A 1978 talk by Professor Perlis at the APL’78 Conference argues APL’s design philosophy is precise and mathematical, not natural-language-shaped.
Key Takeaways
Perlis frames APL as structurally closer to French (terse, formal, rule-bound) than to English (natural, verbose, accumulative).
He explicitly rejects calls to add FORTRAN-style control flow (while, if-then-else, for) to APL, calling them incompatible with the array paradigm.
FORTRAN is characterized as the computing world’s lingua franca: universal, cross-platform, multi-manufacturer, present in books at every level in every language.
The talk predates J, K, and modern array languages but directly foreshadows the ongoing debate about whether array languages should absorb imperative constructs.
Hacker News Comment Review
Commenters treat this as a useful historical document for understanding array language philosophy, not just APL trivia – the core design tensions Perlis names are still live in 2024.
The main point of disagreement: @lokedhs pushes back on Perlis’s categorical dismissal of control flow additions, suggesting the argument hasn’t aged as cleanly as the rest of the talk.
The FORTRAN-as-lingua-franca passage landed with readers as unexpectedly prescient; Perlis called its durability correctly even as he was arguing against its design values.
Notable Comments
@lokedhs: disagrees specifically with Perlis calling APLGOL (APL + structured control flow) “ridiculous” – frames it as the talk’s weakest claim.
@w4yai: confirms this is a transcribed 1978 conference talk, not a written essay – important framing for reading Perlis’s rhetorical style.