Add @disjoint_base decorator in the stdlib

[JelleZijlstra]

No description provided.

[AlexWaygood]

I almost wonder if it's worth adding a variant is_disjoint_base_if, which you can pass a condition to providing the type checker can statically infer its truthiness. E.g.

@disjoint_base_if(sys.version_info< (3, 11))classFoo(tuple): ...

Maybe that would just be overcomplicating the proposal, though

[JelleZijlstra]

I've been thinking about a more general version, where we add a do-nothing identity decorator so you can write @(disjoint_base if sys.version_info >= (3, 12) else identity). Maybe not worth it though.

Pyright seems very unhappy here presumably because I added one too many sys.version_info branches, will try to figure out which one pushed it over the edge.

Some of the remaining stubtest errors are false positives, fixing in python/mypy#19701

[AlexWaygood]

I think for pure-Python classes that are disjoint bases because they have __slots__, we should generally just add the __slots__ rather than adding the @disjoint_base decorator? That will help type checkers in other ways too (#8832). I highlighted a few here, but I think there are probably some more

[JelleZijlstra]

I think for pure-Python classes that are disjoint bases because they have slots, we should generally just add the slots rather than adding the @disjoint_base decorator? That will help type checkers in other ways too (#8832). I highlighted a few here, but I think there are probably some more

Yes, I agree, but I'd rather not have to look at all of these classes manually.

[AlexWaygood]

I've been thinking about a more general version, where we add a do-nothing identity decorator so you can write @(disjoint_base if sys.version_info >= (3, 12) else identity). Maybe not worth it though.

yes, that's quite appealing, actually, from my perspective as a typeshed maintainer -- it would also be useful for deprecations, since we have an increasing number of sys.version_info branches where the only difference between the branches is that a function is marked with @deprecated in one branch but not in the other.

Although that's actually quite convenient for us over at ty, because the deprecation notice is often also reflected in the docstring, and docstring-adder will add the version-specific docstrings to the two branches.

[AlexWaygood]

LGTM if you fix the TarInfo CI failure!

[JelleZijlstra]

I think I need #14626 for CI to go green. I'll also wait for my mypy PR to be merged so I can pin stubtest to a commit from mypy master rather than a random one from my fork.

[JelleZijlstra]

Had to make some more tweaks to accommodate recent stubtest changes, but it's looking green now. I ignored one issue with _random.Random in 3.10 only that seemed annoying, we can deal with that later.

[github-actions]

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

ibis (https://github.com/ibis-project/ibis) - ...typeshed_to_test/stdlib/builtins.pyi:115: note: "Any" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:117: note: "Any" defined here- ...typeshed_to_test/stdlib/builtins.pyi:115: note: "__init__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:117: note: "__init__" of "object" defined here strawberry (https://github.com/strawberry-graphql/strawberry) - ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here- ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here prefect (https://github.com/PrefectHQ/prefect) - ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here- ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here- ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here- ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here- ...typeshed_to_test/stdlib/builtins.pyi:137: note: "__init_subclass__" of "object" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:139: note: "__init_subclass__" of "object" defined here zulip (https://github.com/zulip/zulip) - ...typeshed_to_test/stdlib/builtins.pyi:115: note: "SubTest" defined here+ ...typeshed_to_test/stdlib/builtins.pyi:117: note: "SubTest" defined here discord.py (https://github.com/Rapptz/discord.py) - ...typeshed_to_test/stdlib/typing.pyi:1042: note: "update" of "TypedDict" defined here+ ...typeshed_to_test/stdlib/typing.pyi:1043: note: "update" of "TypedDict" defined here- discord/ext/commands/hybrid.py:834: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/hybrid.py:834: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]- discord/ext/commands/hybrid.py:858: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/hybrid.py:858: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]- discord/ext/commands/hybrid.py:883: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/hybrid.py:883: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]- discord/ext/commands/hybrid.py:935: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/hybrid.py:935: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]- discord/ext/commands/bot.py:290: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/bot.py:290: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]- discord/ext/commands/bot.py:314: error: Overlap between argument names and ** TypedDict items: "name", "with_app_command" [misc]+ discord/ext/commands/bot.py:314: error: Overlap between argument names and ** TypedDict items: "with_app_command", "name" [misc]

[tungol]

It's great to see all those allowlist entries go away