fix(ci): set TMS_CUDA_MAJOR via uv extra-build-variables - #3905
Merged
Conversation
torch-memory-saver's build backend needs TMS_CUDA_MAJOR when uv invokes its build backend during lockfile generation. Declare it once via uv's extra-build-variables so uv lock/sync pick it up automatically in every CI system, instead of deriving it via shell/sed in each caller (#3862, whose GitHub Actions workaround is reverted here as no longer needed). Signed-off-by: Kajal Jain <kajalj@nvidia.com>
terrykong
reviewed
Aug 29, 2026
terrykong
left a comment
terrykong
left a comment
Collaborator
There was a problem hiding this comment.
does the env need to be re locked?
chtruong814
approved these changes
Aug 29, 2026
Contributor
Author
|
/ok to test e1d9abc |
terrykong
approved these changes
Aug 29, 2026
terrykong
left a comment
terrykong
left a comment
Collaborator
There was a problem hiding this comment.
approved. i locked locally and i guess the env doesn't need to be re-locked. looks like env vars don't get saved in lock so technically aren't part of what it determines as equivalency
Contributor
Author
UV.lock is valid - checked |
kajalj22
enabled auto-merge (squash)
August 29, 2026 02:27
kajalj22
disabled auto-merge
August 29, 2026 02:27
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
TMS_CUDA_MAJORwhen uv builds it duringuv lock. Declare it once via[tool.uv.extra-build-variables]inpyproject.tomlso every caller ofuv lock/uv sync(GitHub Actions, nemo-ci, local dev) picks it up automatically — no per-CI-system shell/sed logic needed.uv lock), which this supersedes.Test plan
uv locksucceeds and lockfile includes torch-memory-saver without manually exportingTMS_CUDA_MAJOR