Remove ucx dependency for now

[jorisvandenbossche]

See #962 (comment)

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

[conda-forge-webservices]

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

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[h-vetinari]

Thanks for the PR. You'll have to adapt build.sh as well (it might work without adaptation but it would be confusing). The commit history resp. git blame should be relatively self-explanatory I hope. :)

[jorisvandenbossche]

@conda-forge-admin, please rerender

[jakirkham]

cc @kkraus14

[jakirkham]

I'm hoping we can solve this on the ucx end instead of needing this change. There is a 1.14.0 release coming up, which would be a good time to coordinate package changes.

We already have a change for libnuma in the queue ( conda-forge/ucx-split-feedstock#112 ).

For example issue ( conda-forge/ucx-split-feedstock#66 ) proposes folding the CPU & GPU builds into one build and then making cudatoolkit a run_constrained dependency (so optional) instead of run (required), which drops the need for the selector and stops pulling in cudatoolkit (by default)

[kkraus14]

I agree that I'd much prefer to fix the UCX package than disable it in the libarrow builds if possible, but understand if there's a short term fix we need to make here in the meantime.

[jorisvandenbossche]

Yes, this PR is only meant for "until the ucx issues is solved". But to me it seems quite unsure what would be the exact timeline for getting those fixes in for ucx 1.14 (also just being unfamiliar with the package, I can't judge that), while I would consider the pyarrow issue as a serious regression in user experience (hugely blowing up every environment with pyarrow on linux), something we should fix on the short-term IMO.

[jorisvandenbossche]

The timeout on aarch64, is that a known issue? Or could someone restart that build?

[h-vetinari]

Let's undo this regression (even though somewhat subjective), and then readd it once ucx is ready.

[github-actions]

Hi! This is the friendly conda-forge automerge bot!

I considered the following status checks when analyzing this PR:

• linter: passed
• azure: passed

Thus the PR was passing and merged! Have a great day!

[h-vetinari]

FYI: now that conda-forge/ucx-split-feedstock#111 has landed, I added a PR to re-enable ucx support in #1011