There's a couple of PEP8 violations in the test - some trailing whitespace at the end of the lines, and no newline at the end of the file (see https://travis-ci.org/matplotlib/matplotlib/jobs/205418695#L1677)

Can you please add a test to https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/tests/test_mathtext.py#L17 to cover this case?

Over all 👍 thanks for working on this!

To get the test images you will have to add the test, run the test suite, let if it fail on a missing image, then copy output into tests/baseline_images/test_mathtext and commit them (see http://matplotlib.org/devel/testing.html#writing-an-image-comparison-test)

You will also need to be sure to be testing against the correct version of freetype (see http://matplotlib.org/devel/contributing.html#retrieving-and-installing-the-latest-version-of-the-code)

''power-cycled' to trigger travis.

It needs to be a raw string or escaped backslashes, not both.

Am I missing something? The reStructuredText documents make it sound like you need to make the text is a raw string to properly pass the backslashes out of python, but that you still need to escape the backslashs for the reStructuredText to be parsed correctly (http://docutils.sourceforge.net/docs/user/rst/quickref.html#escaping).

At this point, we don't actually render any of that text through Sphinx. After #8247 is merged, then maybe it will be clearer what should be done.

Do you want to have a go at rebasing this? The sphinx-gallery stuff is merged now, so we should be able to see whether the escaping is going to be a problem.

It looks like some of the new required test images did not get added

_________________ test_mathtext_rendering[png-mathtext-cm-82] __________________
[gw1] linux -- Python 3.4.6 /home/travis/build/matplotlib/matplotlib/venv/bin/python
self = <matplotlib.testing.decorators._ImageComparisonBase object at 0x7fffe12af160>
baseline = 'mathtext_cm_82', extension = 'png'
    def copy_baseline(self, baseline, extension):
        baseline_path = os.path.join(self.baseline_dir, baseline)
        orig_expected_fname = baseline_path + '.' + extension
        if extension == 'eps' and not os.path.exists(orig_expected_fname):
            orig_expected_fname = baseline_path + '.pdf'
        expected_fname = make_test_filename(os.path.join(
            self.result_dir, os.path.basename(orig_expected_fname)), 'expected')
        if os.path.exists(orig_expected_fname):
            shutil.copyfile(orig_expected_fname, expected_fname)
        else:
            reason = ("Do not have baseline image {0} because this "
                      "file does not exist: {1}".format(expected_fname,
                                                        orig_expected_fname))
>           raise ImageComparisonFailure(reason)
E           matplotlib.testing.exceptions.ImageComparisonFailure: Do not have baseline image /home/travis/build/matplotlib/matplotlib/result_images/test_mathtext/mathtext_cm_82-expected.png because this file does not exist: /home/travis/build/matplotlib/matplotlib/lib/matplotlib/tests/baseline_images/test_mathtext/mathtext_cm_82.png
lib/matplotlib/testing/decorators.py:271: ImageComparisonFailure

I suspect the issue is missing optional dependencies on the local system where the test images were generated. @watkinrt I can try to generate them and push to this branch or you can sort out which deps you are missing (my guess is LaTeX?), which ever you would prefer.

@watkinrt I can try to generate them and push to this branch or you can sort out which deps you are missing (my guess is LaTeX?), which ever you would prefer.

I just realized I hadn't fetched the master branch from the main matplotlib repository in a while; I image that is why I was missing some of the new test images. I've updated my working copy and just pushed it. I haven't been able to successfully run the full test suite on this commit as something is messed up with the my development copy of matplotlib. If someone else has time, would you mind running the tests? I'm not sure when I'll be able to get around to fixing my development copy.

Sorry for all of my newbiew mistakes; this is the first large scale repository/team I've worked on.

It's preferable to rebase instead of merging master. However, the tests are failing because it looks like you forgot to commit the image for the new test.

@watkinrt I'll try to generate those test images for you now and push to your branch.

With your permission I can do the rebase for you (and force-push to your branch).

Don't worry about small process mistakes, we were all new once!

I pushed a commit and then almost immediately force-pushed a fixed version of it (accidentally committed a temporary file). Hopefully no one fetched between the two pushes, but if you did, that is the source of any issue.

Looks like this is passing now and the test images look good!

Great! Thanks for closing all of this out. Hopefully that wraps everything up for this pull request.

@QuLogic I think this is ready to go, can you give this another review or dismiss your change request?

@watkinrt Thanks and congratulations on what (I think) is your first Matplotlib contribution 🎉 !

Thanks for taking on one of the more arcane bits of the code base and bearing with us through a very long review process!

Hopefully we will hear from you again.

Wow, this is great, @watkinrt ! I had no idea you had finished this almost 2 years ago!