Summary
Talk about rust-lang/rust#96912 (and generally how projection equality should look like)
Rationale
for<'a> <_ as Trait<'a>>::Assoc: Other bounds resulting in an error is a problem i keep seeing encountered
- came across it in
bevy_ecs
- known
generic_associated_types issue
- blocker for supporting some kind of
F: Fn(T) -> _ for an unconstrained output type as the only reason to want to use -> _ would involve the fn being higher ranked which basically guarantees encountering this bug playground
Also there's a weird hacky change to selection in this PR to fix a typenum regression which definitely seems good to talk about before merged 🤔
Summary
Talk about rust-lang/rust#96912 (and generally how projection equality should look like)
Rationale
for<'a> <_ as Trait<'a>>::Assoc: Otherbounds resulting in an error is a problem i keep seeing encounteredbevy_ecsgeneric_associated_typesissueF: Fn(T) -> _for an unconstrained output type as the only reason to want to use-> _would involve the fn being higher ranked which basically guarantees encountering this bug playgroundAlso there's a weird hacky change to selection in this PR to fix a typenum regression which definitely seems good to talk about before merged 🤔