MNT: Compat with pytest 8.1

[pllim]

To avoid this in pytest 8.1.dev:

pluggy._manager.PluginValidationError: Plugin 'pytest_mpl' for hook 'pytest_report_header'
hookimpl definition: pytest_report_header(config, startdir)
Argument(s) {'startdir'} are declared in the hookimpl but can not be found in the hookspec

See pytest-dev/pytest#11779 (comment)

Fix#216 , xref pytest-dev/pytest#11714 (comment)

[bsipocz]

Could you also revert bd7b5c6, so this could be tested?

(and suggestion for the maintainers: add a job for testing with the 8.0.x branch (RC) just as well with pytestdev)

[pllim]

Good catch; reverted. Thanks!

[pllim]

I have no idea what this test failure is in pytest-dev job

E ValueError: I/O operation on closed file
FAILED ../../tests/test_baseline_path.py::test_config[None-None-dir3-dir3-True]

[pllim]

That commit I reverted was from #215 by @ConorMacBride . I have no idea what he meant by "Skip pytestdev until hook wrapper issue is fixed" as there is no further explanation in the PR nor any linked issue.

[ConorMacBride]

The hook wrapper issue is #216

[ConorMacBride]

When I tested a fix for #216 last month, the tests passed with the pytest main branch installed. I did not see any mention of this issue, which seems strange as I would have expected the deprecation warning to be raised as an exception. Have you seen this pytest-mpl hook raise an exception?

[bsipocz]

I would have expected the deprecation warning to be raised as an exception

We didn't see deprecation warnings in other plugins either and similar failures popped up when they became actual errors. I haven't dug deep enough to get a full understanding but the plugins/pytest doesn't always behave the same predictable way as a normal test dependency library.

[pllim]

pytestdev is green but the log has traceback, is this expected? I don't maintain this plugin so I don't know what the test is supposed to do.

https://github.com/matplotlib/pytest-mpl/actions/runs/7464235520/job/20310772695?pr=219

[pllim]

Would be nice to get this merged sooner than later if this patch is acceptable. astropy devinfra job is failing without this.

[pllim]

@ConorMacBride , does this look good to you now? Can we merge this? I don't have write access to this repo but astropy depends on it.

[ConorMacBride]

Thanks @pllim and @bsipocz!