David Barron: a life in software, 1935–2012

Authors


Professor David Barron died on 2nd January 2012. All of us associated with Software: Practice and Experience remember him as the person who got the journal started in 1971. He joined with C.A. Lang to persuade Wiley to publish the journal, and he was the person who lined up highly regarded researchers to submit articles. Together with Lang, he was the co-editor of the journal from its first issue in 1971 until the end of 1984. After 1984, he continued as a member of the editorial board for many more years. This memoir of David Barron has been authored by Martin Campbell-Kelly, who says that he shared a mutual interest with David in the history of programming that extended over 30 years. We thank Martin for providing this tribute to David Barron.

R. Nigel Horspool and Andy Wellings

Editors, Software: Practice and Experience

When the first computer science degrees were offered in the late 1960s, there was an almost complete absence of programming and software textbooks. David Barron was one of the first authors in the field, and his books were grasped by thankful instructors everywhere. But whereas the names of most other textbook authors have faded with the years, that of Barron is etched indelibly in the mind of everyone who used his books. They were concise and witty and had that most elusive quality—class. In the discipline of computer science, noted for its prosaic subject matter, Barron had a gift for elevating the pedestrian to the positively interesting. His books were practically literature.

David William Barron was born on 9 January 1935 in Blackburn, in the north of England. He was the first son and the youngest of three children of G.W. Barron and his wife Mary. His parents were schoolteachers, and he grew up in a secure, lower-middle-class, suburban, semi-detached home. He was an early beneficiary of the 1944 Education Act, winning at the age of 11 a scholarship to Queen Elizabeth Grammar School, Blackburn. As a student, he was a good all-rounder. He became captain of the fencing team. In the sixth form, he specialized in mathematics and the sciences but gave vent to his literary side by editing the school magazine—an early indication of a ‘two cultures’ outlook that was to become a hallmark of his academic trajectory.

For a sixth-form book prize, Barron chose the first semi-popular account of computing published in Britain, Vivian Bowden's Faster than Thought (1953) [1]. Between its covers, he would have made his first sightings of the Electronic Delay Storage Automatic Calculator (EDSAC), Maurice Wilkes, and Christopher Strachey—all to have a major impact on his life. In the early 1950s, the importance of computers was not as obvious as it sometimes seems in retrospect, however; the computer revolution had barely begun, and Faster than Thought did not deflect him from a decision to study physics at university. He won a scholarship to Downing College, Cambridge University, and graduated with first class honors in the physics tripos in 1956. He then became a research student in the Cavendish Laboratory, joining the radio-physics research group led by J.A. Ratcliff and being supervised by Henry Rishbeth. Barron's topic was the propagation of radio waves in the ionosphere, and his research involved the numerical solution of some very awkward differential equations. It was the need for computational relief that took him to the Mathematical Laboratory and the EDSAC.

biography image

David Barron as a young man.

1 THE MATHEMATICAL LABORATORY

Barron found an extraordinarily convivial atmosphere in the laboratory. He was allocated a study place by Wilkes and made to feel welcome. Barron and Wilkes in fact had something in common—they had both undertaken their PhDs in the Cavendish Laboratory's radio-physics research group; although Wilkes had carried out his work before the war, he retained an active interest. Barron had a shy temperament and was initially somewhat in awe of the laboratory staff, particularly David Wheeler, who was his senior by about 10 years. Wheeler had created the EDSAC programming system in 1949; it was one of the most brilliant programming accomplishments of all time, and Barron knew (like everyone else) that it was a bravura performance he could never hope to match. Another staff member he came to know well was Stanley Gill, who had also performed fine work on the EDSAC, writing subroutines for the numerical integration of differential equations, using what came to be known as the Runge–Kutta–Gill method. Whereas Wheeler was to spend all of his academic life in the laboratory, Gill was more urbane and career minded and left the laboratory soon after Barron's arrival. (Gill became a major force on the British computer scene but sadly died in 1975 at the height of his influence.)

Barron discovered that, even using the EDSAC and Gill's subroutines, each step in a numerical integration took 20 min, and results had to be coaxed from the machine in long, nighttime sessions. During these sessions, he came to know the other laboratory staff well. His scientific interest was rapidly turning towards computing rather than radio-physics, but he plodded on determinedly and completed his dissertation. The research culminated in a paper coauthored with his supervisor and published in the Journal of Astronomical and Terrestrial Physics in 1960 [2]. The paper turned out to be something of a classic and was widely cited, although as Rishbeth once remarked, it did his career more good than Barron's, once he had decided to leave the field.

One way or another, Barron had become a fixture in the laboratory. After his research studentship expired, Wilkes found him a research fellowship with the remit to ‘make himself useful’ and to assist Wheeler in commissioning the EDSAC's successor. In 1961, he was appointed to a university lectureship, and at the same time, he became a fellow of Downing College.

2 EDSAC 2, TITAN, AND CPL

The EDSAC 2 was the world's first computer to be controlled by a microprogram, an invention of Wilkes. The microcode was written by Wheeler with the assistance of Barron. The microcode was stored in a read-only core memory of an unusual construction. The code was hardwired by individually threading cores and then bonding the memory permanently with epoxy resin. There was no second chance to get the code right. Barron discovered, all too late, that one of his routines contained an error. Barron and Wheeler—pragmatists both—‘corrected’ the problem by adjusting the user manual, which now contained a footnote to the effect ‘after executing this instruction, 9 should be added to the instruction counter’. Barron's other contributions to EDSAC 2 included collaborations with Peter Swinnerton-Dyer on the use of magnetic tapes for numerical computation and with David Hartley on debugging procedures [3, 4]. He also repeated the calculations undertaken for his dissertation prior to submitting his paper with Rishbeth; work that had taken a year on EDSAC took a week on EDSAC 2.

After EDSAC 2 came into service, in 1958, thoughts soon turned to a successor. Although the demand for computing in the university had increased massively in the 10 years that the laboratory had been providing a computer service, funding had not kept pace. The laboratory would have liked a Ferranti Atlas computer (costing £2m plus) like Manchester University, but this could not be afforded. However, Peter Hall, head of Ferranti's computer division, arranged a deal whereby Ferranti would supply a cutdown version of the Atlas (notably without virtual memory) at cost price provided the laboratory collaborated with Ferranti to modify the hardware and develop the systems software. The machine was named Titan.

Writing the software for Titan was clearly an ambitious task, although the difficulty of writing large software systems had yet to emerge as a universal experience. Wilkes put Barron in charge of developing the software, placing the people and resources of the laboratory at his disposal. The project fell into two parts—an operating system and a programming language. It was a heady responsibility, and Barron probably never fully realized just how far his star had risen to be given such a responsibility when he had not yet turned 30. Interestingly, the South African author J.M. Coetzee, who was briefly employed by ICT (the successor to Ferranti) as a programmer on the project, recorded his experiences in his autobiographical novel Youth. This gives a rare glimpse into the Cambridge milieu, as seen from below. Coetzee felt he was an insignificant presence in the company of such august academics:

For brilliant they are indeed. Sometimes he shakes his head in disbelief at what is happening. Here he is, an undistinguished graduate from a second-class university in the colonies, being permitted to address by first name men who, once they get talking, leave him dizzied in their wake. Problems over which he has dully wrestled for weeks are solved by them in a flash. More often than not, behind what he had thought were problems they see what are the real problems, which they pretend for his sake he has seen too [5].

Developing a multiprogram, batch-processing operating system was in itself no small challenge. However, the plans were thrown into disarray in autumn 1963 when Wilkes returned from one of his periodic trips to the Massachusetts Institute of Technology with a gleam in his eye. He had seen MIT's Project MAC: timesharing was the new religion, and that was what the laboratory was going to do. It was to Barron's great credit that he was able to regroup and produce a timesharing system that eventually provided a university-wide computing service.

Titan was developed in an era before the use of standard programming languages had become universally accepted. A new computer usually betokened a new programming language, and Barron, Hartley, and Wheeler began to explore the possibilities in 1961 [6]. In the summer of 1962, Wilkes invited Christopher Strachey to join the laboratory as a research fellow. Jointly, they wrote a critique of Algol 60 and decided the laboratory could do better [7]. Strachey organized a study group consisting of himself, Barron, and Hartley to design a new programming language CPL—at that time standing for Cambridge Programming Language. It happened that at the Institute of Computer Science (ICS), London University, John Buxton and Eric Nixon were also intending to develop a programming language for a soon-to-be-acquired Atlas computer. They decided to join the Cambridge study group, and CPL was now understood to mean the Combined Programming Language.

Strachey was the guiding light of the group, and the others followed in his brilliant wake. Meetings were held regularly, sometimes in Cambridge, sometimes at the ICS, and sometimes in Strachey's family home in Bedford Square, Bloomsbury. Progress was rapid, and Barron, as the group's amanuensis, wrote a paper for the Computer Journal ‘The Main Features of CPL’ [8]. Barron and Strachey collaborated well, sometimes working weekends in the Strachey family retreat in Haslemere, Sussex. In summer 1963, they jointly gave a series of nine lectures at the Summer School in Non-numerical Computation organized by Leslie Fox at Oxford University. When written up by Barron, the lectures were effectively a programming primer for CPL [9].

Not uncommonly at the time, CPL's specification ran ahead of the realities of implementation. The result was an elegant, powerful, multipurpose language that was beyond the laboratory's resources to implement. The possibility of an implementation took a further setback after Barron ceased direct participation in the project because he had to devote himself full time to the operating software for Titan. He continued to attend the monthly meetings, but mainly in his administrative role as director of the Titan software effort. Although a CPL implementation was never produced at Cambridge, the language itself was highly influential*. As is well known, CPL inspired Martin Richards to develop a systems implementation subset, BCPL; this in turn inspired Ken Thompson to develop B as an implementation language for Unix; and from that effort emerged C—the most important programming language in the post-Algol, post-Fortran world [6].

3 TEXTBOOKS

Although Barron lacked the programming star quality of his mentors Wheeler and Strachey, he had his own brilliance as a communicator of programming ideas. He was exceptionally fluent: he could lecture extempore, and he had become author of first resort for programming manuals and documentation in the laboratory. Textbook writing was a natural next move.

In the late 1960s, undergraduate programs in computer science were springing up around the world, generally along the lines of the Association for Computing Machinery's Curriculum 68 [13]. A problem faced everywhere was the lack of suitable textbooks—indeed, the Curriculum 68 bibliography consisted largely of articles rather than textbooks. Barron's books thus helped fill a massive void. Beyond Cambridge University (and later Southampton University), he was known foremost as the author of elegant, witty, and economically written textbooks. His first two books were early entries in the famous Macdonald Monograph series—whose dust jackets portrayed a length of seven-track paper tape of the era [13, 14]. The first was Recursive Techniques in Programming (1968). It was highly accessible and was aimed at programmers rather than mathematicians. Barron recalled that although the book went out of print sometime in the 1980s, he continued to receive permission fees for reproducing extracts for many years after. His second volume in the series Assemblers and Loaders (1969) addressed a mundane although essential topic; he had a gift for making such unpromising material if not riveting, then certainly readable.

Barron was probably the first computer science author to make use of epigrams—witty quotations at the head of each chapter. They were a delight to savor, and to read them is to imagine Barron sitting in his armchair, a tumbler of whisky to hand, leafing through a dictionary of quotations in search of something to pique the reader. He would sometimes use an epigram as an oblique comment on the material he was obliged to communicate. For example, in Computer Operating Systems (1971), in a chapter on job control languages, it was no doubt as a riposte to the syntactically rebarbative OS/360 Job Control Language that he quoted the music hall artist George Robey: ‘Let us temper our hilarity with a modicum of reserve’ [15]. In his Study of Programming Languages (1977), he opened a chapter on suggestions for further reading with ‘Guide me, O thou great Redeemer’ [16]. It was enough to make even a jaundiced undergraduate smile.

4 SOUTHAMPTON UNIVERSITY

In 1967, at the age of 32, Barron made the surprising decision to become the founding professor of computer science at Southampton University. It was a fork in the road of his professional and private life.

Barron could have stayed at Cambridge University with some confidence that he would enjoy a distinguished future. As a researcher, he had been able to keep up with Wilkes, Wheeler, Gill, and the other denizens of the Mathematical Laboratory. Even if he was not in the absolute front rank of research, he had compensating qualities. He was a witty and sociable colleague; he was easily the best writer and communicator in the department, a brilliant lecturer, and a capable administrator. Coming from the background he did, at the time he did, a scholarly life at Cambridge University would have been an achievement of which to be proud. However, after a decade in the laboratory, Barron was beginning to feel somewhat smothered by the close-knit atmosphere and was ready for a change. Even so, he made the move with some misgivings that he would reflect upon in later years. A much easier decision for him was to leave behind his bachelor existence and marry Val Webber, chief computer operator in the laboratory. They went on to have two children.

Barron, celebrated as the youngest professor of computer science in the country, was initially situated within the mathematics department at Southampton University. This was quite common in the traditional universities; it tended to be only in the newly founded universities, with fewer vested interests, that computer science departments were set up ab initio. For those in Barron's situation, it could take years to establish an independent department and escape the gravitational pull of mathematics. Also, like others in his situation, Barron was saddled with multiple responsibilities. He was simultaneously head of a computer services facility and a small staff, and he was required to teach computing across the university—which he loved to do, especially when he was able to reach beyond science and engineering. In time, he would be expected to develop undergraduate and postgraduate courses in computing. And there was the matter of research.

Barron's greatest personal influences were Wheeler and Strachey, and he was not being falsely modest when he said he was not truly in their league. He was an admirer of broad scholarship, and his heroes included Don Knuth and the ‘Unix Three’, Ken Thompson, Dennis Ritchie, and Brian Kernighan. He possibly never had the temperament, introspection, or dedication to match their output, but in any case, life at Southampton University was sufficiently distracting, often pleasantly so, that he never had to fully confront his scholarly demons.

Finding a research direction and funding was a challenge for all newly formed computing departments at this time. Apart from providing a basic computing service, Southampton University lacked any computing research tradition or direction. Having spent a decade in one of the world's leading computer laboratories at Cambridge University, Barron recognized that the university needed more than benign good wishes if computing was to prosper. In 1971, he made the point in his inaugural lecture. As was his custom, although unprecedented for an inaugural lecture, he spoke extempore. He candidly summarized the state of computing research at the university:

There is a chicken-and-egg phenomenon here, since without equipment you cannot build up the reputation that justifies equipment on the expensive scale required. The effect is that in Computer Science research there is a de-facto league system, with Southampton at present probably somewhere in the second or third division. One of the most disturbing aspects of this situation is that there is no established mechanism available for promotion (or relegation!) [17]

This was a thinly veiled demand to the powers that if the university wanted to be a force in computing, then it needed to nurture the subject, not least financially.

5 SOFTWARE PRACTICE AND EXPERIENCE

Because of his relative isolation and the lack of an infrastructure, Barron's research horizons were limited. At Cambridge, research had been focused on building computers, and this was the glue that bound researchers together. But by the 1970s, those days were over—computers were now bought not made. Research was becoming more specialized, academic, and fragmented. Barron did not have much enthusiasm for this smaller canvas.

His primary research interest at the end of the 1960s had been large software systems. This was the era of the software crisis and the aftermath of the NATO Software Engineering conference held in Garmisch in 1968. If writing big software systems was now beyond Barron's reach at Southampton, he could at least become an advocate for doing big software properly. He was not alone in thinking that progress in software depended on a marriage of theoretical and practical viewpoints. He liked to quote Strachey:

It has long been my personal view that the separation of practical and theoretical work is artificial and injurious. Much of the practical work done in computing, in both software and hardware design, is unsound and clumsy because the people who do it have not any clear understanding of the fundamental design principles of their work. Most of the abstract mathematical and theoretical work is sterile because it has no point of contact with real computing [18].

But whereas Strachey leaned to the theoretical, Barron tended to the pragmatic. He did not believe that formal methods were the answer to the software crisis, and in any case, there was no shortage of theoretical computer scientists. But he did think that pragmatics were underappreciated. As he explained in his inaugural lecture,

There is an analogy with the study of the flow of water. At one extreme you have the theoretical fluid dynamicist, in the middle you have the hydraulic engineer, and at the other extreme is the plumber. (I like to think of myself as the computing equivalent of the hydraulic engineer, though some of my friends would say that what I do is more akin to plumbing.) [17]

He found a kindred spirit in a former colleague at Cambridge University, Charles Lang (a pioneer in computer-aided design and later a Cambridge entrepreneur). Lang had managed to interest Wiley in publishing a quarterly journal on software pragmatics, and he invited Barron to become joint editor in chief. After a difficult year soliciting papers—and triumphantly securing contributions from Don Knuth and Niklaus Wirth—the first issue appeared in 1971. As a literary enterprise, Software Practice and Experience bore Barron's stamp:

We were determined to put across the image of a quality product, hence the decision to eschew double-column layout, setting in a single measure across the page with generous margins on good quality paper, and choice of an eye-catching simple red cover. From time to time there have been moves to change things, but the production staff at Wiley will not readily forget the editorial ire which greeted their attempts to economize by using thinner paper, nor the scorn which their ideas of ‘jazzing up’ the cover were received [19].

Barron was in the editorial chair until 1984 and the journal—now with 12 issues a year—is in its 40th year of publication.

6 PASCAL

Barron was strongly guided by personal intuitions and a view of computing as a literary endeavor as well as a technical enterprise. He viewed programs as texts and believed that the expressiveness of languages was more important than their efficiency, particularly in an educational context. At Southampton University, this showed most strongly in his advocacy of Pascal.

In 1975, he dragooned the computer center (now under its own director) into providing a Pascal service and eventually into developing a compiler for a newly acquired ICL 2970 computer. The university hosted the First International Symposium on Pascal in March 1977, at which he gave the welcome speech. He subsequently edited a contributed volume Pascal—The Language and Its Implementation (1981) and coauthored a Pascal-based textbook on data structures and algorithms Advanced Programming (1984), with his former research student and protégé Judith Bishop (who independently wrote a series of Pascal textbooks) [20, 21].

7 AN ATYPICAL PROFESSOR

One of Barron's characteristics was a playfulness and gentle combativeness that never left him. Beyond the university, he was a member of the Science Research Council and was sometimes in conflict with it. In 1975, the council decided to close the Atlas Research Laboratory at Chilton, in Barron's opinion one of the most important autonomous computing research laboratories anywhere. He was driven to go public and excoriate the council for its decision. In an article, ‘Atlas Computer Laboratory 1961–1975 R.I.P’, he took a typical swipe at the bureaucrats:

An account of the manner of the Atlas Laboratory's going may be of interest to readers interested in the politics of science and government, and to devotees of the novels of C. P. Snow [22].

Quite possibly, the literary allusion was lost on the bureaucrats. At all events, he thought it probably cost him his membership of the council, but to him it was a price worth paying.

By the 1980s, computing at Southampton had escaped mathematics and become the Department of Computer Studies (not computer science—a telling choice of nomenclature). In the mid-1980s, a merger was mooted with the department of electronics. Although mergers of computing and electrical engineering were common in the USA, this was unusual in the UK. Except for Barron, his department was universally hostile to the proposal, but he was convinced it was the way forward—it was a case of ‘trust me’, and the merger went ahead. Although he disliked administration generally, he rather enjoyed pulling the bureaucratic strings for his own agenda—he served as chairman for 4 years. He fought for a new building and named it the Mountbatten Building—for another of his heroes. The Department of Electronics and Computer Science has since become one of the UK's outstanding schools of computing.

By the mid-1990s, as the end of his career came into view, Barron settled into a lower gear. He became a member of the human–computer interaction research group led by Wendy Hall, another of his protégés. He had always been interested in typography and fonts—he fondly recalled seeing a hot-metal Linotype in action as a child, at the Lancashire Evening Telegraph (his father had a friend on the staff). He was coauthor, with Mike Rees, of Text Processing with Unix (1987) [23]. Another long-term friend and kindred spirit—David Brailsford at the University of Nottingham—had become an authority on computer-mediated documents and a founding editor of the journal Electronic Publishing—Barron served as its book reviews editor. As electronic publishing and ‘document engineering’ came to the fore, he became an advocate for the SGML. Ever receptive to new programming paradigms, he wrote his final book The World of Scripting Languages (2000) [24].

Without any departmental responsibilities, he was freer than ever to act as a thorn in the side of bureaucracy, whether inside or outside the university. One of his most memorable confrontations occurred in 1999 when the Quality Assurance Agency for Higher Education (QAA) named and shamed the degree quality of some universities, including Southampton. His letter to the editor of The Higher, after ridiculing the very idea of such an agency, concluded:

Call me old-fashioned, but if students want to come, and employers want to employ them, then that's quality in my book. And if the QAA doesn't agree, then sod the QAA [25].

It was a response that perfectly captured his defense of a department and a subject he loved against the pointless bureaucracies that sought to regulate them. He retired from Southampton University the following year.

David Barron died of pneumonia on 2 January 2012 in Southampton. He is survived by Val and their children, Nick and Jacky. A memorial seminar was held at Southampton University on 13 September 2012.

biography image

David Barron at the memorial seminar for Maurice Wilkes, June 2011. As always, showing his penchant for jazzy ties. (Picture courtesy of Simon Moore.)

  • *

    A restricted version of CPL was produced in 1964 by GeorgeColouris at the ICS using the Brooker–Morris Compiler Compiler.

Ancillary