one to set a name pre x for the created binaries which
is used to access them in the buildscript (see the "Accessing binaries" section). Also this pre x is
used as a default name for the binary le. For example on Windows the sample above produces
les foo.exe and bar.exe .
Using binary declaration APIs introduced in 1.3
Important: The approach described in this section is deprecated in Kotlin 1.3.30 and will not
be available since Kotlin 1.3.40. Consider using the binaries block instead.
It is possible to use the binary declaration APIs introduced in 1.3 in addition to the binaries DSL.
One can specify one or more of the outputKinds for a compilation using these APIs. The
following output kinds are available:
— executable for an executable program;
— dynamic for a dynamic library;
— static for a static library;
— framework for an Objective-C framework (only supported for macOS and iOS targets)
This can be done as follows:
kotlin {
linuxX64 { // Use your target instead
compilations.main.outputKinds("executable") // could also be "static", "dynamic"
or "framework".
}
}
248
kotlin {
linuxX64 { // Use your target instead
compilations["main"].outputKinds("executable") // could also be "static",
"dynamic" or "framework".
}
}
This creates binaries with corresponding types and the compilation name as name pre xes in the
binaries container. But note that such binaries are created after project evaluation so they are
available only in an afterEvaluate code block.
Accessing binaries
The binaries DSL allows not only creating binaries but also accessing already created ones to
con gure them or get their properties (e.g. path to an output le). The binaries collection
implements the DomainObjectSet interface and provides methods like all or matching
allowing con guring groups of elements .
Also it's possible to get a certain element of the collection. There are two ways to do this. First,
each binary has a unique name. This name is based on the name pre x (if it's speci ed), build
type and binary kind according to the following pattern: <optional-name-prefix><build-
type><binary-kind> , e.g. releaseFramework or testDebugExecutable .
Note: static and shared libraries has su xes static and shared respectively, e.g.
fooDebugStatic or barReleaseShared
This name can be used to access the binary:
// Fails if there is no such a binary.

binaries['fooDebugExecutable']
binaries.fooDebugExecutable
binaries.getByName('fooDebugExecutable')
// Returns null if there is no such a binary.

binaries.findByName('fooDebugExecutable')

binaries["fooDebugExecutable"]
binaries.getByName("fooDebugExecutable")

binaries.findByName("fooDebugExecutable")
The second way is using typed getters. These getters allow one to access a binary of a certain type
by its name pre x and build type.
249
binaries.getExecutable('foo', DEBUG)
binaries.getExecutable('', DEBUG) // Use an empty string if the name prefix isn't set.
// Similar getters are available for other binary kinds:

// getFramework, getStaticLib and getSharedLib.

binaries.findExecutable('foo', DEBUG)

// findFramework, findStaticLib and findSharedLib.

binaries.getExecutable("foo", DEBUG)
binaries.getExecutable("", DEBUG) // Use an empty string if the name prefix isn't set.

// getFramework, getStaticLib and getSharedLib.

binaries.findExecutable("foo", DEBUG)

// findFramework, findStaticLib and findSharedLib.
Con guring binaries
Binaries have a set of properties allowing one to con gure them. The following options are
available:
— Compilation. Each binary is built on basis of some compilation available in the same target.
The main compilation is used by default but a user can specify another one.
— Linker options. Options passed to a system linker during binary building. One can use this
setting to link against some native library.
— Output le name. By default the output le name is based on binary name pre x or, if the
name pre x isn't speci ed, on a project name. But it's possible to con gure the output le
name independently using the baseName property. Note that nal le name will be formed
by adding system-dependent pre x and post x to this base name. E.g. a libfoo.so is
produced for a Linux shared library with the base name foo .
— Entry point (for executable binaries only). By default the entry point for Kotlin/Native
programs is a main function located in the root package. This setting allows one to change
this default and use a custom function as an entry point. For example it can be used to move
the main function from the root package.
— Access to the output le.

— Access to a link task.
— Access to a run task (for executable binaries only). The kotlin-multiplatform plugin
creates run tasks for all executable binaries of host platforms (Windows, Linux and macOS).
250
Names of such tasks are based on binary names, e.g. runReleaseExecutable<target-
name> or runFooDebugExecutable<target-name> . A run task can be accessed using the
runTask property of an executable binary.
— Framework type (only for Objective-C frameworks). By default a framework built by
Kotlin/Native contains a dynamic library. But it's possible to replace it with a static library.
The following example shows how to use these settings.
binaries {
executable('test', [RELEASE]) {
// Build a binary on the basis of the test compilation.
compilation = compilations.test
// Custom command line options for the linker.

linkerOpts = ['-L/lib/search/path', '-L/another/search/path', '-lmylib']
// Base name for the output file.

baseName = 'foo'
// Custom entry point function.

entryPoint = 'org.example.main'
// Accessing the output file.

println("Executable path: ${outputFile.absolutePath}")
// Accessing the link task.

linkTask.dependsOn(additionalPreprocessingTask)
// Accessing the run task.

// Note that the runTask is null for non-host platforms.
runTask?.dependsOn(prepareForRun)
}
framework('my_framework' [RELEASE]) {
// Include a static library instead of a dynamic one into the framework.
isStatic = true
}
}
// Note that the test task created by default is also a run task.
// So you can access it using the same property.
def testTask = binaries.getExecutable("test", DEBUG).runTask
task processTests {
dependsOn(testTask)
}
251
binaries {
executable("test", listOf(RELEASE)) {
// Build a binary on the basis of the test compilation.
compilation = compilations["test"]
// Custom command line options for the linker.

linkerOpts = mutableListOf("-L/lib/search/path", "-L/another/search/path", "-
lmylib")
// Base name for the output file.

baseName = "foo"
// Custom entry point function.

entryPoint = "org.example.main"
// Accessing the output file.

println("Executable path: ${outputFile.absolutePath}")
// Accessing the link task.

linkTask.dependsOn(additionalPreprocessingTask)
// Accessing the run task.

// Note that the runTask is null for non-host platforms.
runTask?.dependsOn(prepareForRun)
}
framework("my_framework" listOf(RELEASE)) {
// Include a static library instead of a dynamic one into the framework.
isStatic = true
}
}
// Note that the test task created by default is also a run task.
// So you can access it using the same property.
val testTask = binaries.getExecutable("test", DEBUG).runTask
val processTests by tasks.creating
processTests.dependsOn(testTask)
Exporting dependencies in frameworks
When building an Objective-C framework, it is often necessary to pack not just the classes of the
current project, but also the classes of some of its dependencies. The Binaries DSL allows one to
specify which dependencies will be exported in the framework using the export method. Note
that only API dependencies of a corresponding source set can be exported.
252
kotlin {
sourceSets {
macosMain.dependencies {
// Will be exported in the framework.
api project(':dependency')
api 'org.example:exported-library:1.0'
// Will not be exported in the framework.

api 'org.example:not-exported-library:1.0'
}
}
macosX64("macos").binaries {
framework {
export project(':dependency')
export 'org.example:exported-library:1.0'
}
}
}
kotlin {
sourceSets {
macosMain.dependencies {
// Will be exported in the framework.
api(project(":dependency"))
api("org.example:exported-library:1.0")
// Will not be exported in the framework.

api("org.example:not-exported-library:1.0")
}
}
macosX64("macos").binaries {
framework {
export(project(":dependency"))
export("org.example:exported-library:1.0")
}
}
}
As shown in this example, maven dependency also can be exported. But due to current
limitations of Gradle metadata such a dependency should be either a platform one (e.g.
kotlinx-coroutines-core-native_debug_macos_x64 instead of kotlinx-
coroutines-core-native) or be exported transitively (see below).
By default, export works non-transitively. If a library foo depending on library bar is exported,
only methods of foo will be added in the output framework. This behaviour can by changed by
the transitiveExport ag.
253
binaries {
framework {
export project(':dependency')
// Export transitively.
transitiveExport = true
}
}
binaries {
framework {
export(project(":dependency"))
// Export transitively.
transitiveExport = true
}
}
Building universal frameworks
By default, an Objective-C framework produced by Kotlin/Native supports only one platform.

However, such frameworks can be merged into a single universal (fat) binary using the lipo
utility. Particularly, such an operation makes sense for 32-bit and 64-bit iOS frameworks. In this
case the resulting universal framework can be used on both 32-bit and 64-bit devices.
The Gradle plugin provides a separate task that creates a universal framework for iOS targets
from several regular ones. The example below shows how to use this task. Note that the fat
framework must have the same base name as the initial frameworks.
import org.jetbrains.kotlin.gradle.tasks.FatFrameworkTask
kotlin {
// Create and configure the targets.
targets {
iosArm32("ios32")
iosArm64("ios64")
configure([ios32, ios64]) {
binaries.framework {
baseName = "my_framework"
}
}
}
// Create a task building a fat framework.
254
task debugFatFramework(type: FatFrameworkTask) {
// The fat framework must have the same base name as the initial frameworks.
// The default destination directory is '<build directory>/fat-framework'.

destinationDir = file("$buildDir/fat-framework/debug")
// Specify the frameworks to be merged.

from(
targets.ios32.binaries.getFramework("DEBUG"),
targets.ios64.binaries.getFramework("DEBUG")
)
}
}
import org.jetbrains.kotlin.gradle.tasks.FatFrameworkTask
kotlin {
// Create and configure the targets.
val ios32 = iosArm32("ios32")
val ios64 = iosArm64("ios64")
configure(listOf(ios32, ios64)) {
binaries.framework {
}
}

tasks.create("debugFatFramework", FatFrameworkTask::class) {
// The fat framework must have the same base name as the initial frameworks.
255
// The default destination directory is '<build directory>/fat-framework'.
destinationDir = buildDir.resolve("fat-framework/debug")
// Specify the frameworks to be merged.

from(
ios32.binaries.getFramework("DEBUG"),
ios64.binaries.getFramework("DEBUG")
)
}
}
CInterop support
Since Kotlin/Native provides interoperability with native languages, there is a DSL allowing one to
con gure this feature for a speci c compilation.
A compilation can interact with several native libraries. Interoperability with each of them can be
con gured in the cinterops block of the compilation:
kotlin {
linuxX64 { // Replace with a target you need.
compilations.main {
cinterops {
myInterop {
// Def-file describing the native API.
// The default path is src/nativeInterop/cinterop/<interop-name>.def
defFile project.file("def-file.def")
// Package to place the Kotlin API generated.

packageName 'org.sample'
// Options to be passed to compiler by cinterop tool.

compilerOpts '-Ipath/to/headers'
// Directories for header search (an analogue of the -I<path>

compiler option).
includeDirs.allHeaders("path1", "path2")
// Additional directories to search headers listed in the

'headerFilter' def-file option.
// -headerFilterAdditionalSearchPrefix command line option analogue.
includeDirs.headerFilterOnly("path1", "path2")
// A shortcut for includeDirs.allHeaders.

includeDirs("include/directory", "another/directory")
}
anotherInterop { /* ... */ }
}
}
}
}
256
kotlin {
linuxX64 { // Replace with a target you need.
compilations.getByName("main") {
val myInterop by cinterops.creating {
// The default path is src/nativeInterop/cinterop/<interop-name>.def
defFile(project.file("def-file.def"))

packageName("org.sample")
// Options to be passed to compiler by cinterop tool.

compilerOpts("-Ipath/to/headers")
// Directories to look for headers.

includeDirs.apply {
// Directories for header search (an analogue of the -I<path>
compiler option).
allHeaders("path1", "path2")
// Additional directories to search headers listed in the

'headerFilter' def-file option.
headerFilterOnly("path1", "path2")
}
includeDirs("include/directory", "another/directory")
}
val anotherInterop by cinterops.creating { /* ... */ }

}
}
}
Often it's necessary to specify target-speci c linker options for a binary which uses a native
library. It can by done using the linkerOpts property of the binary. See the Con guring
binaries section for details.
257
Other
Destructuring Declarations
Sometimes it is convenient to destructure an object into a number of variables, for example:
val (name, age) = person
This syntax is called a destructuring declaration. A destructuring declaration creates multiple

variables at once. We have declared two new variables: name and age , and can use them
independently:
println(name)
println(age)
A destructuring declaration is compiled down to the following code:
val name = person.component1()

val age = person.component2()
The component1() and component2() functions are another example of the principle of
conventions widely used in Kotlin (see operators like + and * , for-loops etc.). Anything can be
on the right-hand side of a destructuring declaration, as long as the required number of
component functions can be called on it. And, of course, there can be component3() and
component4() and so on.
Note that the componentN() functions need to be marked with the operator keyword to
allow using them in a destructuring declaration.
Destructuring declarations also work in for-loops: when you say:
for ((a, b) in collection) { ... }
Variables a and b get the values returned by component1() and component2() called on
elements of the collection.
Example: Returning Two Values from a Function

Let's say we need to return two things from a function. For example, a result object and a status
of some sort. A compact way of doing this in Kotlin is to declare a data class and return its
instance:
258
data class Result(val result: Int, val status: Status)
fun function(...): Result {
// computations
return Result(result, status)

}
// Now, to use this function:

val (result, status) = function(...)
Since data classes automatically declare componentN() functions, destructuring declarations

work here.
NOTE: we could also use the standard class Pair and have function() return Pair<Int,
Status> , but it's often better to have your data named properly.
Example: Destructuring Declarations and Maps

Probably the nicest way to traverse a map is this:
for ((key, value) in map) {

// do something with the key and the value
}
To make this work, we should
— present the map as a sequence of values by providing an iterator() function;
— present each of the elements as a pair by providing functions component1() and

component2() .
And indeed, the standard library provides such extensions:
operator fun <K, V> Map<K, V>.iterator(): Iterator<Map.Entry<K, V>> =

entrySet().iterator()
operator fun <K, V> Map.Entry<K, V>.component1() = getKey()
operator fun <K, V> Map.Entry<K, V>.component2() = getValue()
So you can freely use destructuring declarations in for-loops with maps (as well as collections of
data class instances etc).
Underscore for unused variables (since 1.1)

If you don't need a variable in the destructuring declaration, you can place an underscore instead
of its name:
val (_, status) = getResult()
The componentN() operator functions are not called for the components that are skipped in
this way.
Destructuring in Lambdas (since 1.1)
259
You can use the destructuring declarations syntax for lambda parameters. If a lambda has a
parameter of the Pair type (or Map.Entry , or any other type that has the appropriate
componentN functions), you can introduce several new parameters instead of one by putting
them in parentheses:
map.mapValues { entry -> "${entry.value}!" }

map.mapValues { (key, value) -> "$value!" }
Note the di erence between declaring two parameters and declaring a destructuring pair instead
of a parameter:
{ a -> ... } // one parameter

{ a, b -> ... } // two parameters
{ (a, b) -> ... } // a destructured pair
{ (a, b), c -> ... } // a destructured pair and another parameter
If a component of the destructured parameter is unused, you can replace it with the underscore
to avoid inventing its name:
map.mapValues { (_, value) -> "$value!" }
You can specify the type for the whole destructured parameter or for a speci c component
separately:
map.mapValues { (_, value): Map.Entry<Int, String> -> "$value!" }
map.mapValues { (_, value: String) -> "$value!" }
260
Type Checks and Casts: 'is' and 'as'
is and !is Operators

We can check whether an object conforms to a given type at runtime by using the is operator
or its negated form !is :
if (obj is String) {
print(obj.length)
}
if (obj !is String) { // same as !(obj is String)

print("Not a String")
}
else {
print(obj.length)
}
Smart Casts
In many cases, one does not need to use explicit cast operators in Kotlin, because the compiler
tracks the is -checks and explicit casts for immutable values and inserts (safe) casts
automatically when needed:
fun demo(x: Any) {

if (x is String) {
print(x.length) // x is automatically cast to String
}
}
The compiler is smart enough to know a cast to be safe if a negative check leads to a return:
if (x !is String) return
or in the right-hand side of && and || :
// x is automatically cast to string on the right-hand side of `||`

if (x !is String || x.length == 0) return
// x is automatically cast to string on the right-hand side of `&&`

if (x is String && x.length > 0) {
}
Such smart casts work for when-expressions and while-loops as well:
when (x) {
is Int -> print(x + 1)
is String -> print(x.length + 1)
is IntArray -> print(x.sum())
}
261
Note that smart casts do not work when the compiler cannot guarantee that the variable cannot
change between the check and the usage. More speci cally, smart casts are applicable according
to the following rules:
— val local variables - always except for local delegated properties;
— val properties - if the property is private or internal or the check is performed in the same
module where the property is declared. Smart casts aren't applicable to open properties or
properties that have custom getters;
— var local variables - if the variable is not modi ed between the check and the usage, is not
captured in a lambda that modi es it, and is not a local delegated property;
— var properties - never (because the variable can be modi ed at any time by other code).
"Unsafe" cast operator

Usually, the cast operator throws an exception if the cast is not possible. Thus, we call it unsafe.
The unsafe cast in Kotlin is done by the in x operator as (see operator precedence):
val x: String = y as String
Note that null cannot be cast to String as this type is not nullable, i.e. if y is null, the code
above throws an exception. In order to match Java cast semantics we have to have nullable type
at cast right hand side, like:
val x: String? = y as String?
"Safe" (nullable) cast operator

To avoid an exception being thrown, one can use a safe cast operator as? that returns null on
failure:
val x: String? = y as? String
Note that despite the fact that the right-hand side of as? is a non-null type String the result of
the cast is nullable.
Type erasure and generic type checks

Kotlin ensures type safety of operations involving generics at compile time, while, at runtime,
instances of generic types hold no information about their actual type arguments. For example,
List<Foo> is erased to just List<*> . In general, there is no way to check whether an instance
belongs to a generic type with certain type arguments at runtime.
Given that, the compiler prohibits is-checks that cannot be performed at runtime due to type
erasure, such as ints is List<Int> or list is T (type parameter). You can, however,
check an instance against a star-projected type:
262
if (something is List<*>) {
something.forEach { println(it) } // The items are typed as Àny?`
}
Similarly, when you already have the type arguments of an instance checked statically (at compile
time), you can make an is-check or a cast that involves the non-generic part of the type. Note
that angle brackets are omitted in this case:
fun handleStrings(list: List<String>) {

if (list is ArrayList) {
// `list` is smart-cast to ÀrrayList<String>`
}
}
The same syntax with omitted type arguments can be used for casts that do not take type
arguments into account: list as ArrayList .
Inline functions with rei ed type parameters have their actual type arguments inlined at each call
site, which enables arg is T checks for the type parameters, but if arg is an instance of a
generic type itself, its type arguments are still erased. Example:
inline fun <reified A, reified B> Pair<*, *>.asPairOf(): Pair<A, B>? {

if (first !is A || second !is B) return null
return first as A to second as B
}
val somePair: Pair<Any?, Any?> = "items" to listOf(1, 2, 3)
val stringToSomething = somePair.asPairOf<String, Any>()

val stringToInt = somePair.asPairOf<String, Int>()
val stringToList = somePair.asPairOf<String, List<*>>()
val stringToStringList = somePair.asPairOf<String, List<String>>() // Breaks type safety!
Unchecked casts
As said above, type erasure makes checking actual type arguments of a generic type instance
impossible at runtime, and generic types in the code might be connected to each other not
closely enough for the compiler to ensure type safety.
Even so, sometimes we have high-level program logic that implies type safety instead. For
example:
fun readDictionary(file: File): Map<String, *> = file.inputStream().use {

TODO("Read a mapping of strings to arbitrary elements.")
}
// We saved a map with Ìnt`s into that file

val intsFile = File("ints.dictionary")
// Warning: Unchecked cast: `Map<String, *>` to `Map<String, Int>`

val intsDictionary: Map<String, Int> = readDictionary(intsFile) as Map<String, Int>
263
The compiler produces a warning for the cast in the last line. The cast cannot be fully checked at
runtime and provides no guarantee that the values in the map are Int .
To avoid unchecked casts, you can redesign the program structure: in the example above, there
could be interfaces DictionaryReader<T> and DictionaryWriter<T> with type-safe
implementations for di erent types. You can introduce reasonable abstractions to move
unchecked casts from calling code to the implementation details. Proper use of generic variance
can also help.
For generic functions, using rei ed type parameters makes the casts such as arg as T checked,
unless arg 's type has its own type arguments that are erased.
An unchecked cast warning can be suppressed by annotating the statement or the declaration
where it occurs with @Suppress("UNCHECKED_CAST") :
inline fun <reified T> List<*>.asListOfType(): List<T>? =

if (all { it is T })
@Suppress("UNCHECKED_CAST")
this as List<T> else
null
On the JVM, the array types ( Array<Foo> ) retain the information about the erased type of their
elements, and the type casts to an array type are partially checked: the nullability and actual type
arguments of the elements type are still erased. For example, the cast foo as
Array<List<String>?> will succeed if foo is an array holding any List<*> , nullable or not.
264
This Expression
To denote the current receiver, we use this expressions:
— In a member of a class, this refers to the current object of that class.
— In an extension function or a function literal with receiver this denotes the receiver
parameter that is passed on the left-hand side of a dot.
If this has no quali ers, it refers to the innermost enclosing scope. To refer to this in other
scopes, label quali ers are used:
Quali ed this
To access this from an outer scope (a class, or extension function, or labeled function literal
with receiver) we write this@label where @label is a label on the scope this is meant to be
from:
class A { // implicit label @A

inner class B { // implicit label @B
fun Int.foo() { // implicit label @foo
val a = this@A // A's this
val b = this@B // B's this
val c = this // foo()'s receiver, an Int

val c1 = this@foo // foo()'s receiver, an Int
val funLit = lambda@ fun String.() {

val d = this // funLit's receiver
}
val funLit2 = { s: String ->

// foo()'s receiver, since enclosing lambda expression
// doesn't have any receiver
val d1 = this
}
}
}
}
265
Equality
In Kotlin there are two types of equality:
— Structural equality (a check for equals() ).
— Referential equality (two references point to the same object);
Structural equality
Structural equality is checked by the == operation (and its negated counterpart != ). By
convention, an expression like a == b is translated to:
a?.equals(b) ?: (b === null)
I.e. if a is not null , it calls the equals(Any?) function, otherwise (i.e. a is null ) it checks
that b is referentially equal to null .
Note that there's no point in optimizing your code when comparing to null explicitly: a ==
null will be automatically translated to a === null .
To provide a custom equals check implementation, override the equals(other: Any?):

Boolean function. Functions with the same name and other signatures, like equals(other:
Foo) , don't a ect equality checks with the operators == and != .
Structural equality has nothing to do with comparison de ned by the Comparable<...>

interface, so only a custom equals(Any?) implementation may a ect the behavior of the
operator.
Floating point numbers equality

When an equality check operands are statically known to be Float or Double (nullable or not),
the check follows the IEEE 754 Standard for Floating-Point Arithmetic.
Otherwise, the structural equality is used, which disagrees with the standard so that NaN is equal
to itself, and -0.0 is not equal to 0.0 .
See: Floating Point Numbers Comparison.
Referential equality
Referential equality is checked by the === operation (and its negated counterpart !== ). a ===
b evaluates to true if and only if a and b point to the same object. For values which are
represented as primitive types at runtime (for example, Int ), the === equality check is
equivalent to the == check.
266
Operator overloading
Kotlin allows us to provide implementations for a prede ned set of operators on our types.
These operators have xed symbolic representation (like + or * ) and xed precedence. To
implement an operator, we provide a member function or an extension function with a xed
name, for the corresponding type, i.e. left-hand side type for binary operations and argument
type for unary ones. Functions that overload operators need to be marked with the operator
modi er.
Further we describe the conventions that regulate operator overloading for di erent operators.
Unary operations
Unary pre x operators

Expression Translated to
+a a.unaryPlus()
-a a.unaryMinus()
!a a.not()
This table says that when the compiler processes, for example, an expression +a , it performs the
following steps:
— Determines the type of a , let it be T ;
— Looks up a function unaryPlus() with the operator modi er and no parameters for the
receiver T , i.e. a member function or an extension function;
— If the function is absent or ambiguous, it is a compilation error;

— If the function is present and its return type is R , the expression +a has type R ;
Note that these operations, as well as all the others, are optimized for Basic types and do not
introduce overhead of function calls for them.
As an example, here's how you can overload the unary minus operator:
data class Point(val x: Int, val y: Int)
operator fun Point.unaryMinus() = Point(-x, -y)
val point = Point(10, 20)
fun main() {
println(-point) // prints "Point(x=-10, y=-20)"
}
Increments and decrements
267
a++ a.inc() + see below
a-- a.dec() + see below
The inc() and dec() functions must return a value, which will be assigned to the variable on
which the ++ or -- operation was used. They shouldn't mutate the object on which the inc or
dec was invoked.
The compiler performs the following steps for resolution of an operator in the post x form, e.g.
a++ :
— Determines the type of a , let it be T ;
— Looks up a function inc() with the operator modi er and no parameters, applicable to
the receiver of type T ;
— Checks that the return type of the function is a subtype of T .
The e ect of computing the expression is:
— Store the initial value of a to a temporary storage a0 ;
— Assign the result of a.inc() to a ;
— Return a0 as a result of the expression.
For a-- the steps are completely analogous.
For the pre x forms ++a and --a resolution works the same way, and the e ect is:
— Assign the result of a.inc() to a ;
— Return the new value of a as a result of the expression.
Binary operations
Arithmetic operators
a + b a.plus(b)
a - b a.minus(b)
a * b a.times(b)
a / b a.div(b)
a % b a.rem(b), a.mod(b) (deprecated)
a..b a.rangeTo(b)
For the operations in this table, the compiler just resolves the expression in the Translated to
column.
268
Note that the rem operator is supported since Kotlin 1.1. Kotlin 1.0 uses the mod operator,
which is deprecated in Kotlin 1.1.
Example
Below is an example Counter class that starts at a given value and can be incremented using the
overloaded + operator:
data class Counter(val dayIndex: Int) {

operator fun plus(increment: Int): Counter {
return Counter(dayIndex + increment)
}
}
'In' operator
a in b b.contains(a)
a !in b !b.contains(a)
For in and !in the procedure is the same, but the order of arguments is reversed.
Indexed access operator

a[i] a.get(i)
a[i, j] a.get(i, j)
a[i_1, ..., i_n] a.get(i_1, ..., i_n)
a[i] = b a.set(i, b)
a[i, j] = b a.set(i, j, b)
a[i_1, ..., i_n] = b a.set(i_1, ..., i_n, b)
Square brackets are translated to calls to get and set with appropriate numbers of
arguments.
Invoke operator
a() a.invoke()
a(i) a.invoke(i)
a(i, j) a.invoke(i, j)
a(i_1, ..., i_n) a.invoke(i_1, ..., i_n)
Parentheses are translated to calls to invoke with appropriate number of arguments.
Augmented assignments
269
a += b a.plusAssign(b)
a -= b a.minusAssign(b)
a *= b a.timesAssign(b)
a /= b a.divAssign(b)
a %= b a.remAssign(b), a.modAssign(b) (deprecated)
For the assignment operations, e.g. a += b , the compiler performs the following steps:
— If the function from the right column is available

— If the corresponding binary function (i.e. plus() for plusAssign() ) is available too,
report error (ambiguity),
— Make sure its return type is Unit , and report an error otherwise,
— Generate code for a.plusAssign(b) ;
— Otherwise, try to generate code for a = a + b (this includes a type check: the type of a +
b must be a subtype of a ).
Note: assignments are NOT expressions in Kotlin.
Equality and inequality operators

a == b a?.equals(b) ?: (b === null)
a != b !(a?.equals(b) ?: (b === null))
These operators only work with the function equals(other: Any?): Boolean, which can be
overridden to provide custom equality check implementation. Any other function with the same
name (like equals(other: Foo) ) will not be called.
Note: === and !== (identity checks) are not overloadable, so no conventions exist for them.
The == operation is special: it is translated to a complex expression that screens for null 's.
null == null is always true, and x == null for a non-null x is always false and won't
invoke x.equals() .
Comparison operators
a > b a.compareTo(b) > 0
a = b a.compareTo(b) >= 0
a <= b a.compareTo(b) <= 0
All comparisons are translated into calls to compareTo , that is required to return Int .
270
Property delegation operators
provideDelegate , getValue and setValue operator functions are described in Delegated

properties.
In x calls for named functions

We can simulate custom in x operations by using in x function calls.
271
Null Safety
Nullable types and Non-Null Types

Kotlin's type system is aimed at eliminating the danger of null references from code, also known
as the The Billion Dollar Mistake.
One of the most common pitfalls in many programming languages, including Java, is that
accessing a member of a null reference will result in a null reference exception. In Java this would
be the equivalent of a NullPointerException or NPE for short.
Kotlin's type system is aimed to eliminate NullPointerException 's from our code. The only
possible causes of NPE's may be:
— An explicit call to throw NullPointerException() ;
— Usage of the !! operator that is described below;
— Some data inconsistency with regard to initialization, such as when:

— An uninitialized this available in a constructor is passed and used somewhere ("leaking
this");
— A superclass constructor calls an open member whose implementation in the derived class
uses uninitialized state;
— Java interoperation:
— Attempts to access a member on a null reference of a platform type;
— Generic types used for Java interoperation with incorrect nullability, e.g. a piece of Java
code might add null into a Kotlin MutableList<String> , meaning that
MutableList<String?> should be used for working with it;
— Other issues caused by external Java code.
In Kotlin, the type system distinguishes between references that can hold null (nullable
references) and those that can not (non-null references). For example, a regular variable of type
String can not hold null:
var a: String = "abc"

a = null // compilation error
To allow nulls, we can declare a variable as nullable string, written String? :
var b: String? = "abc"

b = null // ok
print(b)
Now, if you call a method or access a property on a , it's guaranteed not to cause an NPE, so you
can safely say:
val l = a.length
272
But if you want to access the same property on b , that would not be safe, and the compiler
reports an error:
val l = b.length // error: variable 'b' can be null
But we still need to access that property, right? There are a few ways of doing that.
Checking for null in conditions

First, you can explicitly check if b is null, and handle the two options separately:
val l = if (b != null) b.length else -1
The compiler tracks the information about the check you performed, and allows the call to
length inside the if. More complex conditions are supported as well:
val b: String? = "Kotlin"

if (b != null && b.length > 0) {
print("String of length ${b.length}")
} else {
print("Empty string")
}
Note that this only works where b is immutable (i.e. a local variable which is not modi ed
between the check and the usage or a member val which has a backing eld and is not
overridable), because otherwise it might happen that b changes to null after the check.
Safe Calls
Your second option is the safe call operator, written ?. :
val a = "Kotlin"
val b: String? = null
println(b?.length)
println(a?.length) // Unnecessary safe call
This returns b.length if b is not null, and null otherwise. The type of this expression is
Int? .
Safe calls are useful in chains. For example, if Bob, an Employee, may be assigned to a
Department (or not), that in turn may have another Employee as a department head, then to
obtain the name of Bob's department head (if any), we write the following:
bob?.department?.head?.name
Such a chain returns null if any of the properties in it is null.
To perform a certain operation only for non-null values, you can use the safe call operator
together with let:
273
val listWithNulls: List<String?> = listOf("Kotlin", null)
for (item in listWithNulls) {
item?.let { println(it) } // prints Kotlin and ignores null
}
A safe call can also be placed on the left side of an assignment. Then, if one of the receivers in the
safe calls chain is null, the assignment is skipped, and the expression on the right is not evaluated
at all:
// If either `person` or `person.department` is null, the function is not called:

person?.department?.head = managersPool.getManager()
Elvis Operator
When we have a nullable reference r , we can say "if r is not null, use it, otherwise use some
non-null value x ":
val l: Int = if (b != null) b.length else -1
Along with the complete if-expression, this can be expressed with the Elvis operator, written
?: :
val l = b?.length ?: -1
If the expression to the left of ?: is not null, the elvis operator returns it, otherwise it returns
the expression to the right. Note that the right-hand side expression is evaluated only if the left-
hand side is null.
Note that, since throw and return are expressions in Kotlin, they can also be used on the right
hand side of the elvis operator. This can be very handy, for example, for checking function
arguments:
fun foo(node: Node): String? {

val parent = node.getParent() ?: return null
val name = node.getName() ?: throw IllegalArgumentException("name expected")
// ...
}
The !! Operator
The third option is for NPE-lovers: the not-null assertion operator ( !! ) converts any value to a
non-null type and throws an exception if the value is null. We can write b!! , and this will return
a non-null value of b (e.g., a String in our example) or throw an NPE if b is null:
val l = b!!.length
Thus, if you want an NPE, you can have it, but you have to ask for it explicitly, and it does not
appear out of the blue.
274
Safe Casts
Regular casts may result into a ClassCastException if the object is not of the target type.
Another option is to use safe casts that return null if the attempt was not successful:
val aInt: Int? = a as? Int
Collections of Nullable Type

If you have a collection of elements of a nullable type and want to lter non-null elements, you
can do so by using filterNotNull :
val nullableList: List<Int?> = listOf(1, 2, null, 4)

val intList: List<Int> = nullableList.filterNotNull()
275
Exceptions
Exception Classes
All exception classes in Kotlin are descendants of the class Throwable . Every exception has a
message, stack trace and an optional cause.
To throw an exception object, use the throw-expression:
throw Exception("Hi There!")
To catch an exception, use the try-expression:
try {
// some code
}
catch (e: SomeException) {
// handler
}
finally {
// optional finally block
}
There may be zero or more catch blocks. finally block may be omitted. However at least one
catch or finally block should be present.
Try is an expression
try is an expression, i.e. it may have a return value:
val a: Int? = try { parseInt(input) } catch (e: NumberFormatException) { null }
The returned value of a try-expression is either the last expression in the try block or the last
expression in the catch block (or blocks). Contents of the finally block do not a ect the result
of the expression.
Checked Exceptions
Kotlin does not have checked exceptions. There are many reasons for this, but we will provide a
simple example.
The following is an example interface of the JDK implemented by StringBuilder class:
Appendable append(CharSequence csq) throws IOException;
What does this signature say? It says that every time I append a string to something (a
StringBuilder , some kind of a log, a console, etc.) I have to catch those IOExceptions .
Why? Because it might be performing IO ( Writer also implements Appendable )… So it results
into this kind of code all over the place:
276
try {
log.append(message)
}
catch (IOException e) {
// Must be safe
}
And this is no good, see E ective Java, 3rd Edition, Item 77: Don't ignore exceptions.
Bruce Eckel says in Does Java need Checked Exceptions?:
Examination of small programs leads to the conclusion that requiring exception

speci cations could both enhance developer productivity and enhance code quality, but
experience with large software projects suggests a di erent result – decreased productivity
and little or no increase in code quality.
Other citations of this sort:
— Java's checked exceptions were a mistake (Rod Waldho )

— The Trouble with Checked Exceptions (Anders Hejlsberg)
The Nothing type

throw is an expression in Kotlin, so you can use it, for example, as part of an Elvis expression:
val s = person.name ?: throw IllegalArgumentException("Name required")
The type of the throw expression is the special type Nothing . The type has no values and is
used to mark code locations that can never be reached. In your own code, you can use Nothing
to mark a function that never returns:
fun fail(message: String): Nothing {

throw IllegalArgumentException(message)
}
When you call this function, the compiler will know that the execution doesn't continue beyond
the call:
val s = person.name ?: fail("Name required")

println(s) // 's' is known to be initialized at this point
Another case where you may encounter this type is type inference. The nullable variant of this
type, Nothing? , has exactly one possible value, which is null . If you use null to initialize a
value of an inferred type and there's no other information that can be used to determine a more
speci c type, the compiler will infer the Nothing? type:
val x = null // 'x' has type `Nothing?`

val l = listOf(null) // 'l' has type `List<Nothing?>
Java Interoperability
277
Please see the section on exceptions in the Java Interoperability section for information about
Java interoperability.
278
Annotations
Annotation Declaration
Annotations are means of attaching metadata to code. To declare an annotation, put the
annotation modi er in front of a class:
annotation class Fancy
Additional attributes of the annotation can be speci ed by annotating the annotation class with
meta-annotations:
— @Target speci es the possible kinds of elements which can be annotated with the annotation
(classes, functions, properties, expressions etc.);
— @Retention speci es whether the annotation is stored in the compiled class les and
whether it's visible through re ection at runtime (by default, both are true);
— @Repeatable allows using the same annotation on a single element multiple times;
— @MustBeDocumented speci es that the annotation is part of the public API and should be
included in the class or method signature shown in the generated API documentation.
@Target(AnnotationTarget.CLASS, AnnotationTarget.FUNCTION,
AnnotationTarget.VALUE_PARAMETER, AnnotationTarget.EXPRESSION)
@Retention(AnnotationRetention.SOURCE)
@MustBeDocumented
annotation class Fancy
Usage
@Fancy class Foo {

@Fancy fun baz(@Fancy foo: Int): Int {
return (@Fancy 1)
}
}
If you need to annotate the primary constructor of a class, you need to add the constructor
keyword to the constructor declaration, and add the annotations before it:
class Foo @Inject constructor(dependency: MyDependency) { ... }
You can also annotate property accessors:
class Foo {
var x: MyDependency? = null
@Inject set
}
Constructors
Annotations may have constructors that take parameters.
279
annotation class Special(val why: String)
@Special("example") class Foo {}
Allowed parameter types are:
— types that correspond to Java primitive types (Int, Long etc.);

— strings;
— classes ( Foo::class );
— enums;
— other annotations;
— arrays of the types listed above.
Annotation parameters cannot have nullable types, because the JVM does not support storing
null as a value of an annotation attribute.
If an annotation is used as a parameter of another annotation, its name is not pre xed with the @
character:
annotation class ReplaceWith(val expression: String)
annotation class Deprecated(

val message: String,
val replaceWith: ReplaceWith = ReplaceWith(""))
@Deprecated("This function is deprecated, use === instead", ReplaceWith("this ===

other"))
If you need to specify a class as an argument of an annotation, use a Kotlin class (KClass). The
Kotlin compiler will automatically convert it to a Java class, so that the Java code will be able to see
the annotations and arguments normally.
import kotlin.reflect.KClass
annotation class Ann(val arg1: KClass<*>, val arg2: KClass<out Any>)
@Ann(String::class, Int::class) class MyClass
Lambdas
Annotations can also be used on lambdas. They will be applied to the invoke() method into
which the body of the lambda is generated. This is useful for frameworks like Quasar, which uses
annotations for concurrency control.
annotation class Suspendable
val f = @Suspendable { Fiber.sleep(10) }
280
Annotation Use-site Targets
When you're annotating a property or a primary constructor parameter, there are multiple Java
elements which are generated from the corresponding Kotlin element, and therefore multiple
possible locations for the annotation in the generated Java bytecode. To specify how exactly the
annotation should be generated, use the following syntax:
class Example(@field:Ann val foo, // annotate Java field

@get:Ann val bar, // annotate Java getter
@param:Ann val quux) // annotate Java constructor parameter
The same syntax can be used to annotate the entire le. To do this, put an annotation with the
target file at the top level of a le, before the package directive or before all imports if the le
is in the default package:
@file:JvmName("Foo")
package org.jetbrains.demo
If you have multiple annotations with the same target, you can avoid repeating the target by
adding brackets after the target and putting all the annotations inside the brackets:
class Example {
@set:[Inject VisibleForTesting]
var collaborator: Collaborator
}
The full list of supported use-site targets is:
— file ;
— property (annotations with this target are not visible to Java);
— field ;
— get (property getter);
— set (property setter);
— receiver (receiver parameter of an extension function or property);
— param (constructor parameter);
— setparam (property setter parameter);
— delegate (the eld storing the delegate instance for a delegated property).
To annotate the receiver parameter of an extension function, use the following syntax:
fun @receiver:Fancy String.myExtension() { ... }
If you don't specify a use-site target, the target is chosen according to the @Target annotation
of the annotation being used. If there are multiple applicable targets, the rst applicable target
from the following list is used:
281
— param ;
— property ;
— field .
Java Annotations
Java annotations are 100% compatible with Kotlin:
import org.junit.Test
import org.junit.Assert.*
import org.junit.Rule
import org.junit.rules.*
class Tests {
// apply @Rule annotation to property getter
@get:Rule val tempFolder = TemporaryFolder()
@Test fun simple() {

val f = tempFolder.newFile()
assertEquals(42, getTheAnswer())
}
}
Since the order of parameters for an annotation written in Java is not de ned, you can't use a
regular function call syntax for passing the arguments. Instead, you need to use the named
argument syntax:
// Java
public @interface Ann {
int intValue();
String stringValue();
}
// Kotlin
@Ann(intValue = 1, stringValue = "abc") class C
Just like in Java, a special case is the value parameter; its value can be speci ed without an
explicit name:
// Java
public @interface AnnWithValue {
String value();
}
// Kotlin
@AnnWithValue("abc") class C
Arrays as annotation parameters
If the value argument in Java has an array type, it becomes a vararg parameter in Kotlin:
282
// Java
public @interface AnnWithArrayValue {
String[] value();
}
// Kotlin
@AnnWithArrayValue("abc", "foo", "bar") class C
For other arguments that have an array type, you need to use the array literal syntax (since Kotlin
1.2) or arrayOf(...) :
// Java
public @interface AnnWithArrayMethod {
String[] names();
}
// Kotlin 1.2+:
@AnnWithArrayMethod(names = ["abc", "foo", "bar"])
class C
// Older Kotlin versions:

@AnnWithArrayMethod(names = arrayOf("abc", "foo", "bar"))
class D
Accessing properties of an annotation instance
Values of an annotation instance are exposed as properties to Kotlin code:
// Java
public @interface Ann {
int value();
}
// Kotlin
fun foo(ann: Ann) {
val i = ann.value
}
283
Re ection
Re ection is a set of language and library features that allows for introspecting the structure of
your own program at runtime. Kotlin makes functions and properties rst-class citizens in the
language, and introspecting them (i.e. learning a name or a type of a property or function at
runtime) is closely intertwined with simply using a functional or reactive style.
On the Java platform, the runtime component required for using the re ection features
is distributed as a separate JAR le (kotlin-reflect.jar). This is done to reduce the
required size of the runtime library for applications that do not use re ection features. If
you do use re ection, please make sure that the .jar le is added to the classpath of your
project.
Class References
The most basic re ection feature is getting the runtime reference to a Kotlin class. To obtain the
reference to a statically known Kotlin class, you can use the class literal syntax:
val c = MyClass::class
The reference is a value of type KClass.
Note that a Kotlin class reference is not the same as a Java class reference. To obtain a Java class
reference, use the .java property on a KClass instance.
Bound Class References (since 1.1)

You can get the reference to a class of a speci c object with the same ::class syntax by using
the object as a receiver:
val widget: Widget = ...

assert(widget is GoodWidget) { "Bad widget: ${widget::class.qualifiedName}" }
You obtain the reference to an exact class of an object, for instance GoodWidget or
BadWidget , despite the type of the receiver expression ( Widget ).
Callable references
References to functions, properties, and constructors, apart from introspecting the program
structure, can also be called or used as instances of function types.
The common supertype for all callable references is KCallable<out R>, where R is the return
value type, which is the property type for properties, and the constructed type for constructors.
Function References
When we have a named function declared like this:
284
fun isOdd(x: Int) = x % 2 != 0
We can easily call it directly ( isOdd(5) ), but we can also use it as a function type value, e.g. pass
it to another function. To do this, we use the :: operator:
val numbers = listOf(1, 2, 3)

println(numbers.filter(::isOdd))
Here ::isOdd is a value of function type (Int) -> Boolean .
Function references belong to one of the KFunction<out R> subtypes, depending on the
parameter count, e.g. KFunction3<T1, T2, T3, R> .
:: can be used with overloaded functions when the expected type is known from the context.
For example:
fun isOdd(x: Int) = x % 2 != 0

fun isOdd(s: String) = s == "brillig" || s == "slithy" || s == "tove"
val numbers = listOf(1, 2, 3)

println(numbers.filter(::isOdd)) // refers to isOdd(x: Int)
Alternatively, you can provide the necessary context by storing the method reference in a
variable with an explicitly speci ed type:
val predicate: (String) -> Boolean = ::isOdd // refers to isOdd(x: String)
If we need to use a member of a class, or an extension function, it needs to be quali ed, e.g.
String::toCharArray .
Note that even if you initialize a variable with a reference to an extension function, the inferred
function type will have no receiver (it will have an additional parameter accepting a receiver
object). To have a function type with receiver instead, specify the type explicitly:
val isEmptyStringList: List<String>.() -> Boolean = List<String>::isEmpty
Example: Function Composition
Consider the following function:
fun <A, B, C> compose(f: (B) -> C, g: (A) -> B): (A) -> C {
return { x -> f(g(x)) }
}
It returns a composition of two functions passed to it: compose(f, g) = f(g(*)) . Now, you
can apply it to callable references:
285
fun length(s: String) = s.length
val oddLength = compose(::isOdd, ::length)

val strings = listOf("a", "ab", "abc")
println(strings.filter(oddLength))
Property References
To access properties as rst-class objects in Kotlin, we can also use the :: operator:
val x = 1
fun main() {
println(::x.get())
println(::x.name)
}
The expression ::x evaluates to a property object of type KProperty<Int> , which allows us
to read its value using get() or retrieve the property name using the name property. For more
information, please refer to the docs on the KProperty class.
For a mutable property, e.g. var y = 1 , ::y returns a value of type

KMutableProperty<Int>, which has a set() method:
var y = 1
fun main() {
::y.set(2)
println(y)
}
A property reference can be used where a function with one parameter is expected:
val strs = listOf("a", "bc", "def")

println(strs.map(String::length))
To access a property that is a member of a class, we qualify it:
class A(val p: Int)

val prop = A::p
println(prop.get(A(1)))
For an extension property:
val String.lastChar: Char

get() = this[length - 1]
fun main() {
println(String::lastChar.get("abc"))
}
Interoperability With Java Re ection
286
On the Java platform, standard library contains extensions for re ection classes that provide a
mapping to and from Java re ection objects (see package kotlin.reflect.jvm ). For example,
to nd a backing eld or a Java method that serves as a getter for a Kotlin property, you can say
something like this:
import kotlin.reflect.jvm.*
class A(val p: Int)
fun main() {
println(A::p.javaGetter) // prints "public final int A.getP()"
println(A::p.javaField) // prints "private final int A.p"
}
To get the Kotlin class corresponding to a Java class, use the .kotlin extension property:
fun getKClass(o: Any): KClass<Any> = o.javaClass.kotlin
Constructor References
Constructors can be referenced just like methods and properties. They can be used wherever an
object of function type is expected that takes the same parameters as the constructor and
returns an object of the appropriate type. Constructors are referenced by using the :: operator
and adding the class name. Consider the following function that expects a function parameter
with no parameters and return type Foo :
class Foo
fun function(factory: () -> Foo) {

val x: Foo = factory()
}
Using ::Foo , the zero-argument constructor of the class Foo, we can simply call it like this:
function(::Foo)
Callable references to constructors are typed as one of the KFunction<out R> subtypes ,
depending on the parameter count.
Bound Function and Property References (since 1.1)

You can refer to an instance method of a particular object:

println(numberRegex.matches("29"))
val isNumber = numberRegex::matches

println(isNumber("29"))
287
Instead of calling the method matches directly we are storing a reference to it. Such reference
is bound to its receiver. It can be called directly (like in the example above) or used whenever an
expression of function type is expected:

val strings = listOf("abc", "124", "a70")
println(strings.filter(numberRegex::matches))
Compare the types of bound and the corresponding unbound references. Bound callable
reference has its receiver "attached" to it, so the type of the receiver is no longer a parameter:
val isNumber: (CharSequence) -> Boolean = numberRegex::matches
val matches: (Regex, CharSequence) -> Boolean = Regex::matches
Property reference can be bound as well:
val prop = "abc"::length

println(prop.get())
Since Kotlin 1.2, explicitly specifying this as the receiver is not necessary: this::foo and
::foo are equivalent.
Bound constructor references
A bound callable reference to a constructor of an inner class can be obtained by providing an

instance of the outer class:
class Outer {
inner class Inner
}
val o = Outer()
val boundInnerCtor = o::Inner
288
Scope Functions
The Kotlin standard library contains several functions whose sole purpose is to execute a block of
code within the context of an object. When you call such a function on an object with a lambda
expression provided, it forms a temporary scope. In this scope, you can access the object without
its name. Such functions are called scope functions. There are ve of them: let , run , with ,
apply , and also .
Basically, these functions do the same: execute a block of code on an object. What's di erent is
how this object becomes available inside the block and what is the result of the whole
expression.
Here's a typical usage of a scope function:
Person("Alice", 20, "Amsterdam").let {

println(it)
it.moveTo("London")
it.incrementAge()
println(it)
}
If you write the same without let , you'll have to introduce a new variable and repeat its name
whenever you use it.
val alice = Person("Alice", 20, "Amsterdam")

println(alice)
alice.moveTo("London")
alice.incrementAge()
println(alice)
The scope functions do not introduce any new technical capabilities, but they can make your code
more concise and readable.
Due to the similar nature of scope functions, choosing the right one for your case can be a bit
tricky. The choice mainly depends on your intent and the consistency of use in your project.
Below we'll provide detailed descriptions of the distinctions between scope functions and the
conventions on their usage.
Distinctions
Because the scope functions are all quite similar in nature, it's important to understand the
di erences between them. There are two main di erences between each scope function:
— The way to refer to the context object

— The return value.
Context object: this or it
289
Inside the lambda of a scope function, the context object is available by a short reference instead
of its actual name. Each scope function uses one of two ways to access the context object: as a
lambda receiver ( this ) or as a lambda argument ( it ). Both provide the same capabilities, so
we'll describe the pros and cons of each for di erent cases and provide recommendations on
their use.
fun main() {
val str = "Hello"
// this
str.run {
println("The receiver string length: $length")
//println("The receiver string length: ${this.length}") // does the same
}
// it
str.let {
println("The receiver string's length is ${it.length}")
}
}
this
run , with , and apply refer to the context object as a lambda receiver - by keyword this .
Hence, in their lambdas, the object is available as it would be in ordinary class functions. In most
cases, you can omit this when accessing the members of the receiver object, making the code
shorter. On the other hand, if this is omitted, it can be hard to distinguish between the
receiver members and external objects or functions. So, having the context object as a receiver
( this ) is recommended for lambdas that mainly operate on the object members: call its
functions or assign properties.
val adam = Person("Adam").apply {

age = 20 // same as this.age = 20 or adam.age = 20
city = "London"
}
it
In turn, let and also have the context object as a lambda argument. If the argument name is
not speci ed, the object is accessed by the implicit default name it . it is shorter than this
and expressions with it are usually easier for reading. However, when calling the object
functions or properties you don't have the object available implicitly like this . Hence, having
the context object as it is better when the object is mostly used as an argument in function
calls. it is also better if you use multiple variables in the code block.
290
fun getRandomInt(): Int {
return Random.nextInt(100).also {
writeToLog("getRandomInt() generated value $it")
}
}
val i = getRandomInt()
Additionally, when you pass the context object as an argument, you can provide a custom name
for the context object inside the scope.

return Random.nextInt(100).also { value ->
writeToLog("getRandomInt() generated value $value")
}
}
Return value
The scope functions di er by the result they return:
— apply and also return the context object.
— let , run , and with return the lambda result.
These two options let you choose the proper function depending on what you do next in your
code.
Context object
The return value of apply and also is the context object itself. Hence, they can be included
into call chains as side steps: you can continue chaining function calls on the same object after
them.
val numberList = mutableListOf<Double>()

numberList.also { println("Populating the list") }
.apply {
add(2.71)
add(3.14)
add(1.0)
}
.also { println("Sorting the list") }
.sort()
They also can be used in return statements of functions returning the context object.

return Random.nextInt(100).also {
writeToLog("getRandomInt() generated value $it")
}
}
291
Lambda result
let , run , and with return the lambda result. So, you can use them when assigning the result
to a variable, chaining operations on the result, and so on.
val numbers = mutableListOf("one", "two", "three")

val countEndsWithE = numbers.run {
add("four")
add("five")
count { it.endsWith("e") }
}
println("There are $countEndsWithE elements that end with e.")
Additionally, you can ignore the return value and use a scope function to create a temporary
scope for variables.

with(numbers) {
val firstItem = first()
val lastItem = last()
println("First item: $firstItem, last item: $lastItem")
}
Functions
To help you choose the right scope function for your case, we'll describe them in detail and
provide usage recommendations. Technically, functions are interchangeable in many cases, so
the examples show the conventions that de ne the common usage style.
let
The context object is available as an argument ( it ). The return value is the lambda result.
let can be used to invoke one or more functions on results of call chains. For example, the
following code prints the results of two operations on a collection:
val numbers = mutableListOf("one", "two", "three", "four", "five")

val resultList = numbers.map { it.length }.filter { it > 3 }
println(resultList)
With let , you can rewrite it:

numbers.map { it.length }.filter { it > 3 }.let {
println(it)
// and more function calls if needed
}
If the code block contains a single function with it as an argument, you can use the method
reference ( :: ) instead of the lambda:

numbers.map { it.length }.filter { it > 3 }.let(::println)
292
let is often used for executing a code block only with non-null values. To perform actions on a
non-null object, use the safe call operator ?. on it and call let with the actions in its lambda.
val str: String? = "Hello"

//processNonNullString(str) // compilation error: str can be null
val length = str?.let {
println("let() called on $it")
processNonNullString(it) // OK: 'it' is not null inside '?.let { }'
it.length
}
Another case for using let is introducing local variables with a limited scope for improving code
readability. To de ne a new variable for the context object, provide its name as the lambda
argument so that it can be used instead of the default it .

val modifiedFirstItem = numbers.first().let { firstItem ->
println("The first item of the list is '$firstItem'")
if (firstItem.length >= 5) firstItem else "!" + firstItem + "!"
}.toUpperCase()
println("First item after modifications: '$modifiedFirstItem'")
with
A non-extension function: the context object is passed as an argument, but inside the lambda,
it's available as a receiver ( this ). The return value is the lambda result.
We recommend with for calling functions on the context object without providing the lambda
result. In the code, with can be read as “with this object, do the following.”

with(numbers) {
println("'with' is called with argument $this")
println("It contains $size elements")
}
Another use case for with is introducing a helper object whose properties or functions will be
used for calculating a value.

val firstAndLast = with(numbers) {
"The first element is ${first()}," +
" the last element is ${last()}"
}
println(firstAndLast)
run
The context object is available as a receiver ( this ). The return value is the lambda result.
293
run does the same as with but invokes as let - as an extension function of the context
object.
run is useful when your lambda contains both the object initialization and the computation of
the return value.
val service = MultiportService("https://example.kotlinlang.org", 80)
val result = service.run {

port = 8080
query(prepareRequest() + " to port $port")
}
// the same code written with let() function:

val letResult = service.let {
it.port = 8080
it.query(it.prepareRequest() + " to port ${it.port}")
}
Besides calling run on a receiver object, you can use it as a non-extension function. Non-
extension run lets you execute a block of several statements where an expression is required.
val hexNumberRegex = run {

val digits = "0-9"
val hexDigits = "A-Fa-f"
val sign = "+-"
Regex("[$sign]?[$digits$hexDigits]+")
}
for (match in hexNumberRegex.findAll("+1234 -FFFF not-a-number")) {

println(match.value)
}
apply
The context object is available as a receiver ( this ). The return value is the object itself.
Use apply for code blocks that don't return a value and mainly operate on the members of the
receiver object. The common case for apply is the object con guration. Such calls can be read
as “apply the following assignments to the object.”
val adam = Person("Adam").apply {

age = 32
city = "London"
}
Having the receiver as the return value, you can easily include apply into call chains for more
complex processing.
also
294
The context object is available as an argument ( it ). The return value is the object itself.
also is good for performing some actions that take the context object as an argument. Use
also for additional actions that don't alter the object, such as logging or printing debug
information. Usually, you can remove the calls of also from the call chain without breaking the
program logic.
When you see also in the code, you can read it as “and also do the following”.

numbers
.also { println("The list elements before adding new one: $it") }
.add("four")
Function selection
To help you choose the right scope function for your purpose, we provide the table of key
di erences between them.
Function Object reference Return value Is extension function

let it Lambda result Yes
run this Lambda result Yes
run - Lambda result No: called without the context object
with this Lambda result No: takes the context object as an argument.
apply this Context object Yes
also it Context object Yes
Here is a short guide for choosing scope functions depending on the intended purpose:
— Executing a lambda on non-null objects: let
— Introducing an expression as a variable in local scope: let
— Object con guration: apply
— Object con guration and computing the result: run
— Running statements where an expression is required: non-extension run
— Additional e ects: also
— Grouping function calls on an object: with
The use cases of di erent functions overlap, so that you can choose the functions based on the
speci c conventions used in your project or team.
Although the scope functions are a way of making the code more concise, avoid overusing them:
it can decrease your code readability and lead to errors. Avoid nesting scope functions and be
careful when chaining them: it's easy to get confused about the current context object and the
value of this or it .
295
takeIf and takeUnless
In addition to scope functions, the standard library contains the functions takeIf and
takeUnless . These functions let you embed checks of the object state in call chains.
When called on an object with a predicate provided, takeIf returns this object if it matches the
predicate. Otherwise, it returns null . So, takeIf is a ltering function for a single object. In
turn, takeUnless returns the object if it doesn't match the predicate and null if it does. The
object is available as a lambda argument ( it ).
val number = Random.nextInt(100)
val evenOrNull = number.takeIf { it % 2 == 0 }

val oddOrNull = number.takeUnless { it % 2 == 0 }
println("even: $evenOrNull, odd: $oddOrNull")
When chaining other functions after takeIf and takeUnless , don't forget to perform the null
check or the safe call ( ?. ) because their return value is nullable.
val str = "Hello"

val caps = str.takeIf { it.isNotEmpty() }?.toUpperCase()
/val caps = str.takeIf { it.isNotEmpty() }.toUpperCase() //compilation error
println(caps)
takeIf and takeUnless are especially useful together with scope functions. A good case is
chaining them with let for running a code block on objects that match the given predicate. To
do this, call takeIf on the object and then call let with a safe call ( ? ). For objects that don't
match the predicate, takeIf returns null and let isn't invoked.
fun displaySubstringPosition(input: String, sub: String) {

input.indexOf(sub).takeIf { it >= 0 }?.let {
println("The substring $sub is found in $input.")
println("Its start position is $it.")
}
}
displaySubstringPosition("010000011", "11")
This is how the same function looks without the standard library functions:
fun displaySubstringPosition(input: String, sub: String) {

val index = input.indexOf(sub)
if (index >= 0) {
println("The substring $sub is found in $input.")
println("Its start position is $index.")
}
}
296
Type-Safe Builders
By using well-named functions as builders in combination with function literals with receiver it is
possible to create type-safe, statically-typed builders in Kotlin.
Type-safe builders allow creating Kotlin-based domain-speci c languages (DSLs) suitable for
building complex hierarchical data structures in a semi-declarative way. Some of the example use
cases for the builders are:
— Generating markup with Kotlin code, such as HTML or XML;

— Programmatically laying out UI components: Anko
— Con guring routes for a web server: Ktor.
A type-safe builder example

Consider the following code:
import com.example.html.* // see declarations below
fun result() =
html {
head {
title {+"XML encoding with Kotlin"}
}
body {
h1 {+"XML encoding with Kotlin"}
p {+"this format can be used as an alternative markup to XML"}
// an element with attributes and text content

a(href = "http://kotlinlang.org") {+"Kotlin"}
// mixed content
p {
+"This is some"
b {+"mixed"}
+"text. For more see the"
297
+"text. For more see the"
a(href = "http://kotlinlang.org") {+"Kotlin"}
+"project"
}
p {+"some text"}
// content generated by
p {
for (arg in args)
+arg
}
}
}
This is completely legitimate Kotlin code. You can play with this code online (modify it and run in
the browser) here.
How it works
Let's walk through the mechanisms of implementing type-safe builders in Kotlin. First of all, we
need to de ne the model we want to build, in this case we need to model HTML tags. It is easily
done with a bunch of classes. For example, HTML is a class that describes the <html> tag, i.e. it
de nes children like <head> and <body> . (See its declaration below.)
Now, let's recall why we can say something like this in the code:
html {
// ...
}
html is actually a function call that takes a lambda expression as an argument. This function is
de ned as follows:

val html = HTML()
html.init()
return html
}
This function takes one parameter named init , which is itself a function. The type of the
function is HTML.() -> Unit , which is a function type with receiver. This means that we need to
pass an instance of type HTML (a receiver) to the function, and we can call members of that
instance inside the function. The receiver can be accessed through the this keyword:
html {
this.head { ... }
this.body { ... }
}
( head and body are member functions of HTML .)
Now, this can be omitted, as usual, and we get something that looks very much like a builder
already:
298
html {
head { ... }
body { ... }
}
So, what does this call do? Let's look at the body of html function as de ned above. It creates a
new instance of HTML , then it initializes it by calling the function that is passed as an argument
(in our example this boils down to calling head and body on the HTML instance), and then it
returns this instance. This is exactly what a builder should do.
The head and body functions in the HTML class are de ned similarly to html . The only
di erence is that they add the built instances to the children collection of the enclosing HTML
instance:
fun head(init: Head.() -> Unit) : Head {

val head = Head()
head.init()
children.add(head)
return head
}
fun body(init: Body.() -> Unit) : Body {

val body = Body()
body.init()
children.add(body)
return body
}
Actually these two functions do just the same thing, so we can have a generic version, initTag :
protected fun <T : Element> initTag(tag: T, init: T.() -> Unit): T {

tag.init()
children.add(tag)
return tag
}
So, now our functions are very simple:
fun head(init: Head.() -> Unit) = initTag(Head(), init)
fun body(init: Body.() -> Unit) = initTag(Body(), init)
And we can use them to build <head> and <body> tags.
One other thing to be discussed here is how we add text to tag bodies. In the example above we
say something like:
html {
head {
title {+"XML encoding with Kotlin"}
}
// ...
}
299
So basically, we just put a string inside a tag body, but there is this little + in front of it, so it is a
function call that invokes a pre x unaryPlus() operation. That operation is actually de ned by
an extension function unaryPlus() that is a member of the TagWithText abstract class (a
parent of Title ):
operator fun String.unaryPlus() {

children.add(TextElement(this))
}
So, what the pre x + does here is wrapping a string into an instance of TextElement and
adding it to the children collection, so that it becomes a proper part of the tag tree.
All this is de ned in a package com.example.html that is imported at the top of the builder
example above. In the last section you can read through the full de nition of this package.
Scope control: @DslMarker (since 1.1)

When using DSLs, one might have come across the problem that too many functions can be
called in the context. We can call methods of every available implicit receiver inside a lambda and
therefore get an inconsistent result, like the tag head inside another head :
html {
head {
head {} // should be forbidden
}
// ...
}
In this example only members of the nearest implicit receiver this@head must be available;
head() is a member of the outer receiver this@html , so it must be illegal to call it.
To address this problem, in Kotlin 1.1 a special mechanism to control receiver scope was
introduced.
To make the compiler start controlling scopes we only have to annotate the types of all receivers
used in the DSL with the same marker annotation. For instance, for HTML Builders we declare an
annotation @HTMLTagMarker :
@DslMarker
annotation class HtmlTagMarker
An annotation class is called a DSL marker if it is annotated with the @DslMarker annotation.
In our DSL all the tag classes extend the same superclass Tag . It's enough to annotate only the
superclass with @HtmlTagMarker and after that the Kotlin compiler will treat all the inherited
classes as annotated:
@HtmlTagMarker
abstract class Tag(val name: String) { ... }
300
We don't have to annotate the HTML or Head classes with @HtmlTagMarker because their
superclass is already annotated:
class HTML() : Tag("html") { ... }

class Head() : Tag("head") { ... }
After we've added this annotation, the Kotlin compiler knows which implicit receivers are part of
the same DSL and allows to call members of the nearest receivers only:
html {
head {
head { } // error: a member of outer receiver
}
// ...
}
Note that it's still possible to call the members of the outer receiver, but to do that you have to
specify this receiver explicitly:
html {
head {
[email protected] { } // possible
}
// ...
}
Full de nition of the com.example.html package

This is how the package com.example.html is de ned (only the elements used in the example
above). It builds an HTML tree. It makes heavy use of extension functions and lambdas with
receiver.
Note that the @DslMarker annotation is available only since Kotlin 1.1.
package com.example.html
interface Element {
fun render(builder: StringBuilder, indent: String)
}
class TextElement(val text: String) : Element {

override fun render(builder: StringBuilder, indent: String) {
builder.append("$indent$text\n")
}
}
@DslMarker
annotation class HtmlTagMarker
@HtmlTagMarker
abstract class Tag(val name: String) : Element {
val children = arrayListOf<Element>()
val attributes = hashMapOf<String, String>()
301
protected fun <T : Element> initTag(tag: T, init: T.() -> Unit): T {
tag.init()
children.add(tag)
return tag
}
override fun render(builder: StringBuilder, indent: String) {

builder.append("$indent<$name${renderAttributes()}>\n")
for (c in children) {
c.render(builder, indent + " ")
}
builder.append("$indent</$name>\n")
}
private fun renderAttributes(): String {

val builder = StringBuilder()
for ((attr, value) in attributes) {
builder.append(" $attr=\"$value\"")
}
return builder.toString()
}
override fun toString(): String {

val builder = StringBuilder()
render(builder, "")
return builder.toString()
}
}
abstract class TagWithText(name: String) : Tag(name) {

operator fun String.unaryPlus() {
children.add(TextElement(this))
}
}
class HTML : TagWithText("html") {

fun head(init: Head.() -> Unit) = initTag(Head(), init)
fun body(init: Body.() -> Unit) = initTag(Body(), init)

}
class Head : TagWithText("head") {

fun title(init: Title.() -> Unit) = initTag(Title(), init)
}
class Title : TagWithText("title")
abstract class BodyTag(name: String) : TagWithText(name) {

fun b(init: B.() -> Unit) = initTag(B(), init)
fun p(init: P.() -> Unit) = initTag(P(), init)
fun h1(init: H1.() -> Unit) = initTag(H1(), init)
fun a(href: String, init: A.() -> Unit) {
val a = initTag(A(), init)
a.href = href
}
}
class Body : BodyTag("body")

class B : BodyTag("b")
class P : BodyTag("p")
class H1 : BodyTag("h1")
302
class H1 : BodyTag("h1")
class A : BodyTag("a") {
var href: String
get() = attributes["href"]!!
set(value) {
attributes["href"] = value
}
}

val html = HTML()
html.init()
return html
}
303
Experimental API Markers
The annotations for marking and using experimental APIs (@Experimental and
@UseExperimental) are experimental in Kotlin 1.3. See details below.
The Kotlin standard library provides developers with a mechanism for creating and using
experimental APIs. This mechanism lets library authors inform users that certain components of
their API, such as classes or functions, are unstable and are likely to change in the future. Such
changes may require rewriting and recompiling the client code. To prevent potential
compatibility issues, the compiler warns users of the experimental status of such APIs and may
require them to give their explicit consent to use the API.
Using experimental APIs

If a class or a function from a library is marked by its author as experimental, using it in your
code will produce warnings or compilation errors unless you explicitly accept their experimental
status. There are several ways to accept the experimental status of API elements; all of them are
applicable without technical limitations. You are free to choose the way that you nd best for
your situation.
Propagating use
When you use an experimental API in the code intended for third-party use (a library), you can
mark your API as experimental as well. To do this, annotate your declaration with the
experimental marker annotation of the API used in its body. This enables you to use the API
elements annotated with this marker.
// library code
@Experimental
@Retention(AnnotationRetention.BINARY)
@Target(AnnotationTarget.CLASS, AnnotationTarget.FUNCTION)
annotation class ExperimentalDateTime // Experimental API marker
@ExperimentalDateTime
class DateProvider // Experimental class
304
// client code
fun getYear(): Int {
val dateProvider: DateProvider // error: DateProvider is experimental
// ...
}
fun getDate(): Date {
val dateProvider: DateProvider // OK: the function is marked as experimental
// ...
}
fun displayDate() {
println(getDate()) // error: getDate() is experimental, acceptance is required
}
As you can see in this example, the annotated function appears to be a part of the
@ExperimentalDateTime experimental API. So, the described way of acceptance propagates
the experimental status to the code that uses an experimental API; its clients will be required to
accept it as well. To use multiple experimental APIs, annotate the declaration with all their
markers.
Non-propagating use
In modules that don't provide their own API, such as application modules, you can use
experimental APIs without propagating the experimental status to your code. In this case, mark
your code with the @UseExperimental(Marker::class) annotation specifying the marker
annotation of the experimental API:
// library code
@Experimental
annotation class ExperimentalDateTime // Experimental API marker
class DateProvider // Experimental class
//client code
@UseExperimental(ExperimentalDateTime::class)
fun getDate(): Date { // uses DateProvider; doesn't expose the experimental
status
val dateProvider: DateProvider
// ...
}
fun displayDate() {
println(getDate()) // OK: getDate() is not experimental
}
When somebody calls the function getDate() , they won't be informed about the experimental
API used in its body.
305
To use an experimental API in all functions and classes in a le, add the le-level annotation
@file:UseExperimental to the top of the le before the package speci cation and imports.
//client code
@file:UseExperimental(ExperimentalDateTime::class)
Module-wide use
If you don't want to annotate every usage of experimental APIs in your code, you can accept the
experimental status for your whole module. Module-wide use of experimental APIs can be
propagating and non-propagating as well:
— To accept the experimental status without propagation, compile the module with the
argument -Xuse-experimental , specifying the fully quali ed name of the experimental
API marker you use: -Xuse-experimental=org.mylibrary.ExperimentalMarker .
Compiling with this argument has the same e ect as if every declaration in the module had
the annotation @UseExperimental(ExperimentalMarker::class) .
— To accept and propagate the experimental status to your whole module, compile the module
with the argument -Xexperimental=org.mylibrary.ExperimentalMarker . In this
case, every declaration in the module becomes experimental. The use of the module requires
the acceptance of its experimental status as well.
If you build your module with Gradle, you can add arguments like this:
compileKotlin {
kotlinOptions {
freeCompilerArgs += "-Xuse-experimental=org.mylibrary.ExperimentalMarker"
}
}
tasks.withType<KotlinCompile>().all {
kotlinOptions.freeCompilerArgs += "-Xuse-
experimental=org.mylibrary.ExperimentalMarker"
}
For Maven, it would be:
306
<build>
<plugins>
<plugin>
<groupId>org.jetbrains.kotlin</groupId>
<artifactId>kotlin-maven-plugin</artifactId>
<version>${kotlin.version}</version>
<executions>...</executions>
<configuration>
<args>
<arg>-Xuse-experimental=org.mylibrary.ExperimentalMarker</arg>
</args>
</configuration>
</plugin>
</plugins>
</build>
To accept the usage of multiple experimental APIs on the module level, add one of the described
arguments for each experimental API marker used in your module.
Marking experimental API
Creating marker annotations
If you want to declare your module's API as experimental, create an annotation class to use as its
experimental marker. This class must be annotated with @Experimental:
@Experimental
annotation class ExperimentalDateTime
Experimental marker annotations must meet several requirements:
— BINARY retention
— No EXPRESSION and FILE among targets
— No parameters.
A marker annotation can have one of two severity levels of informing about experimental API
usage:
— Experimental.Level.ERROR . Acceptance is mandatory. Otherwise, the code that uses

marked API won't compile. This level is used by default.
— Experimental.Level.WARNING . Acceptance is not mandatory, but advisable. Without it,
the compiler raises a warning. To set the desired level, specify the level parameter of the
@Experimental annotation.
@Experimental(level = Experimental.Level.WARNING)
307
If you publish several features in the experimental state, declare a marker for each. Separate
markers make the use of experimental features safer for your clients: they'll be able to use only
the features that they explicitly accept. This also lets you graduate the features to stable
independently.
Marking API elements
To mark an API element as experimental, annotate its declaration with your experimental marker
annotation:
class DateProvider
fun getTime(): Time {}
Module-wide markers
If you consider all the APIs of your module experimental, you can mark the entire module as such
with the compiler argument -Xexperimental as described in Module-wide use.
Graduation of experimental API

Once your experimental API graduates and is released in its nal state, remove its marker
annotation from declarations so that the clients can use it without restriction. However, you
should leave the marker classes in modules so that the existing client code remains compatible.
To let the API users update their modules accordingly (remove the markers from their code and
recompile), mark the annotations as @Deprecated and provide the explanation in its message.
@Deprecated("This experimental API marker is not used anymore. Remove its usages from
your code.")
@Experimental
Experimental status of experimental API markers

The described mechanism for marking and using experimental APIs is itself experimental in
Kotlin 1.3. This means that in future releases it may be changed in ways that make it
incompatible. To make the users of annotations @Experimental and UseExperimental
aware of their experimental status, the compiler raises warnings when compiling the code with
these annotations:
This class can only be used with the compiler argument '-Xuse-
experimental=kotlin.Experimental'
To remove the warnings, add the compiler argument -Xuse-

experimental=kotlin.Experimental .
308
Reference
Keywords and Operators
Hard Keywords
The following tokens are always interpreted as keywords and cannot be used as identi ers:
— as
— is used for type casts

— speci es an alias for an import
— as? is used for safe type casts
— break terminates the execution of a loop
— class declares a class
— continue proceeds to the next step of the nearest enclosing loop
— do begins a do/while loop (loop with postcondition)
— else de nes the branch of an if expression which is executed when the condition is false
— false speci es the 'false' value of the Boolean type
— for begins a for loop
— fun declares a function
— if begins an if expression
— in
— speci es the object being iterated in a for loop

— is used as an in x operator to check that a value belongs to a range, a collection or another
entity that de nes the 'contains' method
— is used in when expressions for the same purpose
— marks a type parameter as contravariant
— !in
is used as an operator to check that a value does NOT belong to a range, a collection or
—
another entity that de nes the 'contains' method
309
— interface declares an interface
— is
— checks that a value has a certain type

— !is
— checks that a value does NOT have a certain type

— null is a constant representing an object reference that doesn't point to any object
— object declares a class and its instance at the same time
— package speci es the package for the current le
— return returns from the nearest enclosing function or anonymous function
— super
— refers to the superclass implementation of a method or property

— calls the superclass constructor from a secondary constructor
— this
— refers to the current receiver

— calls another constructor of the same class from a secondary constructor
— throw throws an exception
— true speci es the 'true' value of the Boolean type
— try begins an exception handling block
— typealias declares a type alias
— val declares a read-only property or local variable
— var declares a mutable property or local variable
— when begins a when expression (executes one of the given branches)
— while begins a while loop (loop with precondition)
Soft Keywords
The following tokens act as keywords in the context when they are applicable and can be used as
identi ers in other contexts:
— by
— delegates the implementation of an interface to another object

— delegates the implementation of accessors for a property to another object
310
— catch begins a block that handles a speci c exception type
— constructor declares a primary or secondary constructor
— delegate is used as an annotation use-site target
— dynamic references a dynamic type in Kotlin/JS code
— field is used as an annotation use-site target
— file is used as an annotation use-site target
— finally begins a block that is always executed when a try block exits
— get
— declares the getter of a property

— is used as an annotation use-site target
— import imports a declaration from another package into the current le
— init begins an initializer block
— param is used as an annotation use-site target
— property is used as an annotation use-site target
— receiver is used as an annotation use-site target
— set
— declares the setter of a property

— is used as an annotation use-site target
— setparam is used as an annotation use-site target
— where speci es constraints for a generic type parameter
Modi er Keywords
The following tokens act as keywords in modi er lists of declarations and can be used as
identi ers in other contexts:
— actual denotes a platform-speci c implementation in multiplatform projects
— abstract marks a class or member as abstract
— annotation declares an annotation class
— companion declares a companion object
— const marks a property as a compile-time constant
— crossinline forbids non-local returns in a lambda passed to an inline function
— data instructs the compiler to generate canonical members for a class
— enum declares an enumeration
— expect marks a declaration as platform-speci c, expecting an implementation in platform
311
modules.
— external marks a declaration as implemented not in Kotlin (accessible through JNI or in
JavaScript)
— final forbids overriding a member
— infix allows calling a function in in x notation
— inline tells the compiler to inline the function and the lambdas passed to it at the call site
— inner allows referring to the outer class instance from a nested class
— internal marks a declaration as visible in the current module
— lateinit allows initializing a non-null property outside of a constructor
— noinline turns o inlining of a lambda passed to an inline function

— open allows subclassing a class or overriding a member
— operator marks a function as overloading an operator or implementing a convention
— out marks a type parameter as covariant
— override marks a member as an override of a superclass member
private marks a declaration as visible in the current class or le

—
— protected marks a declaration as visible in the current class and its subclasses
— public marks a declaration as visible anywhere
— reified marks a type parameter of an inline function as accessible at runtime
— sealed declares a sealed class (a class with restricted subclassing)
— suspend marks a function or lambda as suspending (usable as a coroutine)
— tailrec marks a function as tail-recursive (allowing the compiler to replace recursion with
iteration)
— vararg allows passing a variable number of arguments for a parameter
Special Identi ers

The following identi ers are de ned by the compiler in speci c contexts and can be used as
regular identi ers in other contexts:
— field is used inside a property accessor to refer to the backing eld of the property
— it is used inside a lambda to refer to its parameter implicitly
Operators and Special Symbols

Kotlin supports the following operators and special symbols:
— + , - , * , / , % - mathematical operators
— * is also used to pass an array to a vararg parameter
312
— =
— assignment operator
— is used to specify default values for parameters
— += , -= , *= , /= , %= - augmented assignment operators
— ++ , -- - increment and decrement operators
— && , || , ! - logical 'and', 'or', 'not' operators (for bitwise operations, use corresponding in x
functions)
— == , != - equality operators (translated to calls of equals() for non-primitive types)
— === , !== - referential equality operators
— < , > , <= , >= - comparison operators (translated to calls of compareTo() for non-primitive
types)
— [ , ] - indexed access operator (translated to calls of get and set )
— !! asserts that an expression is non-null
— ?. performs a safe call (calls a method or accesses a property if the receiver is non-null)
— ?: takes the right-hand value if the left-hand value is null (the elvis operator)
— :: creates a member reference or a class reference
— .. creates a range
— : separates a name from a type in declarations
— ? marks a type as nullable
— ->
— separates the parameters and body of a lambda expression

— separates the parameters and return type declaration in a function type
— separates the condition and body of a when expression branch
— @
— introduces an annotation
— introduces or references a loop label
— introduces or references a lambda label
— references a 'this' expression from an outer scope
— references an outer superclass
— ; separates multiple statements on the same line
— $ references a variable or expression in a string template
— _
313
— substitutes an unused parameter in a lambda expression
— substitutes an unused parameter in a destructuring declaration
314
Grammar
Description
Notation
The notation used on this page corresponds to the ANTLR 4 notation with a few exceptions for
better readability:
— omitted lexer rule actions and commands,

— omitted lexical modes.
Short description:
— operator | denotes alternative,
— operator * denotes iteration (zero or more),
— operator + denotes iteration (one or more),
— operator ? denotes option (zero or one),
— operator .. denotes range (from left to right),
— operator ~ denotes negation.
Grammar source les
Kotlin grammar source les (in ANTLR format) are located in the Kotlin speci cation repository:
— KotlinLexer.g4 describes lexical structure;

— UnicodeClasses.g4 describes the characters that can be used in identi ers (these rules are
omitted on this page for better readability);
— KotlinParser.g4 describes syntax.
The grammar on this page corresponds to the grammar les above.
Symbols and naming
Terminal symbol names start with an uppercase letter, e.g. Identi er.
Non-terminal symbol names start with a lowercase letter, e.g. kotlinFile.
Symbol de nitions may be documented with attributes:
— start attribute denotes a symbol that represents the whole source le (see kotlinFile and
script),
— helper attribute denotes a lexer fragment rule (used only inside other terminal symbols).
Also for better readability some simpli cations are made:
315
— lexer rules consisting of one string literal element are inlined to the use site,
— new line tokens are excluded (new lines are not allowed in some places, see source grammar
les for details).
Scope
The grammar corresponds to the latest stable version of the Kotlin compiler excluding lexer and
parser rules for experimental features that are disabled by default.
Syntax grammar
General
Relevant pages: Packages

start
kotlinFile
: shebangLine? fileAnnotation* packageHeader importList topLevelObject* EOF
;
start
script
: shebangLine? fileAnnotation* packageHeader importList (statement semi)* EOF
;
shebangLine
(used by kotlinFile, script)
: ShebangLine
;
fileAnnotation
: ('@' | AT_PRE_WS) 'file' ':' (('[' unescapedAnnotation+ ']') | unescapedAnnotation)
;
See Packages
packageHeader
: ('package' identifier semi?)?
;
See Imports
importList
: importHeader*
;
importHeader
(used by importList)
: 'import' identifier (('.' '*') | importAlias)? semi?
;
importAlias
(used by importHeader)
: 'as' simpleIdentifier
;
topLevelObject
(used by kotlinFile)
: declaration semis?
;
typeAlias
(used by declaration)
: modifiers? 'typealias' simpleIdentifier typeParameters? '=' type
;
declaration
(used by topLevelObject, classMemberDeclaration, statement)
: classDeclaration
| objectDeclaration
| functionDeclaration
| propertyDeclaration
| typeAlias
;
316
Classes
See Classes and Inheritance

classDeclaration
: modifiers? ('class' | 'interface')
simpleIdentifier typeParameters?
primaryConstructor?
(':' delegationSpecifiers)?
typeConstraints?
(classBody | enumClassBody)?
;
primaryConstructor
(used by classDeclaration)
: (modifiers? 'constructor')? classParameters
;
classBody
(used by classDeclaration, companionObject, objectDeclaration, enumEntry, objectLiteral)
: '{' classMemberDeclarations '}'
;
classParameters
(used by primaryConstructor)
: '(' (classParameter (',' classParameter)*)? ')'
;
classParameter
(used by classParameters)
: modifiers? ('val' | 'var')? simpleIdentifier ':' type ('=' expression)?
;
delegationSpecifiers
(used by classDeclaration, companionObject, objectDeclaration, objectLiteral)
: annotatedDelegationSpecifier (',' annotatedDelegationSpecifier)*
;
delegationSpecifier
(used by annotatedDelegationSpecifier)
: constructorInvocation
| explicitDelegation
| userType
| functionType
;
constructorInvocation
(used by delegationSpecifier, unescapedAnnotation)
: userType valueArguments
;
annotatedDelegationSpecifier
(used by delegationSpecifiers)
: annotation* delegationSpecifier
;
explicitDelegation
(used by delegationSpecifier)
: (userType | functionType) 'by' expression
;
See Generic classes
typeParameters
(used by typeAlias, classDeclaration, functionDeclaration, propertyDeclaration)
: '<' typeParameter (',' typeParameter)* '>'
;
typeParameter
(used by typeParameters)
: typeParameterModifiers? simpleIdentifier (':' type)?
;
See Generic constraints
typeConstraints
(used by classDeclaration, functionDeclaration, propertyDeclaration, anonymousFunction)
: 'where' typeConstraint (',' typeConstraint)*
;
typeConstraint
(used by typeConstraints)
: annotation* simpleIdentifier ':' type
;
Class members
317
classMemberDeclarations
(used by classBody, enumClassBody)
: (classMemberDeclaration semis?)*
;
classMemberDeclaration
(used by classMemberDeclarations)
: declaration
| companionObject
| anonymousInitializer
| secondaryConstructor
;
anonymousInitializer
(used by classMemberDeclaration)
: 'init' block
;
companionObject
: modifiers? 'companion' 'object' simpleIdentifier?
(':' delegationSpecifiers)?
classBody?
;
functionValueParameters
(used by functionDeclaration, secondaryConstructor)
: '(' (functionValueParameter (',' functionValueParameter)*)? ')'
;
functionValueParameter
(used by functionValueParameters)
: parameterModifiers? parameter ('=' expression)?
;
functionDeclaration
: modifiers? 'fun' typeParameters?
(receiverType '.')?
simpleIdentifier functionValueParameters
(':' type)? typeConstraints?
functionBody?
;
functionBody
(used by functionDeclaration, getter, setter, anonymousFunction)
: block
| '=' expression
;
variableDeclaration
(used by multiVariableDeclaration, propertyDeclaration, forStatement, lambdaParameter, whenSubject)
: annotation* simpleIdentifier (':' type)?
;
multiVariableDeclaration
(used by propertyDeclaration, forStatement, lambdaParameter)
: '(' variableDeclaration (',' variableDeclaration)* ')'
;
See Properties and Fields
propertyDeclaration
: modifiers? ('val' | 'var') typeParameters?
(receiverType '.')?
(multiVariableDeclaration | variableDeclaration)
typeConstraints?
(('=' expression) | propertyDelegate)? ';'?
((getter? (semi? setter)?) | (setter? (semi? getter)?))
;
propertyDelegate
(used by propertyDeclaration)
: 'by' expression
;
getter
: modifiers? 'get'
| modifiers? 'get' '(' ')'
(':' type)?
functionBody
;
setter
: modifiers? 'set'
| modifiers? 'set' '(' parameterWithOptionalType ')' (':' type)?
functionBody
;
parametersWithOptionalType
(used by anonymousFunction)
318
: '(' (parameterWithOptionalType (',' parameterWithOptionalType)*)? ')'
;
parameterWithOptionalType
(used by setter, parametersWithOptionalType)
: parameterModifiers? simpleIdentifier (':' type)?
;
parameter
(used by functionValueParameter, functionTypeParameters)
: simpleIdentifier ':' type
;
See Object expressions and Declarations
objectDeclaration
: modifiers? 'object' simpleIdentifier (':' delegationSpecifiers)? classBody?
;
secondaryConstructor
: modifiers? 'constructor' functionValueParameters
(':' constructorDelegationCall)? block?
;
constructorDelegationCall
(used by secondaryConstructor)
: 'this' valueArguments
| 'super' valueArguments
;
Enum classes
See Enum classes

enumClassBody
(used by classDeclaration)
: '{' enumEntries? (';' classMemberDeclarations)? '}'
;
enumEntries
(used by enumClassBody)
: enumEntry (',' enumEntry)* ','?
;
enumEntry
(used by enumEntries)
: modifiers? simpleIdentifier valueArguments? classBody?
;
Types
See Types
type
(used by typeAlias, classParameter, typeParameter, typeConstraint, functionDeclaration, variableDeclaration,
getter, setter, parameterWithOptionalType, parameter, typeProjection, functionType, functionTypeParameters,
parenthesizedType, infixOperation, asExpression, lambdaParameter, anonymousFunction, superExpression,
typeTest, catchBlock)
: typeModifiers? (parenthesizedType | nullableType | typeReference | functionType)
;
typeReference
(used by type, nullableType, receiverType)
: userType
| 'dynamic'
;
nullableType
(used by type, receiverType)
: (typeReference | parenthesizedType) quest+
;
quest
(used by nullableType)
: '?'
| QUEST_WS
;
userType
(used by delegationSpecifier, constructorInvocation, explicitDelegation, typeReference,
parenthesizedUserType, unescapedAnnotation)
: simpleUserType ('.' simpleUserType)*
;
319
simpleUserType
(used by userType)
: simpleIdentifier typeArguments?
;
typeProjection
(used by typeArguments)
: typeProjectionModifiers? type
| '*'
;
typeProjectionModifiers
(used by typeProjection)
: typeProjectionModifier+
;
typeProjectionModifier
(used by typeProjectionModifiers)
: varianceModifier
| annotation
;
functionType
(used by delegationSpecifier, explicitDelegation, type)
: (receiverType '.')? functionTypeParameters '->' type
;
functionTypeParameters
(used by functionType)
: '(' (parameter | type)? (',' (parameter | type))* ')'
;
parenthesizedType
(used by type, nullableType, receiverType)
: '(' type ')'
;
receiverType
(used by functionDeclaration, propertyDeclaration, functionType, callableReference)
: typeModifiers? (parenthesizedType | nullableType | typeReference)
;
parenthesizedUserType
(used by parenthesizedUserType)
: '(' userType ')'
| '(' parenthesizedUserType ')'
;
Statements
statements
(used by block, lambdaLiteral)
: (statement (semis statement)* semis?)?
;
statement
(used by script, statements, controlStructureBody)
: (label | annotation)* (declaration | assignment | loopStatement | expression)
;
See Returns and jumps
label
(used by statement, unaryPrefix, annotatedLambda)
: simpleIdentifier ('@' | AT_POST_WS)
;
controlStructureBody
(used by forStatement, whileStatement, doWhileStatement, ifExpression, whenEntry)
: block
| statement
;
block
(used by anonymousInitializer, functionBody, secondaryConstructor, controlStructureBody, tryExpression,
catchBlock, finallyBlock)
: '{' statements '}'
;
loopStatement
(used by statement)
: forStatement
| whileStatement
| doWhileStatement
;
forStatement
(used by loopStatement)
: 'for'
'(' annotation* (variableDeclaration | multiVariableDeclaration) 'in' expression ')'
controlStructureBody?
;
320
whileStatement
: 'while' '(' expression ')' controlStructureBody
| 'while' '(' expression ')' ';'
;
doWhileStatement
: 'do' controlStructureBody? 'while' '(' expression ')'
;
assignment
(used by statement)
: directlyAssignableExpression '=' expression
| assignableExpression assignmentAndOperator expression
;
semi
(used by script, packageHeader, importHeader, propertyDeclaration, whenEntry)
: EOF
;
semis
(used by topLevelObject, classMemberDeclarations, statements)
: EOF
;
Expressions
Precedence Title Symbols
Highest Postfix ++, --, ., ?., ?
Prefix -, +, ++, --, !, label
Type RHS :, as, as?
Multiplicative *, /, %
Additive +, -
Range ..
Infix function simpleIdentifier
Elvis ?:
Named checks in, !in, is, !is
Comparison <, >, <=, >=
Equality ==, !==
Conjunction &&
Disjunction ||
Lowest Assignment =, +=, -=, *=, /=, %=
expression
(used by classParameter, explicitDelegation, functionValueParameter, functionBody, propertyDeclaration,
propertyDelegate, statement, forStatement, whileStatement, doWhileStatement, assignment, indexingSuffix,
valueArgument, parenthesizedExpression, collectionLiteral, lineStringExpression, multiLineStringExpression,
ifExpression, whenSubject, whenCondition, rangeTest, jumpExpression)
: disjunction
;
disjunction
(used by expression)
: conjunction ('||' conjunction)*
;
conjunction
(used by disjunction)
: equality ('&&' equality)*
;
equality
(used by conjunction)
: comparison (equalityOperator comparison)*
;
comparison
(used by equality)
: infixOperation (comparisonOperator infixOperation)?
;
infixOperation
(used by comparison)
321
: elvisExpression ((inOperator elvisExpression) | (isOperator type))*
;
elvisExpression
(used by infixOperation)
: infixFunctionCall (elvis infixFunctionCall)*
;
elvis
(used by elvisExpression)
: '?' ':'
;
infixFunctionCall
(used by elvisExpression)
: rangeExpression (simpleIdentifier rangeExpression)*
;
rangeExpression
(used by infixFunctionCall)
: additiveExpression ('..' additiveExpression)*
;
additiveExpression
(used by rangeExpression)
: multiplicativeExpression (additiveOperator multiplicativeExpression)*
;
multiplicativeExpression
(used by additiveExpression)
: asExpression (multiplicativeOperator asExpression)*
;
asExpression
(used by multiplicativeExpression)
: prefixUnaryExpression (asOperator type)?
;
prefixUnaryExpression
(used by asExpression, assignableExpression)
: unaryPrefix* postfixUnaryExpression
;
unaryPrefix
(used by prefixUnaryExpression)
: annotation
| label
| prefixUnaryOperator
;
postfixUnaryExpression
(used by prefixUnaryExpression, directlyAssignableExpression)
: primaryExpression
| primaryExpression postfixUnarySuffix+
;
postfixUnarySuffix
(used by postfixUnaryExpression)
: postfixUnaryOperator
| typeArguments
| callSuffix
| indexingSuffix
| navigationSuffix
;
directlyAssignableExpression
(used by assignment, parenthesizedDirectlyAssignableExpression)
: postfixUnaryExpression assignableSuffix
| simpleIdentifier
| parenthesizedDirectlyAssignableExpression
;
parenthesizedDirectlyAssignableExpression
(used by directlyAssignableExpression)
: '(' directlyAssignableExpression ')'
;
assignableExpression
(used by assignment, parenthesizedAssignableExpression)
: prefixUnaryExpression
| parenthesizedAssignableExpression
;
parenthesizedAssignableExpression
(used by assignableExpression)
: '(' assignableExpression ')'
;
assignableSuffix
(used by directlyAssignableExpression)
: typeArguments
| indexingSuffix
| navigationSuffix
;
indexingSuffix
(used by postfixUnarySuffix, assignableSuffix)
: '[' expression (',' expression)* ']'
322
;
navigationSuffix
(used by postfixUnarySuffix, assignableSuffix)
: memberAccessOperator (simpleIdentifier | parenthesizedExpression | 'class')
;
callSuffix
(used by postfixUnarySuffix)
: typeArguments? valueArguments? annotatedLambda
| typeArguments? valueArguments
;
annotatedLambda
(used by callSuffix)
: annotation* label? lambdaLiteral
;
typeArguments
(used by simpleUserType, postfixUnarySuffix, assignableSuffix, callSuffix)
: '<' typeProjection (',' typeProjection)* '>'
;
valueArguments
(used by constructorInvocation, constructorDelegationCall, enumEntry, callSuffix)
: '(' ')'
| '(' valueArgument (',' valueArgument)* ')'
;
valueArgument
(used by valueArguments)
: annotation? (simpleIdentifier '=')? '*'? expression
;
primaryExpression
(used by postfixUnaryExpression)
: parenthesizedExpression
| simpleIdentifier
| literalConstant
| stringLiteral
| callableReference
| functionLiteral
| objectLiteral
| collectionLiteral
| thisExpression
| superExpression
| ifExpression
| whenExpression
| tryExpression
| jumpExpression
;
parenthesizedExpression
(used by navigationSuffix, primaryExpression)
: '(' expression ')'
;
collectionLiteral
(used by primaryExpression)
: '[' expression (',' expression)* ']'
| '[' ']'
;
literalConstant
: BooleanLiteral
| IntegerLiteral
| HexLiteral
| BinLiteral
| CharacterLiteral
| RealLiteral
| 'null'
| LongLiteral
| UnsignedLiteral
;
stringLiteral
: lineStringLiteral
| multiLineStringLiteral
;
lineStringLiteral
(used by stringLiteral)
: '"' (lineStringContent | lineStringExpression)* '"'
;
multiLineStringLiteral
(used by stringLiteral)
: '"""' (multiLineStringContent | multiLineStringExpression | '"')*
TRIPLE_QUOTE_CLOSE
;
lineStringContent
(used by lineStringLiteral)
323
: LineStrText
| LineStrEscapedChar
| LineStrRef
;
lineStringExpression
(used by lineStringLiteral)
: '${' expression '}'
;
multiLineStringContent
(used by multiLineStringLiteral)
: MultiLineStrText
| '"'
| MultiLineStrRef
;
multiLineStringExpression
: '${' expression '}'
;
lambdaLiteral
(used by annotatedLambda, functionLiteral)
: '{' statements '}'
| '{' lambdaParameters? '->' statements '}'
;
lambdaParameters
(used by lambdaLiteral)
: lambdaParameter (',' lambdaParameter)*
;
lambdaParameter
(used by lambdaParameters)
: variableDeclaration
| multiVariableDeclaration (':' type)?
;
anonymousFunction
(used by functionLiteral)
: 'fun' (type '.')? parametersWithOptionalType
(':' type)? typeConstraints?
functionBody?
;
functionLiteral
: lambdaLiteral
| anonymousFunction
;
objectLiteral
: 'object' ':' delegationSpecifiers classBody
| 'object' classBody
;
thisExpression
: 'this'
| THIS_AT
;
superExpression
: 'super' ('<' type '>')? ('@' simpleIdentifier)?
| SUPER_AT
;
ifExpression
: 'if' '(' expression ')'
(controlStructureBody | ';')
| 'if' '(' expression ')'
controlStructureBody? ';'? 'else' (controlStructureBody | ';')
;
whenSubject
(used by whenExpression)
: '(' (annotation* 'val' variableDeclaration '=')? expression ')'
;
whenExpression
: 'when' whenSubject? '{' whenEntry* '}'
;
whenEntry
(used by whenExpression)
: whenCondition (',' whenCondition)* '->' controlStructureBody semi?
| 'else' '->' controlStructureBody semi?
;
whenCondition
(used by whenEntry)
: expression
324
| rangeTest
| typeTest
;
rangeTest
(used by whenCondition)
: inOperator expression
;
typeTest
(used by whenCondition)
: isOperator type
;
tryExpression
: 'try' block ((catchBlock+ finallyBlock?) | finallyBlock)
;
catchBlock
(used by tryExpression)
: 'catch' '(' annotation* simpleIdentifier ':' type ')' block
;
finallyBlock
(used by tryExpression)
: 'finally' block
;
jumpExpression
: 'throw' expression
| ('return' | RETURN_AT) expression?
| 'continue'
| CONTINUE_AT
| 'break'
| BREAK_AT
;
callableReference
: (receiverType? '::' (simpleIdentifier | 'class'))
;
assignmentAndOperator
(used by assignment)
: '+='
| '-='
| '*='
| '/='
| '%='
;
equalityOperator
(used by equality)
: '!='
| '!=='
| '=='
| '==='
;
comparisonOperator
(used by comparison)
: '<'
| '>'
| '<='
| '>='
;
inOperator
(used by infixOperation, rangeTest)
: 'in'
| NOT_IN
;
isOperator
(used by infixOperation, typeTest)
: 'is'
| NOT_IS
;
additiveOperator
(used by additiveExpression)
: '+'
| '-'
;
multiplicativeOperator
(used by multiplicativeExpression)
: '*'
| '/'
| '%'
;
asOperator
(used by asExpression)
325
: 'as'
| 'as?'
;
prefixUnaryOperator
(used by unaryPrefix)
: '++'
| '--'
| '-'
| '+'
| excl
;
postfixUnaryOperator
(used by postfixUnarySuffix)
: '++'
| '--'
| '!' excl
;
excl
(used by prefixUnaryOperator, postfixUnaryOperator)
: '!'
| EXCL_WS
;
memberAccessOperator
(used by navigationSuffix)
: '.'
| safeNav
| '::'
;
safeNav
(used by memberAccessOperator)
: '?' '.'
;
Modi ers
modifiers
(used by typeAlias, classDeclaration, primaryConstructor, classParameter, companionObject,
functionDeclaration, propertyDeclaration, getter, setter, objectDeclaration, secondaryConstructor,
enumEntry)
: annotation
| modifier+
;
parameterModifiers
(used by functionValueParameter, parameterWithOptionalType)
: annotation
| parameterModifier+
;
modifier
(used by modifiers)
: classModifier
| memberModifier
| visibilityModifier
| functionModifier
| propertyModifier
| inheritanceModifier
| parameterModifier
| platformModifier
;
typeModifiers
(used by type, receiverType)
: typeModifier+
;
typeModifier
(used by typeModifiers)
: annotation
| 'suspend'
;
classModifier
(used by modifier)
: 'enum'
| 'sealed'
| 'annotation'
| 'data'
| 'inner'
;
memberModifier
(used by modifier)
: 'override'
| 'lateinit'
326
;
visibilityModifier
(used by modifier)
: 'public'
| 'private'
| 'internal'
| 'protected'
;
varianceModifier
(used by typeProjectionModifier, typeParameterModifier)
: 'in'
| 'out'
;
typeParameterModifiers
(used by typeParameter)
: typeParameterModifier+
;
typeParameterModifier
(used by typeParameterModifiers)
: reificationModifier
| varianceModifier
| annotation
;
functionModifier
(used by modifier)
: 'tailrec'
| 'operator'
| 'infix'
| 'inline'
| 'external'
| 'suspend'
;
propertyModifier
(used by modifier)
: 'const'
;
inheritanceModifier
(used by modifier)
: 'abstract'
| 'final'
| 'open'
;
parameterModifier
(used by parameterModifiers, modifier)
: 'vararg'
| 'noinline'
| 'crossinline'
;
reificationModifier
(used by typeParameterModifier)
: 'reified'
;
platformModifier
(used by modifier)
: 'expect'
| 'actual'
;
Annotations
annotation
(used by annotatedDelegationSpecifier, typeConstraint, variableDeclaration, typeProjectionModifier,
statement, forStatement, unaryPrefix, annotatedLambda, valueArgument, whenSubject, catchBlock,
modifiers, parameterModifiers, typeModifier, typeParameterModifier)
: singleAnnotation
| multiAnnotation
;
singleAnnotation
(used by annotation)
: annotationUseSiteTarget unescapedAnnotation
| ('@' | AT_PRE_WS) unescapedAnnotation
;
multiAnnotation
(used by annotation)
: annotationUseSiteTarget '[' unescapedAnnotation+ ']'
| ('@' | AT_PRE_WS) '[' unescapedAnnotation+ ']'
;
annotationUseSiteTarget
(used by singleAnnotation, multiAnnotation)
327
: ('@' | AT_PRE_WS)
('field' | 'property' | 'get' | 'set' | 'receiver' | 'param' | 'setparam' | 'delegate') ':'
;
unescapedAnnotation
(used by fileAnnotation, singleAnnotation, multiAnnotation)
: constructorInvocation
| userType
;
Identi ers
simpleIdentifier
(used by importAlias, typeAlias, classDeclaration, classParameter, typeParameter, typeConstraint,
companionObject, functionDeclaration, variableDeclaration, parameterWithOptionalType, parameter,
objectDeclaration, enumEntry, simpleUserType, label, infixFunctionCall, directlyAssignableExpression,
navigationSuffix, valueArgument, primaryExpression, superExpression, catchBlock, callableReference,
identifier)
: Identifier
| 'abstract'
| 'annotation'
| 'by'
| 'catch'
| 'companion'
| 'constructor'
| 'crossinline'
| 'data'
| 'dynamic'
| 'enum'
| 'external'
| 'final'
| 'finally'
| 'get'
| 'import'
| 'infix'
| 'init'
| 'inline'
| 'inner'
| 'internal'
| 'lateinit'
| 'noinline'
| 'open'
| 'operator'
| 'out'
| 'override'
| 'private'
| 'protected'
| 'public'
| 'reified'
| 'sealed'
| 'tailrec'
| 'set'
| 'vararg'
| 'where'
| 'field'
| 'property'
| 'receiver'
| 'param'
| 'setparam'
| 'delegate'
| 'file'
| 'expect'
| 'actual'
| 'const'
| 'suspend'
;
identifier
(used by packageHeader, importHeader)
: simpleIdentifier ('.' simpleIdentifier)*
;
Lexical grammar
General
ShebangLine
(used by shebangLine)
328
: '#!' ~[\r\n]*
;
DelimitedComment
(used by DelimitedComment, Hidden)
: ('/*' (DelimitedComment | .)*? '*/')
;
LineComment
(used by Hidden)
: ('//' ~[\r\n]*)
;
WS
(used by Hidden)
: [\u0020\u0009\u000C]
;
helper
Hidden
(used by EXCL_WS, AT_POST_WS, AT_PRE_WS, AT_BOTH_WS, QUEST_WS, NOT_IS, NOT_IN)
: DelimitedComment
| LineComment
| WS
;
Separators and operations

RESERVED
: '...'
;
EXCL_WS
(used by excl)
: '!' Hidden
;
DOUBLE_ARROW
: '=>'
;
DOUBLE_SEMICOLON
: ';;'
;
HASH
: '#'
;
AT_POST_WS
(used by label)
: '@' Hidden
;
AT_PRE_WS
(used by fileAnnotation, singleAnnotation, multiAnnotation, annotationUseSiteTarget)
: Hidden '@'
;
AT_BOTH_WS
: Hidden '@' Hidden
;
QUEST_WS
(used by quest)
: '?' Hidden
;
SINGLE_QUOTE
: '\''
;
Keywords
RETURN_AT
(used by jumpExpression)
: 'return@' Identifier
;
CONTINUE_AT
: 'continue@' Identifier
;
BREAK_AT
: 'break@' Identifier
;
THIS_AT
(used by thisExpression)
: 'this@' Identifier
;
329
SUPER_AT
(used by superExpression)
: 'super@' Identifier
;
TYPEOF
: 'typeof'
;
NOT_IS
(used by isOperator)
: '!is' Hidden
;
NOT_IN
(used by inOperator)
: '!in' Hidden
;
Literals
helper
DecDigit
(used by DecDigitOrSeparator, DecDigits, IntegerLiteral)
: '0'..'9'
;
helper
DecDigitNoZero
(used by IntegerLiteral)
: '1'..'9'
;
helper
DecDigitOrSeparator
(used by DecDigits, IntegerLiteral)
: DecDigit
| '_'
;
helper
DecDigits
(used by DoubleExponent, FloatLiteral, DoubleLiteral)
: DecDigit DecDigitOrSeparator* DecDigit
| DecDigit
;
helper
DoubleExponent
(used by DoubleLiteral)
: [eE] [+-]? DecDigits
;
RealLiteral
(used by literalConstant)
: FloatLiteral
| DoubleLiteral
;
FloatLiteral
(used by RealLiteral)
: DoubleLiteral [fF]
| DecDigits [fF]
;
DoubleLiteral
(used by RealLiteral, FloatLiteral)
: DecDigits? '.' DecDigits DoubleExponent?
| DecDigits DoubleExponent
;
IntegerLiteral
(used by literalConstant, UnsignedLiteral, LongLiteral)
: DecDigitNoZero DecDigitOrSeparator* DecDigit
| DecDigit
;
helper
HexDigit
(used by HexDigitOrSeparator, HexLiteral, UniCharacterLiteral)
: [0-9a-fA-F]
;
helper
HexDigitOrSeparator
(used by HexLiteral)
: HexDigit
| '_'
;
HexLiteral
: '0' [xX] HexDigit HexDigitOrSeparator* HexDigit
330
| '0' [xX] HexDigit
;
helper
BinDigit
(used by BinDigitOrSeparator, BinLiteral)
: [01]
;
helper
BinDigitOrSeparator
(used by BinLiteral)
: BinDigit
| '_'
;
BinLiteral
: '0' [bB] BinDigit BinDigitOrSeparator* BinDigit
| '0' [bB] BinDigit
;
UnsignedLiteral
: (IntegerLiteral | HexLiteral | BinLiteral) [uU] 'L'?
;
LongLiteral
: (IntegerLiteral | HexLiteral | BinLiteral) 'L'
;
BooleanLiteral
: 'true'
| 'false'
;
CharacterLiteral
: '\'' (EscapeSeq | ~[\n\r'\\]) '\''
;
Identi ers
helper
UnicodeDigit
(used by Identifier)
: UNICODE_CLASS_ND
;
Identifier
(used by simpleIdentifier, RETURN_AT, CONTINUE_AT, BREAK_AT, THIS_AT, SUPER_AT, IdentifierOrSoftKey)
: (Letter | '_') (Letter | '_' | UnicodeDigit)*
| '`' ~([\r\n] | '`')+ '`'
;
Depending on the target and publicity of the declaration, the set of allowed symbols in identi ers
is di erent. This rule contains the union of allowed symbols from all targets. Thus, the code for
any target can be parsed using the grammar.
The allowed symbols in identi ers corresponding to the target and publicity of the declaration
are given below.
Kotlin/JVM (any declaration publicity)
~ ( [\r\n] | '`' | '.' | ';' | ':' | '\' | '/' | '[' | ']' | '<' | '>' )
Kotlin/Android (any declaration publicity)
The allowed symbols are di erent from allowed symbols for Kotlin/JVM and correspond to the
Dalvik Executable format.
Kotlin/JS (private declarations)
~ ( [\r\n] | '`' )
Kotlin/JS (public declarations)
The allowed symbols for public declarations correspond to the ECMA speci cation (section 7.6)
except that ECMA reserved words is allowed.
Kotlin/Native (any declaration publicity)
~ ( [\r\n] | '`' )
331
IdentifierOrSoftKey
(used by FieldIdentifier)
: Identifier
| 'abstract'
| 'annotation'
| 'by'
| 'catch'
| 'companion'
| 'constructor'
| 'crossinline'
| 'data'
| 'dynamic'
| 'enum'
| 'external'
| 'final'
| 'finally'
| 'import'
| 'infix'
| 'init'
| 'inline'
| 'inner'
| 'internal'
| 'lateinit'
| 'noinline'
| 'open'
| 'operator'
| 'out'
| 'override'
| 'private'
| 'protected'
| 'public'
| 'reified'
| 'sealed'
| 'tailrec'
| 'vararg'
| 'where'
| 'get'
| 'set'
| 'field'
| 'property'
| 'receiver'
| 'param'
| 'setparam'
| 'delegate'
| 'file'
| 'expect'
| 'actual'
| 'const'
| 'suspend'
;
FieldIdentifier
(used by LineStrRef, MultiLineStrRef)
: '$' IdentifierOrSoftKey
;
helper
UniCharacterLiteral
(used by EscapeSeq, LineStrEscapedChar)
: '\\' 'u' HexDigit HexDigit HexDigit HexDigit
;
helper
EscapedIdentifier
(used by EscapeSeq, LineStrEscapedChar)
: '\\' ('t' | 'b' | 'r' | 'n' | '\'' | '"' | '\\' | '$')
;
helper
EscapeSeq
(used by CharacterLiteral)
: UniCharacterLiteral
| EscapedIdentifier
;
Characters
helper
Letter
(used by Identifier)
: UNICODE_CLASS_LL
| UNICODE_CLASS_LM
| UNICODE_CLASS_LO
332
| UNICODE_CLASS_LT
| UNICODE_CLASS_LU
| UNICODE_CLASS_NL
;
Strings
LineStrRef
(used by lineStringContent)
: FieldIdentifier
;
See String templates
LineStrText
: ~('\\' | '"' | '$')+
| '$'
;
LineStrEscapedChar
: EscapedIdentifier
| UniCharacterLiteral
;
TRIPLE_QUOTE_CLOSE
: ('"'? '"""')
;
MultiLineStrRef
(used by multiLineStringContent)
: FieldIdentifier
;
MultiLineStrText
(used by multiLineStringContent)
: ~('"' | '$')+
| '$'
;
ErrorCharacter
:.
;
333
Java Interop
Calling Java code from Kotlin
Kotlin is designed with Java Interoperability in mind. Existing Java code can be called from Kotlin
in a natural way, and Kotlin code can be used from Java rather smoothly as well. In this section we
describe some details about calling Java code from Kotlin.
Pretty much all Java code can be used without any issues:
import java.util.*
fun demo(source: List<Int>) {

val list = ArrayList<Int>()
// 'for'-loops work for Java collections:
for (item in source) {
list.add(item)
}
// Operator conventions work as well:
for (i in 0..source.size - 1) {
list[i] = source[i] // get and set are called
}
}
Getters and Setters

Methods that follow the Java conventions for getters and setters (no-argument methods with
names starting with get and single-argument methods with names starting with set ) are
represented as properties in Kotlin. Boolean accessor methods (where the name of the getter
starts with is and the name of the setter starts with set ) are represented as properties which
have the same name as the getter method.
For example:
import java.util.Calendar
fun calendarDemo() {
val calendar = Calendar.getInstance()
if (calendar.firstDayOfWeek == Calendar.SUNDAY) { // call getFirstDayOfWeek()
calendar.firstDayOfWeek = Calendar.MONDAY // call setFirstDayOfWeek()
}
if (!calendar.isLenient) { // call isLenient()
calendar.isLenient = true // call setLenient()
}
}
334
Note that, if the Java class only has a setter, it will not be visible as a property in Kotlin, because
Kotlin does not support set-only properties at this time.
Methods returning void

If a Java method returns void, it will return Unit when called from Kotlin. If, by any chance,
someone uses that return value, it will be assigned at the call site by the Kotlin compiler, since
the value itself is known in advance (being Unit ).
Escaping for Java identi ers that are keywords in Kotlin

Some of the Kotlin keywords are valid identi ers in Java: in, object, is, etc. If a Java library uses
a Kotlin keyword for a method, you can still call the method escaping it with the backtick (`)
character:
foo.ìs`(bar)
Null-Safety and Platform Types

Any reference in Java may be null, which makes Kotlin's requirements of strict null-safety
impractical for objects coming from Java. Types of Java declarations are treated specially in Kotlin
and called platform types. Null-checks are relaxed for such types, so that safety guarantees for
them are the same as in Java (see more below).
Consider the following examples:
val list = ArrayList<String>() // non-null (constructor result)

list.add("Item")
val size = list.size // non-null (primitive int)
val item = list[0] // platform type inferred (ordinary Java object)
When we call methods on variables of platform types, Kotlin does not issue nullability errors at
compile time, but the call may fail at runtime, because of a null-pointer exception or an assertion
that Kotlin generates to prevent nulls from propagating:
item.substring(1) // allowed, may throw an exception if item == null
Platform types are non-denotable, meaning that one can not write them down explicitly in the
language. When a platform value is assigned to a Kotlin variable, we can rely on type inference
(the variable will have an inferred platform type then, as item has in the example above), or we
can choose the type that we expect (both nullable and non-null types are allowed):
val nullable: String? = item // allowed, always works

val notNull: String = item // allowed, may fail at runtime
335
If we choose a non-null type, the compiler will emit an assertion upon assignment. This prevents
Kotlin's non-null variables from holding nulls. Assertions are also emitted when we pass platform
values to Kotlin functions expecting non-null values etc. Overall, the compiler does its best to
prevent nulls from propagating far through the program (although sometimes this is impossible
to eliminate entirely, because of generics).
Notation for Platform Types
As mentioned above, platform types cannot be mentioned explicitly in the program, so there's no
syntax for them in the language. Nevertheless, the compiler and IDE need to display them
sometimes (in error messages, parameter info etc), so we have a mnemonic notation for them:
— T! means " T or T? ",
— (Mutable)Collection<T>! means "Java collection of T may be mutable or not, may be

nullable or not",
— Array<(out) T>! means "Java array of T (or a subtype of T ), nullable or not"
Nullability annotations
Java types which have nullability annotations are represented not as platform types, but as actual
nullable or non-null Kotlin types. The compiler supports several avors of nullability annotations,
including:
— JetBrains ( @Nullable and @NotNull from the org.jetbrains.annotations package)
— Android ( com.android.annotations and android.support.annotations )
— JSR-305 ( javax.annotation , more details below)
— FindBugs ( edu.umd.cs.findbugs.annotations )
— Eclipse ( org.eclipse.jdt.annotation )
— Lombok ( lombok.NonNull ).
You can nd the full list in the Kotlin compiler source code.
Annotating type parameters
It is possible to annotate type arguments of generic types to provide nullability information for
them as well. For example, consider these annotations on a Java declaration:
@NotNull
Set<@NotNull String> toSet(@NotNull Collection<@NotNull String> elements) { ... }
It leads to the following signature seen in Kotlin:
fun toSet(elements: (Mutable)Collection<String>) : (Mutable)Set<String> { ... }
336
Note the @NotNull annotations on String type arguments. Without them, we get platform
types in the type arguments:
fun toSet(elements: (Mutable)Collection<String!>) : (Mutable)Set<String!> { ... }
Annotating type arguments works with Java 8 target or higher and requires the nullability
annotations to support the TYPE_USE target ( org.jetbrains.annotations supports this in
version 15 and above).
Note: due to the current technical limitations, the IDE does not correctly recognize these
annotations on type arguments in compiled Java libraries that are used as dependencies.
JSR-305 Support
The @Nonnull annotation de ned in JSR-305 is supported for denoting nullability of Java types.
If the @Nonnull(when = ...) value is When.ALWAYS , the annotated type is treated as non-
null; When.MAYBE and When.NEVER denote a nullable type; and When.UNKNOWN forces the
type to be platform one.
A library can be compiled against the JSR-305 annotations, but there's no need to make the
annotations artifact (e.g. jsr305.jar ) a compile dependency for the library consumers. The
Kotlin compiler can read the JSR-305 annotations from a library without the annotations present
on the classpath.
Since Kotlin 1.1.50, custom nullability quali ers (KEEP-79) are also supported (see below).
Type quali er nicknames (since 1.1.50)
If an annotation type is annotated with both @TypeQualifierNickname and JSR-305

@Nonnull (or its another nickname, such as @CheckForNull ), then the annotation type is
itself used for retrieving precise nullability and has the same meaning as that nullability
annotation:
337
@TypeQualifierNickname
@Nonnull(when = When.ALWAYS)
@Retention(RetentionPolicy.RUNTIME)
public @interface MyNonnull {
}
@TypeQualifierNickname
@CheckForNull // a nickname to another type qualifier nickname
@Retention(RetentionPolicy.RUNTIME)
public @interface MyNullable {
}
interface A {
@MyNullable String foo(@MyNonnull String x);
// in Kotlin (strict mode): `fun foo(x: String): String?`
String bar(List<@MyNonnull String> x);

// in Kotlin (strict mode): `fun bar(x: List<String>!): String!`
}
Type quali er defaults (since 1.1.50)
@TypeQualifierDefault allows introducing annotations that, when being applied, de ne the

default nullability within the scope of the annotated element.
Such annotation type should itself be annotated with both @Nonnull (or its nickname) and
@TypeQualifierDefault(...) with one or more ElementType values:
— ElementType.METHOD for return types of methods;
— ElementType.PARAMETER for value parameters;
— ElementType.FIELD for elds; and
— ElementType.TYPE_USE (since 1.1.60) for any type including type arguments, upper
bounds of type parameters and wildcard types.
The default nullability is used when a type itself is not annotated by a nullability annotation, and
the default is determined by the innermost enclosing element annotated with a type quali er
default annotation with the ElementType matching the type usage.
338
@Nonnull
@TypeQualifierDefault({ElementType.METHOD, ElementType.PARAMETER})
public @interface NonNullApi {
}
@Nonnull(when = When.MAYBE)
@TypeQualifierDefault({ElementType.METHOD, ElementType.PARAMETER, ElementType.TYPE_USE})
public @interface NullableApi {
}
@NullableApi
interface A {
String foo(String x); // fun foo(x: String?): String?
@NotNullApi // overriding default from the interface

String bar(String x, @Nullable String y); // fun bar(x: String, y: String?): String
// The List<String> type argument is seen as nullable because of `@NullableApi`

// having the `TYPE_USE` element type:
String baz(List<String> x); // fun baz(List<String?>?): String?
// The type of `x` parameter remains platform because there's an explicit

// UNKNOWN-marked nullability annotation:
String qux(@Nonnull(when = When.UNKNOWN) String x); // fun baz(x: String!): String?
}
Note: the types in this example only take place with the strict mode enabled, otherwise, the
platform types remain. See the @UnderMigration annotation and Compiler con guration
sections.
Package-level default nullability is also supported:
// FILE: test/package-info.java
@NonNullApi // declaring all types in package 'test' as non-nullable by default
package test;
@UnderMigration annotation (since 1.1.60)
The @UnderMigration annotation (provided in a separate artifact kotlin-annotations-

jvm ) can be used by library maintainers to de ne the migration status for the nullability type
quali ers.
The status value in @UnderMigration(status = ...) speci es how the compiler treats
inappropriate usages of the annotated types in Kotlin (e.g. using a @MyNullable -annotated
type value as non-null):
— MigrationStatus.STRICT makes annotation work as any plain nullability annotation, i.e.

report errors for the inappropriate usages and a ect the types in the annotated declarations
as they are seen in Kotlin;
— with MigrationStatus.WARN , the inappropriate usages are reported as compilation

warnings instead of errors, but the types in the annotated declarations remain platform; and
339
— MigrationStatus.IGNORE makes the compiler ignore the nullability annotation
completely.
A library maintainer can add @UnderMigration status to both type quali er nicknames and
type quali er defaults:
@Nonnull(when = When.ALWAYS)
@TypeQualifierDefault({ElementType.METHOD, ElementType.PARAMETER})
@UnderMigration(status = MigrationStatus.WARN)
public @interface NonNullApi {
}
// The types in the class are non-null, but only warnings are reported
// because `@NonNullApi` is annotated `@UnderMigration(status = MigrationStatus.WARN)`
@NonNullApi
public class Test {}
Note: the migration status of a nullability annotation is not inherited by its type quali er
nicknames but is applied to its usages in default type quali ers.
If a default type quali er uses a type quali er nickname and they are both @UnderMigration ,
the status from the default type quali er is used.
Compiler con guration
The JSR-305 checks can be con gured by adding the -Xjsr305 compiler ag with the following
options (and their combination):
— -Xjsr305={strict|warn|ignore} to set up the behavior for non- @UnderMigration

annotations. Custom nullability quali ers, especially @TypeQualifierDefault , are already
spread among many well-known libraries, and users may need to migrate smoothly when
updating to the Kotlin version containing JSR-305 support. Since Kotlin 1.1.60, this ag only
a ects non- @UnderMigration annotations.
— -Xjsr305=under-migration:{strict|warn|ignore} (since 1.1.60) to override the

behavior for the @UnderMigration annotations. Users may have di erent view on the
migration status for the libraries: they may want to have errors while the o cial migration
status is WARN , or vice versa, they may wish to postpone errors reporting for some until they
complete their migration.
— -Xjsr305=@<fq.name>:{strict|warn|ignore} (since 1.1.60) to override the behavior

for a single annotation, where <fq.name> is the fully quali ed class name of the annotation.
May appear several times for di erent annotations. This is useful for managing the migration
state for a particular library.
The strict , warn and ignore values have the same meaning as those of
MigrationStatus , and only the strict mode a ects the types in the annotated declarations
as they are seen in Kotlin.
340
Note: the built-in JSR-305 annotations @Nonnull, @Nullable and @CheckForNull are
always enabled and a ect the types of the annotated declarations in Kotlin, regardless of
compiler con guration with the -Xjsr305 ag.
For example, adding -Xjsr305=ignore -Xjsr305=under-migration:ignore -

[email protected]:warn to the compiler arguments makes the compiler
generate warnings for inappropriate usages of types annotated by
@org.library.MyNullable and ignore all other JSR-305 annotations.
For kotlin versions 1.1.50+/1.2, the default behavior is the same to -Xjsr305=warn . The
strict value should be considered experimental (more checks may be added to it in the
future).
Mapped types
Kotlin treats some Java types specially. Such types are not loaded from Java "as is", but are
mapped to corresponding Kotlin types. The mapping only matters at compile time, the runtime
representation remains unchanged. Java's primitive types are mapped to corresponding Kotlin
types (keeping platform types in mind):
Java type Kotlin type

byte kotlin.Byte
short kotlin.Short
int kotlin.Int
long kotlin.Long
char kotlin.Char
float kotlin.Float
double kotlin.Double
boolean kotlin.Boolean
Some non-primitive built-in classes are also mapped:

java.lang.Object kotlin.Any!
java.lang.Cloneable kotlin.Cloneable!
java.lang.Comparable kotlin.Comparable!
java.lang.Enum kotlin.Enum!
java.lang.Annotation kotlin.Annotation!
java.lang.Deprecated kotlin.Deprecated!
java.lang.CharSequence kotlin.CharSequence!
java.lang.String kotlin.String!
java.lang.Number kotlin.Number!
java.lang.Throwable kotlin.Throwable!
Java's boxed primitive types are mapped to nullable Kotlin types:
341
java.lang.Byte kotlin.Byte?
java.lang.Short kotlin.Short?
java.lang.Integer kotlin.Int?
java.lang.Long kotlin.Long?
java.lang.Character kotlin.Char?
java.lang.Float kotlin.Float?
java.lang.Double kotlin.Double?
java.lang.Boolean kotlin.Boolean?
Note that a boxed primitive type used as a type parameter is mapped to a platform type: for
example, List<java.lang.Integer> becomes a List<Int!> in Kotlin.
Collection types may be read-only or mutable in Kotlin, so Java's collections are mapped as
follows (all Kotlin types in this table reside in the package kotlin.collections ):
Java type Kotlin read-only Kotlin mutable type Loaded platform type
type
Iterator<T> Iterator<T> MutableIterator<T> (Mutable)Iterator<T>!
Iterable<T> Iterable<T> MutableIterable<T> (Mutable)Iterable<T>!
Collection<T> Collection<T> MutableCollection<T> (Mutable)Collection<T>!
Set<T> Set<T> MutableSet<T> (Mutable)Set<T>!
List<T> List<T> MutableList<T> (Mutable)List<T>!
ListIterator<T> ListIterator<T> MutableListIterator<T> (Mutable)ListIterator<T>!
Map<K, V> Map<K, V> MutableMap<K, V> (Mutable)Map<K, V>!
Map.Entry<K, V> Map.Entry<K, V> MutableMap.MutableEntry<K,V> (Mutable)Map.
(Mutable)Entry<K, V>!
Java's arrays are mapped as mentioned below:

int[] kotlin.IntArray!
String[] kotlin.Array<(out) String>!
Note: the static members of these Java types are not directly accessible on the companion objects
of the Kotlin types. To call them, use the full quali ed names of the Java types, e.g.
java.lang.Integer.toHexString(foo) .
Java generics in Kotlin

Kotlin's generics are a little di erent from Java's (see Generics). When importing Java types to
Kotlin we perform some conversions:
— Java's wildcards are converted into type projections,

— Foo<? extends Bar> becomes Foo<out Bar!>! ,
— Foo<? super Bar> becomes Foo<in Bar!>! ;
342
— Java's raw types are converted into star projections,
— List becomes List<*>! , i.e. List<out Any?>! .
Like Java's, Kotlin's generics are not retained at runtime, i.e. objects do not carry information
about actual type arguments passed to their constructors, i.e. ArrayList<Integer>() is
indistinguishable from ArrayList<Character>() . This makes it impossible to perform is-
checks that take generics into account. Kotlin only allows is-checks for star-projected generic
types:
if (a is List<Int>) // Error: cannot check if it is really a List of Ints

// but
if (a is List<*>) // OK: no guarantees about the contents of the list
Java Arrays
Arrays in Kotlin are invariant, unlike Java. This means that Kotlin does not let us assign an
Array<String> to an Array<Any> , which prevents a possible runtime failure. Passing an
array of a subclass as an array of superclass to a Kotlin method is also prohibited, but for Java
methods this is allowed (through platform types of the form Array<(out) String>! ).
Arrays are used with primitive datatypes on the Java platform to avoid the cost of
boxing/unboxing operations. As Kotlin hides those implementation details, a workaround is
required to interface with Java code. There are specialized classes for every type of primitive
array ( IntArray , DoubleArray , CharArray , and so on) to handle this case. They are not
related to the Array class and are compiled down to Java's primitive arrays for maximum
performance.
Suppose there is a Java method that accepts an int array of indices:
public class JavaArrayExample {
public void removeIndices(int[] indices) {

// code here...
}
}
To pass an array of primitive values you can do the following in Kotlin:
val javaObj = JavaArrayExample()

val array = intArrayOf(0, 1, 2, 3)
javaObj.removeIndices(array) // passes int[] to method
When compiling to JVM byte codes, the compiler optimizes access to arrays so that there's no
overhead introduced:
val array = arrayOf(1, 2, 3, 4)

array[1] = array[1] * 2 // no actual calls to get() and set() generated
for (x in array) { // no iterator created
print(x)
}
343
Even when we navigate with an index, it does not introduce any overhead:
for (i in array.indices) { // no iterator created

array[i] += 2
}
Finally, in-checks have no overhead either:
if (i in array.indices) { // same as (i >= 0 && i < array.size)

print(array[i])
}
Java Varargs
Java classes sometimes use a method declaration for the indices with a variable number of
arguments (varargs):
public class JavaArrayExample {
public void removeIndicesVarArg(int... indices) {

// code here...
}
}
In that case you need to use the spread operator * to pass the IntArray :
val javaObj = JavaArrayExample()

val array = intArrayOf(0, 1, 2, 3)
javaObj.removeIndicesVarArg(*array)
It's currently not possible to pass null to a method that is declared as varargs.
Operators
Since Java has no way of marking methods for which it makes sense to use the operator syntax,
Kotlin allows using any Java methods with the right name and signature as operator overloads
and other conventions ( invoke() etc.) Calling Java methods using the in x call syntax is not
allowed.
Checked Exceptions
In Kotlin, all exceptions are unchecked, meaning that the compiler does not force you to catch
any of them. So, when you call a Java method that declares a checked exception, Kotlin does not
force you to do anything:
fun render(list: List<*>, to: Appendable) {

for (item in list) {
to.append(item.toString()) // Java would require us to catch IOException here
}
}
Object Methods
344
When Java types are imported into Kotlin, all the references of the type java.lang.Object are
turned into Any . Since Any is not platform-speci c, it only declares toString() ,
hashCode() and equals() as its members, so to make other members of
java.lang.Object available, Kotlin uses extension functions.
wait()/notify()
Methods wait() and notify() are not available on references of type Any . Their usage is
generally discouraged in favor of java.util.concurrent . If you really need to call these
methods, you can cast to java.lang.Object :
(foo as java.lang.Object).wait()
getClass()
To retrieve the Java class of an object, use the java extension property on a class reference:
val fooClass = foo::class.java
The code above uses a bound class reference, which is supported since Kotlin 1.1. You can also
use the javaClass extension property:
val fooClass = foo.javaClass
clone()
To override clone() , your class needs to extend kotlin.Cloneable :
class Example : Cloneable {

override fun clone(): Any { ... }
}
Do not forget about E ective Java, 3rd Edition, Item 13: Override clone judiciously.
nalize()
To override finalize() , all you need to do is simply declare it, without using the override
keyword:
class C {
protected fun finalize() {
// finalization logic
}
}
According to Java's rules, finalize() must not be private.
Inheritance from Java classes
345
At most one Java class (and as many Java interfaces as you like) can be a supertype for a class in
Kotlin.
Accessing static members

Static members of Java classes form "companion objects" for these classes. We cannot pass such
a "companion object" around as a value, but can access the members explicitly, for example:
if (Character.isLetter(a)) { ... }
To access static members of a Java type that is mapped to a Kotlin type, use the full quali ed
name of the Java type: java.lang.Integer.bitCount(foo) .
Java Re ection
Java re ection works on Kotlin classes and vice versa. As mentioned above, you can use
instance::class.java , ClassName::class.java or instance.javaClass to enter
Java re ection through java.lang.Class .
Other supported cases include acquiring a Java getter/setter method or a backing eld for a
Kotlin property, a KProperty for a Java eld, a Java method or constructor for a KFunction
and vice versa.
SAM Conversions
Just like Java 8, Kotlin supports SAM conversions. This means that Kotlin function literals can be
automatically converted into implementations of Java interfaces with a single non-default
method, as long as the parameter types of the interface method match the parameter types of
the Kotlin function.
You can use this for creating instances of SAM interfaces:
val runnable = Runnable { println("This runs in a runnable") }
…and in method calls:
val executor = ThreadPoolExecutor()

// Java signature: void execute(Runnable command)
executor.execute { println("This runs in a thread pool") }
If the Java class has multiple methods taking functional interfaces, you can choose the one you
need to call by using an adapter function that converts a lambda to a speci c SAM type. Those
adapter functions are also generated by the compiler when needed:
executor.execute(Runnable { println("This runs in a thread pool") })
Note that SAM conversions only work for interfaces, not for abstract classes, even if those also
have just a single abstract method.
346
Also note that this feature works only for Java interop; since Kotlin has proper function types,
automatic conversion of functions into implementations of Kotlin interfaces is unnecessary and
therefore unsupported.
Using JNI with Kotlin

To declare a function that is implemented in native (C or C++) code, you need to mark it with the
external modi er:
external fun foo(x: Int): Double
The rest of the procedure works in exactly the same way as in Java.
347
Calling Kotlin from Java
Kotlin code can be easily called from Java. For example, instances of a Kotlin class can be
seamlessly created and operated in Java methods. However, there are certain di erences
between Java and Kotlin that require attention when integrating Kotlin code into Java. On this
page, we'll describe the ways to tailor the interop of your Kotlin code with its Java clients.
Properties
A Kotlin property is compiled to the following Java elements:
— A getter method, with the name calculated by prepending the get pre x;
— A setter method, with the name calculated by prepending the set pre x (only for var
properties);
— A private eld, with the same name as the property name (only for properties with backing
elds).
For example, var firstName: String gets compiled to the following Java declarations:
private String firstName;
public String getFirstName() {

return firstName;
}
public void setFirstName(String firstName) {

this.firstName = firstName;
}
If the name of the property starts with is , a di erent name mapping rule is used: the name of
the getter will be the same as the property name, and the name of the setter will be obtained by
replacing is with set . For example, for a property isOpen , the getter will be called
isOpen() and the setter will be called setOpen() . This rule applies for properties of any type,
not just Boolean .
Package-level functions
All the functions and properties declared in a le app.kt inside a package org.example ,
including extension functions, are compiled into static methods of a Java class named
org.example.AppKt .
// app.kt
package org.example
class Util
fun getTime() { /*...*/ }
348
// Java
new org.example.Util();
org.example.AppKt.getTime();
The name of the generated Java class can be changed using the @JvmName annotation:
@file:JvmName("DemoUtils")
package org.example
class Util
// Java
new org.example.Util();
org.example.DemoUtils.getTime();
Having multiple les which have the same generated Java class name (the same package and the
same name or the same @JvmName annotation) is normally an error. However, the compiler has
the ability to generate a single Java facade class which has the speci ed name and contains all the
declarations from all the les which have that name. To enable the generation of such a facade,
use the @JvmMultifileClass annotation in all of the les.
// oldutils.kt
@file:JvmName("Utils")
@file:JvmMultifileClass
package org.example
// newutils.kt
@file:JvmName("Utils")
@file:JvmMultifileClass
package org.example
fun getDate() { /*...*/ }
// Java
org.example.Utils.getTime();
org.example.Utils.getDate();
Instance elds
If you need to expose a Kotlin property as a eld in Java, annotate it with the @JvmField
annotation. The eld will have the same visibility as the underlying property. You can annotate a
property with @JvmField if it has a backing eld, is not private, does not have open ,
override or const modi ers, and is not a delegated property.
349
class User(id: String) {
@JvmField val ID = id
}
// Java
class JavaClient {
public String getID(User user) {
return user.ID;
}
}
Late-Initialized properties are also exposed as elds. The visibility of the eld will be the same as
the visibility of lateinit property setter.
Static elds
Kotlin properties declared in a named object or a companion object will have static backing elds
either in that named object or in the class containing the companion object.
Usually these elds are private but they can be exposed in one of the following ways:
— @JvmField annotation;
— lateinit modi er;
— const modi er.
Annotating such a property with @JvmField makes it a static eld with the same visibility as the
property itself.
class Key(val value: Int) {

companion object {
@JvmField
val COMPARATOR: Comparator<Key> = compareBy<Key> { it.value }
}
}
// Java
Key.COMPARATOR.compare(key1, key2);
// public static final field in Key class
A late-initialized property in an object or a companion object has a static backing eld with the
same visibility as the property setter.
object Singleton {
lateinit var provider: Provider
}
// Java
Singleton.provider = new Provider();
// public static non-final field in Singleton class
Properties declared as const (in classes as well as at the top level) are turned into static elds in
Java:
350
// file example.kt
object Obj {
const val CONST = 1
}
class C {
companion object {
const val VERSION = 9
}
}
const val MAX = 239
In Java:
int const = Obj.CONST;

int max = ExampleKt.MAX;
int version = C.VERSION;
Static methods
As mentioned above, Kotlin represents package-level functions as static methods. Kotlin can also
generate static methods for functions de ned in named objects or companion objects if you
annotate those functions as @JvmStatic. If you use this annotation, the compiler will generate
both a static method in the enclosing class of the object and an instance method in the object
itself. For example:
class C {
companion object {
@JvmStatic fun callStatic() {}
fun callNonStatic() {}
}
}
Now, callStatic() is static in Java, while callNonStatic() is not:
C.callStatic(); // works fine

C.callNonStatic(); // error: not a static method
C.Companion.callStatic(); // instance method remains
C.Companion.callNonStatic(); // the only way it works
Same for named objects:
object Obj {
@JvmStatic fun callStatic() {}
fun callNonStatic() {}
}
In Java:
Obj.callStatic(); // works fine

Obj.callNonStatic(); // error
Obj.INSTANCE.callNonStatic(); // works, a call through the singleton instance
Obj.INSTANCE.callStatic(); // works too
351
Starting from Kotlin 1.3, @JvmStatic applies to functions de ned in companion objects of
interfaces as well. Such functions compile to static methods in interfaces. Note that static method
in interfaces were introduced in Java 1.8, so be sure to use the corresponding targets.
interface ChatBot {
companion object {
@JvmStatic fun greet(username: String) {
println("Hello, $username")
}
}
}
@JvmStatic annotation can also be applied on a property of an object or a companion object

making its getter and setter methods static members in that object or the class containing the
companion object.
Default methods in interfaces
Default methods are available only for targets JVM 1.8 and above.
The @JvmDefault annotation is experimental in Kotlin 1.3. Its name and behavior may
change, leading to future incompatibility.
Starting from JDK 1.8, interfaces in Java can contain default methods. You can declare a non-
abstract member of a Kotlin interface as default for the Java classes implementing it. To make a
member default, mark it with the @JvmDefault annotation. Here is an example of a Kotlin
interface with a default method:
interface Robot {
@JvmDefault fun move() { println("~walking~") }
fun speak(): Unit
}
The default implementation is available for Java classes implementing the interface.
//Java implementation
public class C3PO implements Robot {
// move() implementation from Robot is available implicitly
@Override
public void speak() {
System.out.println("I beg your pardon, sir");
}
}
C3PO c3po = new C3PO();

c3po.move(); // default implementation from the Robot interface
c3po.speak();
Implementations of the interface can override default methods.
352
//Java
public class BB8 implements Robot {
//own implementation of the default method
@Override
public void move() {
System.out.println("~rolling~");
}
@Override
public void speak() {
System.out.println("Beep-beep");
}
}
For the @JvmDefault annotation to take e ect, the interface must be compiled with an -
Xjvm-default argument. Depending on the case of adding the annotation, specify one of the
argument values:
— -Xjvm-default=enabled should be used if you add only new methods with the
@JvmDefault annotation. This includes adding the entire interface for your API.
— -Xjvm-default=compatibility should be used if you are adding a @JvmDefault to the
methods that were available in the API before. This mode helps avoid compatibility breaks: all
the interface implementations written for the previous versions will be fully compatible with
the new version. However, the compatibility mode may add some overhead to the resulting
bytecode size and a ect the performance.
For more details about compatibility issues, see the @JvmDefault reference page.
Note that if an interface with @JvmDefault methods is used as a delegate, the default method
implementations are called even if the actual delegate type provides its own implementations.
interface Producer {
@JvmDefault fun produce() {
println("interface method")
}
}
class ProducerImpl: Producer {

override fun produce() {
println("class method")
}
}
class DelegatedProducer(val p: Producer): Producer by p {

}
fun main() {
val prod = ProducerImpl()
DelegatedProducer(prod).produce() // prints "interface method"
}
For more details about interface delegation in Kotlin, see Delegation.
353
Visibility
The Kotlin visibilities are mapped to Java in the following way:
— private members are compiled to private members;
— private top-level declarations are compiled to package-local declarations;
— protected remains protected (note that Java allows accessing protected members from
other classes in the same package and Kotlin doesn't, so Java classes will have broader access
to the code);
— internal declarations become public in Java. Members of internal classes go through
name mangling, to make it harder to accidentally use them from Java and to allow overloading
for members with the same signature that don't see each other according to Kotlin rules;
— public remains public .
KClass
Sometimes you need to call a Kotlin method with a parameter of type KClass . There is no
automatic conversion from Class to KClass , so you have to do it manually by invoking the
equivalent of the Class<T>.kotlin extension property:
kotlin.jvm.JvmClassMappingKt.getKotlinClass(MainView.class)
Handling signature clashes with @JvmName

Sometimes we have a named function in Kotlin, for which we need a di erent JVM name in the
byte code. The most prominent example happens due to type erasure:
fun List<String>.filterValid(): List<String>

fun List<Int>.filterValid(): List<Int>
These two functions can not be de ned side-by-side, because their JVM signatures are the same:
filterValid(Ljava/util/List;)Ljava/util/List; . If we really want them to have the
same name in Kotlin, we can annotate one (or both) of them with @JvmName and specify a
di erent name as an argument:
fun List<String>.filterValid(): List<String>
@JvmName("filterValidInt")
fun List<Int>.filterValid(): List<Int>
From Kotlin they will be accessible by the same name filterValid , but from Java it will be
filterValid and filterValidInt .
The same trick applies when we need to have a property x alongside with a function getX() :
354
val x: Int
@JvmName("getX_prop")
get() = 15
fun getX() = 10
To change the names of generated accessor methods for properties without explicitly
implemented getters and setters, you can use @get:JvmName and @set:JvmName :
@get:JvmName("x")
@set:JvmName("changeX")
var x: Int = 23
Overloads generation
Normally, if you write a Kotlin function with default parameter values, it will be visible in Java only
as a full signature, with all parameters present. If you wish to expose multiple overloads to Java
callers, you can use the @JvmOverloads annotation.
The annotation also works for constructors, static methods, and so on. It can't be used on
abstract methods, including methods de ned in interfaces.
class Circle @JvmOverloads constructor(centerX: Int, centerY: Int, radius: Double = 1.0)
{
@JvmOverloads fun draw(label: String, lineWidth: Int = 1, color: String = "red") {
/*...*/ }
}
For every parameter with a default value, this will generate one additional overload, which has
this parameter and all parameters to the right of it in the parameter list removed. In this
example, the following will be generated:
// Constructors:
Circle(int centerX, int centerY, double radius)
Circle(int centerX, int centerY)
// Methods
void draw(String label, int lineWidth, String color) { }
void draw(String label, int lineWidth) { }
void draw(String label) { }
Note that, as described in Secondary Constructors, if a class has default values for all constructor
parameters, a public no-argument constructor will be generated for it. This works even if the
@JvmOverloads annotation is not speci ed.
Checked exceptions
As we mentioned above, Kotlin does not have checked exceptions. So, normally, the Java
signatures of Kotlin functions do not declare exceptions thrown. Thus if we have a function in
Kotlin like this:
355
// example.kt
package demo
fun writeToFile() {
/*...*/
throw IOException()
}
And we want to call it from Java and catch the exception:
// Java
try {
demo.Example.writeToFile();
}
catch (IOException e) { // error: writeToFile() does not declare IOException in the
throws list
// ...
}
we get an error message from the Java compiler, because writeToFile() does not declare
IOException . To work around this problem, use the @Throws annotation in Kotlin:
@Throws(IOException::class)
fun writeToFile() {
/*...*/
throw IOException()
}
Null-safety
When calling Kotlin functions from Java, nobody prevents us from passing null as a non-null
parameter. That's why Kotlin generates runtime checks for all public functions that expect non-
nulls. This way we get a NullPointerException in the Java code immediately.
Variant generics
When Kotlin classes make use of declaration-site variance, there are two options of how their
usages are seen from the Java code. Let's say we have the following class and two functions that
use it:
class Box<out T>(val value: T)
interface Base
class Derived : Base
fun boxDerived(value: Derived): Box<Derived> = Box(value)

fun unboxBase(box: Box<Base>): Base = box.value
A naive way of translating these functions into Java would be this:
Box<Derived> boxDerived(Derived value) { ... }

Base unboxBase(Box<Base> box) { ... }
356
The problem is that in Kotlin we can say unboxBase(boxDerived("s")) , but in Java that
would be impossible, because in Java the class Box is invariant in its parameter T , and thus
Box<Derived> is not a subtype of Box<Base> . To make it work in Java we'd have to de ne
unboxBase as follows:
Base unboxBase(Box<? extends Base> box) { ... }
Here we make use of Java's wildcards types ( ? extends Base ) to emulate declaration-site
variance through use-site variance, because it is all Java has.
To make Kotlin APIs work in Java we generate Box<Super> as Box<? extends Super> for
covariantly de ned Box (or Foo<? super Bar> for contravariantly de ned Foo ) when it
appears as a parameter. When it's a return value, we don't generate wildcards, because otherwise
Java clients will have to deal with them (and it's against the common Java coding style). Therefore,
the functions from our example are actually translated as follows:
// return type - no wildcards

Box<Derived> boxDerived(Derived value) { ... }
// parameter - wildcards
Base unboxBase(Box<? extends Base> box) { ... }
When the argument type is nal, there's usually no point in generating the wildcard, so
Box<String> is always Box<String>, no matter what position it takes.
If we need wildcards where they are not generated by default, we can use the @JvmWildcard
annotation:
fun boxDerived(value: Derived): Box<@JvmWildcard Derived> = Box(value)

// is translated to
// Box<? extends Derived> boxDerived(Derived value) { ... }
On the other hand, if we don't need wildcards where they are generated, we can use
@JvmSuppressWildcards :
fun unboxBase(box: Box<@JvmSuppressWildcards Base>): Base = box.value

// is translated to
// Base unboxBase(Box<Base> box) { ... }
@JvmSuppressWildcards can be used not only on individual type arguments, but on

entire declarations, such as functions or classes, causing all wildcards inside them to be
suppressed.
Translation of type Nothing
357
The type Nothing is special, because it has no natural counterpart in Java. Indeed, every Java
reference type, including java.lang.Void , accepts null as a value, and Nothing doesn't
accept even that. So, this type cannot be accurately represented in the Java world. This is why
Kotlin generates a raw type where an argument of type Nothing is used:
fun emptyList(): List<Nothing> = listOf()

// is translated to
// List emptyList() { ... }
358
JavaScript
Dynamic Type
The dynamic type is not supported in code targeting the JVM
Being a statically typed language, Kotlin still has to interoperate with untyped or loosely typed
environments, such as the JavaScript ecosystem. To facilitate these use cases, the dynamic type
is available in the language:
val dyn: dynamic = ...
The dynamic type basically turns o Kotlin's type checker:
— a value of this type can be assigned to any variable or passed anywhere as a parameter;
— any value can be assigned to a variable of type dynamic or passed to a function that takes
dynamic as a parameter;
— null -checks are disabled for such values.
The most peculiar feature of dynamic is that we are allowed to call any property or function
with any parameters on a dynamic variable:
dyn.whatever(1, "foo", dyn) // 'whatever' is not defined anywhere

dyn.whatever(*arrayOf(1, 2, 3))
On the JavaScript platform this code will be compiled "as is": dyn.whatever(1) in Kotlin
becomes dyn.whatever(1) in the generated JavaScript code.
When calling functions written in Kotlin on values of dynamic type, keep in mind the name
mangling performed by the Kotlin to JavaScript compiler. You may need to use the @JsName
annotation to assign well-de ned names to the functions that you need to call.
A dynamic call always returns dynamic as a result, so we can chain such calls freely:
dyn.foo().bar.baz()
When we pass a lambda to a dynamic call, all of its parameters by default have the type
dynamic :
359
dyn.foo {
x -> x.bar() // x is dynamic
}
Expressions using values of dynamic type are translated to JavaScript "as is", and do not use the
Kotlin operator conventions. The following operators are supported:
— binary: + , - , * , / , % , > , < >= , <= , == , != , === , !== , && , ||
— unary
— pre x: - , + , !
— pre x and post x: ++ , --
— assignments: += , -= , *= , /= , %=
— indexed access:
— read: d[a] , more than one argument is an error
— write: d[a1] = a2 , more than one argument in [] is an error
in , !in and .. operations with values of type dynamic are forbidden.
For a more technical description, see the spec document.
360
Calling JavaScript from Kotlin
Kotlin was designed for easy interoperation with Java platform. It sees Java classes as Kotlin
classes, and Java sees Kotlin classes as Java classes. However, JavaScript is a dynamically-typed
language, which means it does not check types in compile-time. You can freely talk to JavaScript
from Kotlin via dynamic types, but if you want the full power of Kotlin type system, you can create
Kotlin headers for JavaScript libraries.
Inline JavaScript
You can inline some JavaScript code into your Kotlin code using the js("…") function. For example:
fun jsTypeOf(o: Any): String {

return js("typeof o")
}
The parameter of js is required to be a string constant. So, the following code is incorrect:
fun jsTypeOf(o: Any): String {

return js(getTypeof() + " o") // error reported here
}
fun getTypeof() = "typeof"
external modi er
To tell Kotlin that a certain declaration is written in pure JavaScript, you should mark it with
external modi er. When the compiler sees such a declaration, it assumes that the
implementation for the corresponding class, function or property is provided by the developer,
and therefore does not try to generate any JavaScript code from the declaration. This means that
you should omit bodies of external declarations. For example:
external fun alert(message: Any?): Unit
external class Node {

val firstChild: Node
fun append(child: Node): Node
fun removeChild(child: Node): Node
// etc
}
external val window: Window
Note that external modi er is inherited by nested declarations, i.e. in Node class we do not
put external before member functions and properties.
The external modi er is only allowed on package-level declarations. You can't declare an
external member of a non- external class.
361
Declaring (static) members of a class
In JavaScript you can de ne members either on a prototype or a class itself. I.e.:
function MyClass() { ... }

MyClass.sharedMember = function() { /* implementation */ };
MyClass.prototype.ownMember = function() { /* implementation */ };
There's no such syntax in Kotlin. However, in Kotlin we have companion objects. Kotlin treats
companion objects of external class in a special way: instead of expecting an object, it
assumes members of companion objects to be members of the class itself. To describe
MyClass from the example above, you can write:
external class MyClass {

companion object {
fun sharedMember()
}
fun ownMember()
}
Declaring optional parameters
An external function can have optional parameters. How the JavaScript implementation actually
computes default values for these parameters, is unknown to Kotlin, thus it's impossible to use
the usual syntax to declare such parameters in Kotlin. You should use the following syntax:
external fun myFunWithOptionalArgs(x: Int,

y: String = definedExternally,
z: Long = definedExternally)
This means you can call myFunWithOptionalArgs with one required argument and two
optional arguments (their default values are calculated by some JavaScript code).
Extending JavaScript classes
You can easily extend JavaScript classes as they were Kotlin classes. Just de ne an external
class and extend it by non- external class. For example:
external open class HTMLElement : Element() {

/* members */
}
class CustomElement : HTMLElement() {

fun foo() {
alert("bar")
}
}
There are some limitations:
1. When a function of external base class is overloaded by signature, you can't override it in a derived class.
362
2. You can't override a function with default arguments.
Note that you can't extend a non-external class by external classes.
external interfaces
JavaScript does not have the concept of interfaces. When a function expects its parameter to
support foo and bar methods, you just pass objects that actually have these methods. You can
use interfaces to express this for statically-typed Kotlin, for example:
external interface HasFooAndBar {

fun foo()
fun bar()
}
external fun myFunction(p: HasFooAndBar)
Another use case for external interfaces is to describe settings objects. For example:
external interface JQueryAjaxSettings {

var async: Boolean
var cache: Boolean
var complete: (JQueryXHR, String) -> Unit
// etc
}
fun JQueryAjaxSettings(): JQueryAjaxSettings = js("{}")
external class JQuery {

companion object {
fun get(settings: JQueryAjaxSettings): JQueryXHR
}
}
fun sendQuery() {
JQuery.get(JQueryAjaxSettings().apply {
complete = { (xhr, data) ->
window.alert("Request complete")
}
})
}
External interfaces have some restrictions:
1. They can't be used on the right hand side of is checks.
2. as cast to external interface always succeeds (and produces a warning in compile-time).
3. They can't be passed as rei ed type arguments.
4. They can't be used in class literal expressions (i.e. I::class ).
363
Calling Kotlin from JavaScript
Kotlin compiler generates normal JavaScript classes, functions and properties you can freely use
from JavaScript code. Nevertheless, there are some subtle things you should remember.
Isolating declarations in a separate JavaScript object

To prevent spoiling the global object, Kotlin creates an object that contains all Kotlin declarations
from the current module. So if you name your module as myModule , all declarations are
available to JavaScript via myModule object. For example:
fun foo() = "Hello"
Can be called from JavaScript like this:
alert(myModule.foo());
This is not applicable when you compile your Kotlin module to JavaScript module (see JavaScript
Modules for more information on this). In this case there won't be a wrapper object, instead,
declarations will be exposed as a JavaScript module of a corresponding kind. For example, in case
of CommonJS you should write:
alert(require('myModule').foo());
Package structure
Kotlin exposes its package structure to JavaScript, so unless you de ne your declarations in the
root package, you have to use fully-quali ed names in JavaScript. For example:
package my.qualified.packagename
fun foo() = "Hello"
Can be called from JavaScript like this:
alert(myModule.my.qualified.packagename.foo());
@JsName annotation
In some cases (for example, to support overloads), Kotlin compiler mangles names of generated
functions and attributes in JavaScript code. To control the generated names, you can use the
@JsName annotation:
364
// Module 'kjs'
class Person(val name: String) {
fun hello() {
println("Hello $name!")
}
@JsName("helloWithGreeting")
fun hello(greeting: String) {
println("$greeting $name!")
}
}
Now you can use this class from JavaScript in the following way:
var person = new kjs.Person("Dmitry"); // refers to module 'kjs'

person.hello(); // prints "Hello Dmitry!"
person.helloWithGreeting("Servus"); // prints "Servus Dmitry!"
If we didn't specify the @JsName annotation, the name of the corresponding function would
contain a su x calculated from the function signature, for example hello_61zpoe$ .
Note that Kotlin compiler does not apply such mangling to external declarations, so you don't
have to use @JsName on them. Another case worth noticing is inheriting non-external classes
from external classes. In this case any overridden functions won't be mangled as well.
The parameter of @JsName is required to be a constant string literal which is a valid identi er.
The compiler will report an error on any attempt to pass non-identi er string to @JsName . The
following example produces a compile-time error:
@JsName("new C()") // error here

external fun newC()
Representing Kotlin types in JavaScript

— Kotlin numeric types, except for kotlin.Long are mapped to JavaScript Number.
— kotlin.Char is mapped to JavaScript Number representing character code.
— Kotlin can't distinguish between numeric types at run time (except for kotlin.Long ), i.e. the
following code works:
fun f() {
val x: Int = 23
val y: Any = x
println(y as Float)
}
— Kotlin preserves over ow semantics for kotlin.Int , kotlin.Byte , kotlin.Short ,

kotlin.Char and kotlin.Long .
— There's no 64 bit integer number in JavaScript, so kotlin.Long is not mapped to any

JavaScript object, it's emulated by a Kotlin class.
— kotlin.String is mapped to JavaScript String.
365
— kotlin.Any is mapped to JavaScript Object (i.e. new Object() , {} , etc).
— kotlin.Array is mapped to JavaScript Array.
— Kotlin collections (i.e. List , Set , Map , etc) are not mapped to any speci c JavaScript type.
— kotlin.Throwable is mapped to JavaScript Error.
— Kotlin preserves lazy object initialization in JavaScript.

— Kotlin does not implement lazy initialization of top-level properties in JavaScript.
Starting with version 1.1.50 primitive array translation utilizes JavaScript TypedArray:
— kotlin.ByteArray , -.ShortArray , -.IntArray , -.FloatArray , and -

.DoubleArray are mapped to JavaScript Int8Array, Int16Array, Int32Array, Float32Array, and
Float64Array correspondingly.
— kotlin.BooleanArray is mapped to JavaScript Int8Array with a property $type$ ==
"BooleanArray"
— kotlin.CharArray is mapped to JavaScript UInt16Array with a property $type$ ==

"CharArray"
— kotlin.LongArray is mapped to JavaScript Array of kotlin.Long with a property

$type$ == "LongArray" .
366
JavaScript Modules
Kotlin allows you to compile your Kotlin projects to JavaScript modules for popular module
systems. Here is the list of available options:
1. Plain. Don't compile for any module system. As usual, you can access a module by its name in the global scope. This
option is used by default.
2. Asynchronous Module De nition (AMD), which is in particular used by require.js library.
3. CommonJS convention, widely used by node.js/npm ( require function and module.exports object)
4. Uni ed Module De nitions (UMD), which is compatible with both AMD and CommonJS, and works as "plain" when
neither AMD nor CommonJS is available at runtime.
Choosing the Target Module System

Choosing the target module system depends on your build environment:
From IntelliJ IDEA
Setup per module: Open File -> Project Structure…, nd your module in Modules and select
"Kotlin" facet under it. Choose appropriate module system in "Module kind" eld.
Setup for the whole project: Open File -> Settings, select "Build, Execution, Deployment" ->
"Compiler" -> "Kotlin compiler". Choose appropriate module system in "Module kind" eld.
From Maven
To select module system when compiling via Maven, you should set moduleKind con guration
property, i.e. your pom.xml should look like this:
<plugin>
<executions>
<execution>
<id>compile</id>
<goals>
<goal>js</goal>
</goals>
</execution>
</executions>

<configuration>
<moduleKind>commonjs</moduleKind>
</configuration>

</plugin>
Available values are: plain , amd , commonjs , umd .
From Gradle
367
To select module system when compiling via Gradle, you should set moduleKind property, i.e.
compileKotlin2Js.kotlinOptions.moduleKind = "commonjs"
Available values are similar to Maven.
@JsModule annotation
To tell Kotlin that an external class, package, function or property is a JavaScript module, you
can use @JsModule annotation. Consider you have the following CommonJS module called
"hello":
module.exports.sayHello = function(name) { alert("Hello, " + name); }
You should declare it like this in Kotlin:
@JsModule("hello")
external fun sayHello(name: String)
Applying @JsModule to packages
Some JavaScript libraries export packages (namespaces) instead of functions and classes. In terms
of JavaScript, it's an object that has members that are classes, functions and properties. Importing
these packages as Kotlin objects often looks unnatural. The compiler allows to map imported
JavaScript packages to Kotlin packages, using the following notation:
@file:JsModule("extModule")
package ext.jspackage.name
external fun foo()
external class C
where the corresponding JavaScript module is declared like this:
module.exports = {
foo: { /* some code here */ },
C: { /* some code here */ }
}
Important: les marked with @file:JsModule annotation can't declare non-external members.
The example below produces compile-time error:
package ext.jspackage.name
external fun foo()
fun bar() = "!" + foo() + "!" // error here
Importing deeper package hierarchies
368
In the previous example the JavaScript module exports a single package. However, some
JavaScript libraries export multiple packages from within a module. This case is also supported by
Kotlin, though you have to declare a new .kt le for each package you import.
For example, let's make our example a bit more complicated:
module.exports = {
mylib: {
pkg1: {
foo: function() { /* some code here */ },
bar: function() { /* some code here */ }
},
pkg2: {
baz: function() { /* some code here */ }
}
}
}
To import this module in Kotlin, you have to write two Kotlin source les:
@file:JsQualifier("mylib.pkg1")
package extlib.pkg1
external fun foo()
external fun bar()
and
@file:JsQualifier("mylib.pkg2")
package extlib.pkg2
external fun baz()
@JsNonModule annotation
When a declaration has @JsModule , you can't use it from Kotlin code when you don't compile it
to a JavaScript module. Usually, developers distribute their libraries both as JavaScript modules
and downloadable .js les that you can copy to project's static resources and include via
<script> element. To tell Kotlin that it's ok to use a @JsModule declaration from non-module
environment, you should put @JsNonModule declaration. For example, given JavaScript code:
function topLevelSayHello(name) { alert("Hello, " + name); }

if (module && module.exports) {
module.exports = topLevelSayHello;
}
can be described like this:
369
@JsModule("hello")
@JsNonModule
@JsName("topLevelSayHello")
external fun sayHello(name: String)
Notes
Kotlin is distributed with kotlin.js standard library as a single le, which is itself compiled as
an UMD module, so you can use it with any module system described above. Also it is available
on NPM as kotlin package
370
JavaScript Re ection
At this time, JavaScript does not support the full Kotlin re ection API. The only supported part of
the API is the ::class syntax which allows you to refer to the class of an instance, or the class
corresponding to the given type. The value of a ::class expression is a stripped-down KClass
implementation that only supports the simpleName and isInstance members.
In addition to that, you can use KClass.js to access the JsClass instance corresponding to the class.
The JsClass instance itself is a reference to the constructor function. This can be used to
interoperate with JS functions that expect a reference to a constructor.
Examples:
class A
class B
class C
inline fun <reified T> foo() {

println(T::class.simpleName)
}
val a = A()
println(a::class.simpleName) // Obtains class for an instance; prints "A"
println(B::class.simpleName) // Obtains class for a type; prints "B"
println(B::class.js.name) // prints "B"
foo<C>() // prints "C"
371
JavaScript DCE
Since version 1.1.4, Kotlin/JS includes a dead code elimination (DCE) tool. This tool allows to strip
out unused properties, functions and classes from the generated JS. There are several ways you
get unused declarations:
— Functions can be inlined and never get called directly (which happens always except for few
situations).
— You are using a shared library which provides much more functions than you actually need.
For example, standard library ( kotlin.js ) contains functions for manipulating lists, arrays,
char sequences, adapters for DOM, etc, which together gives about 1.3 mb le. A simple
"Hello, world" application only requires console routines, which is only few kilobytes for the
entire le.
Dead code elimination is often also called 'tree shaking'.
How to use
DCE tool is currently available from Gradle.
To activate DCE tool, add the following line to build.gradle :
apply plugin: 'kotlin-dce-js'
Note that if you are using multi-project build, you should apply plugin to the main project that is
an entry point to your application.
By default, the resulting set of JavaScript les (your application together with all dependencies)
can be found at path $BUILD_DIR/min/ , where $BUILD_DIR is the path to generated
JavaScript (usually, build/classes/main ).
Con guring
To con gure DCE on the main source set, you can use the runDceKotlinJs task (and
corresponding runDce<sourceSetName>KotlinJs for other source sets).
Sometimes you are going to use a Kotlin declaration directly from JavaScript, and it's being
stripped out by DCE. You may want to keep this declaration. To do so, you can use the following
syntax in build.gradle :
runDceKotlinJs.keep "declarationToKeep"[, "declarationToKeep", ...]
Where declarationToKeep has the following syntax:
moduleName.dot.separated.package.name.declarationName
For example, consider a module is named kotlin-js-example and it contains a function

named toKeep in package org.jetbrains.kotlin.examples . Use the following line:
372
runDceKotlinJs.keep "kotlin-js-example_main.org.jetbrains.kotlin.examples.toKeep"
Note that if your function has parameters, its name will be mangled, so the mangled name
should be used in the keep directive.
Development mode
Running DCE takes a bit of extra time each build, and the output size does not matter during
development. You can improve development builds time by making the DCE tool skip actual dead
code elimination with the dceOptions.devMode ag of the DCE tasks.
For example, to disable DCE based on a custom condition for the main source set and always for
the test code, add the following lines to the build script:
runDceKotlinJs.dceOptions.devMode = isDevMode
runDceTestKotlinJs.dceOptions.devMode = true
Example
A full example that shows how to integrate Kotlin with DCE and webpack to get a small bundle,
can be found here.
Notes
— As for 1.1.x versions, DCE tool is an experimental feature. This does not mean we are going to
remove it, or that it's unusable for production. This means that we can change names of
con guration parameters, default settings, etc.
— Currently you should not use DCE tool if your project is a shared library. It's only applicable
when you are developing an application (which may use shared libraries). The reason is: DCE
does not know which parts of the library are going to be used by the user's application.
— DCE does not perform mini cation (ugli cation) of your code by removing unnecessary
whitespaces and shortening identi ers. You should use existing tools, like UglifyJS
(https://github.com/mishoo/UglifyJS2) or Google Closure Compiler
(https://developers.google.com/closure/compiler/) for this purpose.
373
Native
Concurrency in Kotlin/Native
Kotlin/Native runtime doesn't encourage a classical thread-oriented concurrency model with
mutually exclusive code blocks and conditional variables, as this model is known to be error-
prone and unreliable. Instead, we suggest a collection of alternative approaches, allowing you to
use hardware concurrency and implement blocking IO. Those approaches are as follows, and
they will be elaborated on in further sections:
— Workers with message passing

— Object subgraph ownership transfer
— Object subgraph freezing
— Object subgraph detachment
— Raw shared memory using C globals
— Coroutines for blocking operations (not covered in this document)
Workers
Instead of threads Kotlin/Native runtime o ers the concept of workers: concurrently executed
control ow streams with an associated request queue. Workers are very similar to the actors in
the Actor Model. A worker can exchange Kotlin objects with another worker, so that at any
moment each mutable object is owned by a single worker, but ownership can be transferred. See
section Object transfer and freezing.
Once a worker is started with the Worker.start function call, it can be addressed with its own
unique integer worker id. Other workers, or non-worker concurrency primitives, such as OS
threads, can send a message to the worker with the execute call.
374
val future = execute(TransferMode.SAFE, { SomeDataForWorker() }) {
// data returned by the second function argument comes to the
// worker routine as 'input' parameter.
input ->
// Here we create an instance to be returned when someone consumes result future.
WorkerResult(input.stringParam + " result")
}
future.consume {
// Here we see result returned from routine above. Note that future object or
// id could be transferred to another worker, so we don't have to consume future
// in same execution context it was obtained.
result -> println("result is $result")
}
The call to execute uses a function passed as its second parameter to produce an object
subgraph (i.e. set of mutually referring objects) which is then passed as a whole to that worker, it
is then no longer available to the thread that initiated the request. This property is checked if the
rst parameter is TransferMode.SAFE by graph traversal and is just assumed to be true, if it is
TransferMode.UNSAFE . The last parameter to execute is a special Kotlin lambda, which is
not allowed to capture any state, and is actually invoked in the target worker's context. Once
processed, the result is transferred to whatever consumes it in the future, and it is attached to
the object graph of that worker/thread.
If an object is transferred in UNSAFE mode and is still accessible from multiple concurrent
executors, program will likely crash unexpectedly, so consider that last resort in optimizing, not a
general purpose mechanism.
For a more complete example please refer to the workers example in the Kotlin/Native
repository.
Object transfer and freezing
An important invariant that Kotlin/Native runtime maintains is that the object is either owned by
a single thread/worker, or it is immutable (shared XOR mutable). This ensures that the same data
has a single mutator, and so there is no need for locking to exist. To achieve such an invariant, we
use the concept of not externally referred object subgraphs. This is a subgraph which has no
external references from outside of the subgraph, which could be checked algorithmically with
O(N) complexity (in ARC systems), where N is the number of elements in such a subgraph. Such
subgraphs are usually produced as a result of a lambda expression, for example some builder,
and may not contain objects, referred to externally.
375
Freezing is a runtime operation making a given object subgraph immutable, by modifying the
object header so that future mutation attempts throw an InvalidMutabilityException . It is
deep, so if an object has a pointer to other objects - transitive closure of such objects will be
frozen. Freezing is a one way transformation, frozen objects cannot be unfrozen. Frozen objects
have a nice property that due to their immutability, they can be freely shared between multiple
workers/threads without breaking the "mutable XOR shared" invariant.
If an object is frozen it can be checked with an extension property isFrozen , and if it is, object
sharing is allowed. Currently, Kotlin/Native runtime only freezes the enum objects after creation,
although additional autofreezing of certain provably immutable objects could be implemented in
the future.
Object subgraph detachment
An object subgraph without external references can be disconnected using

DetachedObjectGraph<T> to a COpaquePointer value, which could be stored in void*
data, so the disconnected object subgraphs can be stored in a C data structure, and later attached
back with DetachedObjectGraph<T>.attach() in an arbitrary thread or a worker.
Combining it with raw memory sharing it allows side channel object transfer between concurrent
threads, if the worker mechanisms are insu cient for a particular task.
Raw shared memory
Considering the strong ties between Kotlin/Native and C via interoperability, in conjunction with
the other mechanisms mentioned above it is possible to build popular data structures, like
concurrent hashmap or shared cache with Kotlin/Native. It is possible to rely upon shared C data,
and store in it references to detached object subgraphs. Consider the following .def le:
package = global
---
typedef struct {
int version;
void* kotlinObject;
} SharedData;
SharedData sharedData;
After running the cinterop tool it can share Kotlin data in a versionized global structure, and
interact with it from Kotlin transparently via autogenerated Kotlin like this:
376
class SharedData(rawPtr: NativePtr) : CStructVar(rawPtr) {
var version: Int
var kotlinObject: COpaquePointer?
}
So in combination with the top level variable declared above, it can allow looking at the same
memory from di erent threads and building traditional concurrent structures with platform-
speci c synchronization primitives.
Global variables and singletons
Frequently, global variables are a source of unintended concurrency issues, so Kotlin/Native

implements the following mechanisms to prevent the unintended sharing of state via global
objects:
— global variables, unless specially marked, can be only accessed from the main thread (that is,
the thread Kotlin/Native runtime was rst initialized), if other thread access such a global,
IncorrectDereferenceException is thrown
— for global variables marked with the @kotlin.native.ThreadLocal annotation each
threads keeps thread-local copy, so changes are not visible between threads
— for global variables marked with the @kotlin.native.SharedImmutable annotation
value is shared, but frozen before publishing, so each threads sees the same value
— singleton objects unless marked with @kotlin.native.ThreadLocal are frozen and
shared, lazy values allowed, unless cyclic frozen structures were attempted to be created
— enums are always frozen
Combined, these mechanisms allow natural race-freeze programming with code reuse across
platforms in MPP projects.
377
Immutability in Kotlin/Native
Kotlin/Native implements strict mutability checks, ensuring the important invariant that the
object is either immutable or accessible from the single thread at that moment in time ( mutable
XOR global ).
Immutability is a runtime property in Kotlin/Native, and can be applied to an arbitrary object

subgraph using the kotlin.native.concurrent.freeze function. It makes all the objects
reachable from the given one immutable, such a transition is a one-way operation (i.e., objects
cannot be unfrozen later). Some naturally immutable objects such as kotlin.String ,
kotlin.Int , and other primitive types, along with AtomicInt and AtomicReference are
frozen by default. If a mutating operation is applied to a frozen object, an
InvalidMutabilityException is thrown.
To achieve mutable XOR global invariant, all globally visible state (currently, object
singletons and enums) are automatically frozen. If object freezing is not desired, a
kotlin.native.ThreadLocal annotation can be used, which will make the object state
thread local, and so, mutable (but the changed state is not visible to other threads).
Top level/global variables of non-primitive types are by default accessible in the main thread (i.e.,
the thread which initialized Kotlin/Native runtime rst) only. Access from another thread will lead
to an IncorrectDereferenceException being thrown. To make such variables accessible in
other threads, you can use either the @ThreadLocal annotation, and mark the value thread
local or @SharedImmutable , which will make the value frozen and accessible from other
threads.
Class AtomicReference can be used to publish the changed frozen state to other threads, and
so build patterns like shared caches.
378
Kotlin/Native libraries
Kotlin compiler speci cs

To produce a library with the Kotlin/Native compiler use the -produce library or -p
library ag. For example:
$ kotlinc foo.kt -p library -o bar
the above command will produce a bar.klib with the compiled contents of foo.kt .
To link to a library use the -library <name> or -l <name> ag. For example:
$ kotlinc qux.kt -l bar
the above command will produce a program.kexe out of qux.kt and bar.klib
cinterop tool speci cs

The cinterop tool produces .klib wrappers for native libraries as its main output. For example,
using the simple libgit2.def native library de nition le provided in your Kotlin/Native
distribution
$ cinterop -def samples/gitchurn/src/nativeInterop/cinterop/libgit2.def -compilerOpts -

I/usr/local/include -o libgit2
we will obtain libgit2.klib .
See more details in INTEROP.md
klib utility
The klib library management utility allows you to inspect and install the libraries.
The following commands are available.
To list library contents:
$ klib contents <name>
To inspect the bookkeeping details of the library
$ klib info <name>
To install the library to the default location use
$ klib install <name>
To remove the library from the default repository use
$ klib remove <name>
379
All of the above commands accept an additional -repository <directory> argument for
specifying a repository di erent to the default one.
$ klib <command> <name> -repository <directory>
Several examples
First let's create a library. Place the tiny library source code into kotlinizer.kt :
package kotlinizer
val String.kotlinized
get() = "Kotlin $this"
$ kotlinc kotlinizer.kt -p library -o kotlinizer
The library has been created in the current directory:
$ ls kotlinizer.klib
kotlinizer.klib
Now let's check out the contents of the library:
$ klib contents kotlinizer
We can install kotlinizer to the default repository:
$ klib install kotlinizer
Remove any traces of it from the current directory:
$ rm kotlinizer.klib
Create a very short program and place it into a use.kt :
import kotlinizer.*

println("Hello, ${"world".kotlinized}!")
}
Now compile the program linking with the library we have just created:
$ kotlinc use.kt -l kotlinizer -o kohello
And run the program:
$ ./kohello.kexe
Hello, Kotlin world!
Have fun!
Advanced topics
380
Library search sequence
When given a -library foo ag, the compiler searches the foo library in the following order:
* Current compilation directory or an absolute path.
* All repositories specified with `-repo` flag.
* Libraries installed in the default repository (For now the default is `~/.konan`,
however it could be changed by setting **KONAN_DATA_DIR** environment variable).
* Libraries installed in `$installation/klib` directory.
The library format

Kotlin/Native libraries are zip les containing a prede ned directory structure, with the following
layout:
foo.klib when unpacked as foo/ gives us:
- foo/
- targets/
- $platform/
- kotlin/
- Kotlin compiled to LLVM bitcode.
- native/
- Bitcode files of additional native objects.
- $another_platform/
- There can be several platform specific kotlin and native pairs.
- linkdata/
- A set of ProtoBuf files with serialized linkage metadata.
- resources/
- General resources such as images. (Not used yet).
- manifest - A file in *java property* format describing the library.
An example layout can be found in klib/stdlib directory of your installation.
381
Platform libraries
Overview
To provide access to user's native operating system services, Kotlin/Native distribution
includes a set of prebuilt libraries speci c to each target. We call them Platform Libraries.
POSIX bindings
For all Unix or Windows based targets (including Android and iPhone ) we provide the
posix platform lib. It contains bindings to platform's implementation of POSIX standard.
To use the library just
import platform.posix.*
The only target for which it is not available is WebAssembly.
Note that the content of platform.posix is NOT identical on di erent platforms, in the same
way as di erent POSIX implementations are a little di erent.
Popular native libraries
There are many more platform libraries available for host and cross-compilation targets.
Kotlin/Native distribution provides access to OpenGL , zlib and other popular native
libraries on applicable platforms.
On Apple platforms objc library is provided for interoperability with Objective-C.
Inspect the contents of dist/klib/platform/$target of the distribution for the details.
Availability by default
The packages from platform libraries are available by default. No special link ags need to be
speci ed to use them. Kotlin/Native compiler automatically detects which of the platform
libraries have been accessed and automatically links the needed libraries.
On the other hand, the platform libs in the distribution are merely just wrappers and bindings to
the native libraries. That means the native libraries themselves ( .so , .a , .dylib , .dll etc)
should be installed on the machine.
Examples
Kotlin/Native installation provides a wide spectrum of examples demonstrating the use of
platform libraries. See samples for details.
382
Kotlin/Native interoperability
Introduction
Kotlin/Native follows the general tradition of Kotlin to provide excellent existing platform software
interoperability. In the case of a native platform, the most important interoperability target is a C
library. So Kotlin/Native comes with a cinterop tool, which can be used to quickly generate
everything needed to interact with an external library.
The following work ow is expected when interacting with the native library.
— create a .def le describing what to include into bindings

— use the cinterop tool to produce Kotlin bindings
— run Kotlin/Native compiler on an application to produce the nal executable
The interoperability tool analyses C headers and produces a "natural" mapping of the types,
functions, and constants into the Kotlin world. The generated stubs can be imported into an IDE
for the purpose of code completion and navigation.
Interoperability with Swift/Objective-C is provided too and covered in a separate document

OBJC_INTEROP.md.
Platform libraries
Note that in many cases there's no need to use custom interoperability library creation
mechanisms described below, as for APIs available on the platform standardized bindings called
platform libraries could be used. For example, POSIX on Linux/macOS platforms, Win32 on
Windows platform, or Apple frameworks on macOS/iOS are available this way.
Simple example
Install libgit2 and prepare stubs for the git library:
cd samples/gitchurn
../../dist/bin/cinterop -def src/main/c_interop/libgit2.def \
-compilerOpts -I/usr/local/include -o libgit2
Compile the client:
../../dist/bin/kotlinc src/main/kotlin \
-library libgit2 -o GitChurn
Run the client:
./GitChurn.kexe ../..
Creating bindings for a new library
383
To create bindings for a new library, start by creating a .def le. Structurally it's a simple
property le, which looks like this:
headers = png.h
headerFilter = png.h
package = png
Then run the cinterop tool with something like this (note that for host libraries that are not
included in the sysroot search paths, headers may be needed):
cinterop -def png.def -compilerOpts -I/usr/local/include -o png
This command will produce a png.klib compiled library and png-build/kotlin directory
containing Kotlin source code for the library.
If the behavior for a certain platform needs to be modi ed, you can use a format like
compilerOpts.osx or compilerOpts.linux to provide platform-speci c values to the
options.
Note, that the generated bindings are generally platform-speci c, so if you are developing for
multiple targets, the bindings need to be regenerated.
After the generation of bindings, they can be used by the IDE as a proxy view of the native library.
For a typical Unix library with a con g script, the compilerOpts will likely contain the output of
a con g script with the --cflags ag (maybe without exact paths).
The output of a con g script with --libs will be passed as a -linkedArgs kotlinc ag
value (quoted) when compiling.
Selecting library headers
When library headers are imported to a C program with the #include directive, all of the
headers included by these headers are also included in the program. So all header dependencies
are included in generated stubs as well.
This behavior is correct but it can be very inconvenient for some libraries. So it is possible to
specify in the .def le which of the included headers are to be imported. The separate
declarations from other headers can also be imported in case of direct dependencies.
Filtering headers by globs
It is possible to lter headers by globs. The headerFilter property value from the .def le is
treated as a space-separated list of globs. If the included header matches any of the globs, then
the declarations from this header are included into the bindings.
384
The globs are applied to the header paths relative to the appropriate include path elements, e.g.
time.h or curl/curl.h . So if the library is usually included with #include
<SomeLbrary/Header.h> , then it would probably be correct to lter headers with
headerFilter = SomeLibrary/**
If a headerFilter is not speci ed, then all headers are included.
Filtering by module maps
Some libraries have proper module.modulemap or module.map les in its headers. For
example, macOS and iOS system libraries and frameworks do. The module map le describes the
correspondence between header les and modules. When the module maps are available, the
headers from the modules that are not included directly can be ltered out using the
experimental excludeDependentModules option of the .def le:
headers = OpenGL/gl.h OpenGL/glu.h GLUT/glut.h

compilerOpts = -framework OpenGL -framework GLUT
excludeDependentModules = true
When both excludeDependentModules and headerFilter are used, they are applied as an
intersection.
C compiler and linker options
Options passed to the C compiler (used to analyze headers, such as preprocessor de nitions) and
the linker (used to link nal executables) can be passed in the de nition le as compilerOpts
and linkerOpts respectively. For example
compilerOpts = -DFOO=bar
linkerOpts = -lpng
Target-speci c options, only applicable to the certain target can be speci ed as well, such as
compilerOpts = -DBAR=bar
compilerOpts.linux_x64 = -DFOO=foo1
compilerOpts.mac_x64 = -DFOO=foo2
and so, C headers on Linux will be analyzed with -DBAR=bar -DFOO=foo1 and on macOS with
-DBAR=bar -DFOO=foo2 . Note that any de nition le option can have both common and the
platform-speci c part.
Adding custom declarations
385
Sometimes it is required to add custom C declarations to the library before generating bindings
(e.g., for macros). Instead of creating an additional header le with these declarations, you can
include them directly to the end of the .def le, after a separating line, containing only the
separator sequence --- :
headers = errno.h
---
static inline int getErrno() {

return errno;
}
Note that this part of the .def le is treated as part of the header le, so functions with the
body should be declared as static . The declarations are parsed after including the les from
the headers list.
Including static library in your klib
Sometimes it is more convenient to ship a static library with your product, rather than assume it
is available within the user's environment. To include a static library into .klib use
staticLibrary and libraryPaths clauses. For example:
staticLibraries = libfoo.a
libraryPaths = /opt/local/lib /usr/local/opt/curl/lib
When given the above snippet the cinterop tool will search libfoo.a in /opt/local/lib
and /usr/local/opt/curl/lib , and if it is found include the library binary into klib .
When using such klib in your program, the library is linked automatically.
Using bindings
Basic interop types
All the supported C types have corresponding representations in Kotlin:
— Signed, unsigned integral, and oating point types are mapped to their Kotlin counterpart with
the same width.
— Pointers and arrays are mapped to CPointer<T>? .
— Enums can be mapped to either Kotlin enum or integral values, depending on heuristics and
the de nition le hints.
— Structs / unions are mapped to types having elds available via the dot notation, i.e.
someStructInstance.field1 .
— typedef are represented as typealias .
386
Also, any C type has the Kotlin type representing the lvalue of this type, i.e., the value located in
memory rather than a simple immutable self-contained value. Think C++ references, as a similar
concept. For structs (and typedef s to structs) this representation is the main one and has the
same name as the struct itself, for Kotlin enums it is named ${type}Var , for CPointer<T> it
is CPointerVar<T> , and for most other types it is ${type}Var .
For types that have both representations, the one with a "lvalue" has a mutable .value
property for accessing the value.
Pointer types
The type argument T of CPointer<T> must be one of the "lvalue" types described above, e.g.,
the C type struct S* is mapped to CPointer<S> , int8_t* is mapped to
CPointer<int_8tVar> , and char** is mapped to CPointer<CPointerVar<ByteVar>> .
C null pointer is represented as Kotlin's null , and the pointer type CPointer<T> is not
nullable, but the CPointer<T>? is. The values of this type support all the Kotlin operations
related to handling null , e.g. ?: , ?. , !! etc.:
val path = getenv("PATH")?.toKString() ?: ""
Since the arrays are also mapped to CPointer<T> , it supports the [] operator for accessing
values by index:
fun shift(ptr: CPointer<BytePtr>, length: Int) {

for (index in 0 .. length - 2) {
ptr[index] = ptr[index + 1]
}
}
The .pointed property for CPointer<T> returns the lvalue of type T , pointed by this
pointer. The reverse operation is .ptr : it takes the lvalue and returns the pointer to it.
void* is mapped to COpaquePointer – the special pointer type which is the supertype for
any other pointer type. So if the C function takes void* , then the Kotlin binding accepts any
CPointer .
Casting a pointer (including COpaquePointer ) can be done with .reinterpret<T> , e.g.:
val intPtr = bytePtr.reinterpret<IntVar>()
or
val intPtr: CPointer<IntVar> = bytePtr.reinterpret()
As is with C, these reinterpret casts are unsafe and can potentially lead to subtle memory
problems in the application.
387
Also there are unsafe casts between CPointer<T>? and Long available, provided by the
.toLong() and .toCPointer<T>() extension methods:
val longValue = ptr.toLong()

val originalPtr = longValue.toCPointer<T>()
Note that if the type of the result is known from the context, the type argument can be omitted
as usual due to the type inference.
Memory allocation
The native memory can be allocated using the NativePlacement interface, e.g.
val byteVar = placement.alloc<ByteVar>()
or
val bytePtr = placement.allocArray<ByteVar>(5)
The most "natural" placement is in the object nativeHeap . It corresponds to allocating native
memory with malloc and provides an additional .free() operation to free allocated
memory:
val buffer = nativeHeap.allocArray<ByteVar>(size)

<use buffer>
nativeHeap.free(buffer)
However, the lifetime of allocated memory is often bound to the lexical scope. It is possible to
de ne such scope with memScoped { ... } . Inside the braces, the temporary placement is
available as an implicit receiver, so it is possible to allocate native memory with alloc and
allocArray , and the allocated memory will be automatically freed after leaving the scope.
For example, the C function returning values through pointer parameters can be used like
val fileSize = memScoped {

val statBuf = alloc<stat>()
val error = stat("/", statBuf.ptr)
statBuf.st_size
}
Passing pointers to bindings
Although C pointers are mapped to the CPointer<T> type, the C function pointer-typed
parameters are mapped to CValuesRef<T> . When passing CPointer<T> as the value of such
a parameter, it is passed to the C function as is. However, the sequence of values can be passed
instead of a pointer. In this case the sequence is passed "by value", i.e., the C function receives
the pointer to the temporary copy of that sequence, which is valid only until the function returns.
388
The CValuesRef<T> representation of pointer parameters is designed to support C array
literals without explicit native memory allocation. To construct the immutable self-contained
sequence of C values, the following methods are provided:
— ${type}Array.toCValues() , where type is the Kotlin primitive type
— Array<CPointer<T>?>.toCValues() , List<CPointer<T>?>.toCValues()
— cValuesOf(vararg elements: ${type}) , where type is a primitive or pointer
For example:
C:
void foo(int* elements, int count);

...
int elements[] = {1, 2, 3};
foo(elements, 3);
Kotlin:
foo(cValuesOf(1, 2, 3), 3)
Working with the strings
Unlike other pointers, the parameters of type const char* are represented as a Kotlin
String . So it is possible to pass any Kotlin string to a binding expecting a C string.
There are also some tools available to convert between Kotlin and C strings manually:
— fun CPointer<ByteVar>.toKString(): String
— val String.cstr: CValuesRef<ByteVar> .
To get the pointer, .cstr should be allocated in native memory, e.g.
val cString = kotlinString.cstr.getPointer(nativeHeap)
In all cases, the C string is supposed to be encoded as UTF-8.
To skip automatic conversion and ensure raw pointers are used in the bindings, a
noStringConversion statement in the .def le could be used, i.e.
noStringConversion = LoadCursorA LoadCursorW
This way any value of type CPointer<ByteVar> can be passed as an argument of const
char* type. If a Kotlin string should be passed, code like this could be used:
memScoped {
LoadCursorA(null, "cursor.bmp".cstr.ptr) // for ASCII version
LoadCursorW(null, "cursor.bmp".wcstr.ptr) // for Unicode version
}
389
Scope-local pointers
It is possible to create a scope-stable pointer of C representation of CValues<T> instance using

the CValues<T>.ptr extension property, available under memScoped { ... } . It allows
using the APIs which require C pointers with a lifetime bound to a certain MemScope . For
example:
memScoped {
items = arrayOfNulls<CPointer<ITEM>?>(6)
arrayOf("one", "two").forEachIndexed { index, value -> items[index] = value.cstr.ptr
}
menu = new_menu("Menu".cstr.ptr, items.toCValues().ptr)
...
}
In this example, all values passed to the C API new_menu() have a lifetime of the innermost
memScope it belongs to. Once the control ow leaves the memScoped scope the C pointers
become invalid.
Passing and receiving structs by value
When a C function takes or returns a struct / union T by value, the corresponding argument type
or return type is represented as CValue<T> .
CValue<T> is an opaque type, so the structure elds cannot be accessed with the appropriate
Kotlin properties. It should be possible, if an API uses structures as handles, but if eld access is
required, there are the following conversion methods available:
— fun T.readValue(): CValue<T> . Converts (the lvalue) T to a CValue<T> . So to

construct the CValue<T> , T can be allocated, lled, and then converted to CValue<T> .
— CValue<T>.useContents(block: T.() -> R): R . Temporarily places the CValue<T>

to memory, and then runs the passed lambda with this placed value T as receiver. So to read
a single eld, the following code can be used:
val fieldValue = structValue.useContents { field }
Callbacks
To convert a Kotlin function to a pointer to a C function,

staticCFunction(::kotlinFunction) can be used. It is also able to provide the lambda
instead of a function reference. The function or lambda must not capture any values.
If the callback doesn't run in the main thread, it is mandatory to init the Kotlin/Native runtime by
calling kotlin.native.initRuntimeIfNeeded() .
Passing user data to callbacks
390
Often C APIs allow passing some user data to callbacks. Such data is usually provided by the user
when con guring the callback. It is passed to some C function (or written to the struct) as e.g.
void* . However, references to Kotlin objects can't be directly passed to C. So they require
wrapping before con guring the callback and then unwrapping in the callback itself, to safely
swim from Kotlin to Kotlin through the C world. Such wrapping is possible with StableRef
class.
To wrap the reference:
val stableRef = StableRef.create(kotlinReference)

val voidPtr = stableRef.asCPointer()
where the voidPtr is a COpaquePointer and can be passed to the C function.
To unwrap the reference:
val stableRef = voidPtr.asStableRef<KotlinClass>()

val kotlinReference = stableRef.get()
where kotlinReference is the original wrapped reference.
The created StableRef should eventually be manually disposed using the .dispose()
method to prevent memory leaks:
stableRef.dispose()
After that it becomes invalid, so voidPtr can't be unwrapped anymore.
See the samples/libcurl for more details.
Macros
Every C macro that expands to a constant is represented as a Kotlin property. Other macros are
not supported. However, they can be exposed manually by wrapping them with supported
declarations. E.g. function-like macro FOO can be exposed as function foo by adding the
custom declaration to the library:
headers = library/base.h
---
static inline int foo(int arg) {

return FOO(arg);
}
De nition le hints
The .def le supports several options for adjusting the generated bindings.
391
— excludedFunctions property value speci es a space-separated list of the names of
functions that should be ignored. This may be required because a function declared in the C
header is not generally guaranteed to be really callable, and it is often hard or impossible to
gure this out automatically. This option can also be used to workaround a bug in the interop
itself.
— strictEnums and nonStrictEnums properties values are space-separated lists of the

enums that should be generated as a Kotlin enum or as integral values correspondingly. If the
enum is not included into any of these lists, then it is generated according to the heuristics.
— noStringConversion property value is space-separated lists of the functions whose

const char* parameters shall not be autoconverted as Kotlin string
Portability
Sometimes the C libraries have function parameters or struct elds of a platform-dependent

type, e.g. long or size_t . Kotlin itself doesn't provide neither implicit integer casts nor C-style
integer casts (e.g. (size_t) intValue ), so to make writing portable code in such cases easier,
the convert method is provided:
fun ${type1}.convert<${type2}>(): ${type2}
where each of type1 and type2 must be an integral type, either signed or unsigned.
.convert<${type}> has the same semantics as one of the .toByte , .toShort , .toInt ,
.toLong , .toUByte , .toUShort , .toUInt or .toULong methods, depending on type .
The example of using convert :
fun zeroMemory(buffer: COpaquePointer, size: Int) {

memset(buffer, 0, size.convert<size_t>())
}
Also, the type parameter can be inferred automatically and so may be omitted in some cases.
Object pinning
Kotlin objects could be pinned, i.e. their position in memory is guaranteed to be stable until
unpinned, and pointers to such objects inner data could be passed to the C functions. For
example
392
fun readData(fd: Int): String {
val buffer = ByteArray(1024)
buffer.usePinned { pinned ->
while (true) {
val length = recv(fd, pinned.addressOf(0), buffer.size.convert(), 0).toInt()
if (length <= 0) {
break
}
// Now `buffer` has raw data obtained from the `recv()` call.
}
}
}
Here we use service function usePinned , which pins an object, executes block and unpins it on
normal and exception paths.
393
Kotlin/Native interoperability with Swift/Objective-C
This document covers some details of Kotlin/Native interoperability with Swift/Objective-C.
Usage
Kotlin/Native provides bidirectional interoperability with Objective-C. Objective-C frameworks
and libraries can be used in Kotlin code if properly imported to the build (system frameworks are
imported by default). See e.g. "Using cinterop" in Gradle plugin documentation. A Swift library
can be used in Kotlin code if its API is exported to Objective-C with @objc . Pure Swift modules
are not yet supported.
Kotlin modules can be used in Swift/Objective-C code if compiled into a framework (see "Targets
and output kinds" section in Gradle plugin documentation). See calculator sample for an
example.
Mappings
The table below shows how Kotlin concepts are mapped to Swift/Objective-C and vice versa.
Kotlin Swift Objective-C Notes

class class @interface note
interface protocol @protocol
constructor/create Initializer Initializer note
Property Property Property note
Method Method Method note note
@Throws throws error:(NSError**)error note
Extension Extension Category member note
companion member <- Class method or property Class method or property
null nil nil
Singleton Singleton() [Singleton singleton] note
Primitive type Primitive type / NSNumber note
Unit return type Void void
String String NSString
String NSMutableString NSMutableString note
List Array NSArray
MutableList NSMutableArray NSMutableArray
Set Set NSSet
MutableSet NSMutableSet NSMutableSet note
Map Dictionary NSDictionary
MutableMap NSMutableDictionary NSMutableDictionary note
Function type Function type Block pointer type note
Name translation
394
Objective-C classes are imported into Kotlin with their original names. Protocols are imported as
interfaces with Protocol name su x, i.e. @protocol Foo -> interface FooProtocol .
These classes and interfaces are placed into a package speci ed in build con guration
( platform.* packages for precon gured system frameworks).
The names of Kotlin classes and interfaces are pre xed when imported to Objective-C. The pre x
is derived from the framework name.
Initializers
Swift/Objective-C initializers are imported to Kotlin as constructors and factory methods named
create . The latter happens with initializers declared in the Objective-C category or as a Swift
extension, because Kotlin has no concept of extension constructors.
Kotlin constructors are imported as initializers to Swift/Objective-C.
Top-level functions and properties
Top-level Kotlin functions and properties are accessible as members of special classes. Each
Kotlin le is translated into such a class. E.g.
// MyLibraryUtils.kt
package my.library
fun foo() {}
can be called from Swift like
MyLibraryUtilsKt.foo()
Method names translation
Generally Swift argument labels and Objective-C selector pieces are mapped to Kotlin parameter
names. Anyway these two concepts have di erent semantics, so sometimes Swift/Objective-C
methods can be imported with a clashing Kotlin signature. In this case the clashing methods can
be called from Kotlin using named arguments, e.g.:
[player moveTo:LEFT byMeters:17]

[player moveTo:UP byInches:42]
in Kotlin it would be:
player.moveTo(LEFT, byMeters = 17)

player.moveTo(UP, byInches = 42)
Errors and exceptions
395
Kotlin has no concept of checked exceptions, all Kotlin exceptions are unchecked. Swift has only
checked errors. So if Swift or Objective-C code calls a Kotlin method which throws an exception to
be handled, then the Kotlin method should be marked with a @Throws annotation. In this case
all Kotlin exceptions (except for instances of Error , RuntimeException and subclasses) are
translated into a Swift error/ NSError .
Note that the opposite reversed translation is not implemented yet: Swift/Objective-C error-
throwing methods aren't imported to Kotlin as exception-throwing.
Category members
Members of Objective-C categories and Swift extensions are imported to Kotlin as extensions.
That's why these declarations can't be overridden in Kotlin. And the extension initializers aren't
available as Kotlin constructors.
Kotlin singletons
Kotlin singleton (made with an object declaration, including companion object ) is imported
to Swift/Objective-C as a class with a single instance. The instance is available through the factory
method, i.e. as [MySingleton mySingleton] in Objective-C and MySingleton() in Swift.
NSNumber
Kotlin primitive type boxes are mapped to special Swift/Objective-C classes. For example,
kotlin.Int box is represented as KotlinInt class instance in Swift (or ${prefix}Int
instance in Objective-C, where prefix is the framework names pre x). These classes are
derived from NSNumber , so the instances are proper NSNumber s supporting all corresponding
operations.
NSNumber type is not automatically translated to Kotlin primitive types when used as a
Swift/Objective-C parameter type or return value. The reason is that NSNumber type doesn't
provide enough information about a wrapped primitive value type, i.e. NSNumber is statically
not known to be a e.g. Byte , Boolean , or Double . So Kotlin primitive values should be cast
to/from NSNumber manually (see below).
NSMutableString
NSMutableString Objective-C class is not available from Kotlin. All instances of

NSMutableString are copied when passed to Kotlin.
Collections
396
Kotlin collections are converted to Swift/Objective-C collections as described in the table above.
Swift/Objective-C collections are mapped to Kotlin in the same way, except for NSMutableSet
and NSMutableDictionary . NSMutableSet isn't converted to a Kotlin MutableSet . To
pass an object for Kotlin MutableSet , you can create this kind of Kotlin collection explicitly by
either creating it in Kotlin with e.g. mutableSetOf() , or using the KotlinMutableSet class
in Swift (or ${prefix}MutableSet in Objective-C, where prefix is the framework names
pre x). The same holds for MutableMap .
Function types
Kotlin function-typed objects (e.g. lambdas) are converted to Swift functions / Objective-C blocks.
However there is a di erence in how types of parameters and return values are mapped when
translating a function and a function type. In the latter case primitive types are mapped to their
boxed representation. Kotlin Unit return value is represented as a corresponding Unit
singleton in Swift/Objective-C. The value of this singleton can be retrieved in the same way as it is
for any other Kotlin object (see singletons in the table above). To sum the things up:
fun foo(block: (Int) -> Unit) { ... }
would be represented in Swift as
func foo(block: (KotlinInt) -> KotlinUnit)
and can be called like
foo {
bar($0 as! Int32)
return KotlinUnit()
}
Casting between mapped types

When writing Kotlin code, an object may need to be converted from a Kotlin type to the
equivalent Swift/Objective-C type (or vice versa). In this case a plain old Kotlin cast can be used,
e.g.
val nsArray = listOf(1, 2, 3) as NSArray

val string = nsString as String
val nsNumber = 42 as NSNumber
Subclassing
Subclassing Kotlin classes and interfaces from Swift/Objective-C
Kotlin classes and interfaces can be subclassed by Swift/Objective-C classes and protocols.
Subclassing Swift/Objective-C classes and protocols from Kotlin
397
Swift/Objective-C classes and protocols can be subclassed with a Kotlin final class. Non- final
Kotlin classes inheriting Swift/Objective-C types aren't supported yet, so it is not possible to
declare a complex class hierarchy inheriting Swift/Objective-C types.
Normal methods can be overridden using the override Kotlin keyword. In this case the
overriding method must have the same parameter names as the overridden one.
Sometimes it is required to override initializers, e.g. when subclassing UIViewController .

Initializers imported as Kotlin constructors can be overridden by Kotlin constructors marked with
the @OverrideInit annotation:
class ViewController : UIViewController {

@OverrideInit constructor(coder: NSCoder) : super(coder)
...
}
The overriding constructor must have the same parameter names and types as the overridden
one.
To override di erent methods with clashing Kotlin signatures, you can add a
@Suppress("CONFLICTING_OVERLOADS") annotation to the class.
By default the Kotlin/Native compiler doesn't allow calling a non-designated Objective-C initializer
as a super(...) constructor. This behaviour can be inconvenient if the designated initializers
aren't marked properly in the Objective-C library. Adding a
disableDesignatedInitializerChecks = true to the .def le for this library would
disable these compiler checks.
C features
See INTEROP.md for an example case where the library uses some plain C features (e.g. unsafe
pointers, structs etc.).
398
CocoaPods integration
Starting with 1.3.30, an experimental integration with CocoaPods is added to Kotlin/Native. This
feature allows you to represent a Kotlin/Native Gradle-project as a CocoaPods dependency. Such
a representation provides the following advantages:
— Such a dependency can be included in a Pod le of an Xcode project and automatically built
(and rebuilt) along with this project. As a result, importing to Xcode is simpli ed since there is
no need to write corresponding Gradle tasks and Xcode build steps manually.
— When building from Xcode, you can use CocoaPods libraries without writing .def les
manually and setting cinterop tool parameters. In this case, all required parameters can be
obtained from the Xcode project con gured by CocoaPods.
For an example of CocoaPods integration, refer to the cocoapods sample.
CocoaPods Gradle plugin

The CocoaPods support is implemented in a separate Gradle plugin:
org.jetbrains.kotlin.native.cocoapods .
Note: The plugin is based on the multiplatform project model and requires applying the
org.jetbrains.kotlin.multiplatform plugin. See details about the multiplatform
plugin at the corresponding page.
When applied, the CocoaPods plugin does the following:
1. Adds both debug and release frameworks as output binaries for all iOS and macOS targets.
2. Creates a podspec task which generates a podspec le for the given project.
The podspec generated includes a path to an output framework and script phases which
automate building this framework during a build process of an Xcode project. Some elds of the
podspec le can be con gured using the kotlin.cocoapods { ... } code block.
// Apply plugins.
plugins {
id("org.jetbrains.kotlin.multiplatform") version "1.3.30"
id("org.jetbrains.kotlin.native.cocoapods") version "1.3.30"
}
// CocoaPods requires the podspec to have a version.

version = "1.0"
kotlin {
cocoapods {
// Configure fields required by CocoaPods.
summary = "Some description for a Kotlin/Native module"
homepage = "Link to a Kotlin/Native module homepage"
}
}
399
The following podspec elds are required by CocoaPods:
— version
— summary
— homepage
A version of the Gradle project is used as a value for the version eld. Fields summary and
homepage can be con gured using the cocoapods code block.
This podspec le can be referenced from a Pod le of an Xcode project. After that the framework
built from the Kotlin/Native module can be used from this Xcode project. If necessary, this
framework is automatically rebuilt during Xcode build process.
Work ow
To import a Kotlin/Native module in an existing Xcode project:
1. Make sure that you have CocoaPods installed. We recommend using CocoaPods 1.6.1 or later.
2. Con gure a Gradle project: apply the org.jetbrains.kotlin.native.cocoapods

plugin, add and con gure the targets, and specify the required podspec elds.
3. Run the podspec task. The podspec le described above will be generated.
In order to avoid compatibility issues during an Xcode build, the plugin requires using a
Gradle wrapper. To generate the wrapper automatically during execution of the podspec
task, run it with the parameter -
Pkotlin.native.cocoapods.generate.wrapper=true.
4. Add a reference to the generated podspec in a Pod le of the Xcode project.
target 'my-ios-app' do
pod 'my_kotlin_library', :path => 'path/to/my-kotlin-library'
end
5. Run pod install for the Xcode project.
After completing these steps, you can open the generated workspace (see CocoaPods
documentation) and run an Xcode build.
Interoperability
The CocoaPods plugin also allows using CocoaPods libraries without manual con guring cinterop
parameters (see the corresponding section of the multiplatform plugin documentation). The
cocoapods { ... } code block allows you to add dependencies on CocoaPods libraries.
400
kotlin {
cocoapods {
// Configure a dependency on AFNetworking.
// The CocoaPods version notation is supported.
// The dependency will be added to all macOS and iOS targets.
pod("AFNetworking", "~> 3.2.0")
}
}
To use these dependencies from a Kotlin code, import a package cocoapods.<library-

name> . In the example above, it's cocoapods.AFNetworking .
The dependencies declared in this way are added in the podspec le and downloaded during the
execution of pod install .
Important: To correctly import the dependencies into the Kotlin/Native module, the Pod le
must contain either use_modular_headers! or use_frameworks! directive.
Search paths for libraries added in the Kotlin/Native module in this way are obtained from
properties of the Xcode projects con gured by CocoaPods. Thus if the module uses CocoaPods
libraries, it can be build only from Xcode.
Current Limitations
— If a Kotlin/Native module uses a CocoaPods library, you can built this module only from an
Xcode project. Otherwise the CocoaPods library cannot be resolved by the Kotlin/Native
infrastructure.
— Subspecs are not supported.
401
Kotlin/Native Gradle plugin
IMPORTANT NOTICE
This document describes Kotlin/Native experimental Gradle plugin, which is not the plugin yet
supported by IDE or in multiplatform projects. See MPP Gradle plugin documentation for more
information.
Overview
You may use the Gradle plugin to build Kotlin/Native projects. Builds of the plugin are available at
the Gradle plugin portal, so you can apply it using Gradle plugin DSL:
plugins {
id "org.jetbrains.kotlin.platform.native" version "1.3.0-rc-146"
}
You also can get the plugin from a Bintray repository. In addition to releases, this repo contains
old and development versions of the plugin which are not available at the plugin portal. To get
the plugin from the Bintray repo, include the following snippet in your build script:
buildscript {
repositories {
mavenCentral()
maven {
url "https://dl.bintray.com/jetbrains/kotlin-native-dependencies"
}
}
dependencies {
classpath "org.jetbrains.kotlin:kotlin-native-gradle-plugin:1.3.0-rc-146"
}
}
apply plugin: 'org.jetbrains.kotlin.platform.native'
By default the plugin downloads the Kotlin/Native compiler during the rst run. If you have
already downloaded the compiler manually you can specify the path to its root directory using
org.jetbrains.kotlin.native.home project property (e.g. in gradle.properties ).
org.jetbrains.kotlin.native.home=/home/user/kotlin-native-0.8
In this case the compiler will not be downloaded by the plugin.
Source management
402
Source management in the kotlin.platform.native plugin is uniform with other Kotlin
plugins and is based on source sets. A source set is a group of Kotlin/Native source which may
contain both common and platform-speci c code. The plugin provides a top-level script block
sourceSets allowing you to con gure source sets. Also it creates the default source sets main
and test (for production and test code respectively).
By default the production sources are located in src/main/kotlin and the test sources - in
src/test/kotlin .
sourceSets {
// Adding target-independent sources.
main.kotlin.srcDirs += 'src/main/mySources'
// Adding Linux-specific code. It will be compiled in Linux binaries only.

main.target('linux_x64').srcDirs += 'src/main/linux'
}
Targets and output kinds
By default the plugin creates software components for the main and test source sets. You can
access them via the components container provided by Gradle or via the component property
of a corresponding source set:
// Main component.
components.main
sourceSets.main.component
// Test component.
components.test
sourceSets.test.component
Components allow you to specify:
— Targets (e.g. Linux/x64 or iOS/arm64 etc)

— Output kinds (e.g. executable, library, framework etc)
— Dependencies (including interop ones)
Targets can be speci ed by setting a corresponding component property:
components.main {
// Compile this component for 64-bit MacOS, Linux and Windows.
targets = ['macos_x64', 'linux_x64', 'mingw_x64']
}
The plugin uses the same notation as the compiler. By default, test component uses the same
targets as speci ed for the main one.
Output kinds can also be speci ed using a special property:
403
components.main {
// Compile the component into an executable and a Kotlin/Native library.
outputKinds = [EXECUTABLE, KLIBRARY]
}
All constants used here are available inside a component con guration script block. The plugin
supports producing binaries of the following kinds:
— EXECUTABLE - an executable le;
— KLIBRARY - a Kotlin/Native library (*.klib);
— FRAMEWORK - an Objective-C framework;
— DYNAMIC - shared native library;
— STATIC - static native library.
Also each native binary is built in two variants (build types): debug (debuggable, not optimized)
and release (not debuggable, optimized). Note that Kotlin/Native libraries have only debug
variant because optimizations are preformed only during compilation of a nal binary
(executable, static lib etc) and a ect all libraries used to build it.
Compile tasks
The plugin creates a compilation task for each combination of the target, output kind, and build
type. The tasks have the following naming convention:
compile<ComponentName><BuildType><OutputKind><Target>KotlinNative
For example compileDebugKlibraryMacos_x64KotlinNative ,

compileTestDebugKotlinNative .
The name contains the following parts (some of them may be empty):
— <ComponentName> - name of a component. Empty for the main component.
— <BuildType> - Debug or Release .
— <OutputKind> - output kind name, e.g. Executabe or Dynamic . Empty if the component
has only one output kind.
— <Target> - target the component is built for, e.g. Macos_x64 or Wasm32 . Empty if the
component is built only for one target.
Also the plugin creates a number of aggregate tasks allowing you to build all the binaries for a
build type (e.g. assembleAllDebug ) or all the binaries for a particular target (e.g.
assembleAllWasm32 ).
Basic lifecycle tasks like assemble , build , and clean are also available.
Running tests
404
The plugin builds a test executable for all the targets speci ed for the test component. If the
current host platform is included in this list the test running tasks are also created. To run tests,
execute the standard lifecycle check task:
./gradlew check
Dependencies
The plugin allows you to declare dependencies on les and other projects using traditional
Gradle's mechanism of con gurations. The plugin supports Kotlin multiplatform projects allowing
you to declare the expectedBy dependencies
dependencies {
implementation files('path/to/file/dependencies')
implementation project('library')
testImplementation project('testLibrary')
expectedBy project('common')
}
It's possible to depend on a Kotlin/Native library published earlier in a maven repo. The plugin
relies on Gradle's metadata support so the corresponding feature must be enabled. Add the
following line in your settings.gradle :
enableFeaturePreview('GRADLE_METADATA')
Now you can declare a dependency on a Kotlin/Native library in the traditional

group:artifact:version notation:
dependencies {
implementation 'org.sample.test:mylibrary:1.0'
testImplementation 'org.sample.test:testlibrary:1.0'
}
Dependency declaraion is also possible in the component block:
components.main {
dependencies {
implementation 'org.sample.test:mylibrary:1.0'
}
}
components.test {
dependencies {
implementation 'org.sample.test:testlibrary:1.0'
}
}
Using cinterop
It's possible to declare a cinterop dependency for a component:
405
components.main {
dependencies {
cinterop('mystdio') {
// src/main/c_interop/mystdio.def is used as a def file.
// Set up compiler options

compilerOpts '-I/my/include/path'
// It's possible to set up different options for different targets

target('linux') {
compilerOpts '-I/linux/include/path'
}
}
}
}
Here an interop library will be built and added in the component dependencies.
Often it's necessary to specify target-speci c linker options for a Kotlin/Native binary using an
interop. It can be done using the target script block:
components.main {
target('linux') {
linkerOpts '-L/path/to/linux/libs'
}
}
Also the allTargets block is available.
components.main {
// Configure all targets.
allTargets {
linkerOpts '-L/path/to/libs'
}
}
Publishing
In the presence of maven-publish plugin the publications for all the binaries built are created.
The plugin uses Gradle metadata to publish the artifacts so this feature must be enabled (see the
dependencies section).
Now you can publish the artifacts with the standard Gradle publish task:
./gradlew publish
Only EXECUTABLE and KLIBRARY binaries are published currently.
The plugin allows you to customize the pom generated for the publication with the pom code
block available for every component:
406
components.main {
pom {
withXml {
def root = asNode()
root.appendNode('name', 'My library')
root.appendNode('description', 'A Kotlin/Native library')
}
}
}
Serialization plugin
The plugin is shipped with a customized version of the kotlinx.serialization plugin. To

use it you don't have to add new buildscript dependencies, just apply the plugins and add a
dependency on the serialization library:
apply plugin: 'org.jetbrains.kotlin.platform.native'

apply plugin: 'kotlinx-serialization-native'
dependencies {
implementation 'org.jetbrains.kotlinx:kotlinx-serialization-runtime-native'
}
The the example project for details.
DSL example
In this section a commented DSL is shown. See also the example projects that use this plugin, e.g.
Kotlinx.coroutines, MPP http client
plugins {
id "org.jetbrains.kotlin.platform.native" version "1.3.0-rc-146"
}
sourceSets.main {
// Plugin uses Gradle's source directory sets here,
// so all the DSL methods available in SourceDirectorySet can be called here.
// Platform independent sources.
kotlin.srcDirs += 'src/main/customDir'
// Linux-specific sources
target('linux').srcDirs += 'src/main/linux'
}
components.main {
// Set up targets
targets = ['linux_x64', 'macos_x64', 'mingw_x64']
// Set up output kinds

outputKinds = [EXECUTABLE, KLIBRARY, FRAMEWORK, DYNAMIC, STATIC]
// Specify custom entry point for executables

entryPoint = "org.test.myMain"
// Target-specific options
407
target('linux_x64') {
linkerOpts '-L/linux/lib/path'
}
// Targets independent options

allTargets {
linkerOpts '-L/common/lib/path'
}
dependencies {
// Dependency on a published Kotlin/Native library.

implementation 'org.test:mylib:1.0'
// Dependency on a project
implementation project('library')
// Cinterop dependency
cinterop('interop-name') {
// The default path is src/main/c_interop/<interop-name>.def
defFile project.file("deffile.def")

packageName 'org.sample'
// Options to be passed to compiler and linker by cinterop tool.

compilerOpts 'Options for native stubs compilation'
linkerOpts 'Options for native stubs'
// Additional headers to parse.

headers project.files('header1.h', 'header2.h')
// Directories to look for headers.

includeDirs {
// All objects accepted by the Project.file method may be used with both
options.
// Directories for header search (an analogue of the -I<path> compiler

option).
allHeaders 'path1', 'path2'
// Additional directories to search headers listed in the 'headerFilter'

def-file option.
headerFilterOnly 'path1', 'path2'
}
includeDirs "include/directory" "another/directory"
// Pass additional command line options to the cinterop tool.

extraOpts '-shims', 'true'
// Additional configuration for Linux.

target('linux') {
compilerOpts 'Linux-specific options'
}
}
}
// Additional pom settings for publication.
408
pom {
withXml {
def root = asNode()
root.appendNode('name', 'My library')
root.appendNode('description', 'A Kotlin/Native library')
}
}
// Additional options passed to the compiler.

extraOpts '--time'
}
409
Debugging
Currently the Kotlin/Native compiler produces debug info compatible with the DWARF 2
speci cation, so modern debugger tools can perform the following operations:
— breakpoints
— stepping
— inspection of type information
— variable inspection
Producing binaries with debug info with Kotlin/Native compiler
To produce binaries with the Kotlin/Native compiler it's su cient to use the -g option on the
command line.
Example:
0:b-debugger-fixes:minamoto@unit-703(0)# cat - > hello.kt

println("Hello world")
println("I need your clothes, your boots and your motocycle")
}
0:b-debugger-fixes:minamoto@unit-703(0)# dist/bin/konanc -g hello.kt -o terminator
KtFile: hello.kt
0:b-debugger-fixes:minamoto@unit-703(0)# lldb terminator.kexe
(lldb) target create "terminator.kexe"
Current executable set to 'terminator.kexe' (x86_64).
(lldb) b kfun:main(kotlin.Array<kotlin.String>)
Breakpoint 1: where = terminator.kexe`kfun:main(kotlin.Array<kotlin.String>) + 4 at
hello.kt:2, address = 0x00000001000012e4
(lldb) r
Process 28473 launched: '/Users/minamoto/ws/.git-trees/debugger-fixes/terminator.kexe'
(x86_64)
Process 28473 stopped
* thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.1
frame #0: 0x00000001000012e4 terminator.kexe`kfun:main(kotlin.Array<kotlin.String>)
at hello.kt:2
1 fun main(args: Array<String>) {
-> 2 println("Hello world")
3 println("I need your clothes, your boots and your motocycle")
4 }
(lldb) n
Hello world
* thread #1, queue = 'com.apple.main-thread', stop reason = step over
frame #0: 0x00000001000012f0 terminator.kexe`kfun:main(kotlin.Array<kotlin.String>)
at hello.kt:3
2 println("Hello world")
-> 3 println("I need your clothes, your boots and your motocycle")
4 }
(lldb)
Breakpoints
410
Modern debuggers provide several ways to set a breakpoint, see below for a tool-by-tool
breakdown:
lldb
— by name
(lldb) b -n kfun:main(kotlin.Array<kotlin.String>)
-n is optional, this ag is applied by default
— by location ( lename, line number)
(lldb) b -f hello.kt -l 1
— by address
(lldb) b -a 0x00000001000012e4
Breakpoint 2: address = 0x00000001000012e4
— by regex, you might nd it useful for debugging generated artifacts, like lambda etc. (where
used # symbol in name).
3: regex = 'main\(', locations = 1

3.1: where = terminator.kexe`kfun:main(kotlin.Array<kotlin.String>) + 4 at hello.kt:2,
address = terminator.kexe[0x00000001000012e4], unresolved, hit count = 0
gdb
— by regex
(gdb) rbreak main(

Breakpoint 1 at 0x1000109b4
struct ktype:kotlin.Unit &kfun:main(kotlin.Array<kotlin.String>);
— by name unusable, because : is a separator for the breakpoint by location
(gdb) b kfun:main(kotlin.Array<kotlin.String>)
No source file named kfun.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (kfun:main(kotlin.Array<kotlin.String>)) pending
— by location
(gdb) b hello.kt:1
Breakpoint 2 at 0x100001704: file /Users/minamoto/ws/.git-trees/hello.kt, line 1.
— by address
411
(gdb) b *0x100001704
Note: breakpoint 2 also set at pc 0x100001704.
Breakpoint 3 at 0x100001704: file /Users/minamoto/ws/.git-trees/hello.kt, line 2.
Stepping
Stepping functions works mostly the same way as for C/C++ programs
Variable inspection
Variable inspections for var variables works out of the box for primitive types. For non-primitive
types there are custom pretty printers for lldb in konan_lldb.py :
λ cat main.kt | nl
2 var x = 1
3 var y = 2
4 var p = Point(x, y)
5 println("p = $p")
6 }
7 data class Point(val x: Int, val y: Int)
λ lldb ./program.kexe -o 'b main.kt:5' -o

(lldb) target create "./program.kexe"
Current executable set to './program.kexe' (x86_64).
(lldb) b main.kt:5
Breakpoint 1: where = program.kexe`kfun:main(kotlin.Array<kotlin.String>) + 289 at
main.kt:5, address = 0x000000000040af11
(lldb) r
* thread #1, name = 'program.kexe', stop reason = breakpoint 1.1
frame #0: program.kexe`kfun:main(kotlin.Array<kotlin.String>) at main.kt:5
2 var x = 1
3 var y = 2
4 var p = Point(x, y)
-> 5 println("p = $p")
6 }
7
8 data class Point(val x: Int, val y: Int)
Process 4985 launched: './program.kexe' (x86_64)

(lldb) fr var
(int) x = 1
(int) y = 2
(ObjHeader *) p = 0x00000000007643d8
(lldb) command script import dist/tools/konan_lldb.py
(lldb) fr var
(int) x = 1
(int) y = 2
(ObjHeader *) p = Point(x=1, y=2)
(lldb) p p
(ObjHeader *) $2 = Point(x=1, y=2)
(lldb)
412
Getting representation of the object variable (var) could also be done using the built-in runtime
function Konan_DebugPrint (this approach also works for gdb, using a module of command
syntax):
0:b-debugger-fixes:minamoto@unit-703(0)# cat ../debugger-plugin/1.kt | nl -p

1 fun foo(a:String, b:Int) = a + b
2 fun one() = 1
3 fun main(arg:Array<String>) {
4 var a_variable = foo("(a_variable) one is ", 1)
5 var b_variable = foo("(b_variable) two is ", 2)
6 var c_variable = foo("(c_variable) two is ", 3)
7 var d_variable = foo("(d_variable) two is ", 4)
8 println(a_variable)
9 println(b_variable)
10 println(c_variable)
11 println(d_variable)
12 }
0:b-debugger-fixes:minamoto@unit-703(0)# lldb ./program.kexe -o 'b -f 1.kt -l 9' -o r
(lldb) target create "./program.kexe"
Current executable set to './program.kexe' (x86_64).
(lldb) b -f 1.kt -l 9
Breakpoint 1: where = program.kexe`kfun:main(kotlin.Array<kotlin.String>) + 463 at
1.kt:9, address = 0x0000000100000dbf
(lldb) r
(a_variable) one is 1
* thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.1
frame #0: 0x0000000100000dbf program.kexe`kfun:main(kotlin.Array<kotlin.String>) at
1.kt:9
6 var c_variable = foo("(c_variable) two is ", 3)
7 var d_variable = foo("(d_variable) two is ", 4)
8 println(a_variable)
-> 9 println(b_variable)
10 println(c_variable)
11 println(d_variable)
12 }
Process 80496 launched: './program.kexe' (x86_64)

(lldb) expression -- Konan_DebugPrint(a_variable)
(a_variable) one is 1(KInt) $0 = 0
(lldb)
Known issues
— performance of Python bindings.
Note: Supporting the DWARF 2 speci cation means that the debugger tool recognizes Kotlin as
C89, because before the DWARF 5 speci cation, there is no identi er for the Kotlin language type
in speci cation.
413
Q: How do I run my program?
A: De ne a top level function fun main(args: Array<String>) or just fun main() if you
are not interested in passed arguments, please ensure it's not in a package. Also compiler switch
-entry could be used to make any function taking Array<String> or no arguments and
return Unit as an entry point.
Q: What is Kotlin/Native memory management model?
A: Kotlin/Native provides an automated memory management scheme, similar to what Java or

Swift provides. The current implementation includes an automated reference counter with a
cycle collector to collect cyclical garbage.
Q: How do I create a shared library?
A: Use the -produce dynamic compiler switch, or compilations.main.outputKinds

'DYNAMIC' in Gradle, i.e.
targets {
fromPreset(presets.iosArm64, 'mylib') {
compilations.main.outputKinds 'DYNAMIC'
}
}
It will produce a platform-speci c shared object (.so on Linux, .dylib on macOS, and .dll on
Windows targets) and a C language header, allowing the use of all public APIs available in your
Kotlin/Native program from C/C++ code. See samples/python_extension for an example of
using such a shared object to provide a bridge between Python and Kotlin/Native.
Q: How do I create a static library or an object le?
A: Use the -produce static compiler switch, or compilations.main.outputKinds

'STATIC' in Gradle, i.e.
targets {
fromPreset(presets.iosArm64, 'mylib') {
compilations.main.outputKinds 'STATIC'
}
}
It will produce a platform-speci c static object (.a library format) and a C language header,
allowing you to use all the public APIs available in your Kotlin/Native program from C/C++ code.
Q: How do I run Kotlin/Native behind a corporate proxy?
414
A: As Kotlin/Native needs to download a platform speci c toolchain, you need to specify -
Dhttp.proxyHost=xxx -Dhttp.proxyPort=xxx as the compiler's or gradlew arguments,
or set it via the JAVA_OPTS environment variable.
Q: How do I specify a custom Objective-C pre x/name for my Kotlin framework?
A: Use the -module-name compiler option or matching Gradle DSL statement, i.e.
targets {
fromPreset(presets.iosArm64, 'myapp') {
compilations.main.outputKinds 'FRAMEWORK'
compilations.main.extraOpts '-module-name', 'TheName'
}
}
Q: How do I enable bitcode for my Kotlin framework?
A: By default gradle plugin adds it on iOS target.
— For debug build it embeds placeholder LLVM IR data as a marker.

— For release build it embeds bitcode as data.
Or commandline arguments: -Xembed-bitcode (for release) and -Xembed-bitcode-

marker (debug)
Setting this in a Gradle DSL:
targets {
fromPreset(presets.iosArm64, 'myapp') {
compilations.main.outputKinds 'FRAMEWORK'
compilations.main.embedBitcode BitcodeEmbeddingMode.BITCODE // for release
binaries
// or BitcodeEmbeddingMode.MARKER for debug binaries
}
}
These options have nearly the same e ect as clang's -fembed-bitcode / -fembed-bitcode-
marker and swiftc's -embed-bitcode / -embed-bitcode-marker .
Q: Why do I see InvalidMutabilityException?
A: It likely happens, because you are trying to mutate a frozen object. An object can transfer to
the frozen state either explicitly, as objects reachable from objects on which the
kotlin.native.concurrent.freeze is called, or implicitly (i.e. reachable from enum or
global singleton object - see the next question).
Q: How do I make a singleton object mutable?
415
A: Currently, singleton objects are immutable (i.e. frozen after creation), and it's generally
considered good practise to have the global state immutable. If for some reason you need a
mutable state inside such an object, use the @konan.ThreadLocal annotation on the object.
Also the kotlin.native.concurrent.AtomicReference class could be used to store
di erent pointers to frozen objects in a frozen object and automatically update them.
Q: How can I compile my project against the Kotlin/Native master?
A: One of the following should be done:

For the CLI, you can compile using gradle as stated in the README (and if you get errors, you can try to do a
./gradlew clean):
For Gradle, you can use Gradle composite builds like this:
416
Coroutines
Kotlin, as a language, provides only minimal low-level APIs in its standard library to enable
various other libraries to utilize coroutines. Unlike many other languages with similar capabilities,
async and await are not keywords in Kotlin and are not even part of its standard library.
Moreover, Kotlin's concept of suspending function provides a safer and less error-prone
abstraction for asynchronous operations than futures and promises.
kotlinx.coroutines is a rich library for coroutines developed by JetBrains. It contains a

number of high-level coroutine-enabled primitives that this guide covers, including launch ,
async and others.
This is a guide on core features of kotlinx.coroutines with a series of examples, divided up

into di erent topics.
In order to use coroutines as well as follow the examples in this guide, you need to add a
dependency on kotlinx-coroutines-core module as explained in the project README.
Table of contents
— Coroutine basics
— Cancellation and timeouts
— Composing suspending functions
— Coroutine context and dispatchers
— Exception handling and supervision
— Channels (experimental)
— Shared mutable state and concurrency
— Select expression (experimental)
Additional references
— Guide to UI programming with coroutines
— Guide to reactive streams with coroutines
— Coroutines design document (KEEP)
— Full kotlinx.coroutines API reference
417
Table of contents
— Coroutine basics
— Your rst coroutine
— Bridging blocking and non-blocking worlds
— Waiting for a job
— Structured concurrency
— Scope builder
— Extract function refactoring
— Coroutines ARE light-weight
— Global coroutines are like daemon threads
Coroutine basics
This section covers basic coroutine concepts.
Your rst coroutine
Run the following code:
import kotlinx.coroutines.*
fun main() {
GlobalScope.launch { // launch a new coroutine in background and continue
delay(1000L) // non-blocking delay for 1 second (default time unit is ms)
println("World!") // print after delay
}
println("Hello,") // main thread continues while coroutine is delayed
Thread.sleep(2000L) // block main thread for 2 seconds to keep JVM alive
}
You can get full code here.
You will see the following result:
Hello,
World!
Essentially, coroutines are light-weight threads. They are launched with launch coroutine builder in
a context of some CoroutineScope. Here we are launching a new coroutine in the GlobalScope,
meaning that the lifetime of the new coroutine is limited only by the lifetime of the whole
application.
You can achieve the same result replacing GlobalScope.launch { ... } with thread {
... } and delay(...) with Thread.sleep(...) . Try it.
418
If you start by replacing GlobalScope.launch by thread , the compiler produces the
following error:
Error: Kotlin: Suspend functions are only allowed to be called from a coroutine or another
suspend function
That is because delay is a special suspending function that does not block a thread, but suspends
coroutine and it can be only used from a coroutine.
Bridging blocking and non-blocking worlds
The rst example mixes non-blocking delay(...) and blocking Thread.sleep(...) in the
same code. It is easy to lose track of which one is blocking and which one is not. Let's be explicit
about blocking using runBlocking coroutine builder:
fun main() {
delay(1000L)
println("World!")
}
println("Hello,") // main thread continues here immediately
runBlocking { // but this expression blocks the main thread
delay(2000L) // ... while we delay for 2 seconds to keep JVM alive
}
}
The result is the same, but this code uses only non-blocking delay. The main thread invoking
runBlocking blocks until the coroutine inside runBlocking completes.
This example can be also rewritten in a more idiomatic way, using runBlocking to wrap the
execution of the main function:
fun main() = runBlocking<Unit> { // start main coroutine

delay(1000L)
println("World!")
}
println("Hello,") // main coroutine continues here immediately
delay(2000L) // delaying for 2 seconds to keep JVM alive
}
419
Here runBlocking<Unit> { ... } works as an adaptor that is used to start the top-level
main coroutine. We explicitly specify its Unit return type, because a well-formed main function
in Kotlin has to return Unit .
This is also a way to write unit tests for suspending functions:
class MyTest {
@Test
fun testMySuspendingFunction() = runBlocking<Unit> {
// here we can use suspending functions using any assertion style that we like
}
}
Waiting for a job
Delaying for a time while another coroutine is working is not a good approach. Let's explicitly
wait (in a non-blocking way) until the background Job that we have launched is complete:
val job = GlobalScope.launch { // launch a new coroutine and keep a reference to its Job
delay(1000L)
println("World!")
}
println("Hello,")
job.join() // wait until child coroutine completes
Now the result is still the same, but the code of the main coroutine is not tied to the duration of
the background job in any way. Much better.
Structured concurrency
There is still something to be desired for practical usage of coroutines. When we use
GlobalScope.launch , we create a top-level coroutine. Even though it is light-weight, it still
consumes some memory resources while it runs. If we forget to keep a reference to the newly
launched coroutine it still runs. What if the code in the coroutine hangs (for example, we
erroneously delay for too long), what if we launched too many coroutines and ran out of
memory? Having to manually keep references to all the launched coroutines and join them is
error-prone.
There is a better solution. We can use structured concurrency in our code. Instead of launching
coroutines in the GlobalScope, just like we usually do with threads (threads are always global), we
can launch coroutines in the speci c scope of the operation we are performing.
420
In our example, we have main function that is turned into a coroutine using runBlocking
coroutine builder. Every coroutine builder, including runBlocking , adds an instance of
CoroutineScope to the scope of its code block. We can launch coroutines in this scope without
having to join them explicitly, because an outer coroutine ( runBlocking in our example)
does not complete until all the coroutines launched in its scope complete. Thus, we can make our
example simpler:
fun main() = runBlocking { // this: CoroutineScope

launch { // launch a new coroutine in the scope of runBlocking
delay(1000L)
println("World!")
}
println("Hello,")
}
Scope builder
In addition to the coroutine scope provided by di erent builders, it is possible to declare your
own scope using coroutineScope builder. It creates a coroutine scope and does not complete
until all launched children complete. The main di erence between runBlocking and
coroutineScope is that the latter does not block the current thread while waiting for all children
to complete.
fun main() = runBlocking { // this: CoroutineScope

launch {
delay(200L)
println("Task from runBlocking")
}
coroutineScope { // Creates a coroutine scope

launch {
delay(500L)
println("Task from nested launch")
}
delay(100L)
println("Task from coroutine scope") // This line will be printed before the
nested launch
}
println("Coroutine scope is over") // This line is not printed until the nested
launch completes
}
421
Extract function refactoring
Let's extract the block of code inside launch { ... } into a separate function. When you
perform "Extract function" refactoring on this code you get a new function with suspend
modi er. That is your rst suspending function. Suspending functions can be used inside
coroutines just like regular functions, but their additional feature is that they can, in turn, use
other suspending functions, like delay in this example, to suspend execution of a coroutine.
fun main() = runBlocking {

launch { doWorld() }
println("Hello,")
}
// this is your first suspending function

suspend fun doWorld() {
delay(1000L)
println("World!")
}
But what if the extracted function contains a coroutine builder which is invoked on the current
scope? In this case suspend modi er on the extracted function is not enough. Making
doWorld extension method on CoroutineScope is one of the solutions, but it may not always
be applicable as it does not make API clearer. The idiomatic solution is to have either an explicit
CoroutineScope as a eld in a class containing the target function or an implicit one when the
outer class implements CoroutineScope . As a last resort, CoroutineScope(coroutineContext)
can be used, but such approach is structurally unsafe because you no longer have control on the
scope of execution of this method. Only private APIs can use this builder.
Coroutines ARE light-weight
Run the following code:

repeat(100_000) { // launch a lot of coroutines
launch {
delay(1000L)
print(".")
}
}
}
422
It launches 100K coroutines and, after a second, each coroutine prints a dot. Now, try that with
threads. What would happen? (Most likely your code will produce some sort of out-of-memory
error)
Global coroutines are like daemon threads
The following code launches a long-running coroutine in GlobalScope that prints "I'm sleeping"
twice a second and then returns from the main function after some delay:
GlobalScope.launch {
repeat(1000) { i ->
println("I'm sleeping $i ...")
delay(500L)
}
}
delay(1300L) // just quit after delay
You can run and see that it prints three lines and terminates:
I'm sleeping 0 ...

I'm sleeping 1 ...
I'm sleeping 2 ...
Active coroutines that were launched in GlobalScope do not keep the process alive. They are like
daemon threads.
423
Table of contents
— Cancellation and timeouts

— Cancelling coroutine execution
— Cancellation is cooperative
— Making computation code cancellable
— Closing resources with finally
— Run non-cancellable block

— Timeout
Cancellation and timeouts

This section covers coroutine cancellation and timeouts.
Cancelling coroutine execution
In a long-running application you might need ne-grained control on your background

coroutines. For example, a user might have closed the page that launched a coroutine and now
its result is no longer needed and its operation can be cancelled. The launch function returns a
Job that can be used to cancel the running coroutine:
val job = launch {

repeat(1000) { i ->
println("job: I'm sleeping $i ...")
delay(500L)
}
}
delay(1300L) // delay a bit
println("main: I'm tired of waiting!")
job.cancel() // cancels the job
job.join() // waits for job's completion
println("main: Now I can quit.")
It produces the following output:
job: I'm sleeping 0 ...

main: I'm tired of waiting!
main: Now I can quit.
As soon as main invokes job.cancel , we don't see any output from the other coroutine
because it was cancelled. There is also a Job extension function cancelAndJoin that combines
cancel and join invocations.
424
Cancellation is cooperative
Coroutine cancellation is cooperative. A coroutine code has to cooperate to be cancellable. All the
suspending functions in kotlinx.coroutines are cancellable. They check for cancellation of
coroutine and throw CancellationException when cancelled. However, if a coroutine is working in
a computation and does not check for cancellation, then it cannot be cancelled, like the following
example shows:
val startTime = System.currentTimeMillis()

val job = launch(Dispatchers.Default) {
var nextPrintTime = startTime
var i = 0
while (i < 5) { // computation loop, just wastes CPU
// print a message twice a second
if (System.currentTimeMillis() >= nextPrintTime) {
println("job: I'm sleeping ${i++} ...")
nextPrintTime += 500L
}
}
}
job.cancelAndJoin() // cancels the job and waits for its completion
Run it to see that it continues to print "I'm sleeping" even after cancellation until the job
completes by itself after ve iterations.
Making computation code cancellable
There are two approaches to making computation code cancellable. The rst one is to
periodically invoke a suspending function that checks for cancellation. There is a yield function
that is a good choice for that purpose. The other one is to explicitly check the cancellation status.
Let us try the latter approach.
Replace while (i < 5) in the previous example with while (isActive) and rerun it.
425
val startTime = System.currentTimeMillis()
val job = launch(Dispatchers.Default) {
var nextPrintTime = startTime
var i = 0
while (isActive) { // cancellable computation loop
// print a message twice a second
if (System.currentTimeMillis() >= nextPrintTime) {
println("job: I'm sleeping ${i++} ...")
nextPrintTime += 500L
}
}
}
As you can see, now this loop is cancelled. isActive is an extension property available inside the
coroutine via the CoroutineScope object.
Closing resources with finally
Cancellable suspending functions throw CancellationException on cancellation which can be

handled in the usual way. For example, try {...} finally {...} expression and Kotlin
use function execute their nalization actions normally when a coroutine is cancelled:
val job = launch {

try {
repeat(1000) { i ->
delay(500L)
}
} finally {
println("job: I'm running finally")
}
}
Both join and cancelAndJoin wait for all nalization actions to complete, so the example above
produces the following output:
426
main: I'm tired of waiting!
job: I'm running finally
main: Now I can quit.
Run non-cancellable block
Any attempt to use a suspending function in the finally block of the previous example causes
CancellationException, because the coroutine running this code is cancelled. Usually, this is not a
problem, since all well-behaving closing operations (closing a le, cancelling a job, or closing any
kind of a communication channel) are usually non-blocking and do not involve any suspending
functions. However, in the rare case when you need to suspend in a cancelled coroutine you can
wrap the corresponding code in withContext(NonCancellable) {...} using withContext
function and NonCancellable context as the following example shows:
val job = launch {

try {
repeat(1000) { i ->
delay(500L)
}
} finally {
withContext(NonCancellable) {
println("job: I'm running finally")
delay(1000L)
println("job: And I've just delayed for 1 sec because I'm non-cancellable")
}
}
}
Timeout
The most obvious practical reason to cancel execution of a coroutine is because its execution
time has exceeded some timeout. While you can manually track the reference to the
corresponding Job and launch a separate coroutine to cancel the tracked one after delay, there is
a ready to use withTimeout function that does it. Look at the following example:
withTimeout(1300L) {
repeat(1000) { i ->
delay(500L)
}
}
427
It produces the following output:
I'm sleeping 0 ...

I'm sleeping 1 ...
I'm sleeping 2 ...
Exception in thread "main" kotlinx.coroutines.TimeoutCancellationException: Timed out
waiting for 1300 ms
The TimeoutCancellationException that is thrown by withTimeout is a subclass of

CancellationException. We have not seen its stack trace printed on the console before. That is
because inside a cancelled coroutine CancellationException is considered to be a normal
reason for coroutine completion. However, in this example we have used withTimeout right
inside the main function.
Since cancellation is just an exception, all resources are closed in the usual way. You can wrap the
code with timeout in a try {...} catch (e: TimeoutCancellationException) {...}
block if you need to do some additional action speci cally on any kind of timeout or use the
withTimeoutOrNull function that is similar to withTimeout but returns null on timeout instead
of throwing an exception:
val result = withTimeoutOrNull(1300L) {

repeat(1000) { i ->
delay(500L)
}
"Done" // will get cancelled before it produces this result
}
println("Result is $result")
There is no longer an exception when running this code:
I'm sleeping 0 ...

I'm sleeping 1 ...
I'm sleeping 2 ...
Result is null
428
Table of contents
— Channels
— Channel basics
— Closing and iteration over channels
— Building channel producers
— Pipelines
— Prime numbers with pipeline
— Fan-out
— Fan-in
— Bu ered channels
— Channels are fair
— Ticker channels
Channels
Deferred values provide a convenient way to transfer a single value between coroutines.
Channels provide a way to transfer a stream of values.
Channel basics
A Channel is conceptually very similar to BlockingQueue . One key di erence is that instead of
a blocking put operation it has a suspending send, and instead of a blocking take operation it
has a suspending receive.
val channel = Channel<Int>()

launch {
// this might be heavy CPU-consuming computation or async logic, we'll just send five
squares
for (x in 1..5) channel.send(x * x)
}
// here we print five received integers:
repeat(5) { println(channel.receive()) }
println("Done!")
The output of this code is:
1
4
9
16
25
Done!
429
Closing and iteration over channels
Unlike a queue, a channel can be closed to indicate that no more elements are coming. On the
receiver side it is convenient to use a regular for loop to receive elements from the channel.
Conceptually, a close is like sending a special close token to the channel. The iteration stops as
soon as this close token is received, so there is a guarantee that all previously sent elements
before the close are received:
val channel = Channel<Int>()

launch {
for (x in 1..5) channel.send(x * x)
channel.close() // we're done sending
}
// here we print received values using `for` loop (until the channel is closed)
for (y in channel) println(y)
println("Done!")
Building channel producers
The pattern where a coroutine is producing a sequence of elements is quite common. This is a
part of producer-consumer pattern that is often found in concurrent code. You could abstract such
a producer into a function that takes channel as its parameter, but this goes contrary to common
sense that results must be returned from functions.
There is a convenient coroutine builder named produce that makes it easy to do it right on
producer side, and an extension function consumeEach, that replaces a for loop on the
consumer side:
val squares = produceSquares()

squares.consumeEach { println(it) }
println("Done!")
Pipelines
A pipeline is a pattern where one coroutine is producing, possibly in nite, stream of values:
fun CoroutineScope.produceNumbers() = produce<Int> {

var x = 1
while (true) send(x++) // infinite stream of integers starting from 1
}
And another coroutine or coroutines are consuming that stream, doing some processing, and
producing some other results. In the example below, the numbers are just squared:
430
fun CoroutineScope.square(numbers: ReceiveChannel<Int>): ReceiveChannel<Int> = produce {
for (x in numbers) send(x * x)
}
The main code starts and connects the whole pipeline:
val numbers = produceNumbers() // produces integers from 1 and on

val squares = square(numbers) // squares integers
for (i in 1..5) println(squares.receive()) // print first five
println("Done!") // we are done
coroutineContext.cancelChildren() // cancel children coroutines
All functions that create coroutines are de ned as extensions on CoroutineScope, so that we
can rely on structured concurrency to make sure that we don't have lingering global
coroutines in our application.
Prime numbers with pipeline
Let's take pipelines to the extreme with an example that generates prime numbers using a
pipeline of coroutines. We start with an in nite sequence of numbers.
fun CoroutineScope.numbersFrom(start: Int) = produce<Int> {

var x = start
while (true) send(x++) // infinite stream of integers from start
}
The following pipeline stage lters an incoming stream of numbers, removing all the numbers
that are divisible by the given prime number:
fun CoroutineScope.filter(numbers: ReceiveChannel<Int>, prime: Int) = produce<Int> {

for (x in numbers) if (x % prime != 0) send(x)
}
Now we build our pipeline by starting a stream of numbers from 2, taking a prime number from
the current channel, and launching new pipeline stage for each prime number found:
numbersFrom(2) -> filter(2) -> filter(3) -> filter(5) -> filter(7) ...
The following example prints the rst ten prime numbers, running the whole pipeline in the
context of the main thread. Since all the coroutines are launched in the scope of the main
runBlocking coroutine we don't have to keep an explicit list of all the coroutines we have started.
We use cancelChildren extension function to cancel all the children coroutines after we have
printed the rst ten prime numbers.
431
var cur = numbersFrom(2)
for (i in 1..10) {
val prime = cur.receive()
println(prime)
cur = filter(cur, prime)
}
coroutineContext.cancelChildren() // cancel all children to let main finish
2
3
5
7
11
13
17
19
23
29
Note that you can build the same pipeline using iterator coroutine builder from the standard
library. Replace produce with iterator , send with yield , receive with next ,
ReceiveChannel with Iterator , and get rid of the coroutine scope. You will not need
runBlocking either. However, the bene t of a pipeline that uses channels as shown above is
that it can actually use multiple CPU cores if you run it in Dispatchers.Default context.
Anyway, this is an extremely impractical way to nd prime numbers. In practice, pipelines do

involve some other suspending invocations (like asynchronous calls to remote services) and
these pipelines cannot be built using sequence / iterator , because they do not allow
arbitrary suspension, unlike produce , which is fully asynchronous.
Fan-out
Multiple coroutines may receive from the same channel, distributing work between themselves.
Let us start with a producer coroutine that is periodically producing integers (ten numbers per
second):
fun CoroutineScope.produceNumbers() = produce<Int> {

var x = 1 // start from 1
while (true) {
send(x++) // produce next
delay(100) // wait 0.1s
}
}
Then we can have several processor coroutines. In this example, they just print their id and
received number:
432
fun CoroutineScope.launchProcessor(id: Int, channel: ReceiveChannel<Int>) = launch {
for (msg in channel) {
println("Processor #$id received $msg")
}
}
Now let us launch ve processors and let them work for almost a second. See what happens:
val producer = produceNumbers()

repeat(5) { launchProcessor(it, producer) }
delay(950)
producer.cancel() // cancel producer coroutine and thus kill them all
The output will be similar to the the following one, albeit the processor ids that receive each
speci c integer may be di erent:
Processor #2 received 1
Note that cancelling a producer coroutine closes its channel, thus eventually terminating iteration
over the channel that processor coroutines are doing.
Also, pay attention to how we explicitly iterate over channel with for loop to perform fan-out in
launchProcessor code. Unlike consumeEach , this for loop pattern is perfectly safe to use
from multiple coroutines. If one of the processor coroutines fails, then others would still be
processing the channel, while a processor that is written via consumeEach always consumes
(cancels) the underlying channel on its normal or abnormal completion.
Fan-in
Multiple coroutines may send to the same channel. For example, let us have a channel of strings,
and a suspending function that repeatedly sends a speci ed string to this channel with a speci ed
delay:
suspend fun sendString(channel: SendChannel<String>, s: String, time: Long) {

while (true) {
delay(time)
channel.send(s)
}
}
433
Now, let us see what happens if we launch a couple of coroutines sending strings (in this example
we launch them in the context of the main thread as main coroutine's children):
val channel = Channel<String>()

launch { sendString(channel, "foo", 200L) }
launch { sendString(channel, "BAR!", 500L) }
repeat(6) { // receive first six
println(channel.receive())
}
coroutineContext.cancelChildren() // cancel all children to let main finish
The output is:
foo
foo
BAR!
foo
foo
BAR!
Bu ered channels
The channels shown so far had no bu er. Unbu ered channels transfer elements when sender
and receiver meet each other (aka rendezvous). If send is invoked rst, then it is suspended until
receive is invoked, if receive is invoked rst, it is suspended until send is invoked.
Both Channel() factory function and produce builder take an optional capacity parameter to
specify bu er size. Bu er allows senders to send multiple elements before suspending, similar to
the BlockingQueue with a speci ed capacity, which blocks when bu er is full.
Take a look at the behavior of the following code:
val channel = Channel<Int>(4) // create buffered channel

val sender = launch { // launch sender coroutine
repeat(10) {
println("Sending $it") // print before sending each element
channel.send(it) // will suspend when buffer is full
}
}
// don't receive anything... just wait....
delay(1000)
sender.cancel() // cancel sender coroutine
It prints "sending" ve times using a bu ered channel with capacity of four:
434
Sending 0
Sending 1
Sending 2
Sending 3
Sending 4
The rst four elements are added to the bu er and the sender suspends when trying to send the
fth one.
Channels are fair
Send and receive operations to channels are fair with respect to the order of their invocation
from multiple coroutines. They are served in rst-in rst-out order, e.g. the rst coroutine to
invoke receive gets the element. In the following example two coroutines "ping" and "pong"
are receiving the "ball" object from the shared "table" channel.
data class Ball(var hits: Int)

val table = Channel<Ball>() // a shared table
launch { player("ping", table) }
launch { player("pong", table) }
table.send(Ball(0)) // serve the ball
delay(1000) // delay 1 second
coroutineContext.cancelChildren() // game over, cancel them
}
suspend fun player(name: String, table: Channel<Ball>) {

for (ball in table) { // receive the ball in a loop
ball.hits++
println("$name $ball")
delay(300) // wait a bit
table.send(ball) // send the ball back
}
}
The "ping" coroutine is started rst, so it is the rst one to receive the ball. Even though "ping"
coroutine immediately starts receiving the ball again after sending it back to the table, the ball
gets received by the "pong" coroutine, because it was already waiting for it:
ping Ball(hits=1)
pong Ball(hits=2)
ping Ball(hits=3)
pong Ball(hits=4)
Note that sometimes channels may produce executions that look unfair due to the nature of the
executor that is being used. See this issue for details.
Ticker channels
435
Ticker channel is a special rendezvous channel that produces Unit every time given delay
passes since last consumption from this channel. Though it may seem to be useless standalone, it
is a useful building block to create complex time-based produce pipelines and operators that do
windowing and other time-dependent processing. Ticker channel can be used in select to
perform "on tick" action.
To create such channel use a factory method ticker. To indicate that no further elements are
needed use ReceiveChannel.cancel method on it.
Now let's see how it works in practice:
import kotlinx.coroutines.channels.*
fun main() = runBlocking<Unit> {

val tickerChannel = ticker(delayMillis = 100, initialDelayMillis = 0) // create
ticker channel
var nextElement = withTimeoutOrNull(1) { tickerChannel.receive() }
println("Initial element is available immediately: $nextElement") // initial delay
hasn't passed yet
nextElement = withTimeoutOrNull(50) { tickerChannel.receive() } // all subsequent

elements has 100ms delay
println("Next element is not ready in 50 ms: $nextElement")
nextElement = withTimeoutOrNull(60) { tickerChannel.receive() }

println("Next element is ready in 100 ms: $nextElement")
// Emulate large consumption delays

println("Consumer pauses for 150ms")
delay(150)
// Next element is available immediately
println("Next element is available immediately after large consumer delay:
$nextElement")
// Note that the pause between `receive` calls is taken into account and next element
arrives faster
436
println("Next element is ready in 50ms after consumer pause in 150ms: $nextElement")
tickerChannel.cancel() // indicate that no more elements are needed

}
It prints following lines:
Initial element is available immediately: kotlin.Unit

Next element is not ready in 50 ms: null
Next element is ready in 100 ms: kotlin.Unit
Consumer pauses for 150ms
Next element is available immediately after large consumer delay: kotlin.Unit
Next element is ready in 50ms after consumer pause in 150ms: kotlin.Unit
Note that ticker is aware of possible consumer pauses and, by default, adjusts next produced
element delay if a pause occurs, trying to maintain a xed rate of produced elements.
Optionally, a mode parameter equal to TickerMode.FIXED_DELAY can be speci ed to maintain a

xed delay between elements.
437
Table of contents
— Composing suspending functions

— Sequential by default
— Concurrent using async
— Lazily started async
— Async-style functions
— Structured concurrency with async
Composing suspending functions

This section covers various approaches to composition of suspending functions.
Sequential by default
Assume that we have two suspending functions de ned elsewhere that do something useful like
some kind of remote service call or computation. We just pretend they are useful, but actually
each one just delays for a second for the purpose of this example:
suspend fun doSomethingUsefulOne(): Int {

delay(1000L) // pretend we are doing something useful here
return 13
}
suspend fun doSomethingUsefulTwo(): Int {

delay(1000L) // pretend we are doing something useful here, too
return 29
}
What do we do if need to invoke them sequentially – rst doSomethingUsefulOne and then

doSomethingUsefulTwo and compute the sum of their results? In practice we do this if we use
the results of the rst function to make a decision on whether we need to invoke the second one
or to decide on how to invoke it.
We use a normal sequential invocation, because the code in the coroutine, just like in the regular
code, is sequential by default. The following example demonstrates it by measuring the total time
it takes to execute both suspending functions:
val time = measureTimeMillis {

val one = doSomethingUsefulOne()
val two = doSomethingUsefulTwo()
println("The answer is ${one + two}")
}
println("Completed in $time ms")
It produces something like this:
438
The answer is 42
Completed in 2017 ms
Concurrent using async
What if there are no dependencies between invocation of doSomethingUsefulOne and

doSomethingUsefulTwo and we want to get the answer faster, by doing both concurrently?
This is where async comes to help.
Conceptually, async is just like launch. It starts a separate coroutine which is a light-weight thread
that works concurrently with all the other coroutines. The di erence is that launch returns a
Job and does not carry any resulting value, while async returns a Deferred – a light-weight non-
blocking future that represents a promise to provide a result later. You can use .await() on a
deferred value to get its eventual result, but Deferred is also a Job , so you can cancel it if
needed.

val one = async { doSomethingUsefulOne() }
val two = async { doSomethingUsefulTwo() }
println("The answer is ${one.await() + two.await()}")
}
The answer is 42
This is twice as fast, because we have concurrent execution of two coroutines. Note that
concurrency with coroutines is always explicit.
Lazily started async
There is a laziness option to async using an optional start parameter with a value of
CoroutineStart.LAZY. It starts coroutine only when its result is needed by some await or if a start
function is invoked. Run the following example:

val one = async(start = CoroutineStart.LAZY) { doSomethingUsefulOne() }
val two = async(start = CoroutineStart.LAZY) { doSomethingUsefulTwo() }
// some computation
one.start() // start the first one
two.start() // start the second one
}
439
The answer is 42
So, here the two coroutines are de ned but not executed as in the previous example, but the
control is given to the programmer on when exactly to start the execution by calling start. We
rst start one , then start two , and then await for the individual coroutines to nish.
Note that if we have called await in println and omitted start on individual coroutines, then
we would have got the sequential behaviour as await starts the coroutine execution and waits for
the execution to nish, which is not the intended use-case for laziness. The use-case for
async(start = CoroutineStart.LAZY) is a replacement for the standard lazy function
in cases when computation of the value involves suspending functions.
Async-style functions
We can de ne async-style functions that invoke doSomethingUsefulOne and

doSomethingUsefulTwo asynchronously using async coroutine builder with an explicit
GlobalScope reference. We name such functions with "Async" su x to highlight the fact that they
only start asynchronous computation and one needs to use the resulting deferred value to get
the result.
// The result type of somethingUsefulOneAsync is Deferred<Int>

fun somethingUsefulOneAsync() = GlobalScope.async {
doSomethingUsefulOne()
}
// The result type of somethingUsefulTwoAsync is Deferred<Int>

fun somethingUsefulTwoAsync() = GlobalScope.async {
doSomethingUsefulTwo()
}
Note that these xxxAsync functions are not suspending functions. They can be used from
anywhere. However, their use always implies asynchronous (here meaning concurrent) execution
of their action with the invoking code.
The following example shows their use outside of coroutine:
440
// note that we don't have `runBlocking` to the right of `main` in this example
fun main() {
// we can initiate async actions outside of a coroutine
val one = somethingUsefulOneAsync()
val two = somethingUsefulTwoAsync()
// but waiting for a result must involve either suspending or blocking.
// here we use `runBlocking { ... }` to block the main thread while waiting for
the result
runBlocking {
}
}
}
This programming style with async functions is provided here only for illustration, because it
is a popular style in other programming languages. Using this style with Kotlin coroutines is
strongly discouraged for the reasons that are explained below.
Consider what happens if between val one = somethingUsefulOneAsync() line and

one.await() expression there is some logic error in the code and the program throws an
exception and the operation that was being performed by the program aborts. Normally, a global
error-handler could catch this exception, log and report the error for developers, but the
program could otherwise continue doing other operations. But here we have
somethingUsefulOneAsync still running in background, despite the fact, that operation that
had initiated it aborts. This problem does not happen with structured concurrency, as shown in
the section below.
Structured concurrency with async
Let us take Concurrent using async example and extract a function that concurrently performs
doSomethingUsefulOne and doSomethingUsefulTwo and returns the sum of their results.
Because async coroutines builder is de ned as extension on CoroutineScope we need to have it
in the scope and that is what coroutineScope function provides:
suspend fun concurrentSum(): Int = coroutineScope {

val one = async { doSomethingUsefulOne() }
val two = async { doSomethingUsefulTwo() }
one.await() + two.await()
}
This way, if something goes wrong inside the code of concurrentSum function and it throws an
exception, all the coroutines that were launched in its scope are cancelled.
441
println("The answer is ${concurrentSum()}")
}
We still have concurrent execution of both operations as evident from the output of the above
main function:
The answer is 42
Cancellation is always propagated through coroutines hierarchy:
fun main() = runBlocking<Unit> {

try {
failedConcurrentSum()
} catch(e: ArithmeticException) {
println("Computation failed with ArithmeticException")
}
}
suspend fun failedConcurrentSum(): Int = coroutineScope {

val one = async<Int> {
try {
delay(Long.MAX_VALUE) // Emulates very long computation
42
} finally {
println("First child was cancelled")
}
442
}
val two = async<Int> {
println("Second child throws an exception")
throw ArithmeticException()
}
one.await() + two.await()
}
Note, how both rst async and awaiting parent are cancelled on the one child failure:
Second child throws an exception

First child was cancelled
Computation failed with ArithmeticException
443
Table of contents
— Coroutine context and dispatchers

— Dispatchers and threads
— Uncon ned vs con ned dispatcher
— Debugging coroutines and threads
— Jumping between threads
— Job in the context
— Children of a coroutine
— Parental responsibilities
— Naming coroutines for debugging
— Combining context elements
— Coroutine scope
— Thread-local data
Coroutine context and dispatchers

Coroutines always execute in some context which is represented by the value of
CoroutineContext type, de ned in the Kotlin standard library.
The coroutine context is a set of various elements. The main elements are the Job of the
coroutine, which we've seen before, and its dispatcher, which is covered in this section.
Dispatchers and threads
Coroutine context includes a coroutine dispatcher (see CoroutineDispatcher) that determines

what thread or threads the corresponding coroutine uses for its execution. Coroutine dispatcher
can con ne coroutine execution to a speci c thread, dispatch it to a thread pool, or let it run
uncon ned.
All coroutine builders like launch and async accept an optional CoroutineContext parameter that
can be used to explicitly specify the dispatcher for new coroutine and other context elements.
Try the following example:
444
launch { // context of the parent, main runBlocking coroutine
println("main runBlocking : I'm working in thread
${Thread.currentThread().name}")
}
launch(Dispatchers.Unconfined) { // not confined -- will work with main thread
println("Unconfined : I'm working in thread
}
launch(Dispatchers.Default) { // will get dispatched to DefaultDispatcher
println("Default : I'm working in thread
}
launch(newSingleThreadContext("MyOwnThread")) { // will get its own new thread
println("newSingleThreadContext: I'm working in thread
}
It produces the following output (maybe in di erent order):
Unconfined : I'm working in thread main

Default : I'm working in thread DefaultDispatcher-worker-1
newSingleThreadContext: I'm working in thread MyOwnThread
main runBlocking : I'm working in thread main
When launch { ... } is used without parameters, it inherits the context (and thus
dispatcher) from the CoroutineScope that it is being launched from. In this case, it inherits the
context of the main runBlocking coroutine which runs in the main thread.
Dispatchers.Uncon ned is a special dispatcher that also appears to run in the main thread, but it
is, in fact, a di erent mechanism that is explained later.
The default dispatcher, that is used when coroutines are launched in GlobalScope, is represented
by Dispatchers.Default and uses shared background pool of threads, so
launch(Dispatchers.Default) { ... } uses the same dispatcher as
GlobalScope.launch { ... } .
newSingleThreadContext creates a thread for the coroutine to run. A dedicated thread is a very
expensive resource. In a real application it must be either released, when no longer needed,
using close function, or stored in a top-level variable and reused throughout the application.
Uncon ned vs con ned dispatcher
The Dispatchers.Uncon ned coroutine dispatcher starts coroutine in the caller thread, but only
until the rst suspension point. After suspension it resumes in the thread that is fully determined
by the suspending function that was invoked. Uncon ned dispatcher is appropriate when
coroutine does not consume CPU time nor updates any shared data (like UI) that is con ned to a
speci c thread.
445
On the other side, by default, a dispatcher for the outer CoroutineScope is inherited. The default
dispatcher for runBlocking coroutine, in particular, is con ned to the invoker thread, so inheriting
it has the e ect of con ning execution to this thread with a predictable FIFO scheduling.
launch(Dispatchers.Unconfined) { // not confined -- will work with main thread

println("Unconfined : I'm working in thread ${Thread.currentThread().name}")
delay(500)
println("Unconfined : After delay in thread ${Thread.currentThread().name}")
}
launch { // context of the parent, main runBlocking coroutine
println("main runBlocking: I'm working in thread ${Thread.currentThread().name}")
delay(1000)
println("main runBlocking: After delay in thread ${Thread.currentThread().name}")
}
Produces the output:
Unconfined : I'm working in thread main

main runBlocking: I'm working in thread main
Unconfined : After delay in thread kotlinx.coroutines.DefaultExecutor
main runBlocking: After delay in thread main
So, the coroutine that had inherited context of runBlocking {...} continues to execute in
the main thread, while the uncon ned one had resumed in the default executor thread that
delay function is using.
Uncon ned dispatcher is an advanced mechanism that can be helpful in certain corner cases
where dispatching of coroutine for its execution later is not needed or produces undesirable
side-e ects, because some operation in a coroutine must be performed right away.
Uncon ned dispatcher should not be used in general code.
Debugging coroutines and threads
Coroutines can suspend on one thread and resume on another thread. Even with a single-
threaded dispatcher it might be hard to gure out what coroutine was doing, where, and when.
The common approach to debugging applications with threads is to print the thread name in the
log le on each log statement. This feature is universally supported by logging frameworks. When
using coroutines, the thread name alone does not give much of a context, so
kotlinx.coroutines includes debugging facilities to make it easier.
Run the following code with -Dkotlinx.coroutines.debug JVM option:
446
val a = async {
log("I'm computing a piece of the answer")
6
}
val b = async {
log("I'm computing another piece of the answer")
7
}
log("The answer is ${a.await() * b.await()}")
There are three coroutines. The main coroutine (#1) – runBlocking one, and two coroutines
computing deferred values a (#2) and b (#3). They are all executing in the context of
runBlocking and are con ned to the main thread. The output of this code is:
[main @coroutine#2] I'm computing a piece of the answer

[main @coroutine#3] I'm computing another piece of the answer
[main @coroutine#1] The answer is 42
The log function prints the name of the thread in square brackets and you can see, that it is the
main thread, but the identi er of the currently executing coroutine is appended to it. This
identi er is consecutively assigned to all created coroutines when debugging mode is turned on.
You can read more about debugging facilities in the documentation for newCoroutineContext
function.
Jumping between threads
Run the following code with -Dkotlinx.coroutines.debug JVM option (see debug):
newSingleThreadContext("Ctx1").use { ctx1 ->

newSingleThreadContext("Ctx2").use { ctx2 ->
runBlocking(ctx1) {
log("Started in ctx1")
withContext(ctx2) {
log("Working in ctx2")
}
log("Back to ctx1")
}
}
}
It demonstrates several new techniques. One is using runBlocking with an explicitly speci ed
context, and the other one is using withContext function to change a context of a coroutine while
still staying in the same coroutine as you can see in the output below:
447
[Ctx1 @coroutine#1] Started in ctx1
[Ctx2 @coroutine#1] Working in ctx2
[Ctx1 @coroutine#1] Back to ctx1
Note that this example also uses use function from the Kotlin standard library to release
threads that are created with newSingleThreadContext when they are no longer needed.
Job in the context
The coroutine's Job is part of its context. The coroutine can retrieve it from its own context using
coroutineContext[Job] expression:
println("My job is ${coroutineContext[Job]}")
It produces something like that when running in debug mode:

My job is "coroutine#1":BlockingCoroutine{Active}@6d311334
Note that isActive in CoroutineScope is just a convenient shortcut for

coroutineContext[Job]?.isActive == true .
Children of a coroutine
When a coroutine is launched in the CoroutineScope of another coroutine, it inherits its context
via CoroutineScope.coroutineContext and the Job of the new coroutine becomes a child of the
parent coroutine's job. When the parent coroutine is cancelled, all its children are recursively
cancelled, too.
However, when GlobalScope is used to launch a coroutine, it is not tied to the scope it was
launched from and operates independently.
448
// launch a coroutine to process some kind of incoming request
val request = launch {
// it spawns two other jobs, one with GlobalScope
GlobalScope.launch {
println("job1: I run in GlobalScope and execute independently!")
delay(1000)
println("job1: I am not affected by cancellation of the request")
}
// and the other inherits the parent context
launch {
delay(100)
println("job2: I am a child of the request coroutine")
delay(1000)
println("job2: I will not execute this line if my parent request is cancelled")
}
}
delay(500)
request.cancel() // cancel processing of the request
delay(1000) // delay a second to see what happens
println("main: Who has survived request cancellation?")
job1: I run in GlobalScope and execute independently!

job2: I am a child of the request coroutine
job1: I am not affected by cancellation of the request
main: Who has survived request cancellation?
Parental responsibilities
A parent coroutine always waits for completion of all its children. Parent does not have to
explicitly track all the children it launches and it does not have to use Job.join to wait for them at
the end:
// launch a coroutine to process some kind of incoming request

val request = launch {
repeat(3) { i -> // launch a few children jobs
launch {
delay((i + 1) * 200L) // variable delay 200ms, 400ms, 600ms
println("Coroutine $i is done")
}
}
println("request: I'm done and I don't explicitly join my children that are still
active")
}
request.join() // wait for completion of the request, including all its children
println("Now processing of the request is complete")
The result is going to be:
449
request: I'm done and I don't explicitly join my children that are still active
Coroutine 0 is done
Coroutine 1 is done
Coroutine 2 is done
Now processing of the request is complete
Naming coroutines for debugging
Automatically assigned ids are good when coroutines log often and you just need to correlate log
records coming from the same coroutine. However, when coroutine is tied to the processing of a
speci c request or doing some speci c background task, it is better to name it explicitly for
debugging purposes. CoroutineName context element serves the same function as a thread
name. It'll get displayed in the thread name that is executing this coroutine when debugging
mode is turned on.
The following example demonstrates this concept:
log("Started main coroutine")

// run two background value computations
val v1 = async(CoroutineName("v1coroutine")) {
delay(500)
log("Computing v1")
252
}
val v2 = async(CoroutineName("v2coroutine")) {
delay(1000)
log("Computing v2")
6
}
log("The answer for v1 / v2 = ${v1.await() / v2.await()}")
The output it produces with -Dkotlinx.coroutines.debug JVM option is similar to:
[main @main#1] Started main coroutine

[main @v1coroutine#2] Computing v1
[main @v2coroutine#3] Computing v2
[main @main#1] The answer for v1 / v2 = 42
Combining context elements
Sometimes we need to de ne multiple elements for coroutine context. We can use + operator
for that. For example, we can launch a coroutine with an explicitly speci ed dispatcher and an
explicitly speci ed name at the same time:
launch(Dispatchers.Default + CoroutineName("test")) {
println("I'm working in thread ${Thread.currentThread().name}")
}
450
The output of this code with -Dkotlinx.coroutines.debug JVM option is:
I'm working in thread DefaultDispatcher-worker-1 @test#2
Coroutine scope
Let us put our knowledge about contexts, children and jobs together. Assume that our
application has an object with a lifecycle, but that object is not a coroutine. For example, we are
writing an Android application and launch various coroutines in the context of an Android activity
to perform asynchronous operations to fetch and update data, do animations, etc. All of these
coroutines must be cancelled when activity is destroyed to avoid memory leaks. We, of course,
can manipulate contexts and jobs manually to tie activity's and coroutines lifecycles, but
kotlinx.coroutines provides an abstraction that encapsulates that: CoroutineScope. You
should be already familiar with coroutine scope as all coroutine builders are declared as
extensions on it.
We manage a lifecycle of our coroutines by creating an instance of CoroutineScope that is tied to

the lifecycle of our activity. CoroutineScope instance can be created by CoroutineScope() or
MainScope() factory functions. The former creates a general-purpose scope, while the latter
creates scope for UI applications and uses Dispatchers.Main as default dispatcher:
class Activity {
private val mainScope = MainScope()
fun destroy() {
mainScope.cancel()
}
// to be continued ...
Alternatively, we can implement CoroutineScope interface in this Activity class. The best way
to do it is to use delegation with default factory functions. We also can combine the desired
dispatcher (we used Dispatchers.Default in this example) with the scope:
class Activity : CoroutineScope by CoroutineScope(Dispatchers.Default) {

// to be continued ...
Now, we can launch coroutines in the scope of this Activity without having to explicitly specify
their context. For the demo, we launch ten coroutines that delay for a di erent time:
451
// class Activity continues
fun doSomething() {
// launch ten coroutines for a demo, each working for a different time
repeat(10) { i ->
launch {
delay((i + 1) * 200L) // variable delay 200ms, 400ms, ... etc
println("Coroutine $i is done")
}
}
}
} // class Activity ends
In our main function we create activity, call our test doSomething function, and destroy activity
after 500ms. This cancels all the coroutines that were launched which we can con rm by noting
that it does not print onto the screen anymore if we wait:
val activity = Activity()

activity.doSomething() // run test function
println("Launched coroutines")
delay(500L) // delay for half a second
println("Destroying activity!")
activity.destroy() // cancels all coroutines
delay(1000) // visually confirm that they don't work
The output of this example is:
Launched coroutines
Coroutine 0 is done
Coroutine 1 is done
Destroying activity!
As you can see, only the rst two coroutines had printed a message and the others were
cancelled by a single invocation of job.cancel() in Activity.destroy() .
Thread-local data
Sometimes it is convenient to have an ability to pass some thread-local data, but, for coroutines,
which are not bound to any particular thread, it is hard to achieve it manually without writing a
lot of boilerplate.
For ThreadLocal, asContextElement extension function is here for the rescue. It creates an
additional context element, which keeps the value of the given ThreadLocal and restores it
every time the coroutine switches its context.
It is easy to demonstrate it in action:
452
threadLocal.set("main")
println("Pre-main, current thread: ${Thread.currentThread()}, thread local value:
'${threadLocal.get()}'")
val job = launch(Dispatchers.Default + threadLocal.asContextElement(value = "launch")) {
println("Launch start, current thread: ${Thread.currentThread()}, thread local value:
yield()
println("After yield, current thread: ${Thread.currentThread()}, thread local value:
}
job.join()
println("Post-main, current thread: ${Thread.currentThread()}, thread local value:
In this example we launch a new coroutine in a background thread pool using

Dispatchers.Default, so it works on a di erent threads from a thread pool, but it still has the
value of thread local variable, that we've speci ed using
threadLocal.asContextElement(value = "launch") , no matter on what thread the
coroutine is executed. Thus, output (with debug) is:
Pre-main, current thread: Thread[main @coroutine#1,5,main], thread local value: 'main'

Launch start, current thread: Thread[DefaultDispatcher-worker-1 @coroutine#2,5,main],
thread local value: 'launch'
After yield, current thread: Thread[DefaultDispatcher-worker-2 @coroutine#2,5,main],
thread local value: 'launch'
Post-main, current thread: Thread[main @coroutine#1,5,main], thread local value: 'main'
Note how easily one may forget the corresponding context element and then still safely access
thread local. To avoid such situations, it is recommended to use ensurePresent method and fail-
fast on improper usages.
ThreadLocal has rst-class support and can be used with any primitive
kotlinx.coroutines provides. It has one key limitation: when thread-local is mutated, a new
value is not propagated to the coroutine caller (as context element cannot track all
ThreadLocal object accesses) and updated value is lost on the next suspension. Use
withContext to update the value of the thread-local in a coroutine, see asContextElement for
more details.
Alternatively, a value can be stored in a mutable box like class Counter(var i: Int) ,
which is, in turn, stored in a thread-local variable. However, in this case you are fully responsible
to synchronize potentially concurrent modi cations to the variable in this mutable box.
For advanced usage, for example for integration with logging MDC, transactional contexts or any
other libraries which internally use thread-locals for passing data, see documentation for
ThreadContextElement interface that should be implemented.
453
Table of contents
— Exception handling
— Exception propagation
— CoroutineExceptionHandler
— Cancellation and exceptions
— Exceptions aggregation
— Supervision
— Supervision job
— Supervision scope
— Exceptions in supervised coroutines
Exception handling
This section covers exception handling and cancellation on exceptions. We already know that
cancelled coroutine throws CancellationException in suspension points and that it is ignored by
coroutines machinery. But what happens if an exception is thrown during cancellation or
multiple children of the same coroutine throw an exception?
Exception propagation
Coroutine builders come in two avors: propagating exceptions automatically (launch and actor)
or exposing them to users (async and produce). The former treat exceptions as unhandled,
similar to Java's Thread.uncaughtExceptionHandler , while the latter are relying on the user
to consume the nal exception, for example via await or receive (produce and receive are
covered later in Channels section).
It can be demonstrated by a simple example that creates coroutines in the GlobalScope:
454

val job = GlobalScope.launch {
println("Throwing exception from launch")
throw IndexOutOfBoundsException() // Will be printed to the console by
Thread.defaultUncaughtExceptionHandler
}
job.join()
println("Joined failed job")
val deferred = GlobalScope.async {
println("Throwing exception from async")
throw ArithmeticException() // Nothing is printed, relying on user to call await
}
try {
deferred.await()
println("Unreached")
} catch (e: ArithmeticException) {
println("Caught ArithmeticException")
}
}
The output of this code is (with debug):
Throwing exception from launch

Exception in thread "DefaultDispatcher-worker-2 @coroutine#2"
java.lang.IndexOutOfBoundsException
Joined failed job
Throwing exception from async
Caught ArithmeticException
CoroutineExceptionHandler
But what if one does not want to print all exceptions to the console? CoroutineExceptionHandler
context element is used as generic catch block of coroutine where custom logging or exception
handling may take place. It is similar to using Thread.uncaughtExceptionHandler.
On JVM it is possible to rede ne global exception handler for all coroutines by registering
CoroutineExceptionHandler via ServiceLoader. Global exception handler is similar to
Thread.defaultUncaughtExceptionHandler which is used when no more speci c handlers
are registered. On Android, uncaughtExceptionPreHandler is installed as a global coroutine
exception handler.
CoroutineExceptionHandler is invoked only on exceptions which are not expected to be handled

by the user, so registering it in async builder and the like of it has no e ect.
455
val handler = CoroutineExceptionHandler { _, exception ->
println("Caught $exception")
}
val job = GlobalScope.launch(handler) {
throw AssertionError()
}
val deferred = GlobalScope.async(handler) {
throw ArithmeticException() // Nothing will be printed, relying on user to call
deferred.await()
}
joinAll(job, deferred)
Caught java.lang.AssertionError
Cancellation and exceptions
Cancellation is tightly bound with exceptions. Coroutines internally use

CancellationException for cancellation, these exceptions are ignored by all handlers, so
they should be used only as the source of additional debug information, which can be obtained
by catch block. When a coroutine is cancelled using Job.cancel without a cause, it terminates,
but it does not cancel its parent. Cancelling without cause is a mechanism for parent to cancel its
children without cancelling itself.
val job = launch {

val child = launch {
try {
delay(Long.MAX_VALUE)
} finally {
println("Child is cancelled")
}
}
yield()
println("Cancelling child")
child.cancel()
child.join()
yield()
println("Parent is not cancelled")
}
job.join()
Cancelling child
Child is cancelled
Parent is not cancelled
456
If a coroutine encounters exception other than CancellationException , it cancels its parent
with that exception. This behaviour cannot be overridden and is used to provide stable
coroutines hierarchies for structured concurrency which do not depend on
CoroutineExceptionHandler implementation. The original exception is handled by the parent
when all its children terminate.
This also a reason why, in these examples, CoroutineExceptionHandler is always installed to

a coroutine that is created in GlobalScope. It does not make sense to install an exception
handler to a coroutine that is launched in the scope of the main runBlocking, since the main
coroutine is going to be always cancelled when its child completes with exception despite the
installed handler.

}
launch { // the first child
try {
} finally {
withContext(NonCancellable) {
println("Children are cancelled, but exception is not handled until all
children terminate")
delay(100)
println("The first child finished its non cancellable block")
}
}
}
launch { // the second child
delay(10)
println("Second child throws an exception")
}
}
job.join()
Second child throws an exception

Children are cancelled, but exception is not handled until all children terminate
The first child finished its non cancellable block
Caught java.lang.ArithmeticException
Exceptions aggregation
457
What happens if multiple children of a coroutine throw an exception? The general rule is "the
rst exception wins", so the rst thrown exception is exposed to the handler. But that may cause
lost exceptions, for example if coroutine throws an exception in its finally block. So,
additional exceptions are suppressed.
One of the solutions would have been to report each exception separately, but then
Deferred.await should have had the same mechanism to avoid behavioural inconsistency
and this would cause implementation details of a coroutines (whether it had delegated parts
of its work to its children or not) to leak to its exception handler.
import java.io.*

println("Caught $exception with suppressed
${exception.suppressed.contentToString()}")
}
launch {
try {
} finally {
}
}
launch {
delay(100)
throw IOException()
}
}
job.join()
}
Note: This above code will work properly only on JDK7+ that supports suppressed
exceptions
Caught java.io.IOException with suppressed [java.lang.ArithmeticException]
Note, this mechanism currently works only on Java version 1.7+. Limitation on JS and Native
is temporary and will be xed in the future.
Cancellation exceptions are transparent and unwrapped by default:
458
println("Caught original $exception")
}
val inner = launch {
launch {
launch {
throw IOException()
}
}
}
try {
inner.join()
} catch (e: CancellationException) {
println("Rethrowing CancellationException with original cause")
throw e
}
}
job.join()
Rethrowing CancellationException with original cause

Caught original java.io.IOException
Supervision
As we have studied before, cancellation is a bidirectional relationship propagating through the
whole coroutines hierarchy. But what if unidirectional cancellation is required?
A good example of such a requirement is a UI component with the job de ned in its scope. If any
of the UI's child tasks have failed, it is not always necessary to cancel (e ectively kill) the whole UI
component, but if UI component is destroyed (and its job is cancelled), then it is necessary to fail
all child jobs as their results are no longer required.
Another example is a server process that spawns several children jobs and needs to supervise
their execution, tracking their failures and restarting just those children jobs that had failed.
Supervision job
For these purposes SupervisorJob can be used. It is similar to a regular Job with the only
exception that cancellation is propagated only downwards. It is easy to demonstrate with an
example:

val supervisor = SupervisorJob()
with(CoroutineScope(coroutineContext + supervisor)) {
// launch the first child -- its exception is ignored for this example (don't do
this in practice!)
459
this in practice!)
val firstChild = launch(CoroutineExceptionHandler { _, _ -> }) {
println("First child is failing")
throw AssertionError("First child is cancelled")
}
// launch the second child
val secondChild = launch {
firstChild.join()
// Cancellation of the first child is not propagated to the second child
println("First child is cancelled: ${firstChild.isCancelled}, but second one
is still active")
try {
} finally {
// But cancellation of the supervisor is propagated
println("Second child is cancelled because supervisor is cancelled")
}
}
// wait until the first child fails & completes
firstChild.join()
println("Cancelling supervisor")
supervisor.cancel()
secondChild.join()
}
}
First child is failing

First child is cancelled: true, but second one is still active
Cancelling supervisor
Second child is cancelled because supervisor is cancelled
Supervision scope
For scoped concurrency supervisorScope can be used instead of coroutineScope for the same
purpose. It propagates cancellation only in one direction and cancels all children only if it has
failed itself. It also waits for all children before completion just like coroutineScope does.
460
import kotlin.coroutines.*

try {
supervisorScope {
val child = launch {
try {
println("Child is sleeping")
} finally {
println("Child is cancelled")
}
}
// Give our child a chance to execute and print using yield
yield()
println("Throwing exception from scope")
}
} catch(e: AssertionError) {
println("Caught assertion error")
}
}
Child is sleeping
Throwing exception from scope
Child is cancelled
Caught assertion error
Exceptions in supervised coroutines
Another crucial di erence between regular and supervisor jobs is exception handling. Every child
should handle its exceptions by itself via exception handling mechanisms. This di erence comes
from the fact that child's failure is not propagated to the parent.
import kotlin.coroutines.*

}
supervisorScope {
val child = launch(handler) {
println("Child throws an exception")
}
println("Scope is completing")
}
println("Scope is completed")
}
461
Scope is completing
Child throws an exception
Caught java.lang.AssertionError
Scope is completed
462
Table of contents
— Select expression (experimental)

— Selecting from channels
— Selecting on close
— Selecting to send
— Selecting deferred values
— Switch over a channel of deferred values
Select expression (experimental)

Select expression makes it possible to await multiple suspending functions simultaneously and
select the rst one that becomes available.
Select expressions are an experimental feature of kotlinx.coroutines. Their API is

expected to evolve in the upcoming updates of the kotlinx.coroutines library with
potentially breaking changes.
Selecting from channels
Let us have two producers of strings: fizz and buzz . The fizz produces "Fizz" string every
300 ms:
fun CoroutineScope.fizz() = produce<String> {

while (true) { // sends "Fizz" every 300 ms
delay(300)
send("Fizz")
}
}
And the buzz produces "Buzz!" string every 500 ms:
fun CoroutineScope.buzz() = produce<String> {

while (true) { // sends "Buzz!" every 500 ms
delay(500)
send("Buzz!")
}
}
Using receive suspending function we can receive either from one channel or the other. But
select expression allows us to receive from both simultaneously using its onReceive clauses:
463
suspend fun selectFizzBuzz(fizz: ReceiveChannel<String>, buzz: ReceiveChannel<String>) {
select<Unit> { // <Unit> means that this select expression does not produce any
result
fizz.onReceive { value -> // this is the first select clause
println("fizz -> '$value'")
}
buzz.onReceive { value -> // this is the second select clause
println("buzz -> '$value'")
}
}
}
Let us run it all seven times:
val fizz = fizz()

val buzz = buzz()
repeat(7) {
selectFizzBuzz(fizz, buzz)
}
coroutineContext.cancelChildren() // cancel fizz & buzz coroutines
The result of this code is:
fizz -> 'Fizz'

buzz -> 'Buzz!'
fizz -> 'Fizz'
fizz -> 'Fizz'
buzz -> 'Buzz!'
fizz -> 'Fizz'
buzz -> 'Buzz!'
Selecting on close
The onReceive clause in select fails when the channel is closed causing the corresponding
select to throw an exception. We can use onReceiveOrNull clause to perform a speci c action
when the channel is closed. The following example also shows that select is an expression that
returns the result of its selected clause:
464
suspend fun selectAorB(a: ReceiveChannel<String>, b: ReceiveChannel<String>): String =
select<String> {
a.onReceiveOrNull { value ->
if (value == null)
"Channel 'a' is closed"
else
"a -> '$value'"
}
b.onReceiveOrNull { value ->
if (value == null)
"Channel 'b' is closed"
else
"b -> '$value'"
}
}
Let's use it with channel a that produces "Hello" string four times and channel b that produces
"World" four times:
val a = produce<String> {
repeat(4) { send("Hello $it") }
}
val b = produce<String> {
repeat(4) { send("World $it") }
}
repeat(8) { // print first eight results
println(selectAorB(a, b))
}
coroutineContext.cancelChildren()
The result of this code is quite interesting, so we'll analyze it in mode detail:
a -> 'Hello 0'

a -> 'Hello 1'
b -> 'World 0'
a -> 'Hello 2'
a -> 'Hello 3'
b -> 'World 1'
Channel 'a' is closed
Channel 'a' is closed
There are couple of observations to make out of it.
First of all, select is biased to the rst clause. When several clauses are selectable at the same
time, the rst one among them gets selected. Here, both channels are constantly producing
strings, so a channel, being the rst clause in select, wins. However, because we are using
unbu ered channel, the a gets suspended from time to time on its send invocation and gives a
chance for b to send, too.
The second observation, is that onReceiveOrNull gets immediately selected when the channel is
already closed.
465
Selecting to send
Select expression has onSend clause that can be used for a great good in combination with a
biased nature of selection.
Let us write an example of producer of integers that sends its values to a side channel when
the consumers on its primary channel cannot keep up with it:
fun CoroutineScope.produceNumbers(side: SendChannel<Int>) = produce<Int> {

for (num in 1..10) { // produce 10 numbers from 1 to 10
delay(100) // every 100 ms
select<Unit> {
onSend(num) {} // Send to the primary channel
side.onSend(num) {} // or to the side channel
}
}
}
Consumer is going to be quite slow, taking 250 ms to process each number:
val side = Channel<Int>() // allocate side channel

launch { // this is a very fast consumer for the side channel
side.consumeEach { println("Side channel has $it") }
}
produceNumbers(side).consumeEach {
println("Consuming $it")
delay(250) // let us digest the consumed number properly, do not hurry
}
println("Done consuming")
coroutineContext.cancelChildren()
So let us see what happens:
Consuming 1
Side channel has 2
Side channel has 3
Consuming 4
Side channel has 5
Side channel has 6
Consuming 7
Side channel has 8
Side channel has 9
Consuming 10
Done consuming
Selecting deferred values
Deferred values can be selected using onAwait clause. Let us start with an async function that
returns a deferred string value after a random delay:
466
fun CoroutineScope.asyncString(time: Int) = async {
delay(time.toLong())
"Waited for $time ms"
}
Let us start a dozen of them with a random delay.
fun CoroutineScope.asyncStringsList(): List<Deferred<String>> {

val random = Random(3)
return List(12) { asyncString(random.nextInt(1000)) }
}
Now the main function awaits for the rst of them to complete and counts the number of
deferred values that are still active. Note that we've used here the fact that select expression
is a Kotlin DSL, so we can provide clauses for it using an arbitrary code. In this case we iterate
over a list of deferred values to provide onAwait clause for each deferred value.
val list = asyncStringsList()

val result = select<String> {
list.withIndex().forEach { (index, deferred) ->
deferred.onAwait { answer ->
"Deferred $index produced answer '$answer'"
}
}
}
println(result)
val countActive = list.count { it.isActive }
println("$countActive coroutines are still active")
The output is:
Deferred 4 produced answer 'Waited for 128 ms'

11 coroutines are still active
Switch over a channel of deferred values
Let us write a channel producer function that consumes a channel of deferred string values, waits
for each received deferred value, but only until the next deferred value comes over or the
channel is closed. This example puts together onReceiveOrNull and onAwait clauses in the same
select :
467
fun CoroutineScope.switchMapDeferreds(input: ReceiveChannel<Deferred<String>>) =
produce<String> {
var current = input.receive() // start with first received deferred value
while (isActive) { // loop while not cancelled/closed
val next = select<Deferred<String>?> { // return next deferred value from this
select or null
input.onReceiveOrNull { update ->
update // replaces next value to wait
}
current.onAwait { value ->
send(value) // send value that current deferred has produced
input.receiveOrNull() // and use the next deferred from the input channel
}
}
if (next == null) {
println("Channel was closed")
break // out of loop
} else {
current = next
}
}
}
To test it, we'll use a simple async function that resolves to a speci ed string after a speci ed
time:
fun CoroutineScope.asyncString(str: String, time: Long) = async {

delay(time)
str
}
The main function just launches a coroutine to print results of switchMapDeferreds and
sends some test data to it:
val chan = Channel<Deferred<String>>() // the channel for test

launch { // launch printing coroutine
for (s in switchMapDeferreds(chan))
println(s) // print each received string
}
chan.send(asyncString("BEGIN", 100))
delay(200) // enough time for "BEGIN" to be produced
chan.send(asyncString("Slow", 500))
delay(100) // not enough time to produce slow
chan.send(asyncString("Replace", 100))
delay(500) // give it time before the last one
chan.send(asyncString("END", 500))
delay(1000) // give it time to process
chan.close() // close the channel ...
delay(500) // and wait some time to let it finish
The result of this code:
468
BEGIN
Replace
END
Channel was closed
469
Table of contents
— Shared mutable state and concurrency

— The problem
— Volatiles are of no help
— Thread-safe data structures
— Thread con nement ne-grained
— Thread con nement coarse-grained
— Mutual exclusion
— Actors
Shared mutable state and concurrency

Coroutines can be executed concurrently using a multi-threaded dispatcher like the
Dispatchers.Default. It presents all the usual concurrency problems. The main problem being
synchronization of access to shared mutable state. Some solutions to this problem in the land
of coroutines are similar to the solutions in the multi-threaded world, but others are unique.
The problem
Let us launch a hundred coroutines all doing the same action thousand times. We'll also measure
their completion time for further comparisons:
suspend fun massiveRun(action: suspend () -> Unit) {

val n = 100 // number of coroutines to launch
val k = 1000 // times an action is repeated by each coroutine
coroutineScope { // scope for coroutines
repeat(n) {
launch {
repeat(k) { action() }
}
}
}
}
println("Completed ${n * k} actions in $time ms")
}
We start with a very simple action that increments a shared mutable variable using multi-
threaded Dispatchers.Default.
470
var counter = 0

withContext(Dispatchers.Default) {
massiveRun {
counter++
}
}
println("Counter = $counter")
}
What does it print at the end? It is highly unlikely to ever print "Counter = 100000", because a
hundred coroutines increment the counter concurrently from multiple threads without any
synchronization.
Volatiles are of no help
There is common misconception that making a variable volatile solves concurrency problem.
Let us try it:
@Volatile // in Kotlin `volatile` is an annotation

var counter = 0

massiveRun {
counter++
}
}
}
This code works slower, but we still don't get "Counter = 100000" at the end, because volatile
variables guarantee linearizable (this is a technical term for "atomic") reads and writes to the
corresponding variable, but do not provide atomicity of larger actions (increment in our case).
Thread-safe data structures
The general solution that works both for threads and for coroutines is to use a thread-safe (aka
synchronized, linearizable, or atomic) data structure that provides all the necessarily
synchronization for the corresponding operations that needs to be performed on a shared state.
In the case of a simple counter we can use AtomicInteger class which has atomic
incrementAndGet operations:
471
var counter = AtomicInteger()

massiveRun {
counter.incrementAndGet()
}
}
}
This is the fastest solution for this particular problem. It works for plain counters, collections,
queues and other standard data structures and basic operations on them. However, it does not
easily scale to complex state or to complex operations that do not have ready-to-use thread-safe
implementations.
Thread con nement ne-grained
Thread con nement is an approach to the problem of shared mutable state where all access to the
particular shared state is con ned to a single thread. It is typically used in UI applications, where
all UI state is con ned to the single event-dispatch/application thread. It is easy to apply with
coroutines by using a
single-threaded context.
val counterContext = newSingleThreadContext("CounterContext")

var counter = 0

massiveRun {
// confine each increment to a single-threaded context
withContext(counterContext) {
counter++
}
}
}
}
This code works very slowly, because it does ne-grained thread-con nement. Each individual
increment switches from multi-threaded Dispatchers.Default context to the single-threaded
context using withContext(counterContext) block.
Thread con nement coarse-grained
472
In practice, thread con nement is performed in large chunks, e.g. big pieces of state-updating
business logic are con ned to the single thread. The following example does it like that, running
each coroutine in the single-threaded context to start with.
val counterContext = newSingleThreadContext("CounterContext")

var counter = 0

// confine everything to a single-threaded context
withContext(counterContext) {
massiveRun {
counter++
}
}
}
This now works much faster and produces correct result.
Mutual exclusion
Mutual exclusion solution to the problem is to protect all modi cations of the shared state with a
critical section that is never executed concurrently. In a blocking world you'd typically use
synchronized or ReentrantLock for that. Coroutine's alternative is called Mutex. It has lock
and unlock functions to delimit a critical section. The key di erence is that Mutex.lock() is a
suspending function. It does not block a thread.
There is also withLock extension function that conveniently represents mutex.lock(); try {
... } finally { mutex.unlock() } pattern:
val mutex = Mutex()

var counter = 0

massiveRun {
// protect each increment with lock
mutex.withLock {
counter++
}
}
}
}
473
The locking in this example is ne-grained, so it pays the price. However, it is a good choice for
some situations where you absolutely must modify some shared state periodically, but there is
no natural thread that this state is con ned to.
Actors
An actor is an entity made up of a combination of a coroutine, the state that is con ned and
encapsulated into this coroutine, and a channel to communicate with other coroutines. A simple
actor can be written as a function, but an actor with a complex state is better suited for a class.
There is an actor coroutine builder that conveniently combines actor's mailbox channel into its
scope to receive messages from and combines the send channel into the resulting job object, so
that a single reference to the actor can be carried around as its handle.
The rst step of using an actor is to de ne a class of messages that an actor is going to process.
Kotlin's sealed classes are well suited for that purpose. We de ne CounterMsg sealed class with
IncCounter message to increment a counter and GetCounter message to get its value. The
later needs to send a response. A CompletableDeferred communication primitive, that
represents a single value that will be known (communicated) in the future, is used here for that
purpose.
// Message types for counterActor

sealed class CounterMsg
object IncCounter : CounterMsg() // one-way message to increment counter
class GetCounter(val response: CompletableDeferred<Int>) : CounterMsg() // a request with
reply
Then we de ne a function that launches an actor using an actor coroutine builder:
// This function launches a new counter actor

fun CoroutineScope.counterActor() = actor<CounterMsg> {
var counter = 0 // actor state
for (msg in channel) { // iterate over incoming messages
when (msg) {
is IncCounter -> counter++
is GetCounter -> msg.response.complete(counter)
}
}
}
The main code is straightforward:
474
val counter = counterActor() // create the actor
massiveRun {
counter.send(IncCounter)
}
}
// send a message to get a counter value from an actor
val response = CompletableDeferred<Int>()
counter.send(GetCounter(response))
println("Counter = ${response.await()}")
counter.close() // shutdown the actor
}
It does not matter (for correctness) what context the actor itself is executed in. An actor is a
coroutine and a coroutine is executed sequentially, so con nement of the state to the speci c
coroutine works as a solution to the problem of shared mutable state. Indeed, actors may modify
their own private state, but can only a ect each other through messages (avoiding the need for
any locks).
Actor is more e cient than locking under load, because in this case it always has work to do and
it does not have to switch to a di erent context at all.
Note that an actor coroutine builder is a dual of produce coroutine builder. An actor is
associated with the channel that it receives messages from, while a producer is associated
with the channel that it sends elements to.
475
Tools
Documenting Kotlin Code
The language used to document Kotlin code (the equivalent of Java's JavaDoc) is called KDoc. In
its essence, KDoc combines JavaDoc's syntax for block tags (extended to support Kotlin's speci c
constructs) and Markdown for inline markup.
Generating the Documentation

Kotlin's documentation generation tool is called Dokka. See the Dokka README for usage
instructions.
Dokka has plugins for Gradle, Maven and Ant, so you can integrate documentation generation
into your build process.
KDoc Syntax
Just like with JavaDoc, KDoc comments start with /** and end with */ . Every line of the
comment may begin with an asterisk, which is not considered part of the contents of the
comment.
By convention, the rst paragraph of the documentation text (the block of text until the rst
blank line) is the summary description of the element, and the following text is the detailed
description.
Every block tag begins on a new line and starts with the @ character.
Here's an example of a class documented using KDoc:
476
/**
* A group of *members*.
*
* This class has no useful logic; it's just a documentation example.
*
* @param T the type of a member in this group.
* @property name the name of this group.
* @constructor Creates an empty group.
*/
class Group<T>(val name: String) {
/**
* Adds a [member] to this group.
* @return the new size of the group.
*/
fun add(member: T): Int { ... }
}
Block Tags
KDoc currently supports the following block tags:
@param <name>
Documents a value parameter of a function or a type parameter of a class, property or function.

To better separate the parameter name from the description, if you prefer, you can enclose the
name of the parameter in brackets. The following two syntaxes are therefore equivalent:
@param name description.
@param[name] description.
@return
Documents the return value of a function.
@constructor
Documents the primary constructor of a class.
@receiver
Documents the receiver of an extension function.
@property <name>
Documents the property of a class which has the speci ed name. This tag can be used for
documenting properties declared in the primary constructor, where putting a doc comment
directly before the property de nition would be awkward.
@throws <class>, @exception <class>
Documents an exception which can be thrown by a method. Since Kotlin does not have checked
exceptions, there is also no expectation that all possible exceptions are documented, but you can
still use this tag when it provides useful information for users of the class.
477
@sample <identifier>
Embeds the body of the function with the speci ed quali ed name into the documentation for
the current element, in order to show an example of how the element could be used.
@see <identifier>
Adds a link to the speci ed class or method to the See Also block of the documentation.
@author
Speci es the author of the element being documented.
@since
Speci es the version of the software in which the element being documented was introduced.
@suppress
Excludes the element from the generated documentation. Can be used for elements which are
not part of the o cial API of a module but still have to be visible externally.
KDoc does not support the @deprecated tag. Instead, please use the @Deprecated
annotation.
Inline Markup
For inline markup, KDoc uses the regular Markdown syntax, extended to support a shorthand
syntax for linking to other elements in the code.
Linking to Elements
To link to another element (class, method, property or parameter), simply put its name in square
brackets:
Use the method [foo] for this purpose.
If you want to specify a custom label for the link, use the Markdown reference-style syntax:
Use [this method][foo] for this purpose.
You can also use quali ed names in the links. Note that, unlike JavaDoc, quali ed names always
use the dot character to separate the components, even before a method name:
Use [kotlin.reflect.KClass.properties] to enumerate the properties of the class.
Names in links are resolved using the same rules as if the name was used inside the element
being documented. In particular, this means that if you have imported a name into the current
le, you don't need to fully qualify it when you use it in a KDoc comment.
478
Note that KDoc does not have any syntax for resolving overloaded members in links. Since the
Kotlin documentation generation tool puts the documentation for all overloads of a function on
the same page, identifying a speci c overloaded function is not required for the link to work.
Module and Package Documentation

Documentation for a module as a whole, as well as packages in that module, is provided as a
separate Markdown le, and the paths to that le is passed to Dokka using the -include
command line parameter or the corresponding parameters in Ant, Maven and Gradle plugins.
Inside the le, the documentation for the module as a whole and for individual packages is
introduced by the corresponding rst-level headings. The text of the heading must be "Module
<module name> " for the module, and "Package <package qualified name> " for a package.
Here's an example content of the le:

# Module kotlin-demo
The module shows the Dokka syntax usage.
# Package org.jetbrains.kotlin.demo
Contains assorted useful stuff.
## Level 2 heading
Text after this heading is also part of documentation for òrg.jetbrains.kotlin.demo`
# Package org.jetbrains.kotlin.demo2
Useful stuff in another package.
479
Annotation Processing with Kotlin
Annotation processors (see JSR 269) are supported in Kotlin with the kapt compiler plugin.
In a nutshell, you can use libraries such as Dagger or Data Binding in your Kotlin projects.
Please read below about how to apply the kapt plugin to your Gradle/Maven build.
Using in Gradle
Apply the kotlin-kapt Gradle plugin:
plugins {
id "org.jetbrains.kotlin.kapt" version "1.3.31"
}
plugins {
kotlin("kapt") version "1.3.31"
}
Alternatively, you can use the apply plugin syntax:
apply plugin: 'kotlin-kapt'
Then add the respective dependencies using the kapt con guration in your dependencies
block:
dependencies {
kapt 'groupId:artifactId:version'
}
dependencies {
kapt("groupId:artifactId:version")
}
If you previously used the Android support for annotation processors, replace usages of the
annotationProcessor con guration with kapt . If your project contains Java classes, kapt
will also take care of them.
If you use annotation processors for your androidTest or test sources, the respective kapt
con gurations are named kaptAndroidTest and kaptTest . Note that kaptAndroidTest
and kaptTest extends kapt , so you can just provide the kapt dependency and it will be
available both for production sources and tests.
Annotation processor arguments

Use arguments {} block to pass arguments to annotation processors:
480
kapt {
arguments {
arg("key", "value")
}
}
Gradle build cache support (since 1.2.20)

The kapt annotation processing tasks are not cached in Gradle by default. Annotation processors
run arbitrary code that may not necessarily transform the task inputs into the outputs, might
access and modify the les that are not tracked by Gradle etc. To enable caching for kapt anyway,
add the following lines to the build script:
kapt {
useBuildCache = true
}
Running kapt tasks in parallel (since 1.2.60)

To improve the speed of builds that use kapt, you can enable the Gradle worker API for kapt
tasks. Using the worker API lets Gradle run independent annotation processing tasks from a
single project in parallel, which in some cases signi cantly decreases the execution time.
However, running kapt with Gradle worker API enabled can result in increased memory
consumption due to parallel execution.
To use the Gradle worker API for parallel execution of kapt tasks, add this line to your
gradle.properties le:
kapt.use.worker.api=true
Compile avoidance for kapt (since 1.3.20)

To improve the times of incremental builds with kapt, it can use the Gradle compile avoidance.
With compile avoidance enabled, Gradle can skip annotation processing when rebuilding a
project. Particularly, annotation processing is skipped when:
— The project's source les are unchanged.

— The changes in dependencies are ABI compatible. For example, the only changes are in
method bodies.
However, compile avoidance can't be used for annotation processors discovered in the compile
classpath since any changes in them require running the annotation processing tasks.
To run kapt with compile avoidance:
— Add the annotation processor dependencies to the kapt* con gurations manually as
described above.
— Turn o the discovery of annotation processors in the compile classpath by adding this line to
your gradle.properties le:
481
kapt.include.compile.classpath=false
Incremental annotation processing (since 1.3.30)

Starting from version 1.3.30, kapt supports incremental annotation processing as an
experimental feature. Currently, annotation processing can be incremental only if all annotation
processors being used are incremental.
To enable incremental annotation processing, add this line to your gradle.properties le:
kapt.incremental.apt=true
Note that incremental annotation processing requires incremental compilation to be enabled as

well.
Java compiler options

Kapt uses Java compiler to run annotation processors.
Here is how you can pass arbitrary options to javac:
kapt {
javacOptions {
// Increase the max count of errors from annotation processors.
// Default is 100.
option("-Xmaxerrs", 500)
}
}
Non-existent type correction

Some annotation processors (such as AutoFactory ) rely on precise types in declaration
signatures. By default, Kapt replaces every unknown type (including types for the generated
classes) to NonExistentClass , but you can change this behavior. Add the additional ag to the
build.gradle le to enable error type inferring in stubs:
kapt {
correctErrorTypes = true
}
Using in Maven
Add an execution of the kapt goal from kotlin-maven-plugin before compile :
482
<execution>
<id>kapt</id>
<goals>
<goal>kapt</goal>
</goals>
<configuration>
<sourceDirs>
<sourceDir>src/main/kotlin</sourceDir>
<sourceDir>src/main/java</sourceDir>
</sourceDirs>
<annotationProcessorPaths>

<annotationProcessorPath>
<groupId>com.google.dagger</groupId>
<artifactId>dagger-compiler</artifactId>
<version>2.9</version>
</annotationProcessorPath>
</annotationProcessorPaths>
</configuration>
</execution>
You can nd a complete sample project showing the use of Kotlin, Maven and Dagger in the
Kotlin examples repository.
Please note that kapt is still not supported for IntelliJ IDEA’s own build system. Launch the build
from the “Maven Projects” toolbar whenever you want to re-run the annotation processing.
Using in CLI
Kapt compiler plugin is available in the binary distribution of the Kotlin compiler.
You can attach the plugin by providing the path to its JAR le using the Xplugin kotlinc option:
-Xplugin=$KOTLIN_HOME/lib/kotlin-annotation-processing.jar
Here is a list of the available options:
— sources (required): An output path for the generated les.
— classes (required): An output path for the generated class les and resources.
— stubs (required): An output path for the stub les. In other words, some temporary
directory.
— incrementalData : An output path for the binary stubs.
— apclasspath (repeatable): A path to the annotation processor JAR. Pass as many

apclasspath options as many JARs you have.
— apoptions : A base64-encoded list of the annotation processor options. See AP/javac
options encoding for more information.
— javacArguments : A base64-encoded list of the options passed to javac. See AP/javac
options encoding for more information.
— processors : A comma-speci ed list of annotation processor quali ed class names. If
483
speci ed, kapt does not try to nd annotation processors in apclasspath .
— verbose : Enable verbose output.
— aptMode (required)
— stubs – only generate stubs needed for annotation processing;
— apt – only run annotation processing;
— stubsAndApt – generate stubs and run annotation processing.
— correctErrorTypes : See below. Disabled by default.
The plugin option format is: -P plugin:<plugin id>:<key>=<value> . Options can be

repeated.
An example:
-P plugin:org.jetbrains.kotlin.kapt3:sources=build/kapt/sources
-P plugin:org.jetbrains.kotlin.kapt3:classes=build/kapt/classes
-P plugin:org.jetbrains.kotlin.kapt3:stubs=build/kapt/stubs
-P plugin:org.jetbrains.kotlin.kapt3:apclasspath=lib/ap.jar
-P plugin:org.jetbrains.kotlin.kapt3:apclasspath=lib/anotherAp.jar
-P plugin:org.jetbrains.kotlin.kapt3:correctErrorTypes=true
Generating Kotlin sources

Kapt can generate Kotlin sources. Just write the generated Kotlin source les to the directory
speci ed by processingEnv.options["kapt.kotlin.generated"] , and these les will be
compiled together with the main sources.
You can nd the complete sample in the kotlin-examples Github repository.
Note that Kapt does not support multiple rounds for the generated Kotlin les.
AP/Javac options encoding

apoptions and javacArguments CLI options accept an encoded map of options.
Here is how you can encode options by yourself:
fun encodeList(options: Map<String, String>): String {

val os = ByteArrayOutputStream()
val oos = ObjectOutputStream(os)
oos.writeInt(options.size)
for ((key, value) in options.entries) {
oos.writeUTF(key)
oos.writeUTF(value)
}
oos.flush()
return Base64.getEncoder().encodeToString(os.toByteArray())
}
484
Using Gradle
In order to build a Kotlin project with Gradle, you should set up the kotlin-gradle plugin, apply it
to your project and add kotlin-stdlib dependencies. Those actions may also be performed
automatically in IntelliJ IDEA by invoking Tools | Kotlin | Con gure Kotlin in Project action.
Plugin and Versions

Apply the Kotlin Gradle plugin by using the Gradle plugins DSL. The Kotlin Gradle plugin 1.3.31
works with Gradle 4.1 and later.
plugins {
id 'org.jetbrains.kotlin.<...>' version '1.3.31'
}
plugins {
kotlin("<...>") version "1.3.31"
}
The placeholder <...> should be replaced with one of the plugin names that can be found in
further sections.
Alternatively, apply plugin by adding the kotlin-gradle-plugin dependency to the build

script classpath:
buildscript {
repositories {
mavenCentral()
}
dependencies {
classpath "org.jetbrains.kotlin:kotlin-gradle-plugin:1.3.31"
}
}
plugins {
id "org.jetbrains.kotlin.<...>" version "1.3.31"
}
buildscript {
repositories {
mavenCentral()
}
dependencies {
classpath(kotlin("gradle-plugin", version = "1.3.31"))
}
}
plugins {
kotlin("<...>")
}
This is not required when using Kotlin Gradle plugin 1.1.1 and above with the Gradle plugins DSL,
and with Gradle Kotlin DSL.
485
Building Kotlin Multiplatform Projects
Using the kotlin-multiplatform plugin for building multiplatform projects is described in
Building Multiplatform Projects with Gradle.
Targeting the JVM

To target the JVM, apply the Kotlin JVM plugin. Starting with Kotlin 1.1.1, the plugin can be applied
using the Gradle plugins DSL:
plugins {
id "org.jetbrains.kotlin.jvm" version "1.3.31"
}
plugins {
kotlin("jvm") version "1.3.31"
}
The version should be literal in this block, and it cannot be applied from another build script.
Alternatively, you can use the older apply plugin approach:
apply plugin: 'kotlin'
It's not recommended to apply Kotlin plugins with apply in Gradle Kotlin DSL. The details are
provided below.
Kotlin sources can be mixed with Java sources in the same folder, or in di erent folders. The
default convention is using di erent folders:
project
- src
- main (root)
- kotlin
- java
The corresponding sourceSets property should be updated if not using the default convention:
sourceSets {
main.kotlin.srcDirs += 'src/main/myKotlin'
main.java.srcDirs += 'src/main/myJava'
}
sourceSets["main"].java.srcDir("src/main/myJava")
sourceSets["main"].withConvention(KotlinSourceSet::class) {
kotlin.srcDir("src/main/myKotlin")
}
With Gradle Kotlin DSL, con gure source sets with java.sourceSets { ... } instead.
Targeting JavaScript
When targeting JavaScript, a di erent plugin should be applied:
486
plugins {
id 'kotlin2js' version '1.3.31'
}
plugins {
id("kotlin2js") version "1.3.31"
}
Note that this way of applying the Kotlin/JS plugin requires adding the following code to Gradle
settings le ( settings.gradle ):
pluginManagement {
resolutionStrategy {
eachPlugin {
if (requested.id.id == "kotlin2js") {
useModule("org.jetbrains.kotlin:kotlin-gradle-
plugin:${requested.version}")
}
}
}
}
This plugin only works for Kotlin les so it is recommended to keep Kotlin and Java les separate
(in case if the same project contains Java les). As with targeting the JVM, if not using the default
convention, you should specify the source folder using sourceSets:
sourceSets {
main.kotlin.srcDirs += 'src/main/myKotlin'
}
sourceSets["main"].withConvention(KotlinSourceSet::class) {
kotlin.srcDir("src/main/myKotlin")
}
In addition to the output JavaScript le, the plugin by default creates an additional JS le with
binary descriptors. This le is required if you're building a reusable library that other Kotlin
modules can depend on, and should be distributed together with the result of translation. The
generation is controlled by the kotlinOptions.metaInfo option:
compileKotlin2Js {
kotlinOptions.metaInfo = true
}
tasks {
"compileKotlin2Js"(Kotlin2JsCompile::class) {
kotlinOptions.metaInfo = true
}
}
Targeting Android
Android's Gradle model is a little di erent from ordinary Gradle, so if we want to build an
Android project written in Kotlin, we need kotlin-android plugin instead of kotlin:
487
buildscript {
ext.kotlin_version = '1.3.31'
...
dependencies {
classpath 'com.android.tools.build:gradle:3.2.1'
classpath "org.jetbrains.kotlin:kotlin-gradle-plugin:$kotlin_version"
}
}
plugins {
id 'com.android.application'
id 'kotlin-android'
}
buildscript {
dependencies {
classpath("com.android.tools.build:gradle:3.2.1")
classpath(kotlin("gradle-plugin", version = "1.3.31"))
}
}
plugins {
id("com.android.application")
id("kotlin-android")
}
Kotlin Gradle plugin 1.3.31 works with Android Gradle Plugin 3.0 and later.
Don't forget to con gure the standard library dependency.
Android Studio
If using Android Studio, the following needs to be added under android:
android {
...
sourceSets {
main.java.srcDirs += 'src/main/kotlin'
}
}
android {
...
sourceSets["main"].java.srcDir("src/main/kotlin")
}
This lets Android Studio know that the kotlin directory is a source root, so when the project
model is loaded into the IDE it will be properly recognized. Alternatively, you can put Kotlin
classes in the Java source directory, typically located in src/main/java .
Con guring Dependencies
488
In addition to the kotlin-gradle-plugin dependency shown above, you need to add a
dependency on the Kotlin standard library:
repositories {
mavenCentral()
}
dependencies {
implementation "org.jetbrains.kotlin:kotlin-stdlib"
}
repositories {
mavenCentral()
}
dependencies {
implementation(kotlin("stdlib"))
}
The Kotlin standard library kotlin-stdlib targets Java 6 and above. There are extended
versions of the standard library that add support for some of the features of JDK 7 and JDK 8. To
use these versions, add one of the following dependencies instead of kotlin-stdlib :
implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk7"
implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
In Kotlin 1.1.x, use kotlin-stdlib-jre7 and kotlin-stdlib-jre8 instead.
If you target JavaScript, use the stdlib-js dependency.
implementation "org.jetbrains.kotlin:kotlin-stdlib-js"
implementation(kotlin("stdlib-js"))
If your project uses Kotlin re ection or testing facilities, you need to add the corresponding
dependencies as well:
implementation "org.jetbrains.kotlin:kotlin-reflect"
testImplementation "org.jetbrains.kotlin:kotlin-test"
testImplementation "org.jetbrains.kotlin:kotlin-test-junit"
implementation(kotlin("reflect"))
testImplementation(kotlin("test"))
testImplementation(kotlin("test-junit"))
Starting with Kotlin 1.1.2, the dependencies with group org.jetbrains.kotlin are by default
resolved with the version taken from the applied plugin. You can provide the version manually
using the full dependency notation:
489
implementation "org.jetbrains.kotlin:kotlin-stdlib:$kotlin_version"
implementation(kotlin("stdlib", kotlinVersion))
Annotation Processing
Kotlin supports annonation processing via the Kotlin annotation processing tool( kapt ). Usage of
kapt with Gradle is described on the kapt page.
Incremental Compilation
The Kotlin Gradle plugin supports incremental compilation. Incremental compilation tracks
changes of source les between builds so only les a ected by these changes would be
compiled.
Incremental compilation is supported for Kotlin/JVM and Kotlin/JS projects. It's enabled by default
since Kotlin 1.1.1 for Kotlin/JVM and 1.3.20 for Kotlin/JS.
There are several ways to override the default setting:
— In Gradle con guration les: add the line kotlin.incremental=<value> for Kotlin/JVM or
kotlin.incremental.js=<value> for Kotlin/JS projects either to gradle.properties
or to local.properties le. <value> is a boolean value re ecting the usage of
incremental compilation.
— In Gradle command line parameters: add the parameter -Pkotlin.incremental or -

Pkotlin.incremental.js with the boolean value re ecting the usage of incremental
compilation. Note that in this case the parameter should be added to each subsequent build,
and any build with disabled incremental compilation invalidates incremental caches.
Note that the rst build isn't incremental in any case.
Gradle Build Cache Support (since 1.2.20)

The Kotlin plugin supports Gradle Build Cache (Gradle version 4.3 and above is required; caching
is disabled with lower versions).
To disable the caching for all Kotlin tasks, set the system property ag
kotlin.caching.enabled to false (run the build with the argument -
Dkotlin.caching.enabled=false ).
If you use kapt, note that the kapt annotation processing tasks are not cached by default.
However, you can enable caching for them manually. See the kapt page for details.
Compiler Options
To specify additional compilation options, use the kotlinOptions property of a Kotlin
compilation task.
490
When targeting the JVM, the tasks are called compileKotlin for production code and
compileTestKotlin for test code. The tasks for custom source sets are called accordingly to
the compile<Name>Kotlin pattern.
The names of the tasks in Android Projects contain the build variant names and follow the
pattern compile<BuildVariant>Kotlin , for example, compileDebugKotlin ,
compileReleaseUnitTestKotlin .
When targeting JavaScript, the tasks are called compileKotlin2Js and

compileTestKotlin2Js respectively, and compile<Name>Kotlin2Js for custom source
sets.
To con gure a single task, use its name. Examples:
compileKotlin {
kotlinOptions.suppressWarnings = true
}
//or
compileKotlin {
kotlinOptions {
suppressWarnings = true
}
}
import org.jetbrains.kotlin.gradle.tasks.KotlinCompile
// ...
val compileKotlin: KotlinCompile by tasks
compileKotlin.kotlinOptions.suppressWarnings = true
Note that with Gradle Kotlin DSL, you should get the task from the project's tasks rst.
Use the types Kotlin2JsCompile and KotlinCompileCommon for the JS and Common
targets, accordingly.
It is also possible to con gure all Kotlin compilation tasks in the project:
tasks.withType(org.jetbrains.kotlin.gradle.tasks.KotlinCompile).all {
kotlinOptions { ... }
}
import org.jetbrains.kotlin.gradle.tasks.KotlinCompile
tasks.withType<KotlinCompile> {
kotlinOptions.suppressWarnings = true
}
The complete list of options for the Gradle tasks is the following:
491
Attributes Common for JVM, JS, and JS DCE
Name Description Possible values Default value
allWarningsAsErrors Report an error if there are any warnings false
suppressWarnings Generate no warnings false
verbose Enable verbose logging output false
freeCompilerArgs A list of additional compiler arguments []
Attributes Common for JVM and JS

Name Description Possible values Default
value
apiVersion Allow to use declarations only from the "1.0", "1.1", "1.2", "1.3",
specified version of bundled libraries "1.4 (EXPERIMENTAL)"
languageVersion Provide source compatibility with specified "1.0", "1.1", "1.2", "1.3",
language version "1.4 (EXPERIMENTAL)"
Attributes Speci c for JVM

value
javaParameters Generate metadata for Java 1.8 reflection on method false
parameters
jdkHome Path to JDK home directory to include into classpath, if
differs from default JAVA_HOME
jvmTarget Target version of the generated JVM bytecode (1.6, 1.8, "1.6", "1.8", "9", "1.6"
9, 10, 11 or 12), default is 1.6 "10", "11", "12"
noJdk Don't include Java runtime into classpath false
noReflect Don't include Kotlin reflection implementation into true
classpath
noStdlib Don't include Kotlin runtime into classpath true
Attributes Speci c for JS

value
friendModulesDisabled Disable internal declaration export false
main Whether a main function should be called "call", "noCall" "call"
metaInfo Generate .meta.js and .kjsm files with true
metadata. Use to create a library
moduleKind Kind of a module generated by compiler "plain", "amd", "plain"
"commonjs", "umd"
noStdlib Don't use bundled Kotlin stdlib true
outputFile Output file path
sourceMap Generate source map false
sourceMapEmbedSources Embed source files into source map "never", "always",
"inlining"
sourceMapPrefix Prefix for paths in a source map
target Generate JS files for specific ECMA version "v5" "v5"
typedArrays Translate primitive arrays to JS typed arrays true
Generating Documentation
492
To generate documentation for Kotlin projects, use Dokka; please refer to the Dokka README for
con guration instructions. Dokka supports mixed-language projects and can generate output in
multiple formats, including standard JavaDoc.
OSGi
For OSGi support see the Kotlin OSGi page.
Using Gradle Kotlin DSL

When using Gradle Kotlin DSL, apply the Kotlin plugins using the plugins { ... } block. If
you apply them with apply { plugin(...) } instead, you may encounter unresolved
references to the extensions generated by Gradle Kotlin DSL. To resolve that, you can comment
out the erroneous usages, run the Gradle task kotlinDslAccessorsSnapshot , then
uncomment the usages back and rerun the build or reimport the project into the IDE.
Examples
The following examples show di erent possibilities of con guring the Gradle plugin:
— Kotlin
— Mixed Java and Kotlin
— Android
— JavaScript
493
Using Maven
Plugin and Versions

The kotlin-maven-plugin compiles Kotlin sources and modules. Currently only Maven v3 is
supported.
De ne the version of Kotlin you want to use via a kotlin.version property:
<properties>
<kotlin.version>1.3.31</kotlin.version>
</properties>
Dependencies
Kotlin has an extensive standard library that can be used in your applications. Con gure the
following dependency in the pom le:
<dependencies>
<dependency>
<artifactId>kotlin-stdlib</artifactId>
</dependency>
</dependencies>
If you're targeting JDK 7 or JDK 8, you can use extended versions of the Kotlin standard library
which contain additional extension functions for APIs added in new JDK versions. Instead of
kotlin-stdlib , use kotlin-stdlib-jdk7 or kotlin-stdlib-jdk8 , depending on your
JDK version (for Kotlin 1.1.x use kotlin-stdlib-jre7 and kotlin-stdlib-jre8 as the
jdk counterparts were introduced in 1.2.0).
If your project uses Kotlin re ection or testing facilities, you need to add the corresponding
dependencies as well. The artifact IDs are kotlin-reflect for the re ection library, and
kotlin-test and kotlin-test-junit for the testing libraries.
Compiling Kotlin only source code

To compile source code, specify the source directories in the tag:
<build>
<sourceDirectory>${project.basedir}/src/main/kotlin</sourceDirectory>
<testSourceDirectory>${project.basedir}/src/test/kotlin</testSourceDirectory>
</build>
The Kotlin Maven Plugin needs to be referenced to compile the sources:
494
<build>
<plugins>
<plugin>
<executions>
<execution>
<id>compile</id>
<goals> <goal>compile</goal> </goals>
</execution>
<execution>
<id>test-compile</id>
<goals> <goal>test-compile</goal> </goals>
</execution>
</executions>
</plugin>
</plugins>
</build>
Compiling Kotlin and Java sources

To compile mixed code applications Kotlin compiler should be invoked before Java compiler. In
maven terms that means kotlin-maven-plugin should be run before maven-compiler-plugin using
the following method, making sure that the kotlin plugin is above the maven-compiler-plugin in
your pom.xml le:
495
<build>
<plugins>
<plugin>
<executions>
<execution>
<id>compile</id>
<configuration>
<sourceDirs>
<sourceDir>${project.basedir}/src/main/kotlin</sourceDir>
<sourceDir>${project.basedir}/src/main/java</sourceDir>
</sourceDirs>
</configuration>
</execution>
<execution>
<goals> <goal>test-compile</goal> </goals>
<configuration>
<sourceDirs>
<sourceDir>${project.basedir}/src/test/kotlin</sourceDir>
<sourceDir>${project.basedir}/src/test/java</sourceDir>
</sourceDirs>
</configuration>
</execution>
</executions>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<version>3.5.1</version>
<executions>

<execution>
<id>default-compile</id>
<phase>none</phase>
</execution>

<args>
<arg>-Xjsr305=strict</arg> 
<kotlin2js src="root1" output="out.js" metaInfo="true"/>
</target>
</project>
References
Complete list of elements and attributes are listed below:
Attributes common for kotlinc and kotlin2js

Name Description Required Default Value
src Kotlin source file or directory to compile Yes
nowarn Suppresses all compilation warnings No false
noStdlib Does not include the Kotlin standard library into the classpath No false
failOnError Fails the build if errors are detected during the compilation No true
kotlinc Attributes
Name Description Required Default Value
output Destination directory or .jar file name Yes
classpath Compilation class path No
classpathref Compilation class path reference No
includeRuntime If output is a .jar file, whether Kotlin runtime No true
library is included in the jar
moduleName Name of the module being compiled No The name of the target (if
specified) or the project
kotlin2js Attributes
Name Description Required
output Destination file Yes
libraries Paths to Kotlin libraries No
outputPrefix Prefix to use for generated JavaScript files No
outputSuffix Suffix to use for generated JavaScript files No
sourcemap Whether sourcemap file should be generated No
metaInfo Whether metadata file with binary descriptors should be generated No
main Should compiler generated code call the main function No
Passing raw compiler arguments
To pass custom raw compiler arguments, you can use <compilerarg> elements with either
value or line attributes. This can be done within the <kotlinc> , <kotlin2js> , and
<withKotlin> task elements, as follows:
504
<kotlinc src="${test.data}/hello.kt" output="${temp}/hello.jar">
<compilerarg value="-Xno-inline"/>
<compilerarg line="-Xno-call-assertions -Xno-param-assertions"/>
<compilerarg value="-Xno-optimize"/>
</kotlinc>
The full list of arguments that can be used is shown when you run kotlinc -help .
505
Kotlin and OSGi
To enable Kotlin OSGi support you need to include kotlin-osgi-bundle instead of regular
Kotlin libraries. It is recommended to remove kotlin-runtime , kotlin-stdlib and
kotlin-reflect dependencies as kotlin-osgi-bundle already contains all of them. You
also should pay attention in case when external Kotlin libraries are included. Most regular Kotlin
dependencies are not OSGi-ready, so you shouldn't use them and should remove them from
your project.
Maven
To include the Kotlin OSGi bundle to a Maven project:
<dependencies>
<dependency>
<artifactId>kotlin-osgi-bundle</artifactId>
</dependency>
</dependencies>
To exclude the standard library from external libraries (notice that "star exclusion" works in
Maven 3 only):
<dependency>
<groupId>some.group.id</groupId>
<artifactId>some.library</artifactId>
<version>some.library.version</version>
<exclusions>
<exclusion>
<artifactId>*</artifactId>
</exclusion>
</exclusions>
</dependency>
Gradle
To include kotlin-osgi-bundle to a gradle project:
compile "org.jetbrains.kotlin:kotlin-osgi-bundle:$kotlinVersion"
To exclude default Kotlin libraries that comes as transitive dependencies you can use the
following approach:
dependencies {
compile (
[group: 'some.group.id', name: 'some.library', version: 'someversion'],
.....) {
exclude group: 'org.jetbrains.kotlin'
}
506
FAQ
Why not just add required manifest options to all Kotlin libraries
Even though it is the most preferred way to provide OSGi support, unfortunately it couldn't be
done for now due to so called "package split" issue that couldn't be easily eliminated and such a
big change is not planned for now. There is Require-Bundle feature but it is not the best
option too and not recommended to use. So it was decided to make a separate artifact for OSGi.
507
Compiler Plugins
All-open compiler plugin

Kotlin has classes and their members final by default, which makes it inconvenient to use
frameworks and libraries such as Spring AOP that require classes to be open . The all-open
compiler plugin adapts Kotlin to the requirements of those frameworks and makes classes
annotated with a speci c annotation and their members open without the explicit open
keyword.
For instance, when you use Spring, you don't need all the classes to be open, but only classes
annotated with speci c annotations like @Configuration or @Service . All-open allows to
specify such annotations.
We provide all-open plugin support both for Gradle and Maven with the complete IDE integration.
Note: For Spring you can use the kotlin-spring compiler plugin (see below).
Using in Gradle
Add the plugin artifact to the buildscript dependencies and apply the plugin:
buildscript {
dependencies {
classpath "org.jetbrains.kotlin:kotlin-allopen:$kotlin_version"
}
}
apply plugin: "kotlin-allopen"
As an alternative, you can enable it using the plugins block:
plugins {
id "org.jetbrains.kotlin.plugin.allopen" version "1.3.31"
}
Then specify the list of annotations that will make classes open:
allOpen {
annotation("com.my.Annotation")
// annotations("com.another.Annotation", "com.third.Annotation")
}
If the class (or any of its superclasses) is annotated with com.my.Annotation , the class itself
and all its members will become open.
It also works with meta-annotations:
508
@com.my.Annotation
annotation class MyFrameworkAnnotation
@MyFrameworkAnnotation
class MyClass // will be all-open
MyFrameworkAnnotation is annotated with the all-open meta-annotation

com.my.Annotation , so it becomes an all-open annotation as well.
Using in Maven
Here's how to use all-open with Maven:
<plugin>
<configuration>
<compilerPlugins>

<plugin>all-open</plugin>
</compilerPlugins>
<pluginOptions>

<option>all-open:annotation=com.my.Annotation</option>
<option>all-open:annotation=com.their.AnotherAnnotation</option>
</pluginOptions>
</configuration>
<dependencies>
<dependency>
<artifactId>kotlin-maven-allopen</artifactId>
</dependency>
</dependencies>
</plugin>
Please refer to the "Using in Gradle" section above for the detailed information about how all-
open annotations work.
Spring support
If you use Spring, you can enable the kotlin-spring compiler plugin instead of specifying Spring
annotations manually. The kotlin-spring is a wrapper on top of all-open, and it behaves exactly
the same way.
As with all-open, add the plugin to the buildscript dependencies:
509
buildscript {
dependencies {
classpath "org.jetbrains.kotlin:kotlin-allopen:$kotlin_version"
}
}
apply plugin: "kotlin-spring" // instead of "kotlin-allopen"
Or using the Gradle plugins DSL:
plugins {
id "org.jetbrains.kotlin.plugin.spring" version "1.3.31"
}
In Maven, enable the spring plugin:
<compilerPlugins>
<plugin>spring</plugin>
</compilerPlugins>
The plugin speci es the following annotations: @Component, @Async, @Transactional,

@Cacheable and @SpringBootTest. Thanks to meta-annotations support classes annotated
with @Configuration, @Controller, @RestController, @Service or @Repository are
automatically opened since these annotations are meta-annotated with @Component.
Of course, you can use both kotlin-allopen and kotlin-spring in the same project.
Note that if you use the project template generated by the start.spring.io service, the kotlin-
spring plugin will be enabled by default.
Using in CLI
All-open compiler plugin JAR is available in the binary distribution of the Kotlin compiler. You can
attach the plugin by providing the path to its JAR le using the Xplugin kotlinc option:
-Xplugin=$KOTLIN_HOME/lib/allopen-compiler-plugin.jar
You can specify all-open annotations directly, using the annotation plugin option, or enable
the "preset". The only preset available now for all-open is spring .
# The plugin option format is: "-P plugin:<plugin id>:<key>=<value>".

# Options can be repeated.
-P plugin:org.jetbrains.kotlin.allopen:annotation=com.my.Annotation
-P plugin:org.jetbrains.kotlin.allopen:preset=spring
No-arg compiler plugin

The no-arg compiler plugin generates an additional zero-argument constructor for classes with a
speci c annotation.
510
The generated constructor is synthetic so it can’t be directly called from Java or Kotlin, but it can
be called using re ection.
This allows the Java Persistence API (JPA) to instantiate a class although it doesn't have the zero-
parameter constructor from Kotlin or Java point of view (see the description of kotlin-jpa
plugin below).
Using in Gradle
The usage is pretty similar to all-open.
Add the plugin and specify the list of annotations that must lead to generating a no-arg
constructor for the annotated classes.
buildscript {
dependencies {
classpath "org.jetbrains.kotlin:kotlin-noarg:$kotlin_version"
}
}
apply plugin: "kotlin-noarg"
plugins {
id "org.jetbrains.kotlin.plugin.noarg" version "1.3.31"
}
Then specify the list of no-arg annotations:
noArg {
}
Enable invokeInitializers option if you want the plugin to run the initialization logic from
the synthetic constructor. Starting from Kotlin 1.1.3-2, it is disabled by default because of KT-
18667 and KT-18668 which will be addressed in the future.
noArg {
invokeInitializers = true
}
Using in Maven
511
<plugin>
<configuration>
<compilerPlugins>

<plugin>no-arg</plugin>
</compilerPlugins>
<pluginOptions>
<option>no-arg:annotation=com.my.Annotation</option>


</pluginOptions>
</configuration>
<dependencies>
<dependency>
<artifactId>kotlin-maven-noarg</artifactId>
</dependency>
</dependencies>
</plugin>
JPA support
As with the kotlin-spring plugin, kotlin-jpa is a wrapped on top of no-arg. The plugin speci es
@Entity, @Embeddable and @MappedSuperclass no-arg annotations automatically.
That's how you add the plugin in Gradle:
buildscript {
dependencies {
classpath "org.jetbrains.kotlin:kotlin-noarg:$kotlin_version"
}
}
apply plugin: "kotlin-jpa"
plugins {
id "org.jetbrains.kotlin.plugin.jpa" version "1.3.31"
}
In Maven, enable the jpa plugin:
<compilerPlugins>
<plugin>jpa</plugin>
</compilerPlugins>
Using in CLI
512
As with all-open, add the plugin JAR le to the compiler plugin classpath and specify annotations
or presets:
-Xplugin=$KOTLIN_HOME/lib/noarg-compiler-plugin.jar
-P plugin:org.jetbrains.kotlin.noarg:annotation=com.my.Annotation
-P plugin:org.jetbrains.kotlin.noarg:preset=jpa
SAM-with-receiver compiler plugin

The sam-with-receiver compiler plugin makes the rst parameter of the annotated Java "single
abstract method" (SAM) interface method a receiver in Kotlin. This conversion only works when
the SAM interface is passed as a Kotlin lambda, both for SAM adapters and SAM constructors (see
the documentation for more details).
Here is an example:
public @interface SamWithReceiver {}
@SamWithReceiver
public interface TaskRunner {
void run(Task task);
}
fun test(context: TaskContext) {

val runner = TaskRunner {
// Here 'this' is an instance of 'Task'
println("$name is started")
context.executeTask(this)
println("$name is finished")
}
}
Using in Gradle
The usage is the same to all-open and no-arg, except the fact that sam-with-receiver does not
have any built-in presets, and you need to specify your own list of special-treated annotations.
buildscript {
dependencies {
classpath "org.jetbrains.kotlin:kotlin-sam-with-receiver:$kotlin_version"
}
}
apply plugin: "kotlin-sam-with-receiver"
Then specify the list of SAM-with-receiver annotations:
samWithReceiver {
}
Using in Maven
513
<plugin>
<configuration>
<compilerPlugins>
<plugin>sam-with-receiver</plugin>
</compilerPlugins>
<pluginOptions>
<option>
sam-with-receiver:annotation=com.my.SamWithReceiver
</option>
</pluginOptions>
</configuration>
<dependencies>
<dependency>
<artifactId>kotlin-maven-sam-with-receiver</artifactId>
</dependency>
</dependencies>
</plugin>
Using in CLI
Just add the plugin JAR le to the compiler plugin classpath and specify the list of sam-with-
receiver annotations:
-Xplugin=$KOTLIN_HOME/lib/sam-with-receiver-compiler-plugin.jar
-P plugin:org.jetbrains.kotlin.samWithReceiver:annotation=com.my.SamWithReceiver
514
Code Style Migration Guide
Kotlin Coding Conventions and IntelliJ IDEA formatter

Kotlin Coding Conventions a ect several aspects of writing idiomatic Kotlin, and a set of
formatting recommendations aimed at improving Kotlin code readability is among them.
Unfortunately, the code formatter built into IntelliJ IDEA had to work long before this document
was released and now has a default setup that produces di erent formatting from what is now
recommended.
It may seem a logical next step to remove this obscurity by switching the defaults in IntelliJ IDEA
and make formatting consistent with the Kotlin Coding Conventions. But this would mean that all
the existing Kotlin projects will have a new code style enabled the moment the Kotlin plugin is
installed. Not really the expected result for plugin update, right?
That's why we have the following migration plan instead:
— Enable the o cial code style formatting by default starting from Kotlin 1.3 and only for new
projects (old formatting can be enabled manually)
— Authors of existing projects may choose to migrate to the Kotlin Coding Conventions
— Authors of existing projects may choose to explicitly declare using the old code style in a
project (this way the project won't be a ected by switching to the defaults in the future)
— Switch to the default formatting and make it consistent with Kotlin Coding Conventions in
Kotlin 1.4
Di erences between "Kotlin Coding Conventions" and "IntelliJ IDEA default

code style"
The most notable change is in the continuation indentation policy. There's a nice idea to use the
double indent for showing that a multi-line expression hasn't ended on the previous line. This is
a very simple and general rule, but several Kotlin constructions look a bit awkward when they are
formatted this way. In Kotlin Coding Conventions it's recommended to use a single indent in
cases where the long continuation indent has been forced before
In practice, quite a bit of code is a ected, so this can be considered a major code style update.
Migration to a new code style discussion

A new code style adoption might be a very natural process if it starts with a new project, when
there's no code formatted in the old way. That is why starting from version 1.3, the Kotlin IntelliJ
Plugin creates new projects with formatting from the Code Conventions document which is
enabled by default.
Changing formatting in an existing project is a far more demanding task, and should probably be
started with discussing all the caveats with the team.
515
The main disadvantage of changing the code style in an existing project is that the
blame/annotate VCS feature will point to irrelevant commits more often. While each VCS has
some kind of way to deal with this problem ("Annotate Previous Revision" can be used in IntelliJ
IDEA), it's important to decide if a new style is worth all the e ort. The practice of separating
reformatting commits from meaningful changes can help a lot with later investigations.
Also migrating can be harder for larger teams because committing a lot of les in several
subsystems may produce merging con icts in personal branches. And while each con ict
resolution is usually trivial, it's still wise to know if there are large feature branches currently in
work.
In general, for small projects, we recommend converting all the les at once.
For medium and large projects the decision may be tough. If you are not ready to update many
les right away you may decide to migrate module by module, or continue with gradual migration
for modi ed les only.
Migration to a new code style

Switching to the Kotlin Coding Conventions code style can be done in Settings → Editor →
Code Style → Kotlin dialog. Switch scheme to Project and activate Set from... →
Predefined Style → Kotlin Style Guide .
In order to share those changes for all project developers .idea/codeStyle folder have to be
committed to VCS.
If an external build system is used for con guring the project, and it's been decided not to share
.idea/codeStyle folder, Kotlin Coding Conventions can be forced with an additional property:
In Gradle
Add kotlin.code.style=o cial property to the gradle.properties le at the project root and
commit the le to VCS.
In Maven
Add kotlin.code.style o cial property to root pom.xml project le.
<properties>
<kotlin.code.style>official</kotlin.code.style>
</properties>
Warning: having the kotlin.code.style option set may modify the code style scheme during a
project import and may change the code style settings.
After updating your code style settings, activate “Reformat Code” in the project view on the
desired scope.
516
For a gradual migration, it's possible to enable the "File is not formatted according to project
settings" inspection. It will highlight the places that should be reformatted. After enabling the
"Apply only to modi ed les" option, inspection will show formatting problems only in modi ed
les. Such les are probably going to be committed soon anyway.
Store old code style in project

It's always possible to explicitly set the IntelliJ IDEA code style as the correct code style for the
project. To do so please switch to the Project scheme in Settings → Editor → Code Style
→ Kotlin and select "Kotlin obsolete IntelliJ IDEA codestyle" in the "Use defaults from:" on the Load
tab.
In order to share the changes across the project developers .idea/codeStyle folder, it has to
be committed to VCS. Alternatively kotlin.code.style=obsolete can be used for projects
con gured with Gradle or Maven.
517
Evolution
Kotlin Evolution
Principles of Pragmatic Evolution

Language design is cast in stone,
but this stone is reasonably soft,
and with some effort we can reshape it later.
Kotlin Design Team
Kotlin is designed to be a pragmatic tool for programmers. When it comes to language evolution,
its pragmatic nature is captured by the following principles:
— Keep the language modern over the years.

— Stay in the constant feedback loop with the users.
— Make updating to new versions comfortable for the users.
As this is key to understanding how Kotlin is moving forward, let's expand on these principles.
Keeping the Language Modern. We acknowledge that systems accumulate legacy over time.
What had once been cutting-edge technology can be hopelessly outdated today. We have to
evolve the language to keep it relevant to the needs of the users and up-to-date with their
expectations. This includes not only adding new features, but also phasing out old ones that are
no longer recommended for production use and have altogether become legacy.
Comfortable Updates. Incompatible changes, such as removing things from a language, may
lead to painful migration from one version to the next if carried out without proper care. We will
always announce such changes well in advance, mark things as deprecated and provide
automated migration tools before the change happens. By the time the language is changed we
want most of the code in the world to be already updated and thus have no issues migrating to
the new version.
518
Feedback Loop. Going through deprecation cycles requires signi cant e ort, so we want to
minimize the number of incompatible changes we'll be making in the future. Apart from using
our best judgement, we believe that trying things out in real life is the best way to validate a
design. Before casting things in stone we want them battle-tested. This is why we use every
opportunity to make early versions of our designs available in production versions of the
language, but with experimental status. Experimental features are not stable, they can be changed
at any time, and the users that opt into using them do so explicitly to indicate that they are ready
to deal with the future migration issues. These users provide invaluable feedback that we gather
to iterate on the design and make it rock-solid.
Incompatible Changes
If, upon updating from one version to another, some code that used to work doesn't work any
more, it is an incompatible change in the language (sometimes referred to as "breaking change").
There can be debates as to what "doesn't work any more" means precisely in some cases, but it
de nitely includes the following:
— Code that compiled and ran ne is now rejected with an error (at compile or link time). This
includes removing language constructs and adding new restrictions.
— Code that executed normally is now throwing an exception.
The less obvious cases that belong to the "grey area" include handling corner cases di erently,
throwing an exception of a di erent type than before, changing behavior observable only
through re ection, changes in undocumented/unde ned behavior, renaming binary artifacts, etc.
Sometimes such changes are very important and a ect migration experience dramatically,
sometimes they are insigni cant.
Some examples of what de nitely isn't an incompatible change include
— Adding new warnings.

— Enabling new language constructs or relaxing limitations for existing ones.
— Changing private/internal APIs and other implementation details.
The principles of Keeping the Language Modern and Comfortable Updates suggest that
incompatible changes are sometimes necessary, but they should be introduced carefully. Our
goal is to make the users aware of upcoming changes well in advance to let them migrate their
code comfortably.
Ideally, every incompatible change should be announced through a compile-time warning

reported in the problematic code (usually referred to as a deprecation warning) and accompanied
with automated migration aids. So, the ideal migration work ow goes as follows:
— Update to version A (where the change is announced)

— See warnings about the upcoming change
— Migrate the code with the help of the tooling
519
— Update to version B (where the change happens)
— See no issues at all
In practice some changes can't be accurately detected at compile time, so no warnings can be
reported, but at least the users will be noti ed through Release notes of version A that a change
is coming in version B.
Dealing with compiler bugs
Compilers are complicated software and despite the best e ort of their developers they have
bugs. The bugs that cause the compiler itself to fail or report spurious errors or generate
obviously failing code, though annoying and often embarrassing, are easy to x, because the xes
do not constitute incompatible changes. Other bugs may cause the compiler to generate
incorrect code that does not fail: e.g. by missing some errors in the source or simply generating
wrong instructions. Fixes of such bugs are technically incompatible changes (some code used to
compile ne, but now it won't any more), but we are inclined to xing them as soon as possible to
prevent the bad code patterns from spreading across user code. In our opinion, this serves the
principle of Comfortable Updates, because fewer users have a chance of encountering the issue.
Of course, this applies only to bugs that are found soon after appearing in a released version.
Decision Making
JetBrains, the original creator of Kotlin, is driving its progress with the help of the community and
in accord with the Kotlin Foundation.
All changes to the Kotlin Programming Language are overseen by the Lead Language Designer
(currently Andrey Breslav). The Lead Designer has the nal say in all matters related to language
evolution. Additionally, incompatible changes to fully stable components have to be approved to
by the Language Committee designated under the Kotlin Foundation (currently comprised of
Je rey van Gogh, William R. Cook and Andrey Breslav).
The Language Committee makes nal decisions on what incompatible changes will be made and
what exact measures should be taken to make user updates comfortable. In doing so, it relies on
a set of guidelines available here.
Feature Releases and Incremental Releases

Stable releases with versions 1.2, 1.3, etc. are usually considered to be feature releases bringing
major changes in the language. Normally, we publish incremental releases, numbered 1.2.20,
1.2.30, etc, in between feature releases.
Incremental releases bring updates in the tooling (often including features), performance
improvements and bug xes. We try to keep such versions compatible with each other, so
changes to the compiler are mostly optimizations and warning additions/removals. Experimental
features may, of course, be added, removed or changed at any time.
520
Feature releases often add new features and may remove or change previously deprecated ones.
Feature graduation from experimental to stable also happens in feature releases.
EAP Builds
Before releasing stable versions, we usually publish a number of preview builds dubbed EAP (for
"Early Access Preview") that let us iterate faster and gather feedback from the community. EAPs
of feature releases usually produce binaries that will be later rejected by the stable compiler to
make sure that possible bugs in the binary format survive no longer than the preview period.
Final Release Candidates normally do not bear this limitation.
Experimental features
According to the Feedback Loop principle described above, we iterate on our designs in the open
and release versions of the language where some features have the experimental status and are
supposed to change. Experimental features can be added, changed or removed at any point and
without warning. We make sure that experimental features can't be used accidentally by an
unsuspecting user. Such features usually require some sort of an explicit opt-in either in the code
or in the project con guration.
Experimental features usually graduate to the stable status after some iterations.
Status of di erent components
To check the stability status of di erent components of Kotlin (Kotlin/JVM, JS, Native, various
libraries, etc), please consult this link.
Libraries
A language is nothing without its ecosystem, so we pay extra attention to enabling smooth library
evolution.
Ideally, a new version of a library can be used as a "drop-in replacement" for an older version.
This means that upgrading a binary dependency should not break anything, even if the
application is not recompiled (this is possible under dynamic linking).
On the one hand, to achieve this, the compiler has to provide certain ABI stability guarantees
under the constraints of separate compilation. This is why every change in the language is
examined from the point of view of binary compatibility.
On the other hand, a lot depends on the library authors being careful about which changes are
safe to make. Thus it's very important that library authors understand how source changes a ect
compatibility and follow certain best practices to keep both APIs and ABIs of their libraries stable.
Here are some assumptions that we make when considering language changes from the library
evolution standpoint:
521
— Library code should always specify return types of public/protected functions and properties
explicitly thus never relying on type inference for public API. Subtle changes in type inference
may cause return types to change inadvertently, leading to binary compatibility issues.
— Overloaded functions and properties provided by the same library should do essentially the
same thing. Changes in type inference may result in more precise static types to be known at
call sites causing changes in overload resolution.
Library authors can use the @Deprecated and @Experimental annotations to control the
evolution of their API surface. Note that @Deprecated(level=HIDDEN) can be used to preserve
binary compatibility even for declarations removed from the API.
Also, by convention, packages named "internal" are not considered public API. All API residing in
packages named "experimental" is considered experimental and can change at any moment.
We evolve the Kotlin Standard Library (kotlin-stdlib) for stable platforms according to the
principles stated above. Changes to the contracts for its API undergo the same procedures as
changes in the language itself.
Compiler Keys
Command line keys accepted by the compiler are also a kind of public API, and they are subject
to the same considerations. Supported ags (those that don't have the "-X" or "-XX" pre x) can be
added only in feature releases and should be properly deprecated before removing them. The "-
X" and "-XX" ags are experimental and can be added and removed at any time.
Compatibility Tools
As legacy features get removed and bugs xed, the source language changes, and old code that
has not been properly migrated may not compile any more. The normal deprecation cycle allows
a comfortable period of time for migration, and even when it's over and the change ships in a
stable version, there's still a way to compile unmigrated code.
Compatibility ags
We provide the -language-version and -api-version ags that make a new version emulate the
behaviour of an old one, for compatibility purposes. Normally, at least one previous version is
supported. This e ectively leaves a time span of two full feature release cycles for migration
(which usually amounts to about two years). Using an older kotlin-stdlib or kotlin-re ect with a
newer compiler without specifying compatibility ags is not recommended, and the compiler will
report a warning when this happens.
Actively maintained code bases can bene t from getting bug xes ASAP, without waiting for a full
deprecation cycle to complete. Currently such project can enable the -progressive ag and get
such xes enabled even in incremental releases.
All ags are available on the command line as well as Gradle and Maven.
522
Evolving the binary format
Unlike sources that can be xed by hand in the worst case, binaries are a lot harder to migrate,
and this makes backwards compatibility very important in the case of binaries. Incompatible
changes to binaries can make updates very uncomfortable and thus should be introduced with
even more care than those in the source language syntax.
For fully stable versions of the compiler the default binary compatibility protocol is the following:
— All binaries are backwards compatible, i.e. a newer compiler can read older binaries (e.g. 1.3
understands 1.0 through 1.2),
— Older compilers reject binaries that rely on new features (e.g. a 1.0 compiler rejects binaries
that use coroutines).
— Preferably (but we can't guarantee it), the binary format is mostly forwards compatible with
the next feature release, but not later ones (in the cases when new features are not used, e.g.
1.3 can understand most binaries from 1.4, but not 1.5).
This protocol is designed for comfortable updates as no project can be blocked from updating its
dependencies even if it's using a slightly outdated compiler.
Please note that not all target platforms have reached this level of stability (but Kotlin/JVM has).
523
Stability of Di erent Components
There can be di erent modes of stability depending of how quickly a component is evolving:
— Moving fast (MF): no compatibility should be expected between even incremental releases,
any functionality can be added, removed or changed without warning.
— Additions in Incremental Releases (AIR): things can be added in an incremental release,

removals and changes of behavior should be avoided and announced in a previous
incremental release if necessary.
— Stable Incremental Releases (SIR): incremental releases are fully compatible, only
optimizations and bug xes happen. Any changes can be made in a feature release.
— Fully Stable (FS): incremental releases are fully compatible, only optimizations and bug xes
happen. Feature releases are backwards compatible.
Source and binary compatibility may have di erent modes for the same component, e.g. the
source language can reach full stability before the binary format stabilizes, or vice versa.
The provisions of the Kotlin evolution policy fully apply only to components that have reached
Full Stability (FS). From that point on incompatible changes have to be approved by the Language
Committee.
Component Status Entered at Mode for Mode for

version Sources Binaries
Kotlin/JVM 1.0 FS FS
kotlin-stdlib (JVM) 1.0 FS FS
KDoc syntax 1.0 FS N/A
Coroutines 1.3 FS FS
kotlin-reflect (JVM) 1.0 SIR SIR
Kotlin/JS 1.1 AIR MF
Kotlin/Native 1.3 AIR MF
Kotlin Scripts (*.kts) 1.2 AIR MF
dokka 0.1 MF N/A
Kotlin Scripting APIs 1.2 MF MF
Compiler Plugin API 1.0 MF MF
Serialization 1.3 MF MF
Multiplatform Projects 1.2 MF MF
Inline classes 1.3 MF MF
Unsigned arithmetics 1.3 MF MF
All other experimental features, by N/A MF MF
default
524
Compatibility Guide for Kotlin 1.3
Keeping the Language Modern and Comfortable Updates are among the fundamental principles in
Kotlin Language Design. The former says that constructions which obstruct language evolution
should be removed, and the latter says that this removal should be well-communicated
beforehand to make code migration as smooth as possible.
While most of the language changes were already announced through other channels, like
update changelogs or compiler warnings, this document summarizes them all, providing a
complete reference for migration from Kotlin 1.2 to Kotlin 1.3
Basic terms
In this document we introduce several kinds of compatibility:
— Source: source-incompatible change stops code that used to compile ne (without errors or
warnings) from compiling anymore
— Binary: two binary artifacts are said to be binary-compatible if interchanging them doesn't
lead to loading or linkage errors
— Behavioral: a change is said to be behavioral-incompatible if one and the same program
demonstrates di erent behavior before and after applying the change
One has to remember that those de nitions are given only for pure Kotlin. Compatibility of Kotlin
code from the other languages perspective (e.g. from Java) is out of the scope of this document.
Evaluation order of constructor arguments regarding <clinit> call
Issue: KT-19532
Component: Kotlin/JVM
Incompatible change type: behavioral
Short summary: evaluation order with respect to class initialization is changed in 1.3
Deprecation cycle:
— <1.3: old behavior (see details in the Issue)

— >= 1.3: behavior changed, -Xnormalize-constructor-calls=disable can be used to
temporarily revert to pre-1.3 behavior. Support for this ag is going to be removed in the
next major release.
Missing getter-targeted annotations on annotation constructor parameters
525
Issue: KT-25287
Incompatible change type: behavioral
Short summary: getter-target annotations on annotations constructor parameters will be

properly written to class les in 1.3
Deprecation cycle:
— <1.3: getter-target annotations on annotation constructor parameters are not applied

— >=1.3: getter-target annotations on annotation constructor parameters are properly
applied and written to the generated code
Missing errors in class constructor’s @get: annotations
Issue: KT-19628
Component: Core language
Incompatible change type: Source
Short summary: errors in getter-target annotations will be reported properly in 1.3
Deprecation cycle:
— <1.2: compilation errors in getter-target annotations were not reported, causing incorrect
code to be compiled ne.
— 1.2.x: errors reported only by tooling, the compiler still compiles such code without any
warnings
— >=1.3: errors reported by the compiler too, causing erroneous code to be rejected
Nullability assertions on access to Java types annotated with @NotNull
526
Issue: KT-20830
Incompatible change type: Behavioral
Short summary: nullability assertions for Java-types annotated with not-null annotations will
be generated more aggressively, causing code which passes null here to fail faster.
Deprecation cycle:
— <1.3: the compiler could miss such assertions when type inference was involved, allowing
potential null propagation during compilation against binaries (see Issue for details).
— >=1.3: the compiler generates missed assertions. This can case code which was
(erroneously) passing nulls here fail faster.
-XXLanguage:-StrictJavaNullabilityAssertions can be used to temporarily
return to the pre-1.3 behavior. Support for this ag will be removed in the next major
release.
Unsound smartcasts on enum members
Issue: KT-20772
Short summary: a smartcast on a member of one enum entry will be correctly applied to
only this enum entry
Deprecation cycle:
— <1.3: a smartcast on a member of one enum entry could lead to an unsound smartcast on
the same member of other enum entries.
— >=1.3: smartcast will be properly applied only to the member of one enum entry.
-XXLanguage:-SoundSmartcastForEnumEntries will temporarily return old
behavior. Support for this ag will be removed in the next major release.
val backing eld reassignment in getter
527
Issue: KT-16681
Components: Core language
Short summary: reassignment of the backing eld of val-property in its getter is now
prohibited
Deprecation cycle:
— <1.2: Kotlin compiler allowed to modify backing eld of val in its getter. Not only it
violates Kotlin semantic, but also generates ill-behaved JVM bytecode which reassigns
final eld.
— 1.2.X: deprecation warning is reported on code which reassigns backing eld of val
— >=1.3: deprecation warnings are elevated to errors
Array capturing before the for-loop where it is iterated
Issue: KT-21354
Short summary: if an expression in for-loop range is a local variable updated in a loop body,
this change a ects loop execution. This is inconsistent with iterating over other containers,
such as ranges, character sequences, and collections.
Deprecation cycle:
— <1.2: described code patterns are compiled ne, but updates to local variable a ect loop
execution
— 1.2.X: deprecation warning reported if a range expression in a for-loop is an array-typed
local variable which is assigned in a loop body
— 1.3: change behavior in such cases to be consistent with other containers
Nested classi ers in enum entries
528
Issue: KT-16310
Short summary: since Kotlin 1.3, nested classi ers (classes, object, interfaces, annotation
classes, enum classes) in enum entries are prohibited
Deprecation cycle:
— <1.2: nested classi ers in enum entries are compiled ne, but may fail with exception at
runtime
— 1.2.X: deprecation warnings reported on the nested classi ers
— >=1.3: deprecation warnings elevated to errors
Data class overriding copy
Issue: KT-19618
Components: Core language
Short summary: since Kotlin 1.3, data classes are prohibited to override copy()
Deprecation cycle:
— <1.2: data classes overriding copy() are compiled ne but may fail at runtime/expose
strange behavior
— 1.2.X: deprecation warnings reported on data classes overriding copy()
Inner classes inheriting Throwable that capture generic parameters from the
outer class
529
Issue: KT-17981
Short summary: since Kotlin 1.3, inner classes are not allowed to inherit Throwable
Deprecation cycle:
— <1.2: inner classes inheriting Throwable are compiled ne. If such inner classes happen
to capture generic parameters, it could lead to strange code patterns which fail at
runtime.
— 1.2.X: deprecation warnings reported on inner classes inheriting Throwable
Visibility rules regarding complex class hierarchies with companion objects
Issues: KT-21515, KT-25333
Short summary: since Kotlin 1.3, rules of visibility by short names are stricter for complex
class hierarchies involving companion objects and nested classi ers.
Deprecation cycle:
— <1.2: old visibility rules (see Issue for details)

— 1.2.X: deprecation warnings reported on short names which are not going to be
accessible anymore. Tooling suggests automated migration by adding full name.
— >=1.3: deprecation warnings elevated to errors. O ending code should add full quali ers
or explicit imports
Non-constant vararg annotation parameters
530
Issue: KT-23153
Short summary: since Kotlin 1.3, setting non-constant values as vararg annotation
parameters is prohibited
Deprecation cycle:
— <1.2: the compiler allows to pass non-constant value for vararg annotation parameter, but
actually drops that value during bytecode generation, leading to non-obvious behavior
— 1.2.X: deprecation warnings reported on such code patterns
Local annotation classes
Issue: KT-23277
Short summary: since Kotlin 1.3 local annotation classes are not supported
Deprecation cycle:
— <1.2: the compiler compiled local annotation classes ne

— 1.2.X: deprecation warnings reported on local annotation classes
Smartcasts on local delegated properties
531
Issue: KT-22517
Short summary: since Kotlin 1.3 smartcasts on local delegated properties are not allowed
Deprecation cycle:
— <1.2: the compiler allowed to smartcast local delegated property, which could lead to
unsound smartcast in case of ill-behaved delegates
— 1.2.X: smartcasts on local delegated properries are reported as deprecated (the compiler
issues warnings)
mod operator convention
Issues: KT-24197
Short summary: since Kotlin 1.3 declaration of mod operator is prohibited, as well as calls
which resolve to such declarations
Deprecation cycle:
— 1.1.X, 1.2.X: report warnings on declarations of operator mod, as well as on calls which
resolve to it
— 1.3.X: elevate warnings to error, but still allow to resolve to operator mod declarations
— 1.4.X: do not resolve calls to operator mod anymore
Passing single element to vararg in named form
532
Issues: KT-20588, KT-20589. See also KT-20171
Short summary: in Kotlin 1.3, assigning single element to vararg is deprecated and should
be replaced with consecutive spread and array construction.
Deprecation cycle:
— <1.2: assigning one value element to vararg in named form compiles ne and is treated as
assigning single element to array, causing non-obvious behavior when assigning array to
vararg
— 1.2.X: deprecation warnings are reported on such assignments, users are suggested to
switch to consecutive spread and array construction.
— 1.3.X: warnings are elevated to errors
— >= 1.4: change semantic of assigning single element to vararg, making assignment of array
equivalent to the assignment of a spread of an array
Retention of annotations with target EXPRESSION
Issue: KT-13762
Short summary: since Kotlin 1.3, only SOURCE retention is allowed for annotations with
target EXPRESSION
Deprecation cycle:
— <1.2: annotations with target EXPRESSION and retention other than SOURCE are allowed,
but silently ignored at use-sites
— 1.2.X: deprecation warnings are reported on declarations of such annotations
— >=1.3: warnings are elevated to errors
Annotations with target PARAMETER shouldn't be applicable to parameter's type
533
Issue: KT-9580
Short summary: since Kotlin 1.3, error about wrong annotation target will be properly
reported when annotation with target PARAMETER is applied to parameter's type
Deprecation cycle:
— <1.2: aforementioned code patterns are compiled ne; annotations are silently ignored
and not present in the bytecode
— 1.2.X: deprecation warnings are reported on such usages
— >=1.3: warnings are elevated to errors
Array.copyOfRange throws an exception when indices are out of bounds instead

of enlarging the returned array
Issue: KT-19489
Component: kotlin-stdlib (JVM)
Short summary: since Kotlin 1.3, ensure that the toIndex argument of
Array.copyOfRange, which represents the exclusive end of the range being copied, is not
greater than the array size and throw IllegalArgumentException if it is.
Deprecation cycle:
— <1.3: in case toIndex in the invocation of Array.copyOfRange is greater than the array
size, the missing elements in range ll be lled with nulls, violating soundness of the
Kotlin type system.
— >=1.3: check that toIndex is in the array bounds, and throw exception if it isn't
Progressions of ints and longs with a step of Int.MIN_VALUE and Long.MIN_VALUE

are outlawed and won’t be allowed to be instantiated
534
Issue: KT-17176
Short summary: since Kotlin 1.3, prohibit step value for integer progressions being the
minimum negative value of its integer type (Long or Int), so that calling
IntProgression.fromClosedRange(0, 1, step = Int.MIN_VALUE) will throw
IllegalArgumentException
Deprecation cycle:
— <1.3: it was possible to create an IntProgression with Int.MIN_VALUE step, which

yields two values [0, -2147483648], which is non-obvious behavior
— >=1.3: throw IllegalArgumentException if the step is the minimum negative value of
its integer type
Check for index over ow in operations on very long sequences
Issue: KT-16097
Short summary: since Kotlin 1.3, make sure index, count and similar methods do not
over ow for long sequences. See the Issue for the full list of a ected methods.
Deprecation cycle:
— <1.3: calling such methods on very long sequences could produce negative results due to
integer over ow
— >=1.3: detect over ow in such methods and throw exception immediately
Unify split by an empty match regex result across the platforms
535
Issue: KT-21049
Short summary: since Kotlin 1.3, unify behavior of split method by empty match regex
across all platforms
Deprecation cycle:
— <1.3: behavior of described calls is di erent when comparing JS, JRE 6, JRE 7 versus JRE 8+
— >=1.3: unify behavior across the platforms
Discontinued deprecated artifacts in the compiler distribution
Issue: KT-23799
Component: other
Incompatible change type: Binary
Short summary: Kotlin 1.3 discontinues the following deprecated binary artifacts:
— kotlin-runtime: use kotlin-stdlib instead
— kotlin-stdlib-jre7/8: use kotlin-stdlib-jdk7/8 instead
— kotlin-jslib in the compiler distribution: use kotlin-stdlib-js instead
Deprecation cycle:
— 1.2.X: the artifacts were marked as deprecated, the compiler reported warning on usage
of those artifacts
— >=1.3: the artifacts are discontinued
Annotations in stdlib
536
Issue: KT-21784
Incompatible change type: Binary
Short summary: Kotlin 1.3 removes annotations from the package

org.jetbrains.annotations from stdlib and moves them to the separate artifacts
shipped with the compiler: annotations-13.0.jar and mutability-annotations-
compat.jar
Deprecation cycle:
— <1.3: annotations were shipped with the stdlib artifact

— >=1.3: annotations ship in separate artifacts
537
FAQ
FAQ
What is Kotlin?
Kotlin is an OSS statically typed programming language that targets the JVM, Android, JavaScript
and Native. It’s developed by JetBrains. The project started in 2010 and was open source from
very early on. The rst o cial 1.0 release was in February 2016.
What is the current version of Kotlin?
The currently released version is 1.3.31, published on April 25, 2019.
Is Kotlin free?
Yes. Kotlin is free, has been free and will remain free. It is developed under the Apache 2.0
license and the source code is available on GitHub.
Is Kotlin an object-oriented language or a functional one?
Kotlin has both object-oriented and functional constructs. You can use it in both OO and FP styles,
or mix elements of the two. With rst-class support for features such as higher-order functions,
function types and lambdas, Kotlin is a great choice if you’re doing or exploring functional
programming.
What advantages does Kotlin give me over the Java programming language?
Kotlin is more concise. Rough estimates indicate approximately a 40% cut in the number of lines
of code. It’s also more type-safe, e.g. support for non-nullable types makes applications less
prone to NPE’s. Other features including smart casting, higher-order functions, extension
functions and lambdas with receivers provide the ability to write expressive code as well as
facilitating creation of DSL.
Is Kotlin compatible with the Java programming language?
538
Yes. Kotlin is 100% interoperable with the Java programming language and major emphasis has
been placed on making sure that your existing codebase can interact properly with Kotlin. You
can easily call Kotlin code from Java and Java code from Kotlin. This makes adoption much easier
and lower-risk. There’s also an automated Java-to-Kotlin converter built into the IDE that
simpli es migration of existing code.
What can I use Kotlin for?
Kotlin can be used for any kind of development, be it server-side, client-side web and Android.
With Kotlin/Native currently in the works, support for other platforms such as embedded
systems, macOS and iOS is coming. People are using Kotlin for mobile and server-side
applications, client-side with JavaScript or JavaFX, and data science, just to name a few
possibilities.
Can I use Kotlin for Android development?
Yes. Kotlin is supported as a rst-class language on Android. There are hundreds of applications
already using Kotlin for Android, such as Basecamp, Pinterest and more. For more information
check out the resource on Android development.
Can I use Kotlin for server-side development?
Yes. Kotlin is 100% compatible with the JVM and as such you can use any existing frameworks
such as Spring Boot, vert.x or JSF. In addition there are speci c frameworks written in Kotlin such
as Ktor. For more information check out the resource on server-side development.
Can I use Kotlin for web development?
Yes. In addition to using for backend web, you can also use Kotlin/JS for client-side web. Kotlin
can use de nitions from De nitelyTyped to get static typing for common JavaScript libraries, and
it is compatible with existing module systems such as AMD and CommonJS. For more information
check out the resource on client-side development.
Can I use Kotlin for desktop development?
Yes. You can use any Java UI framework such as JavaFx, Swing or other. In addition there are
Kotlin speci c frameworks such as TornadoFX.
Can I use Kotlin for native development?
Kotlin/Native is currently in the works. It compiles Kotlin to native code that can run without a
VM. There is a Technology Preview released but it is not production-ready yet, and doesn’t yet
target all the platforms that we plan to support for 1.0. For more information check out the blog
post announcing Kotlin/Native.
539
What IDEs support Kotlin?
Kotlin is supported by all major Java IDEs including IntelliJ IDEA, Android Studio, Eclipse and
NetBeans. In addition, a command line compiler is available and provides straightforward
support for compiling and running applications.
What build tools support Kotlin?
On the JVM side, the main build tools include Gradle, Maven, Ant, and Kobalt. There are also
some build tools available that target client-side JavaScript.
What does Kotlin compile down to?
When targeting the JVM, Kotlin produces Java compatible bytecode. When targeting JavaScript,
Kotlin transpiles to ES5.1 and generates code which is compatible with module systems including
AMD and CommonJS. When targeting native, Kotlin will produce platform-speci c code (via LLVM).
Does Kotlin only target Java 6?
No. Kotlin lets you choose between generating Java 6 and Java 8 compatible bytecode. More
optimal byte code may be generated for higher versions of the platform.
Is Kotlin hard?
Kotlin is inspired by existing languages such as Java, C#, JavaScript, Scala and Groovy. We've tried
to ensure that Kotlin is easy to learn, so that people can easily jump on board, reading and
writing Kotlin in a matter of days. Learning idiomatic Kotlin and using some more of its advanced
features can take a little longer, but overall it is not a complicated language.
What companies are using Kotlin?
There are too many companies using Kotlin to list, but some more visible companies that have
publicly declared usage of Kotlin, be this via blog posts, GitHub repositories or talks include
Square, Pinterest, Basecamp or Corda.
Who develops Kotlin?
Kotlin is primarily developed by a team of engineers at JetBrains (current team size is 50+). The
lead language designer is Andrey Breslav. In addition to the core team, there are also over 250
external contributors on GitHub.
Where can I learn more about Kotlin?
The best place to start is this website. From there you can download the compiler, try it online as
well as get access to resources, reference documentation and tutorials.
540
Are there any books on Kotlin?
There are already a number of books available for Kotlin, including Kotlin in Action which is by
Kotlin team members Dmitry Jemerov and Svetlana Isakova, Kotlin for Android Developers
targeted at Android developers.
Are there any online courses available for Kotlin?
There are a few courses available for Kotlin, including a Pluralsight Kotlin Course by Kevin Jones,
an O’Reilly Course by Hadi Hariri and an Udemy Kotlin Course by Peter Sommerho .
There are also many recordings of Kotlin talks available on YouTube and Vimeo.
Does Kotlin have a community?
Yes. Kotlin has a very vibrant community. Kotlin developers hang out on the Kotlin forums,
StackOver ow and more actively on the Kotlin Slack (with close to 20000 members as of October
2018).
Are there Kotlin events?
Yes. There are many User Groups and Meetups now focused exclusively around Kotlin. You can
nd a list on the web site. In addition there are community organised Kotlin Nights events
around the world.
Is there a Kotlin conference?
Yes. The o cial annual KotlinConf is hosted by JetBrains. It has taken place in San-Francisco in
2017 and in Amsterdam in 2018. Kotlin is also being covered in di erent conferences worldwide.
You can nd a list of upcoming talks on the web site.
Is Kotlin on Social Media?
Yes. The most active Kotlin account is on Twitter.
Any other online Kotlin resources?
The web site has a bunch of online resources, including Kotlin Digests by community members, a
newsletter, a podcast and more.
Where can I get an HD Kotlin logo?
Logos can be downloaded here. When using the logos, please follow simple rules in the
guidelines.pdf inside the archive and Kotlin brand usage guidelines.
541
Comparison to Java Programming Language
Some Java issues addressed in Kotlin

Kotlin xes a series of issues that Java su ers from:
— Null references are controlled by the type system.

— No raw types
— Arrays in Kotlin are invariant
— Kotlin has proper function types, as opposed to Java's SAM-conversions
— Use-site variance without wildcards
— Kotlin does not have checked exceptions
What Java has that Kotlin does not

— Checked exceptions
— Primitive types that are not classes
— Static members
— Non-private elds
— Wildcard-types
— Ternary-operator a ? b : c
What Kotlin has that Java does not

— Lambda expressions + Inline functions = performant custom control structures
— Extension functions
— Null-safety
— Smart casts
— String templates
— Properties
— Primary constructors
— First-class delegation
— Type inference for variable and property types
— Singletons
— Declaration-site variance & Type projections
— Range expressions
— Operator overloading
— Companion objects
— Data classes
542
— Separate interfaces for read-only and mutable collections
— Coroutines
543
544

Kotlin Docs

Uploaded by

Copyright:

Available Formats

Kotlin Docs

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

Kotlin Docs

Uploaded by

Copyright:

Available Formats

Kotlin Language Documentation

Using Kotlin for Server-side Development 7

Using Kotlin for Android Development 9

Kotlin JavaScript Overview 11

Kotlin/Native for Native 13

Coroutines for asynchronous programming and more 15

What's New in Kotlin 1.1 19

What's New in Kotlin 1.2 30

What's New in Kotlin 1.3 39

Control Flow: if, when, for, while 87

Returns and Jumps 91

Classes and Objects 94

Classes and Inheritance 94

Properties and Fields 102

Visibility Modifiers 109

Sealed Classes 120

Nested and Inner Classes 128

Enum Classes 129

Object Expressions and Declarations 131

Inline classes 136

Delegated Properties 143

Functions and Lambdas 150

Higher-Order Functions and Lambdas 157

Inline Functions 165

Kotlin Collections Overview 169

Constructing Collections 174

Ranges and Progressions 179

Collection Operations Overview 184

Collection Transformations 186

plus and minus Operators 193

Retrieving Collection Parts 195

Retrieving Single Elements 198

Collection Ordering 201

Collection Aggregate Operations 204

Collection Write Operations 206

Set Specific Operations 213

Map Specific Operations 214

Multiplatform Programming 218

Platform-Specific Declarations 218

Building Multiplatform Projects with Gradle 220

Destructuring Declarations 258

Type Checks and Casts: 'is' and 'as' 261

This Expression 265

Operator overloading 267

Null Safety 272

Scope Functions 289

Type-Safe Builders 297

Experimental API Markers 304

Keywords and Operators 309

Java Interop 334

Calling Java code from Kotlin 334

Calling Kotlin from Java 348

Dynamic Type 359

Calling JavaScript from Kotlin 361

JavaScript Reflection 371

JavaScript DCE 372

Concurrency in Kotlin/Native 374

Immutability in Kotlin/Native 378

Kotlin/Native libraries 379

Advanced topics 380