matplotlib.tests.test_mathtext.test_mathtext_{cm,stix,stixsans}_{37,53}.test are failing

[jlec]

FAIL: matplotlib.tests.test_mathtext.test_mathtext_cm_53.test

Traceback (most recent call last):
File "/usr/lib64/python2.7/site-packages/nose/case.py", line 197, in runTest
self.test(_self.arg)
File "/var/tmp/portage/dev-python/matplotlib-9999/work/matplotlib-9999-python2_7/build/test/lib/matplotlib-1.4.x-py2.7-linux-x86_64.egg/matplotlib/testing/decorators.py", line 45, in failer
result = f(_args, **kwargs)
File "/var/tmp/portage/dev-python/matplotlib-9999/work/matplotlib-9999-python2_7/build/test/lib/matplotlib-1.4.x-py2.7-linux-x86_64.egg/matplotlib/testing/decorators.py", line 189, in do_test
'(RMS %(rms).3f)'%err)
ImageComparisonFailure: images not close: /var/tmp/portage/dev-python/matplotlib-9999/temp/python2.7/result_images/test_mathtext/mathtext_cm_53_pdf.png vs. /var/tmp/portage/dev-python/matplotlib-9999/temp/python2.7/result_images/test_mathtext/mathtext_cm_53-expected_pdf.png (RMS 34.378)

[jlec]

This happens with 1.3 > HEAD
Which more information do you need?

[jlec]

Result:

[jlec]

original:

[mdboom]

This looks like the perennial problem with different versions of freetype (we unfortunately run into this all the time). The text is rendered slightly differently on your machine than in the original images, and I suspect that is enough to throw the test off. So it's benign, but we should probably increase the tolerance on these tests.

[jlec]

I am running freetype-2.5.0.1.

[mdboom]

Yeah -- I've got 2.4.11 (and these images may well have been generated with an even earlier version).

[efiring]

This looks like something that has been addressed. Closing.