SubconParsedType and SubconBuildTypes can and should be invariant

[robin523790]

SubconParsedType and SubconBuildTypes can and should be invariant.

[timrid]

The Question is why should it be invariant? Off the top of my head, I can't recall the exact reasons anymore, but there were reasons why that was added in #1.

But what I'm wondering is why the tests test_buildtypes_contravariance and test_parsetype_covariance don't fail when we simply remove the covariance and contravariance.

[robin523790]

The Question is why should it be invariant? Off the top of my head, I can't recall the exact reasons anymore, but there were reasons why that was added in #1.

All the type: ignore that were necessary before, are a good indicator for "why", I'd say. It looks like only BuildTypes and ParseType need to be covariant and contravariant, respectively. The Subcon... types were likely copy&paste?

But what I'm wondering is why the tests test_buildtypes_contravariance and test_parsetype_covariance don't fail when we simply remove the covariance and contravariance.

Well, these tests consider BuildTypes and ParseType, not the Subcon... pieces.

[timrid]

Well, these tests consider BuildTypes and ParseType, not the Subcon... pieces.

Oh, my mistake. I mixed up SubconParsedType / SubconBuildTypes with ParsedType / BuildTypes.

This looks fine. Thanks.