flix
flix copied to clipboard
experiment: separate unkinded and kinded types
What's the plan here?
Depends on what you think of it.
I think it's going in the right direction. There are a few things that can be cleaned up afterward, as I was trying to make the translation as straight as possible. For example, moving mkNot etc. from resolver to UnkindedType, removing unused validations from Kinder, and providing a way to format unkinded types.
I think this is reviewable at least.
The one big consideration is what to do about unkinded types in error messages. My current idea is that we just use the string from the source instead of trying to format it somehow.
Which ASTs would use UnkindedType?
Which ASTs would use UnkindedType?
Introduced in the Resolver
and eliminated in the Kinder
(the following phase)
@mlutze Conflict, sorry.
You want this merged, right?
You want this merged, right?
Yes, assuming we're in agreement on it in principle.
I think its a good refactoring.
(Lets just hope that we won't need to revert in the future.)
Overall I like the refactoring 👍
Merged. Thanks! Nice cleanup 👍