John C. G. Sturdy M.A., PhD | Cnoc na Gortíní |
Marital Status: single | Maigh Rua |
Clean UK full driving licence (Category C) | Contae Luimnigh |
Eire |
An experienced and versatile Computer Scientist with research background and extensive experience of large-scale commercial and research software development, good knowledge of various computer systems and languages, and experience of teaching small and large groups.
Cúntóir Taighde agus Dearthóir Cursaí, Roinn Riomheolais, Ollscoil Luimnigh1
Postdoctoral Research Fellow, B4STEP programme, CSIS Department, University of Limerick
Part-time Supervisor, Cambridge University Computer Laboratory
Senior Computer Programmer at the Wellcome Trust Sanger Institute
Software Engineer at EDS Unigraphics
Computer Consultant (Software Engineer) at Harlequin Ltd
Various vacation jobs in software engineering, mostly at Acornsoft on BBC Micro; Gap Year job included writing an editor and working on the kernel of a small concurrent Operating System
Research in, and development of, software development tools, taking care to make them clearly comprehensible to other engineers who may work on them in the future.
I am an experienced Lisp (Common Lisp and Emacs Lisp, with about 2.4Mb published as Open Source in the latter) and C programmer (with fair C++ experience) mostly on Unix (largely BSD) but also with a couple of years Windows-NT/MFC experience, and have very good PostScript skills. For the past few years, I have had a high proportion of my Lisp code work first time. I'm experienced in general Unix work (shell, Perl, make, sed etc). I have done some work in Assembly languages (various) and also worked on CPU microcode (various graphics hacks such as Mandelbrot calculations on the Orion-1). My Unix work has included a major multiprocessing project (a pioneering implementation of PostScript for very-high-resolution printing systems) in which I led the work on the parallelism in it.
I have done some BSD sysadmin work in the past (a normal side-effect of doing a PhD in that era!), and am now picking up on this again as needed.
In my PhD I constructed a reifying, reflective, mixed-language programming language interpreter providing access to the underlying abstractions of data and process representation, that could regress to an infinite number of infinite levels of abstraction and representation, the resulting interpreter nevertheless running around 40% of the speed of a plain commercial interpreter for the same languages. This system provided systematically the means to examine and extend a running software system dynamically, putting debuggers and modularly extensible systems on a more formal footing. (Reflective facilities are now beginning to appear commercially, and a relatively very primitive form of reflection is regarded as an advanced feature of Java.) The implementation language for this was Common Lisp, running under BSD Unix. The ideas behind this built partly on the work of Brian Cantwell Smith, and partly on an abstract machine for extensible languages, that I had designed and implemented for my undergraduate dissertation project.
At Harlequin, I researched and constructed an experimental dynamic software migration system for Common Lisp (for an EU Esprit project, Chameleon), capable of suspending a process on one system, packing all the data needed to represent it, transferring it over the network to another machine and continuing to run it where it left off on its previous host. While still on the Lisp projects, I also developed a link loader for loading externally compiled code (COFF objects from the C compiler, for example) into the running Lisp image.
After that joined Harlequin's Electronic Publishing division (and was there for most of my time with this employer), where I worked on many parts of their PostScript-compatible (``clone'') Raster Image Processor, the highlight of which was designing and implementing the conversion of our existing single-process system to a pipelined parallel system, implemented in C. I also wrote large amounts of PostScript embedded in the implementation. I designed and implemented the computer end of the low-level interface to the laser imagesetter hardware. Later parts of the parallelism work involved working on obscure timing-related problems. I later moved over to various aspects of their advanced colour system, learning about colour science to do this.
At Unigraphics, I worked mostly on a series of projects within a very large CAD/CAM system (about 20 years work by an average of 300 people), including an object-oriented database interface to the system.
At the Wellcome Trust Sanger Institute, I worked on the genomic database system AceDB (an object-oriented genomics database system), working mostly on a new database query language for it (in C), and also on the project's web site and its automatic build and test systems (which were mostly in Perl).
I am experienced in writing web pages, taking care to keep them straightforward and fast-loading, and to make them accessible to all, including those using text-only (or speech or braille) browsers. I did some web work professionally while working at the Sanger Institute.
Between 1990 and 2002 I also supervised (tutored) Computer Science students for all years of the Degree course, and for the post-graduate Diploma course, at Cambridge University (as an evening job), thus both keeping in touch with what is now being taught in computer science and also developing my skills in explaining ideas and in evaluating others' understanding of topics. I have supervised a wide range of topics, but mostly those related to programming languages and operating systems.
In my work in B4STEP, I work mainly on software development tools and the psychology underlying them. However, of necessity, I have also spent considerable time co-ordinating our work on the EU project COSPA (Consortium for Open Source in Public Administration), including managing the construction of a knowledge base system for it. The COSPA work also involved a considerable amount of programming (Java and Perl), unfortunately not leading to publishable research output. I also led our independent evaluation of Prof David Parnas' ``Tabular notation'' specifications. As well as these, I lectured the ``Operating Systems Overview'' module of the Computer Systems degree for three years.
The main part of my recent work on software tools resulted in a set of experimental extensions to the GNUemacs editor/working environment, using language-specific plugins to support not only navigation in terms of semantically significant units, but also automated high-level operations for work that programmers have so far done as highly stereotyped sequences of manual actions, such as converting code blocks to procedures (finding automatically what needs to be passed in as parameters, what types the parameters must be, etc)3.
In my gap year job, I worked in a group proving theoretically the kernel of a small concurrent operating system (CrestOS, now long-lost; written in BCPL with an 800-line assembler kernel). This gave me a solid foundation in real-world concurrent programming.
In my B.A. dissertation, I designed a low-level reflective programming language (with a view to it being implementable in hardware), and implemented it on an emulator. Also while an undergraduate I was regarded by computing service staff as being the first person to write readable programs in the local editor scripting language (including a toy FORTRAN subset compiler).
Having discovered Koza's work on Genetic Programming, I started an experimental Lisp-based system for looking at the interaction of communicable information between individuals and the evolved characteristics in the population (for example, whether predators in an ecosystem would evolve differently if hunting strategies were communicable), but did not take this to a publishable stage.
As a spare-time activity, I once developed a user-contributable streetmap system (using technology that would now probably be called ``Wiki'', but before that term was used); I am now re-awakening this as a public project on SourceForge.
Another of my spare-time projects is a multi-lingual vocabulary learning tool (again within the Emacs editor / environment) which displays translations of words as you type them.
At primary school, I invented the resistor, but suspected that other people would have thought of it too, and so did not develop this project beyond basic experiments.
When bored at secondary school, I designed my own microcoded CPU architecture, but never had the chance to implement it.
Also at school I was one of the small group who interfaced the school computer (an Apple II) to an ASR33 teletype, building our own hardware and writing the machine code software to drive it.
[Work] John C. G. Sturdy | Last modified: Sat Aug 11 18:24:13 IST 2007 |