I am replying inline... On Sun, Jan 15, 2023 at 6:09 PM Raul Miller <rauldmil...@gmail.com> wrote: > > On Sun, Jan 15, 2023 at 3:38 PM Jose Mario Quintana > <jose.mario.quint...@gmail.com> wrote: > > I am aware that BQN has first-class functions. Is there any other array > > language that also has them? > > In this context, a mozilla page on "first class functions" is interesting: > > https://developer.mozilla.org/en-US/docs/Glossary/First-class_Function > > "A programming language is said to have First-class functions when > functions in that language are treated like any other variable. For > example, in such a language, a function can be passed as an argument > to other functions, can be returned by another function and can be > assigned as a value to a variable." > > --------- > > Here, BQN uses upper case vs. lower case to distinguish the role of a > function definition (roughly analogous to distinguishing between a > gerund and its verb form).
Right, I personally do not like too much that convention; in general, I use lowercase names for functions and uppercase names for nouns (which I seldom use). > > From a user perspective this is largely a syntactic issue (though, of > course, there's also implementation issues). The key word is "largely." Any gerund still is a noun; thus, gerunds are just representations of their associated functions. > > --------- > > The K/Q approach is also interesting here. (K is a family of languages > rather than a single language -- it's intentionally undocumented with > large differences between releases, and Q is (somewhat more stably) There is a fair amount of q documentation here, Reference card | kdb+ and q documentation - Kdb+ and q documentation (kx.com) <https://code.kx.com/q/ref/> > implemented in K.) Here, arrays are themselves "first class functions" > -- specifically indexing a list has the same syntax as calling a > function. There's no rank (all arrays are rank 1, though they may Right, since there are list of lists of ... ;so, that does seem to be a big issue (although one would have to write, for example, verbs for matrix multiplication and inverse unless a suitable library is available). > contain arrays analogous to J's boxing) and K's monadic verbs support > a multi-argument sort of valence analogous to APL's multi-dimensional > indexing A[1 2 3; 4 5 6;7 8 9] for example. And you might get into > different operators when applying "functions" to values (though, > because of the many versions, it's difficult to describe this > comprehensively). Lo and behold, q verbs are genuinely first-class, they can take and produce verbs and, apparently, is the recommended way to program functionally in q. The following is a toy example to illustrate the method by calculating a sum of squares, q){x y z}[(+/);{x xexp 2}] {x y z}[+/;{x xexp 2}] q)({x y z}[(+/);{x xexp 2}]) 1 2 3 4 / redundant parentheses for clarity 30f or one could use the built-in keyword ' instead of {x y z}. In j, of course, one can do this, +/@:*: +/@:*: (+/@:*:) 1 2 3 4 30 Yet, there is a subtle potentially critical difference, the q version {x y z} is dynamic, but the j version @: is not because @: is a conjunction (not surprisingly, tacit j8xx dynamic versions are possible when black cats are involved). > > --------- > > Anyways, ... I guess one of the things which makes J's gerunds be > considered as "not first class" is that they have a unique syntax > which must be used when we consider them as verb definitions -- more > specifically we can't use some of this syntax with actual verb > definitions. (Interestingly, in some contexts, this distinction > vanishes. So in those contexts we might consider verbs as first class > while thinking of adverbs and conjunctions as not first class.) > Some time ago we had a similar discussion, I mentioned that, in my opinion, the j9xx tacit tools were relatively weak and I proposed an exercise that was just beyond trivial, [Jchat] Was [Jprogramming] Tacit completeness (jsoftware.com) <http://www.jsoftware.com/pipermail/chat/2021-December/009017.html> Unfortunately, the j903 calculus add-on was unusable for that purpose. I wonder if some progress has been made. Perhaps, this thread should be moved to the Chat forum although I would like to see some j90x code here; but, I will let others decide. > Thanks, > > -- > Raul > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
