-
Notifications
You must be signed in to change notification settings - Fork 720
Install conda from miniforge rather than using old conda installation with a slow solver #4090
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
base: main
Are you sure you want to change the base?
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/audio/4090
Note: Links to docs will display an error until the docs builds have been completed. ❌ 6 New FailuresAs of commit 6b7e190 with merge base 2e30055 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
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.
Replace
echo "::group::Build FFmpeg"
.github/scripts/ffmpeg/build_gpu.sh
with
conda install -q -y ffmpeg
It the doc build time will stay around 15 minutes (as it is now), building ffmpeg within torchaudio can be eliminated.
Not working yet- it seems the system pip is still being used, not one associated with miniconda. In general, I find working with all the different kinds of conda spinoffs quite confusing; if I continue to encounter problems with the miniconda approach, I might just switch doing the build with pixi instead. |
Co-authored-by: Pearu Peterson <[email protected]>
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.
LGTM! Thanks, @samanklesaria!
Removing ffmpeg build can be finished in #4083 .
If waiting for green CI is blocking on #4083, I suggest merge this PR as the changes in this PR affects only docs build and it was successful (modulo a known test-infra issue). |
This PR is in response to #4083. The old version of conda used by pytorch-infra can't resolve the environment well enough to install ffmpeg, so we have to resort to the manual building step given by the ffmpeg github workflow. But @pearu thinks this problem could go away if we just used a more modern conda. This PR is a work in progress- it installs a fresh miniconda and attempts to activate it before building the rest of the docs.