The History of Formal Methods 2019 workshop ended with a discussion reflecting on the discipline ... more The History of Formal Methods 2019 workshop ended with a discussion reflecting on the discipline of formal methods. An initial prompting question, “What have formal methods ever done for us?”, was presented, but the discussion evolved from there into consideration of applicability, education, and even Star Trek. The session was chaired and curated by Troy Astarte, who has also prepared this summary of the discussion. It is not a perfect transcription but rather a report on the interesting and stimulating conversation that resulted.
Although there are thousands of programming languages, most of them share common features. This p... more Although there are thousands of programming languages, most of them share common features. This paper reviews some key underlying language concepts and the challenges they present to the task of formally describing language semantics. The responses to these challenges in operational, axiomatic and denotational approaches to semantic description are reviewed. There are interesting overlaps between these responses; similarities are exposed even where accidental notational conventions disguise them so that essential differences can be pinpointed. Depending on the objectives of writing a formal semantic description of a language, one or other approach might be considered the best choice. An argument is made for increasing the use of formal semantics in language design and here it is suggested that the operational approach is the most viable for a complete language description.
199) As computers became multi-component systems in the 1950s, handling the speed differentials e... more 199) As computers became multi-component systems in the 1950s, handling the speed differentials efficiently was identified as a major challenge. The desire for better understanding and control of ‘concurrency’ spread into hardware, software, and formalism. This work-in-progress talk traces some early attempts to find primitives for concurrency. Initially, system programs called ‘monitors’ were used for directly managing synchronisation. Attempts to reframe synchronisation at a higher level led to a series of algorithms; Dijkstra’s semaphores were a reaction to the algorithms’ complexity. Towards the end of the 1960s, new desires for clearer ‘structured programming’ created a milieu in which Dijkstra, Hoare, and Brinch Hansen (among others) aimed for a concurrency primitive which embodied the new view of programming. Via conditional critical regions and Dijkstra’s ‘secretaries’, the monitor appeared to provide the desired encapsulation. A few programming languages adopted the constru...
Although there are thousands of programming languages, many of them share common features. This p... more Although there are thousands of programming languages, many of them share common features. This paper reviews some key underlying language concepts and the challenges they present to the task of formal semantic description. Most material concerns the responses in operational, axiomatic and denotational approaches to the description of programming languages. There are interesting overlaps between the responses of these main approaches to the challenges posed by the language concepts: these similarities are exposed even where accidental details often disguise them; essential differences are also pinpointed.
In the 1960s, a full formal description was seen as a crucial and unavoidable part of creating a ... more In the 1960s, a full formal description was seen as a crucial and unavoidable part of creating a new programming language. A key part of that was a thorough and rigorous description of the semantics. However, in the decades since, the focus on providing this has somewhat diminished. Why was formal semantics once seen as so critical? Why did it not succeed in the ways hoped? These questions are explored by considering the history of modelbased approaches to describing programming languages, with a particular focus on the IBM Laboratory Vienna under Heinz Zemanek, and the Programming Research Group at Oxford University under Christopher Strachey. It is shown that there were a variety of different approaches to the problem, stemming from the different backgrounds of the people and groups involved. The story of formal language description is in some ways also the story of early programming languages and the integration of mathematics into the emerging new field of computer science, resu...
The programming language ALGOL 60 has been used to illustrate several different styles of formal ... more The programming language ALGOL 60 has been used to illustrate several different styles of formal semantic description. This paper identifies the main challenges in providing formal semantics for imperative programming languages and reviews the responses to these challenges in four relatively complete formal descriptions of ALGOL 60. The aim is to identify the key concepts rather than become distracted by the minutiae of notational conventions adopted by their authors. This paper also explores the historical reasons for the development of these descriptions in particular, and gives some general historical background of the groups involved (the IBM laboratories in Vienna and Hursley, and Oxford’s Programming Research Group).
The History of Formal Methods 2019 workshop ended with a discussion reflecting on the discipline ... more The History of Formal Methods 2019 workshop ended with a discussion reflecting on the discipline of formal methods. An initial prompting question, “What have formal methods ever done for us?”, was presented, but the discussion evolved from there into consideration of applicability, education, and even Star Trek. The session was chaired and curated by Troy Astarte, who has also prepared this summary of the discussion. It is not a perfect transcription but rather a report on the interesting and stimulating conversation that resulted.
Although there are thousands of programming languages, most of them share common features. This p... more Although there are thousands of programming languages, most of them share common features. This paper reviews some key underlying language concepts and the challenges they present to the task of formally describing language semantics. The responses to these challenges in operational, axiomatic and denotational approaches to semantic description are reviewed. There are interesting overlaps between these responses; similarities are exposed even where accidental notational conventions disguise them so that essential differences can be pinpointed. Depending on the objectives of writing a formal semantic description of a language, one or other approach might be considered the best choice. An argument is made for increasing the use of formal semantics in language design and here it is suggested that the operational approach is the most viable for a complete language description.
199) As computers became multi-component systems in the 1950s, handling the speed differentials e... more 199) As computers became multi-component systems in the 1950s, handling the speed differentials efficiently was identified as a major challenge. The desire for better understanding and control of ‘concurrency’ spread into hardware, software, and formalism. This work-in-progress talk traces some early attempts to find primitives for concurrency. Initially, system programs called ‘monitors’ were used for directly managing synchronisation. Attempts to reframe synchronisation at a higher level led to a series of algorithms; Dijkstra’s semaphores were a reaction to the algorithms’ complexity. Towards the end of the 1960s, new desires for clearer ‘structured programming’ created a milieu in which Dijkstra, Hoare, and Brinch Hansen (among others) aimed for a concurrency primitive which embodied the new view of programming. Via conditional critical regions and Dijkstra’s ‘secretaries’, the monitor appeared to provide the desired encapsulation. A few programming languages adopted the constru...
Although there are thousands of programming languages, many of them share common features. This p... more Although there are thousands of programming languages, many of them share common features. This paper reviews some key underlying language concepts and the challenges they present to the task of formal semantic description. Most material concerns the responses in operational, axiomatic and denotational approaches to the description of programming languages. There are interesting overlaps between the responses of these main approaches to the challenges posed by the language concepts: these similarities are exposed even where accidental details often disguise them; essential differences are also pinpointed.
In the 1960s, a full formal description was seen as a crucial and unavoidable part of creating a ... more In the 1960s, a full formal description was seen as a crucial and unavoidable part of creating a new programming language. A key part of that was a thorough and rigorous description of the semantics. However, in the decades since, the focus on providing this has somewhat diminished. Why was formal semantics once seen as so critical? Why did it not succeed in the ways hoped? These questions are explored by considering the history of modelbased approaches to describing programming languages, with a particular focus on the IBM Laboratory Vienna under Heinz Zemanek, and the Programming Research Group at Oxford University under Christopher Strachey. It is shown that there were a variety of different approaches to the problem, stemming from the different backgrounds of the people and groups involved. The story of formal language description is in some ways also the story of early programming languages and the integration of mathematics into the emerging new field of computer science, resu...
The programming language ALGOL 60 has been used to illustrate several different styles of formal ... more The programming language ALGOL 60 has been used to illustrate several different styles of formal semantic description. This paper identifies the main challenges in providing formal semantics for imperative programming languages and reviews the responses to these challenges in four relatively complete formal descriptions of ALGOL 60. The aim is to identify the key concepts rather than become distracted by the minutiae of notational conventions adopted by their authors. This paper also explores the historical reasons for the development of these descriptions in particular, and gives some general historical background of the groups involved (the IBM laboratories in Vienna and Hursley, and Oxford’s Programming Research Group).
Uploads
Papers by Troy Astarte