Fix return type of logging.config.BaseConfigurator.as_tuple

[brianschubert]

Fix typo: tuple[Any] -> tuple[Any, ...]

Based on the source, a length-1 tuple isn't intended; the tuple can have any length.

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉

[AlexWaygood]

Nice catch! If you install flake8 and flake8-pyi, and apply this diff, then run flake8 stubs stdlib --select=Y090, you might be able to find a few more of these. I haven't done so in a while.

--- a/.flake8
+++ b/.flake8
@@ -3,8 +3,6 @@
 # Y: Flake8 is only used to run flake8-pyi, everything else is in Ruff
 # F821: Typeshed is a testing ground for flake8-pyi, which monkeypatches F821
 select = NQA, Y, F821
-# Ignore rules normally excluded by default
-extend-ignore = Y090
 per-file-ignores =
   # We should only need to noqa Y and F821 codes in .pyi files
   *.py: NQA

(the flake8 rule is off by default because it has false positives)

[JelleZijlstra]

It might be worth turning it on by default (I know I argued the other way in the past). A single-element tuple is a weird enough type that it seems worth it to force you to write a noqa if it's really what you want.

[AlexWaygood]

there's still lots of hits on typeshed...

What I'd really love is a flake8_pyi_primer that leaves a comment on your PR with all the new Y090 hits that your PR caused relative to the baseline of main. That would not only let us get more signal from Y090, but also allow us to add more rules to flake8-pyi that are usually correct, but have too many false positives to be on by default.

Unfortunately I do not have the time to work on such a tool right now (brb, building a type checker, etc.)