[build] Monkeypatch gen_rst to call inside subprocess#3355
Merged
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3355
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 919d7ed with merge base b2e614d ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
malfet
approved these changes
May 15, 2025
Contributor
malfet
left a comment
There was a problem hiding this comment.
LGTM, just remove reset_seeds function completely
e54c7e8 to
a9972bc
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Replace sphinx gallery's generate_file_rst call with a wrapper that calls that function but in a subprocess, and remove some things that we added in order to handle files changing global state
I tried using multiprocess.pool and apply async but ran into pickling problems. I don't understand why this current implementation doesn't have that issue
parallelism got added in some version of sphinx gallery, but it uses joblib, which seems to result in errors (conflicts with functions in subprocess library? im not sure). Additionally, it has no effect if you set parallel = 1 (it will not put each file run into its own process and run singly) so you need parallel >= 2, but I'm not sure if there are any tutorials that require being run on their own (ex for memory profiling)
Afaict (although I am not familiar with joblib) the way I do it here is similar to how sphinx gallery does it with job lib but they can wrap the actual call instead of needing to replace the function
https://github.com/sphinx-gallery/sphinx-gallery/blob/dd092a09513ea1d0616ac9e59b8d76d5a8217e4a/sphinx_gallery/gen_rst.py#L594-L608
I also tried to call sphinx once per file in #3351 but there's an over head of ~5 min per file due to sphinx doing extra stuff (generating the html for all the other files?) resulting in taking 2x longer