Chapter 1 Insight, inspiration and collaboration C. B. Jones, A. W. Roscoe Abstract Tony Hoare's many contributions to computing science are marked by insight that was grounded in practical programming. Many of his papers have had a profound impact on the evolution of our field; they have moreover provided a source of inspiration to several generations of researchers. We examine the development of his work through a review of the development of some of his most influential pieces of work such as Hoare logic, CSP and Unifying Theories. 1.1 Introduction To many who know Tony Hoare only through his publications, they must often look like polished gems that come from a mind that rarely makes false steps, nor even perhaps has to work at their creation. As so often, this impres- sion is a further compliment to someone who actually adds to very hard work and many discarded attempts the final polish that makes complex ideas rel- atively easy for the reader to comprehend. As indicated on page xi of [HJ89], his ideas typically go through many revisions. The two authors of the current paper each had the honour of Tony Hoare supervising their doctoral studies in Oxford. They know at first hand his kind and generous style and will count it as an achievement if this paper can convey something of the working style of someone big enough to eschew competition and point scoring. Indeed it will be apparent from the following sections how often, having started some new way of thinking or exciting ideas, he happily leaves their exploration and development to others. We have both benefited personally from this. Tony retired from Oxford in 1999 and has had, as we write this, 10 ex- tremely active years at Microsoft, improving that company's software devel- opment techniques, engaging enthusiastically in the debates of the computer 1 2 C. B. Jones, A. W. Roscoe science and software engineering world, promoting Grand Challenges such as the Verifying Compiler, and taking renewed interests in programming logic thanks to topics such as Separation Logic. We, however, have restricted our- selves to studying his work up to 1999 on the grounds that 10 years is already too short a time to understand the impact of academic work. In writing about the various phases and topics of Tony's career we have tried to analyse the influences and developing themes that have run through it. 1.2 Education and early career Charles Antony Richard Hoare, the eldest of five children (he has two brothers and two sisters), was born of British parents on 11 January 1934 in Colombo in what was then called Ceylon (now Sri Lanka). Ceylon was at that time part of the British Empire, and his father and maternal grandfather were both Englishmen engaged in the business of Empire, and from somewhat upper-class backgrounds.1 After his family returned to England at the end of World War II, Tony attended the Dragon School, Oxford and King's School, Canterbury before going to Oxford University to study Greats (formally known as Literae Hu- maniores) at Merton College between 1952 and 1956. Greats is Oxford's clas- sics course, in which students study Latin and Greek for the first two years, and concentrate on philosophy, literature and ancient history for the final two. Tony specialised in modern philosophy, being taught by John Lucas, an expert on logic and G¨odelin particular, who was then a Junior Research Fellow2 at Merton. By the time the authors studied at Oxford from the mid 1970s to early 1980s, Greats had gained the reputation of being one of the best courses to do at Oxford if you wanted to become a computer programmer. So perhaps the training Greats offered in systematic thinking, particularly given Lucas' influence, was in fact the ideal education for an early computer scientist. There was no undergraduate course in computer science at Oxford until about a decade after Hoare returned as a professor. In 1956 Tony was called up into the Royal Navy to do his \National Ser- vice", two years' military service that was compulsory for young men in the UK until the early 1960s. Perhaps thanks to his linguistic background, he went on a course on the Russian language while in the Navy. Tony returned for a further year at Oxford after completing his National Service degree, studying Statistics. During that year he took a course in programming (Mercury Autocode) from Leslie Fox, the founding Director of 1 See thepeerage.com, for example. 2 This is a a type of position given by Oxford Colleges to allow leading young academics to pursue their research. 1 Insight, inspiration and collaboration 3 Oxford University Computing Laboratory, about two years after the Lab- oratory was founded. Fox, one of the great figures of Numerical Analysis, remained Director until he retired in 1983, at which point Hoare took over this role. He then went to Moscow State University as a graduate student and stud- ied Machine Translation, along with probability in the school of the great Russian Mathematician Andrey Kolmogorov. Tony states that it was there, in the context of dictionary processing, that he invented Quicksort while unaware of any sorting algorithms other than bubblesort, which he had re- discovered and decided was too slow. At the same time he began translating Russian literature on computer science into English. On his return to England he joined the small British computer company Elliott Brothers, by whom he had been recruited while still in Moscow. One of the first tasks he was given there was to implement Shellsort in Elliott 803 Autocode. He remarks in [Hoa81b] that he then bet his manager that he had an algorithm that would usually run faster. He remarks how difficult Quicksort was to explain in the language of the time; but he won his sixpenny ($0.025) bet. He famously led the team that wrote one of the first ALGOL 60 compilers, for the Elliott 503 (the curiously numbered successor to the 803), a computer with 8K of 39-bit words and which was advertised as being able to run \as many as 200 programs per day"3. By the time this compiler was released in 1963 Tony had married (in 1962) Jill Pym, a member of his team. The ALGOL compiler was \one pass": in other words it only required a single pass through the source code tree of the object program. There is no doubt that Tony's work on ALGOL helped to define his under- standing of the nature of programming. Indeed, in [Hoa81b], he writes \It was there [an ALGOL 60 course in Brighton by Naur, Dijkstra and Landin which inspired him to choose this language for Elliott] that I first learned about recursive procedures and saw how to program the sorting method which I had earlier found such difficulty in explaining." In [Hoa81b] Tony goes on to explain how his understanding of program- ming and the need for clear semantics of programming languages developed as the result of the failure to deliver an operating system for the Elliott 503 Mark II, and how this, in particular, inspired his work on concurrency: I did not see why the design and implementation of an operating system should be so much more difficult than that of a compiler. This is the reason why I have devoted my later research to problems of parallel programming and language constructs which would assist in clear structuring of operating systems{constructs such as monitors and communicating processes. Tony reached the position of Chief Engineer at Elliott Brothers, but de- cided to leave because of the effects of the company being taken over in 1968. His academic career therefore began after he saw an advertisement for the 3 According to [27] when the compiler was run on the much slower 803 a typical half-page ALGOL program would take half an hour to compile and execute. 4 C. B. Jones, A. W. Roscoe position of Professor at the Queen's University, Belfast. By this time his po- sition in the developing subject of computer science was secure thanks not only to Quicksort but, perhaps more importantly, to the collaborations and contacts he obtained through his ALGOL work and the work he was doing on the ALGOL Working Group (IFIP WG2.1). By the time Tony was recruited to Oxford in 1977 no application was necessary: he was simply contacted and told that he had been elected to the job. 1.3 Programming languages Hoare's most explicit set of positive rules for designers of programming lan- guages was titled \Hints on Programming Language Design". This was origi- nally written for the first (ACM SIGPLAN) POPL conference held in Boston in October 1973. Sadly, the paper did not appear in the proceedings but has been reprinted several times in slightly different forms | probably the most accessible electronic version is [Hoa73a]. Rather than repeat the points in this important paper, can we encourage our readers to study it? This plea is most strongly addressed to anyone who is thinking of designing a new language. The importance that Tony Hoare attaches to programming languages is made abundantly clear in his acceptance speech for the ACM Turing Award.4 This 1980 speech is published as [Hoa81b]. As mentioned in Section 1.2, he remarks there how he could only express Quicksort elegantly after he had seen ALGOL. Hoare also makes clear that he sees it \as the highest goal of programming language design to enable good ideas to be elegantly expressed". Later in the same paper he observes the importance of \programming notations so as to maximise the number of errors which cannot be made, or if made, can be reliably detected at compile time." ALGOL 60 had been devised by a committee; but a committee of the highest calibre.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages34 Page
-
File Size-