build: restrict dependency cooldown to pypi index#672
Merged
gsprochette merged 1 commit intoJun 4, 2026
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
This PR changes the scope of the dependency cooldown introduced in #660 because the cooldown was affecting extra indices which contain wheels without a release date.
Doing
uv pip install torch==2.12 torchvision --index-url https://download.pytorch.org/whl/cu126(or any extra index) would fail because the wheels on the extra index lack a release date and therefore are excluded by the cooldown condition.This means only the blackwell-specific cu130 pypi wheels for torch 2.11 and 2.12 are currently considered compatible.
The fix consists in applying the cooldown only to the pypi index instead of globally
Since pypi doesn't allow wheels to specify external urls, this is carried recursively to all pypi dependencies. In pruna the only extra index url is our pythonanywhere index which we trust, and other indices would have been manually added by the user and therefore their responsability.
This requires uv>=0.11.5 hence the uv version bump in the CI, and the early versions after 0.11.5 raise an experimental warning for the per-index exclude-newer feature.
Type of Change
Testing
uv pip install torch==2.12 torchvision --index-url https://download.pytorch.org/whl/cu126now runs.Checklist
Thanks for contributing to Pruna! We're excited to review your work.
New to contributing? Check out our Contributing Guide for everything you need to get started.