_______________________ test_if_rctemplate_is_up_to_date _______________________
[gw0] linux -- Python 3.5.3 /venv/bin/python3.5
def test_if_rctemplate_is_up_to_date():
# This tests if the matplotlibrc.template file
# contains all valid rcParams.
dep1 = mpl._all_deprecated
dep2 = mpl._deprecated_set
deprecated = list(dep1.union(dep2))
#print(deprecated)
path_to_rc = mpl.matplotlib_fname()
> with open(path_to_rc, "r") as f:
E FileNotFoundError: [Errno 2] No such file or directory: '/venv/lib/python3.5/site-packages/matplotlib/mpl-data/matplotlibrc'
Just a plea - the builds have been failing continuously since I put up this issue. Leaving the builds to fail means it's harder to detect new failures in wheel builds, making it harder to do the eventual release. I am happy to turn off the daily builds if it does not seem worth the effort to support them.
Yep sorry, I wasn't aware that mpl.matplotlib_fname() could actually point to anything other than the shipped matplotlibrc. The file that should be tested is the matplotlibrc.template which resides in the root folder of the source, but after building should be matplotlib\mpl-data\matplotlibrc (without extension). However it seems that this is not always the case?
I built a wheel python3 setup.py bdist_wheel and unzipped it and matplotlib/mpl-data/matplotlibrc was there. So I made a venv and installed it and the file was still there. I'm sorry, but you (or maybe someone on macOS?) will have to investigate this more as I cannot reproduce the problem.
Activity
matthew-brett commented on Mar 26, 2018
Just a plea - the builds have been failing continuously since I put up this issue. Leaving the builds to fail means it's harder to detect new failures in wheel builds, making it harder to do the eventual release. I am happy to turn off the daily builds if it does not seem worth the effort to support them.
matthew-brett commented on Apr 2, 2018
OK - disabling the daily builds for now.
QuLogic commented on Apr 2, 2018
Does #10929 help with this?
ImportanceOfBeingErnest commented on Apr 3, 2018
Yep sorry, I wasn't aware that mpl.matplotlib_fname() could actually point to anything other than the shipped matplotlibrc. The file that should be tested is the matplotlibrc.template which resides in the root folder of the source, but after building should be matplotlib\mpl-data\matplotlibrc (without extension). However it seems that this is not always the case?
matthew-brett commented on Apr 8, 2018
@QuLogic - thanks for the suggestion - but no - current master (which has #10929 merged I believe) still fails with the same error:
https://api.travis-ci.org/v3/job/363745427/log.txt
QuLogic commented on Apr 8, 2018
That's not how that line should look with #10929. I think we need a merge up from v2.2.x to master.
QuLogic commented on Apr 10, 2018
Please try again now that v2.2.x is merged into master.
matthew-brett commented on Apr 11, 2018
Same failure building mpl commit 0a57153 : https://api.travis-ci.org/v3/job/364856291/log.txt
QuLogic commented on Apr 13, 2018
I built a wheel python3 setup.py bdist_wheel and unzipped it and matplotlib/mpl-data/matplotlibrc was there. So I made a venv and installed it and the file was still there. I'm sorry, but you (or maybe someone on macOS?) will have to investigate this more as I cannot reproduce the problem.
matthew-brett commented on Apr 13, 2018
The OSX installation is standard (and passes all the other tests) (and lots of other projects use it and pass their tests).
Did you try running the tests?
jflatow commented on May 11, 2018
Is this the same issue? Copying matplotlibrc from the install dir got deleted here: 9093edd#diff-c28358ce9623af21f7b6e068f69085b7L725