build: keep sccache prefix when CUDA_PYTHON_TOOLCHAIN selects a compiler - #2969
Merged
Merged
Conversation
Contributor
Contributor
Author
|
/ok to test 4dcef12 |
Contributor
|
rwgk
reviewed
Sep 30, 2026
rwgk
approved these changes
Sep 30, 2026
juenglin
added a commit
to juenglin/cuda-python
that referenced
this pull request
Sep 30, 2026
Keep LDCXXSHARED sysconfig extras; NVIDIA#2969 sccache prefix is already in this branch.
2 tasks done
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Follow-up to #2903. An explicit
CUDA_PYTHON_TOOLCHAINwas replacing CI'sCC="sccache cc"/CXX="sccache c++"with a bare compiler, so llvm/gnu builds would skip sccache._with_sccachekeeps a leadingsccachetoken (including absolute paths such as/host/.../sccacheused by cibuildwheel) and only swaps the compiler:CC="sccache cc"+ llvm becomesCC="sccache clang". The same prefix is applied toCXX;LDSHAREDis{CXX} -shared. The default path (unsetCUDA_PYTHON_TOOLCHAIN) still does not touchCC/CXX.Checklist