WebAssembly uses LEB128 for encoding lengths for vectors and so on. The idea is that you can use one byte most of the time, but it can store any number with more bytes. Might be a good solution to problems with low maximum lengths.
@dzaima I've been using the dzaima/BQN versions for a few days, but I don't want to push that until I have good TAO tests.
@dzaima I'm still wondering whether those should work on the entire right argument or split it into cells like Index Of, though.
@Marshall btw, i have this in my local dzref, and then executing )cs "…/dc.bqn"‿"raw"•EX"…/spec/dzref" in a REPL and immediately gives access to the compiler
(and i guess an •ERASE"init" outside for re-executing to be as easy)
@AviFS ⎕R and⎕S are PCRE based, but enhanced by Dyalog to include application of custom regular expressions-style transformations and arbitrary APL code (including preservation of information from call to call) on each match.
@RGS one of my first goals was to reduce the amount of string literals (as each costs 2 quotes), and that's pretty much required to do that (the precise number and use mostly was gotten from trial and error; also reminds me of this)
@Adám yeah but I was wondering if your 1⌽'><' used '><' for a particular reason, e.g. because you have seen 1⌽'><' in a specific domain; and the only thing that occurred to me that has strings enclosed in <> is html
@Marshall almost implemented it concatenating the bytes in reverse. i feel like it should be more efficient, but tests are inconclusive. ..might be thinking about it too much