kubeflow / pipelines

Machine Learning Pipelines for Kubeflow
https://www.kubeflow.org/docs/components/pipelines/
Apache License 2.0
3.62k stars 1.63k forks source link

[feature] SDK Support for Python 3.13 #11365

Closed mynewestgitaccount closed 1 week ago

mynewestgitaccount commented 2 weeks ago

Feature Area

/area sdk

This may apply to other areas too. I only care/know about the SDK.

What feature would you like to see?

Support for Python 3.13 (released October 7, 2024).

What is the use case or pain point?

Python 3.13 is not yet supported, so the latest version of the kfp package enforces <3.13.0. Please support Python 3.13 and increase the upper bound to allow it.

https://github.com/kubeflow/pipelines/blob/1708ce79bbfe1fafc9527769b49d87cd38558439/sdk/python/setup.py#L100

Separate but related issue: The upper bound should be removed entirely, regardless of what Python versions you actually support.

Upper bounds in python_requires create more issues than they solve (see this Python packaging discussion for more information). Instead, Python version support should be indicated exclusively through the Programming Language :: Python :: 3.XX Trove classifiers.

Here, running pip install kfp in a Python 3.13 environment does not fail as the current python_requires might lead you to expect. Instead, it results in an install of kfp==1.8.22 (the last version before the upper bound was added), and issues in such an old version are even harder to understand.

Is there a workaround currently?

No.


Love this idea? Give it a 👍.