Unclear to me why this would be needed, git tracks files, not folders, and all of the files that get put in __pycache__ folders should be excluded by the *.py[ocd] rule on line 27

If we do include __pycache__ as an explicit exclude, I don't think it should be scoped to lib/matplotlib/tests

Do you have an example of a file that gets placed in __pycache__ that is not already excluded? If there is something new in there, I think we should exclude that rather than __pycache__

Mostly wanted it cause this threw me for a loop:

None of the files in there get committed, but it's a bit confusing. Changed to the general form.

What are the contents of that folder?

usually its all .pyc files, and I'm pretty sure git usually will not show that if that is the case (if it did, it would show that all the time)... it does abbreviate to folder in git status when there are multiple untracked files in there, though

What are the contents of that folder?

.pyc files and WindowsPath('lib/matplotlib/tests/__pycache__/test_axes.cpython-311-pytest-7.4.0.pyc.10632')

.pyc files and WindowsPath('lib/matplotlib/tests/__pycache__/test_axes.cpython-311-pytest-7.4.0.pyc.10632')

I can confirm some of these files existing on Windows, though only for pytest 7.4.0. There is also a pytest 7.3.2 file but without the pid at the end. It looks like they write that and replace the pid-less filename. I'm not sure why that never seems to be a problem on Linux.

Perhaps something to do with pytest-dev/pytest@4cd0322 or that Windows can't delete open files (i.e., if you're running with xdist.)

I don’t see any reason to not ignore this folder, explicitly. By construction we don’t want this in the repository. And if, for some reason, there will be other file types in the future it’s better to explicitly ignore the folder, rather than relying on ignoring its content.