[srittau]

Closes #4514 #2630 #2686

I tested this against our internal codebases and against the example in #1464 and it seems to works fine.

[srittau]

@hauntsaninja There is something strange going on with the mro here, see the comment in the stubtest whitelist. In my opinion, this special case (quirk in Python?) should be handled by stubtest (or mypy?), instead of putting the enums on the whitelist. What do you think?

An alternative is to override __new__ explicitly in IntEnum, IntFlag, and another enum I forgot in the stubs. That doesn't feel right to me, though, as the implementation doesn't do that either.

[gvanrossum]

I'm guessing that the Enum metaclass plunks a reference to Enum.__new__ in the __dict__ of all new Enum subclasses. I think here: https://github.com/python/cpython/blob/a33f2c2bae759fc9d06e1c032fd2026135f2df45/Lib/enum.py#L273

[hauntsaninja]

Oh interesting. I'd probably agree that this should be handled by stubtest / mypy.

[srittau]

@gvanrossum @hauntsaninja

1. Good catch!
2. Yikes!
3. In that case I am not sure that stubtest needs fixing. Since every enum sub-class basically gets Enum.__new__ assigned to it, manually overriding it in sub-classes that derive from both int/str and Enum might be the correct solution.

[hauntsaninja]

I was a little worried about this PR, so I ran mypy_primer:

(mypy_primer) ~/dev/mypy_primer master λ python -m primer --new v0.782 --old v0.782 --custom-typeshed-repo ~/dev/typeshed --new-typeshed dundernew --old-typeshed 98667b18 -o concise     
+ isort/output.py:561: error: "_LineWithComments" has no attribute "comments"

- src/aiortc/mediastreams.py:22: note:     def __init__(self, x: Union[str, bytes, SupportsInt, _SupportsIndex] = ...) -> int
+ src/aiortc/mediastreams.py:22: note:     def __new__(cls, x: Union[str, SupportsInt, _SupportsIndex] = ...) -> int

+ sphinx/builders/html/__init__.py:94: error: unused 'type: ignore' comment
+ sphinx/builders/html/__init__.py:117: error: unused 'type: ignore' comment

Seems fine!