-
Notifications
You must be signed in to change notification settings - Fork 22
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
Build Mac wheels in CI #230
Conversation
65d2476
to
10dd8ca
Compare
This PR enable the build of M1 wheels to rely on `python -m build` instead of `python setup.py`. The default is still `python setup.py`. This is the same as #5071 and #5076 that were landed for linux wheels. We need this for torchcodec to support M1 wheels pytorch/torchcodec#230 CC @scotts @atalman
7ae2cc3
to
1032dd1
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What a ride. Thank you so much @scotts for pushing through!
Nothing blocking from me, just nits and one Q
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What a ride. Thank you so much @scotts for pushing through!
Nothing blocking from me, just nits and one Q
cd96123
to
b78a737
Compare
c3a7d00
to
af25dad
Compare
dc2fd48
to
393d108
Compare
This PR enable the build of M1 wheels to rely on `python -m build` instead of `python setup.py`. The default is still `python setup.py`. This is the same as pytorch#5071 and pytorch#5076 that were landed for linux wheels. We need this for torchcodec to support M1 wheels pytorch/torchcodec#230 CC @scotts @atalman
Note that this summary was written a month after opening the PR. As evidenced by the commit history below, getting this in shape has been a journey. A summary of the actual changes we'll commit:
VLOG
andLOG
) from our C++ code. For reasons that we have not been able to fully explain, we get segfaults in those logging calls only when we ran from a Mac wheel. We do not segfault there when we build and test Mac from source, or in any situation with Linux. That might indicate a problem in PyTorch core. If we want logging back, we can decide how best to it then.