Akuli

You really shouldn't have to care. The type annotation in this case should IMO be something simple that doesn't get in the way, not something full of `None, None` that...

So if I understood correctly, your function looks like this: ``` @contextlib.contextmanager def foo() -> Iterator[int]: return bar() ``` and it broke when `bar()` was changed to return an iterator...

It's technically not wrong: ``` >>> import contextlib >>> @contextlib.contextmanager ... def foo(): ... return iter([1]) ... >>> with foo(): ... print("Hi") ... Hi ``` but this is very unlikely...

Minimal reproducer: ```python3 from typing import * T = TypeVar("T") def f(dict_factory: Callable[[list[tuple[str, Any]]], T]) -> T: ... f(dict) ``` Error: `Argument 1 to "f" has incompatible type "Type[Dict[Any, Any]]";...

https://github.com/python/typeshed/blob/9aa66f0c37a633ed70ee3570778a437b8be82c29/stdlib/typing.pyi#L459-L462 We should probably change `_KT` to `_KT | None` in both overloads. Technically the first argument can be anything that overlaps with `_KT`, but in practice, allowing `None` is...

We could probably make it work well enough in practice by special-casing `None`s.

> false positives for places that use `wraps()` but change the signature These are arguably true positives, as one reason to use `wraps()` is that it copies the signature. A...

Detecting this at type check time doesn't seem to be possible, as it depends on `self.watching` and `self.explicit_transaction`, both of which can change after instantiating: https://github.com/andymccurdy/redis-py/blob/9ed5cd7808789f791fdc7ee368bd268307ac9847/redis/client.py#L1587-L1591 This would be yet...

A possible downside: If the tests of a third-party stub break for some reason, e.g. because a new version of the corresponding non-stub package is released, the problem will remain...

This seems to be done now.