Andy Matuschak on Twitter: "@jckarter This makes sense! As I understand: that wouldn't *preclude* future support of non-one-to-one relationships, right?" / Twitter

Can someone refresh my memory on the challenges which associated types avoid (by comparison to type-parameterized protocols)?

2

1

Associated types enforce functional dependency, so you can reliably infer T.Foo from T.

1

4

5

Replying to

This makes sense! As I understand: that wouldn't *preclude* future support of non-one-to-one relationships, right?

Replying to

Yeah, for instance Rust has "multi-dispatch" traits, where types are peers in the conformance: smallcultfollowing.com/babysteps/blog

1

and GHC lets you specify the dependencies alongside the parameter set haskell.org/haskellwiki/Fu

1

Show replies

Conversation