one sticking point for me is the way mixing nested strings pads them out to the same length with spaces; and if it did the padding with nulls instead, it would be possible to split them back to their original lengths. Presumably having the padding for character vectors be space is useful for some situations?
there has been discussion previously on jelly's left-"binding" functions vs apl's right-"binding", what do you think of an apl (so right-"binding") written right-to-left?
@JeffZeitlin Not officially, since there hasn't been much demand for it. That said, you can start dyalog from bash on Windows, and it will run in that console. Be warned, that is is extremely unstable and comes with no guarantee or support whatsoever. E.g. it will quit on any error, so it is useful to start by entering ⎕TRAP←0 'E' '↑⎕DM'.
@Adám yeah, i was also thinking of having identifiers being in hebrew/arabic, maybe abjad names are more "apl-y", eg: see identifier names in ngn/k and arcfide's code
i found myself recently doing, ⌽ <left> ∘ <left> ⍉ <left> ∘ <left> ⌽, so i thought maybe writing apl would be easier if it went right-to-left
@KritixiLithos However, it keeps coming back to APL reading nicely from left to right. If APL executed left to right, or if one attempts to read APL from right to left (e.g. due to the code being in Hebrew/Arabic) then if becomes very "do-y" instead of "is-y"
@KritixiLithos Yeah, me too. My home key is worn smooth. I'd still really love to try my hands on a true, and thoroughly designed, left-to-right LPA.
@KritixiLithos I don't know about Arabic, but Israelis read numbers big-endiany, though the Bible sometimes uses bing-endian, and some-times little-endian spelling.
@KritixiLithos The interpreter would be the same. It is simply an input method, no?
@Adám hm i had a good reason why the interpreter would't be the same (something along multiline dfns?), but i forgot now, i have to sleep now, so i will reply tomorrow
wait maybe it is multiline dfns, well not necessarily multiline, but multi-statement
my idea was to completely leave the interpreter alone, just input text differently. perhaps jumping to the end on ⋄, who knows. these kind of problems are what came up