astropy: ImportError in Python 3.6 importlib
I see this in Python 3.6 test suites. Is this safe to ignore or should be fixed somehow? Should we move this to astropy-helpers
? I am not sure where this is coming from.
home/travis/miniconda/envs/test/lib/python3.6/importlib/_bootstrap.py:205:
ImportWarning: can't resolve package from __spec__ or __package__,
falling back on __name__ and __path__
return f(*args, **kwds)
p.s. Instead of decreasing the number of issues, I am increasing it. Sorry, @bsipocz !
About this issue
- Original URL
- State: closed
- Created 7 years ago
- Reactions: 1
- Comments: 27 (25 by maintainers)
I think this is still an issue until it is fixed upstream, so keeping this open for now.
Sorry I thought it was pllim above 😁 Yes I guess this could happen, but the cython releases is old enough now that we should be safe.
Yes, you can certainly ignore this warning. Ultimately this will go away seamlessly once addressed upstream (and we just keep it open in the meantime to avoid duplicate reports opened for it).
I also see these warnings, so did a quick debugging:
It is a Cython issue (https://github.com/cython/cython/issues/1753 , https://github.com/cython/cython/issues/1720) related to a new warning in Python 3.6 (http://bugs.python.org/issue25791). https://github.com/cython/cython/issues/1720 gives some workaround, i will try.
IMO we should remove it from the ignore list, or not close but update the description here if the warning is still raised in newer versions
There was a fix in Cython 0.27 but disabled since (“Reopening since the implementation in 0.27.0 is not complete and was disabled in 0.27.1” : https://github.com/cython/cython/issues/1715).
Reading through their related issues, the tone seemed to be a bit arrogant and suggested that this won’t go away any time soon.
Maybe we can do a blind import before test collection, just as we do for a few upstream modules to filter out their deprecation warnings? But even if we do, I agree that it can be a bit risky, and anyway, low priority.
So, since this is caused by bug upstream, should we close this issue here?