Conversation

Brendan Zabarauskas

@brendanzab

·

Sep 15, 2021

Mind blown today: Constraint generation + solving is *not* a requirement for implementing Hindley-Milner type checkers — metavariables + unification is enough!

5

4

32

Gergő Érdi @ ICFP

@gergo_erdi

·

Sep 16, 2021

I'm partial to compositional typechecking for HM:

github.com

hm-compo/Compositional.hs at master · gergoerdi/hm-compo

Compositional type checking for a Hindley-Milner type system - hm-compo/Compositional.hs at master · gergoerdi/hm-compo

1

2

Brendan Zabarauskas

@brendanzab

Replying to

@gergo_erdi

Oooh, interesting! Is this based on some research? I looked at the README but couldn't see anything mentioned.

8:07 AM · Sep 16, 2021·Twitter Web App

Gergő Érdi @ ICFP

@gergo_erdi

·

Sep 16, 2021

Replying to

@brendanzab

https://youtube.com/watch?v=bi9rl9DuoYM… https://gergo.erdi.hu/talks/2016-06-compty/CompTy.pdf… https://gergo.erdi.hu/blog/2010-10-23-the_case_for_compositional_type_checking/…

youtube.com

Compositional Type Checking

Speaker: Gergő ÉrdiEvent Page: Produced by Engineers.SGHelp us caption & translate this video!http://amara.org/v/Ol3w/

1

6

Brendan Zabarauskas

@brendanzab

·

Sep 16, 2021

Thanks! Yeah I found your website now! :)