linux-aarch64 / osx-arm64 builds

[conda-forge-admin]

Hi! This is the friendly automated conda-forge-webservice.

I've started rerendering the recipe as instructed in #3.

If I find any needed changes to the recipe, I'll push them to this PR shortly. Thank you for waiting!

Here's a checklist to do before merging.

• Bump the build number if needed

Fixes #3

[conda-forge-admin]

Hi! This is the friendly automated conda-forge-linting service.

I failed to even lint the recipe, probably because of a conda-smithy bug 😢. This likely indicates a problem in your meta.yaml, though. To get a traceback to help figure out what's going on, install conda-smithy and run conda smithy recipe-lint --conda-forge . from the recipe directory.

_{This message was generated by GitHub actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/11420727659.}

[Tobias-Fischer]

@conda-forge-admin please rerender

[Tobias-Fischer]

@conda-forge-admin please rerender

[Tobias-Fischer]

@h-vetinari do you have any clue why CI is failing to checkout the code?

[h-vetinari]

@h-vetinari do you have any clue why CI is failing to checkout the code?

No idea TBH, never seen that failure before...

@conda-forge-admin, please rerender

[h-vetinari]

On aarch, you'll need something like conda-forge/faiss-split-feedstock@36fbef2 (until we solve conda-forge/conda-forge-pinning-feedstock#6283)

[Tobias-Fischer]

@conda-forge-admin, please rerender

[Tobias-Fischer]

@conda-forge-admin, please rerender

[conda-forge-admin]

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

_{This message was generated by GitHub actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/11415225410.}

[Tobias-Fischer]

Do we know which package contains libcuda.so.1? It cannot be loaded in the kaldi related tests on linux-64, something must have changed in the latest kaldi builds

[h-vetinari]

Do we know which package contains libcuda.so.1?

That's one of the only libraries not packaged by conda-forge (due to licensing reasons AFAIU, even in the new CUDA 12+ stack). We'll need to skip the tests that require loading it. Presumably the relevant paths aren't being exercised on the kaldi feedstock.

The only alternative would be to switch to running on https://github.com/Quansight/open-gpu-server/, which has actual GPU hardware with the drivers available. That would be conceivable if too large a percentage of functionality of the package depends on loading libcuda, and we wouldn't want to skip all that.

CC @conda-forge/cuda-nvcc

[h-vetinari]

I just had another thought why the tests passed before: it probably works fine if we use the CPU-only version of kaldi. That would make sense as a test-only restriction.

[Tobias-Fischer]

This is all green now @h-vetinari - please go ahead with your preferred way of pulling in kaldi and then I think we should be ready to merge