Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.Sign up
Called process error output doesn’t all go to stderr #6651
pip 19.1.1, Python: 3.8-dev, Ubuntu xenial
My travis build with Python 3.8 fails: https://travis-ci.org/theislab/scanpy/jobs/550248884#L285
The error occurs when trying to install scipy’s build requirements via
As you can see, the subsequent ignoring of the second line is not reported as an error (as I would expect).
Well, that would fix the part about one of two identical messages being reported as error.
But I still don’t understand what the error here is. It shouldn’t be an error if some environment marker in the build deps doesn’t match, that dep should simply be skipped.
FYI, I posted PR #6655 to address this.
It would remove the confusing
changed the title
Error when using environment markers in build-system.requires
Jun 27, 2019
Check out the link to the build failure. The line
is red, i.e. on stderr, while the subsequent line
is gray, i.e. on stdout.
Okay, thanks. That’s actually just an issue with the coloring getting applied to the log message. All those lines are still going to stderr (the subprocess error gets logged as one giant ~7500 line message). It’s just that only the first line of the error message is colored. That should probably be filed as a separate issue.