re: code, screenreader-unfriendly
Did I mention that I'm trying to write a bug-for-bug-compatible implementation
re: code, screenreader-unfriendly
@joepie91 RIR? or what are you aiming for?
re: code, screenreader-unfriendly
@eater Not Rust, weird performance experiment with V8
re: code, screenreader-unfriendly
@joepie91 oh god oh no, are you re implementing in JS?
re: code, screenreader-unfriendly
@eater That is really not the sort of comment I am looking for when working on a project
re: code, screenreader-unfriendly
@joepie91 ah sorry! I guess I should have used some tone indicators there, it's not meant as "don't do this", it just sounds delightfully cursed to me, and I highly encourage you, I am honestly very curious to what results you might end up with, V8's performance can be really good after all!
so yea, not meant as discouragement! sorry if it came over that way!
re: code, screenreader-unfriendly
@eater Ahh right, yeah, it definitely is cursed, but in the "this just might work but most people wouldn't bother to try" way
re: code, screenreader-unfriendly
@eater (Specifically I'm compiling Nix to JS and then evaluating that JS within V8, to see if its optimizer can make sense of it, since the semantics of the two are surprisingly close)
re: code, screenreader-unfriendly
@joepie91 oooh neat!! I wonder if you could make a little nix playground like this in-browser also,
re: code, screenreader-unfriendly
@eater That'd be one of the things I'm going for as well, yeah
re: code, screenreader-unfriendly
@joepie91 what are you using for parsing? just doing it by hand or using something alike antlr?
re: code, screenreader-unfriendly
@eater treesitter-nix, taped to some custom AST wrangling code
re: code, screenreader-unfriendly
@eater In case you're interested in following along: https://git.cryto.net/joepie91/nix-in-node
re: code, screenreader-unfriendly
@joepie91 oooh awesome, I'll keep an eye out
re: code, screenreader-unfriendly
Given that all __overrides usage seems to have been eliminated from nixpkgs in 2014, I am very tempted to just declare the entire feature deprecated in my implementation, this thing is broken in so many ways