The Genesis of ASL

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 36

Talk of something like ASL goes back at least three or four years. Jake Ansell had voiced the idea at various meetings of the APL Statistics User Group (which he has almost single-handedly nurtured for some considerable time). In early 1988 it began to be taken seriously and things started to happen.

The spark was a gathering at Warwick University to launch its APL2 service. This was to make APL2 on Warwick’s mainframe available over the academic computing network JANET to other universities. (It was beset with problems over translation tables, etc., and was not a success, but that is another story). Several statisticians were present, the idea of a functions library was briefly discussed and they agreed to meet again to consider it carefully.

This meeting was on 18th April 1988 at Warwick University, attended by Jake Ansell, Phil Dawid, Brian Newman, Andy Pole, Ewart Shaw, Norman Thomson and myself. Warwick’s APL2 service prompted the natural suggestion of disseminating the library electronically from a repository at Warwick, an idea which has not actually been discussed lately but remains an option.

It is interesting to look at some of the topics discussed at this April 1988 meeting.

Language. The inevitable question arose of whether to work in ISO standard, first generation APL with enhancements over the standard like diamond separators and replicate, or some form of second generation APL. We agreed to use ISO, with extensions that could be converted between dialects automatically. But we also agreed to allow direct definition and to accept functions in APL2. (Norman Thomson was as persuasive as ever!)

Style. A consistent style was recognised as being important. We agreed that "stylistic questions to be considered include naming of arguments and other variables, and how much internal documentation should be provided in the form of comments". A distinctive naming convention was recommended for global variables, and standard "structured programming" functions like IF and WHILE were suggested.

Graphics. The progress made at Warwick, partly driven by its CTI project, on a single consistent graphics interface was recognised at this early meeting. This is the first time that it was given the name of WAGS - the Warwick APL Graphics Standard.

From these headings it can be seen that some of the most persistent topics of discussion in ASL meetings were already established. We had already hit the problems of Language and Naming that have taken a great deal of work to resolve. However, the major preoccupation in the early days was funding. This was item 1 on the agenda in April 1988. Even though the two suggestions made at the time turned out to be non-starters, it is interesting to take a look at them.

(a) Approach the Computer Board/UGC, whose Computers In Teaching Initiative (CTI) had funded a project at Warwick into the use of APL in Statistics teaching.

(b) Obtain a Science and Engineering Research Council (SERC) grant, perhaps a Research Grant for a PhD student to develop software.

The CTI initiative at that time was inviting proposals for setting up centres for educational software information and exchange in different subject areas. In Mathematics and Statistics, such a centre was subsequently established jointly at the Universities of Birmingham and Glasgow. We would really only have been interested in an APL service, which would have been too narrow a formulation for CTI, so this approach was dropped. However, it was decided that it would be useful in any such proposal to have letters of support from academic statisticians, so this was done and my files include letters from John Copas, Phil Dawid and Alan Hawkes.

The idea of a PhD project also seemed highly unlikely to have enough academic challenge, so the whole functions library idea lapsed until it was revived again at a meeting in October….. which is where the story will continue in our next Newsletter!