Repository navigation
Make pip install work on a stock CUDA machine - #2
Merged
Merged
Conversation
The 0.1.0 sdist on PyPI contains no CUDA sources at all -- no .cu, no headers, no CMakeLists -- so `pip install culof` could never build. Nothing short of a new release fixes that, and 0.1.0 cannot be reuploaded, so this prepares 0.2.0. Finding nvcc is the other half. CMake searches only PATH for it, while the usual Linux install puts the toolkit under /usr/local/cuda and leaves only the driver on PATH. project(... LANGUAGES CUDA) then aborts with "No CMAKE_CUDA_COMPILER could be found" on a machine with a perfectly good toolkit -- reproduced here on an A6000. Look in the standard roots before enabling the language. Also: the README pointed at its images with relative paths, which resolve on GitHub and break on the PyPI project page, so every figure was a broken image there. Absolute raw.githubusercontent.com URLs render in both places. The release workflow publishes via PyPI Trusted Publishing, so no API token is stored in the repository, and it refuses to publish an sdist that is missing any file needed to build -- the exact defect that shipped in 0.1.0. Verified: sdist builds, installs into a clean venv with nvcc off PATH, cuda_available() is true, and all 45 tests pass against the installed package.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The 0.1.0 sdist on PyPI carries no CUDA sources, so
pip install culofcould never build. This prepares 0.2.0 and fixes the two things that made the published package unusable..cu, header andCMakeLists.txt; the release workflow fails if any is missing.PATH; the usual toolkit layout (/usr/local/cuda, driver-only on PATH) aborted atproject()on a machine with a working A6000. Now the standard roots are checked first.Verified end to end: sdist builds, installs into a clean venv with nvcc off PATH,
cuda_available()is true, 45/45 tests pass against the installed package.