Shortened Transcript of Roy Sykes’ Closing Plenary Address at APL 86

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 89

Transcriber’s note: The full transcript of Roy’s speech was thirteen closely written pages (about six full pages of Vector). This abbreviated version omits the jokes and the politenesses to many contributors to the success of APL 86 and tries to do justice to Roy’s serious theme. Words which are verbatim are in quotations.

After a few introductory jokes Roy asked the serious question why APL is not dominant. “It has been estimated that APL accounts for between two and six percent of programs written in the world today.” He offered the following main reasons why we had not yet persuaded the rest of the world to follow us.

Inertia “Organisations have little reason to discard old yet operational software and are not prepared to retrain entrenched hordes of COBOL programmers. The twelve figure investment in computer software and standards [which] considerably exceeds the investment in computer hardware . . . ” is not on our side.

Cost “The memory required by APL has only recently become . . affordable . . . the speed differential between interpreted APL and compiled scalar languages remains. Furthermore APL implementations are still priced at a premium.”

Insularity APL took longer to gain links to computer peripherals, files and databases than the other languages. This has delayed our integration with the rest of computing. “APL programmers have ignored the garden in favour of a single rose bush.”

“There are hosts of other reasons for the orbit in which we find ourselves with our language. We may think we have charm but the rest of the world thinks we have strangeness. We excel in parallel but collide sequentially. The stable molecules we build from our periodic table are seen by others as unstable gluons. Overall APL and its advocates have a reputation for inhabiting an esoteric world of strange symbols, peculiar terms, dense programs and evangelical fervour.”

After congratulating the organisers (but not on the food – “lunch tickets anyone?”) Roy asked whether APL would ever come to fruition.

“I see the seas parting and the promised land beckoning . . . . Our technology has burst explosively into the mainstream of computing. In less than a decade we have seen the integration of APL with scores of other languages and packages.” APL can now use any equipment, read any file and its speed disadvantage is gradually being eroded. The new data structures, primitives and operators “expand the expressive power of APL explosively yet they retain the fundamental flavour of the language.”

Although Roy voted for the motion that APL enhancements do not help the spread of APL (which was carried) he voted against the proposition that APL was not yet suitable for AI (which was lost). “I wonder if the debate had been based on APL 1 (in Jim Brown’s parlance) would this motion have been defeated? Is VS APL sufficient for the task of AI? I think not.” So he revised his opinion and now believes that enhancements do help the spread of APL and implied they will do so increasingly.

“ . . . the most important technological explosion for APL is the most obvious. The advent of the personal computer and the extraordinary implementations of APL based on it . . . If APL\360 was the first revolution in APL then APL*PLUS/PC is the second.”

The biggest factor in the future of APL is education. “IBM has donated tens of thousands of PCs to universities. Thousands are now running APL . . . Fertile minds are being seeded with APL.”

“It seems to me that we have a more diverse and committed commercial sector devoted to APL than ever before. The UK now has an independent base of APL professionals approaching the size of that in the entire United States at the late 1970s peak of timesharing.”

What can we do now? Roy’s answers were; correct the old misunderstandings about APL; show it to people who should appreciate it; use it to write applications; think big – don’t sell just to programmers which is a small market compared with that of users (the largest used APL system is ADRS); use APL for prototypes (slogan: ‘Get it up fast with APL’). And he gave an example: he had produced an address book and telephone control system in about five days work, superior for his purposes to any of the packages on offer. “The fact is: APL can produce good products. Why don’t we produce them? Why don’t we market them? APL will get us there better and faster and the architecture of future machines favours the parallel structure of our language.”

We need to train APL programmers properly. Too much APL is done by amateurs and users who are not trained and their mistakes bring APL a bad reputation. There are still under a thousand professional APL programmers. Perhaps most of all we need to spread APL downwards in the educational world from universities to schools.

Roy’s closing message was that the great period of enhancements to APL is over. “The APL that we are now using is the one that we will be developing products on now and for the foreseeable future. Let us resolve to exploit APL: to expand its usage and application far beyond its current scope. Our livelihoods depend on it. Our dreams will be satisfied by nothing less. Remember APL is not a programming language. It’s a way of life.”