build libwholegraph docs in CI #96
Conversation
jameslamb
added
doc
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
|
/ok to test |
|
/ok to test |
|
/ok to test |
jameslamb
changed the title
|
Looks like this is working as expected:
|
.github/workflows/build.yaml
Outdated
| arch: "amd64" | ||
| branch: ${{ inputs.branch }} | ||
| build_type: ${{ inputs.build_type || 'branch' }} | ||
| container_image: "rapidsai/ci-conda:cuda12.5.1-ubuntu22.04-py3.11" |
There was a problem hiding this comment.
Can we use the latest tag? I think it's an oversight that we are not currently using rapidsai/ci-conda:latest for all our docs builds. We switched to a specific tag for a short time during the Python 3.12 migration, but now we should be able to use latest again. (Don't block on this if this PR is holding up other work.)
There was a problem hiding this comment.
cugraph actually is using :latest and I intentionally chose a specific tag here (though should have picked Python 3.12, that's an oversight).
I was thinking that a specific tag is helpful to be improve the probability that a branch is buildable for hotfixes even a few weeks after a release, if ci-conda:latest by that time points to a CUDA version or operating system or Python version that that release didn't support.
The tradeoff is that you get things like my mistake here, where the image being used isn't the latest one that the library could support.
Given that, do you still think :latest?
Don't block on this if this PR is holding up other work.
Yeah we need this PR in to proceed with cugraph-docs, so I'll merge it whenever the next CI builds passes and put up a follow-up with whatever consensus we reach on the image tag.
There was a problem hiding this comment.
It doesn’t really matter. If you want to pin it, that is fine, but we just need to be responsible for keeping those pins up to date. We also drop versions all the time (we may not need 12.5.1 in our test matrix at some future time) which can break this without warning. There is always a latest tag.
There was a problem hiding this comment.
Yeah fair point. Changed my mind about waiting for a follow-up PR... jobs weren't actually that far along yet so I just pushed this change here: 5df836b
|
/merge |
The branch build triggered by merging #96 failed immediately. > The workflow is not valid. .github/workflows/build.yaml (Line: 47, Col: 12): Job 'docs-build' depends on unknown job 'conda-cpp-build'. ([build link](https://github.com/rapidsai/cugraph-gnn/actions/runs/12379736454)) This fixes that. Authors: - James Lamb (https://github.com/jameslamb) Approvers: - Don Acosta (https://github.com/acostadon) - Bradley Dice (https://github.com/bdice) URL: #97
Fixes #94
Uploads API docs for
libwholegraph, to be used by rapidsai/cugraph-docs#46Also removes
sphinxdependencies... this repo only needs to produce Doxygen docs forlibwholegraph, all the other Sphinx stuff will be done in https://github.com/rapidsai/cugraph-docs.