-
-
Notifications
You must be signed in to change notification settings - Fork 9.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
BLD: rewrite np.distutils.exec_command.exec_command() #7862
Conversation
2b5f14f
to
f2c1d37
Compare
There's wacky failure on one of the Travis builders that I can't really understand... |
Looks odd indeed, I restarted that build to make sure it's a real issue. |
Hmm, it fails reproducibly. Many times |
Will be a nice cleanup, thanks @pitrou. |
Scary piece of untested code to touch though:) |
It does have tests in |
Would be good to check also a Scipy build on Windows, to make sure we don't make the "too many open files" errors worse there. I recently had |
Indeed it does, forgot about that. The |
See numpy/numpy#7614 for context, and numpy/numpy#7862 for the patch submitted upstream. This should fix random failures on our Windows CI builders.
exec_command() is currently a mess of several implementations using outdated Python APIs and various hacks. This rewrites it to use the standard subprocess module. See PR numpy#7614 for previous discussion.
f2c1d37
to
f7a5d04
Compare
I've solved the issue by setting the locale to something existing in the build script. This was really a bug in the build script. |
I've tried too, but I end up with "numpy.distutils.system_info.NotFoundError: no lapack/blas resources found". I guess someone more expert than me will have to try it out. |
I guess there are no known problems on windows? |
thanks this was long overdue refactoring |
Bonus comment from #8767 for posterity: this patch allows parallel builds to actually run in parallel on macOS. |
exec_command() is currently a mess of several implementations using outdated Python APIs and various hacks. This rewrites it to use the standard subprocess module. See PR #7614 for previous discussion.
I've tested this on Python 2.7 and 3.5, Windows and Linux.