Fixing the stubtest issue...

stubtest PR here: python/mypy#16750

This is the script I used: https://raw.githubusercontent.com/hauntsaninja/snippets/main/pep570.py

Looks reasonable, though I wonder if it handles classmethods? You could probably special-case parameters named cls (and maybe even mcls, for metaclass classmethods?) in the same way you special-case parameters named self, right?

Ah did I not push that change? Oops, looks like I did not. Good point on metaclasses, will do that as well

Diff from mypy_primer, showing the effect of this PR on open source code:

prefect (https://github.com/PrefectHQ/prefect)
- src/prefect/utilities/pydantic.py:184: note:     def __new__(cls: type[type], object, /) -> type
+ src/prefect/utilities/pydantic.py:184: note:     def __new__(type[type], object, /) -> type
- src/prefect/utilities/pydantic.py:184: note:     def [_typeshed.Self] __new__(cls: type[_typeshed.Self], str, tuple[type, ...], dict[str, Any], /, **kwds: Any) -> _typeshed.Self
+ src/prefect/utilities/pydantic.py:184: note:     def [_typeshed.Self] __new__(type[_typeshed.Self], str, tuple[type, ...], dict[str, Any], /, **kwds: Any) -> _typeshed.Self
- src/prefect/utilities/pydantic.py:186: note:     def __new__(cls: type[type], object, /) -> type
+ src/prefect/utilities/pydantic.py:186: note:     def __new__(type[type], object, /) -> type
- src/prefect/utilities/pydantic.py:186: note:     def [_typeshed.Self] __new__(cls: type[_typeshed.Self], str, tuple[type, ...], dict[str, Any], /, **kwds: Any) -> _typeshed.Self
+ src/prefect/utilities/pydantic.py:186: note:     def [_typeshed.Self] __new__(type[_typeshed.Self], str, tuple[type, ...], dict[str, Any], /, **kwds: Any) -> _typeshed.Self

Looks like there's only one instance of the old syntax that PyCQA/flake8-pyi#461 detects in the stdlib and this PR doesn't fix:

You could fix that as well by also special-casing parameters named metacls in the script (or just fix it manually with an additional commit).

Maybe it'd be best to revert the handful of changes stubtest has issues with for now? We can easily defer them until there's a mypy release with the stubtest fix

Thanks for checking that! I also found several cases of incorrect positional-only arguments that this PR leaves alone.

It's basically all overloads containing positional-only args. Since this isn't user facing, I'd rather either just wait or use a pre-release mypy for stubtest CI