Repository navigation
Closed
Description
Bug report
Bug summary
Starting with #18545 and the subsequent merge, test_transparent_markers[pdf] has been failing on v3.3.x for Travis macOS. It does not appear to be failing on other systems or other branches. There were no other builds on that branch for 5 days before that, so it may or may not be related to that PR specifically.
I could not reproduce on the Mac Mini either, so I'm not sure how to fix it.
Actual outcome
________________________ test_transparent_markers[pdf] _________________________
[gw0] darwin -- Python 3.8.5 /usr/local/opt/python@3.8/bin/python3.8
expected = '/Users/travis/build/matplotlib/matplotlib/result_images/test_axes/transparent_markers-expected.pdf'
actual = PosixPath('/Users/travis/build/matplotlib/matplotlib/result_images/test_axes/transparent_markers.pdf')
tol = 0
def _raise_on_image_difference(expected, actual, tol):
__tracebackhide__ = True
err = compare_images(expected, actual, tol, in_decorator=True)
if err:
for key in ["actual", "expected", "diff"]:
err[key] = os.path.relpath(err[key])
> raise ImageComparisonFailure(
('images not close (RMS %(rms).3f):'
'\n\t%(actual)s\n\t%(expected)s\n\t%(diff)s') % err)
E matplotlib.testing.exceptions.ImageComparisonFailure: images not close (RMS 4.572):
E result_images/test_axes/transparent_markers_pdf.png
E result_images/test_axes/transparent_markers-expected_pdf.png
E result_images/test_axes/transparent_markers_pdf-failed-diff.png
lib/matplotlib/testing/decorators.py:139: ImageComparisonFailure
------------------------------ Captured log call -------------------------------
DEBUG matplotlib.axes._base:_base.py:2613 title position was updated manually, not adjusting
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IHDR' 16 13
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'iCCP' 41 2354
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:351 iCCP profile name b'default_rgb.icc'
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:352 Compression method 0
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'pHYs' 2407 9
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'tEXt' 2428 31
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IDAT' 2471 8192
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IHDR' 16 13
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'iCCP' 41 2354
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:351 iCCP profile name b'default_rgb.icc'
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:352 Compression method 0
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'pHYs' 2407 9
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'tEXt' 2428 31
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IDAT' 2471 8192
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IHDR' 16 13
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'iCCP' 41 2354
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:351 iCCP profile name b'default_rgb.icc'
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:352 Compression method 0
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'pHYs' 2407 9
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'tEXt' 2428 31
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IDAT' 2471 8192
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IHDR' 16 13
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'iCCP' 41 2354
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:351 iCCP profile name b'default_rgb.icc'
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:352 Compression method 0
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'pHYs' 2407 9
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'tEXt' 2428 31
DEBUG PIL.PngImagePlugin:PngImagePlugin.py:153 STREAM b'IDAT' 2471 8192
Matplotlib version
- Operating system: macOS
- Matplotlib version: v3.3.x
- Matplotlib backend (
print(matplotlib.get_backend())): macosx, I guess, though maybe Agg for the test - Python version: 3.8
Activity
timhoffm commentedon Sep 23, 2020
QuLogic commentedon Sep 29, 2020
This started failing on Fedora Rawhide as well, which just updated to Ghoscript 9.53. This also changed versions between the working and failing macOS builds. The strange thing is that there is no substantial difference in the test and even the baseline image between
v3.3.xandmaster, unless it has to do with the test framework itself.QuLogic commentedon Sep 29, 2020
Ah, it's because ghostscript is installed as a side effect on v3.3.x, but not on master.