... | ... | @@ -19,6 +19,7 @@ |
|
|
- [\#11739](https://gitlab.haskell.org//ghc/ghc/issues/11739) (simplify axioms)
|
|
|
- Fix all the `TypeInType` bugs
|
|
|
- Clean up pure unifier to make the fact that kind coercions *only* affect type variables by using, e.g., `getCastedTyVar_maybe`.
|
|
|
- Is it possible to remove all the pushing in `mkCastTy`? I think so. The pushing doesn't help `splitTyConApp` at all, because pushing a coercion into a `TyConApp` doesn't make it become another `TyConApp`. The pushing *does* help `splitAppTy`, but perhaps all usages of `splitAppTy` already account for casts. *Dangling problem:* what on earth to do about `LRCo`, given that it's not always possible to push? It seems that it may be impossible to make `LRCo` work in a way that respects `eqType` equality, with `eqType`'s ability to move casts about willy-nilly.
|
|
|
|
|
|
**Iceland_jack**: By `[]` as a data family do you mean:
|
|
|
|
... | ... | |