Repository navigation
very small errors in test suite, py3.6 and 3.5 #7851
Description
19:13 $ inkscape --version Inkscape 0.92.0 r
======================================================================
FAIL: matplotlib.tests.test_axes.test_alpha.test(1, '/home/tcaswell/source/p/matplotlib/result_images/test_axes/test_alpha.svg', '/home/tcaswell/source/p/matplotlib/result_images/test_axes/test_alpha-expected.svg')
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/tcaswell/.virtualenvs/dd36/lib/python3.6/site-packages/nose/case.py", line 197, in runTest
self.test(*self.arg)
File "/home/tcaswell/source/p/matplotlib/lib/matplotlib/testing/decorators.py", line 58, in failer
result = f(*args, **kwargs)
File "/home/tcaswell/source/p/matplotlib/lib/matplotlib/testing/decorators.py", line 259, in do_test
'(RMS %(rms).3f)'%err)
matplotlib.testing.exceptions.ImageComparisonFailure: images not close: /home/tcaswell/source/p/matplotlib/result_images/test_axes/test_alpha_svg.png vs. /home/tcaswell/source/p/matplotlib/result_images/test_axes/test_alpha-expected_svg.png (RMS 0.004)
======================================================================
FAIL: matplotlib.tests.test_axes.test_pcolormesh.test(1, '/home/tcaswell/source/p/matplotlib/result_images/test_axes/pcolormesh.svg', '/home/tcaswell/source/p/matplotlib/result_images/test_axes/pcolormesh-expected.svg')
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/tcaswell/.virtualenvs/dd36/lib/python3.6/site-packages/nose/case.py", line 197, in runTest
self.test(*self.arg)
File "/home/tcaswell/source/p/matplotlib/lib/matplotlib/testing/decorators.py", line 58, in failer
result = f(*args, **kwargs)
File "/home/tcaswell/source/p/matplotlib/lib/matplotlib/testing/decorators.py", line 259, in do_test
'(RMS %(rms).3f)'%err)
matplotlib.testing.exceptions.ImageComparisonFailure: images not close: /home/tcaswell/source/p/matplotlib/result_images/test_axes/pcolormesh_svg.png vs. /home/tcaswell/source/p/matplotlib/result_images/test_axes/pcolormesh-expected_svg.png (RMS 0.001)
----------------------------------------------------------------------
Ran 6161 tests in 307.138s
Activity
tacaswell commented on Jan 17, 2017
efiring commented on Jan 17, 2017
It looks to me like a little tolerance is in order!
tacaswell commented on Jan 17, 2017
Looks like there is some rounding differences in how paths are output (scroll to the bottom of the diff)
Fun fact: "You can't comment at this time — your comment is too long (maximum is 65536 characters)."
A very truncated diff:
efiring commented on Jan 17, 2017
This is odd, because formatting is being done with short_float_fmt which is just
return '{0:f}'.format(x).rstrip('0').rstrip('.')
I don't understand how this could be leaving the zeros on the stroke-width value.
tacaswell commented on Jan 17, 2017
we may not have re-generated the svgs after that change?
efiring commented on Jan 17, 2017
#5651 was the PR with that change, back in Dec. 2015. Only two svg files were updated.
jkseppan commented on Jan 27, 2017
It's not only dropping zeros from the end, it's also 387.105882353 vs 387.105882
tacaswell commented on Jan 27, 2017
@QuLogic reported on gitter that the changes are 1 pixel by 1 bit (or around that magnitude).