> anything that’s been around for a long time has some serious warts This is very true. I mean, just look at the x86 processor ISA.

how do you write dsls in racket without the beautiful-racket package?

See https://docs.racket-lang.org/guide/languages.html

after skimming it I do not think is very good. I did not see any mention about lexers, parsers and other important components

Have you looked at this page also: https://beautifulracket.com/appendix/from-br-to-racket-base.html\|https://beautifulracket.com/appendix/from-br-to-racket-base.html

huh… has typed/racket’s type checking of match always been so lacklustre?

(: stuff ((U 1 2) -> Symbol)) (define (stuff x) (match x [1 'one])) Because this really shouldn’t typecheck

I challenge you to produce an input for which that returns something other than a symbol. :slightly_smiling_face:

If the question is “why doesn’t match (or match + typed racket) do pattern exhaustiveness checking” then that’s always been the case, and it’s very hard.

fair enough :wink:

but yeah I was asking about the exhaustiveness checking

@samth at the risk of opening a can of worms, why is it hard?