Rebuild for python 3.14

[regro-cf-autotick-bot]

This PR has been triggered in an effort to update python314.

Notes and instructions for merging this PR:

1. Please merge the PR only after the tests have passed.
2. Feel free to push to the bot's branch to update this PR if needed.

Please note that if you close this PR we presume that the feedstock has been rebuilt, so if you are going to perform the rebuild yourself don't close this PR until the your rebuild has been merged.

---

If this PR was opened in error or needs to be updated please add the bot-rerun label to this PR. The bot will close this PR and schedule another one. If you do not have permissions to add this label, you can use the phrase @conda-forge-admin, please rerun bot in a PR comment to have the conda-forge-admin add it for you.

_{This PR was created by the regro-cf-autotick-bot. The regro-cf-autotick-bot is a service to automatically track the dependency graph, migrate packages, and propose package version updates for conda-forge. Feel free to drop us a line if there are any issues! This PR was generated by https://github.com/regro/cf-scripts/actions/runs/18236997477 - please use this URL for debugging.}

[conda-forge-admin]

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/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

• ℹ️ The recipe is not parsable by parser conda-recipe-manager. The recipe can only be automatically migrated to the new v1 format if it is parseable by conda-recipe-manager.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/19284509137. Examine the logs at this URL for more detail.}

[xhochy]

@setu4993 What's the reason you need to change the requirements? Are they already wrong in the non-migrated version?

[setu4993]

@setu4993 What's the reason you need to change the requirements? Are they already wrong in the non-migrated version?

Yeah, the datasets dependency was pinned to a very old version, which was causing some of the CI failures. I think I can revert the other changes.

Looking through what's happening, I believe we'll need multiprocess v0.70.17 to be published for Python 3.14, because that is pinned on datasets: https://github.com/huggingface/datasets/blob/main/setup.py#L127

[setu4993]

Don't need this anymore because that is from a few months years ago.

[setu4993]

Blocked on: conda-forge/multiprocess-feedstock#57

[setu4993]

This should have been fixed earlier.

[h-vetinari]

I believe we'll need multiprocess v0.70.17 to be published for Python 3.14, because that is pinned on datasets: https://github.com/huggingface/datasets/blob/main/setup.py#L127

I think we should patch out a too-tight pin from datasets rather than mess around with migrating for old multiprocess versions

[setu4993]

I believe we'll need multiprocess v0.70.17 to be published for Python 3.14, because that is pinned on datasets: https://github.com/huggingface/datasets/blob/main/setup.py#L127

I think we should patch out a too-tight pin from datasets rather than mess around with migrating for old multiprocess versions

I don't quite understand how that'd work, but that sounds like a better option, yes.

This is also only for tests, and not a build / run dependency, so I'm also wondering if we can just skip those tests for Python 3.14.

[setu4993]

Dependencies are good now, but the tests are failing because of huggingface/datasets#7813, which is downstream of uqfoundation/dill#725.

[setu4993]

WIP fixes in: huggingface/datasets#7817, uqfoundation/dill#724.

[Ahajha]

Question for anyone, why do these rebuilds even need to happen? Tokenizers build for the 3.9 stable ABI, so theoretically one build should work for any python version? Do we have to have separate packages?

[setu4993]

Question for anyone, why do these rebuilds even need to happen? Tokenizers build for the 3.9 stable ABI, so theoretically one build should work for any python version? Do we have to have separate packages?

I'm not an expert in conda builds, but my understanding is that because these are arch specific builds, and require builds with a specific Python version, they get built and packaged for that version independently. Other recipes that are noarch aren't subject to it.

Either way, the issue here is that the unit tests for the underlying package are failing on Python 3.14 because of an incompatibility with one of the dependencies. So, even if these conda package gets published, some parts of the package will propagate bugs / exceptions.

IMHO, it seems preferable to not publish this package for 3.14 until the underlying issue is resolved.

[h-vetinari]

Tokenizers build for the 3.9 stable ABI, so theoretically one build should work for any python version?

That's great, thanks for the comment! You can try to follow the docs on how to do that.

[reneleonhardt]

The test used code which wasn't released 3 weeks ago, a new run should fix that problem.
https://dev.azure.com/conda-forge/84710dde-1620-425b-80d0-4cf5baca359d/_apis/build/builds/1370818/logs/109

TestTrainFromIterators.test_datasets

>      self._batch_setitems(obj.items(), obj)
E      TypeError: Pickler._batch_setitems() takes 2 positional arguments but 3 were given

huggingface/datasets#7817
https://github.com/huggingface/datasets/releases/tag/4.4.0

[setu4993]

Thanks for the nudge, @reneleonhardt. Just bumped up the version and retrying CI now.