Fix LaTeX error due to labels that become empty in simplify_labels

[lpsinger]

Fixes #8004.

[pllim]

Thanks, @lpsinger ! Any chance we can add a test for this use case too?

[lpsinger]

I added a unit test, but it is failing with this error message:

RuntimeError: Failed to process string with tex because latex could not be found

Are there any CI jobs that run in an environment with LaTeX installed? Is there already a decorator or a pytest mark that I can use to skip this test if LaTeX is not installed?

[pllim]

Hmm. Not that I know of. You can try using pytest.importorskip feature to skip it.

[lpsinger]

I am using checkdep_usetex from Matplotlib now. See this example:

https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/tests/test_usetex.py#L11-L12

[pllim]

Might need to skip it regardless if matplotlib version is too old. The job with Matplotlib 2.1.2 is failing. FYI.

[lpsinger]

Done.

[lpsinger]

Now it's failing with:

NameError: name 'TEX_MISSING' is not defined

I've no idea why, since it is not having a problem with MATPLOTLIB_LT_21.

[lpsinger]

I don't think that it's even respecting @pytest.mark.skipif('MATPLOTLIB_LT_21'). Something very strange and counterintuitive is going on. I use pytest.mark.skipif all the time, and this behavior does not make sense to me. Help!

[pllim]

I wonder if it is incompatible with @ignore_matplotlibrc. I will try to push a commit to your PR and see. I hope you don't mind.

[pep8speaks]

Hello @lpsinger 👋! It looks like you've made some changes in your pull request, so I've checked the code again for style.

There are no PEP8 style issues with this pull request - thanks! 🎉

Comment last updated at 2020-03-19 20:55:04 UTC

[lpsinger]

I will try to push a commit to your PR and see. I hope you don't mind.

Not at all. Please go ahead.

[pllim]

Failures are unrelated but we're also blocked by #10039

[Cadair]

This is an empty maths block right i.e. $...$ where there is nothing in between, as opposed to the LaTeX for display maths?

Could you add a comment about this just to make it more obvious when reading it?

[lpsinger]

Done.

[lpsinger]

@pllim, you're right, @ignore_matplotlibrc does not work with skipif. I think that ignore_matplotlibrc would be better implemented as a pytest fixture. I'll put together a patch for that so that we don't have to work around it.

[bsipocz]

@lpsinger - since your other PR has been merged, could you rebase this and use the new fixture here? Thanks!

[lpsinger]

@lpsinger - since your other PR has been merged, could you rebase this and use the new fixture here? Thanks!

Done.

[pllim]

Thanks!

[bsipocz]

Thank you @lpsinger!