Papers I Enjoyed at APL86

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 69

Dave Ziemann asked me to write a review of APL86 with the above title, so I shall not dwell on other noteworthy or notworthy aspects of the conference. The APL86 committee should be congratulated on organizing and staging a well-run conference and producing an excellent programme and proceedings. In spite of culinary calamities, the hospitality was warm, the ale cool, and the content fulfilling.

Of the fifteen or so sessions and presentations I attended, all had at least some redeeming features. For some, it was merely that the speaker was awake; for others, it was a lively and engaging presentation of cogent and thoughtful ideas. I was fortunate in that my associate, Kyle, also attended most of these sessions with me. Given our propensity to avail ourselves of ale long after the formal sessions had ended, we found we needed spirited presentations just to stay awake. We were complementary: presentation style kept Kyle awake; technical content kept me awake. Unfortunately, we were sometimes fast asleep.

In his opening plenary speech ‘Core APL’ – misbilled as ‘The Split in APL’ – Ken Iverson reiterated his point of view regarding further developments in the language. Never one to be ambivalent, Ken argued forcefully for his ideas of operators and unified syntax and semantics, as contained in ‘A Dictionary of the APL Language’. A relaxed and well-prepared presentation, it nonetheless disappointed many attendees who had come to hear a sage and balanced view of the differences between APL2 and Sharp APL.

Stream 2 on Monday was disappointing, mainly for lack of adequate speaker preparation. Michael Berry’s presentation added nothing to that already contained in the APL86 Conference Proceedings, although his foils were of high quality. Roland Pesch (who also coauthored the Berry paper) gave a somewhat more stimulating talk on the use of ‘nil’ to solve problems associated with empty arrays in Sharp APL. Even in the absence of its implementation, nil rationalizes the treatment of empty frames in Sharp APL somewhat more smoothly than prototypes do in APL2. Nonetheless, IBM’s more complete implementation has yet to be matched by Sharp, and it remains the best working solution.

The content of David Landaeta’s paper ‘a notation for manipulating arrays of operations’, was controversial, particularly regarding his ideas of hyper-operators, which he calls ‘elevators’. Unfortunately, he escalated his talk well beyond its 1.5-hour limit, and, having spent the first half summarizing APL2, had little time to present and clarify his central thesis.

The proponents carried the day in the debate entitled ‘APL enhancements do not help the spread of APL.’ The opponents (Steve Schiavo and John Myrna) failed to adequately address the central issue of the long-term impact of major enhancements, and their smokescreen was blown away by the cogent arguments of the proponents (Richard Nabavi and Valerie Lusmore). It was a sad, if surprising, result that reflected the debating skills of the advocates. The debate had its amusing moments. When the audience was invited to voice its views, one member said “I have a question . . .” to which the chairman replied, “That’s out of order; questions are not allowed.” The attendee, not to be silenced, responded, “Well then, may I reformulate it as a statement?” eliciting a roar from the audience.

Another presentation of merit was Rob Hodgkinson’s report on the notable extensions contained in Sharp APL/HP. Rob gave a provocative and engaging talk on the DEL hyper-operator and its use to construct arrays of APL procedures. His excellent preparation and fluid delivery belied little of his initial nervousness in this, his first conference presentation.

The ever-smooth Jim Brown seemed pervasive at APL86, participating in no less than four sessions dealing with his favourite APL and its applicability to logic programming, multi-user applications, and AI, too. Always well-prepared, Jim’s talks are an optimistic portent of APL’s future direction and usage in state-of-the-art computing.

Another group of talks, given by Clark Wiedmann, Wei-Mee Ching, and others, dealt with the exciting topic of compiling APL. Once considered a remote prospect, at least one compiler (STSC’s) is now commercially available and others are in various stages of development. These papers illustrated dramatic gains in efficiency, in many cases rivaling the performance of traditionally compiled languages such as Fortran.

All in all, APL86 exhibited a bright view of APL’s future. The former insularity of the APL community seems to be giving way to an aggressive pursuit of solutions to the problems holding back APL’s wider acceptance. We hope that Dallas and Sydney will continue this trend.