Open jdumas opened 1 week ago
Actually I would've expected the env variable to be set by the CUDA activation script.
These should be passed to the PyPI build. @baszalmstra do you know something about the 'CUDA_HOME' env variable?
Im also pretty sure this should be aet by the activation script. Could you check by going into the shell and echoing it?
@tdejager do we actually run the activation script and pass those variables to uv?
We do pass env variables to uv for sure, otherwise things like c compilers wouldn't work also. And I believe we use the 'activator' for that.
If I remove the surfel
package, after entering the pixi shell
running echo $env:CUDA_HOME
shows as empty. I have no idea how the manual pip install
is able to succeed. Do you think it's a problem with uv
itself?
EDIT: I've opened https://github.com/astral-sh/uv/issues/7299 after trying to build this dependency with uv
directly.
Checks
pixi --version
.Reproducible example
Issue description
Hi. Consider the provided
pixi.toml
file. I find that I cannot build the environment on my Windows machine. Specifically it fails to build thepip
package from source, with the following error message:I am on pixi 0.29.0. I find that if I run the
pip install
step as separate command (viapixi run surfel-install
), then I can build the CUDA kernels as expected. I would appreciate any guidance you have on this particular issue. I couldn't find any other issue related to a missingCUDA_HOME
env variable. The closest I could find is this conda issue (https://github.com/conda/conda/issues/7757). Since I'm able to build by runningpixi run surfel-install
, it seems it's a bug/missing feature in pixi but let me know if I'm wrong.Expected behavior
Can build provided
pixi.toml
file.