Conor MacBride

I rebased and added a commit to remove PyPI codecov, but that should be a separate PR. I don't like how these new options are configured. I think it would...

I've opened #163 to remove codecov. `py310-test-sphinxdev` is failing there also.

#103 would have worked for my use case, although I think my need to document private methods was due to bad API design! I think #103 is more useful and...

This issue may be related to the `_update_header_data` method not being called when the `data` attribute is updated. https://github.com/astropy/astropy/pull/12058 may be a partial solution to this, however it may also...

Very strange. It seems that passing `format='png'` to savefig changes the hashes. I wonder if a default has changed in 3.5?

The hashes are different on `main` also (testing py310-test-mpl35 on macOS locally) so not related to this PR. Matplotlib 3.5.2 was released 19 days ago so my guess is that...

Yes, the tests pass on MPL 3.5.1. Can you pin this to `3.5.1` rather than `3.5.*`? https://github.com/matplotlib/pytest-mpl/blob/e3876180847c2d04aeda1982372074998a7f8716/tox.ini#L29 Hopefully this is something that the perceptual hashes can solve, as we really...

> @ConorMacBride If the `sha` hashes are sensitive to some change from `mpl>3.5.1`, what does that mean for default hash testing? Switch to `phash`? `sha` hashes are definitely too sensitive....

Given how effective removing the metadata version number was in #163 I think we should definitely remove it by default and make it a breaking change (for `v1.0.0` 😉).

I believe the differences are due to pytest-mpl using the `classic` style by default. [(Matplotlib changed their default style in v2.0)](https://matplotlib.org/stable/users/prev_whats_new/dflt_style_changes.html) To use Matplotlib's current default style in pytest-mpl, test...