GH-49227: [Python] Deprecate pyarrow.gandiva

[AlenkaF]

Rationale for this change

See: #49227 (comment)

What changes are included in this PR?

Deprecation warnings for pyarrow.gandiva added for version 24.0.0

Are these changes tested?

Yes, tested locally that warnings are raised on import. CI tests verify the warnings are filtered in out test suite.

Are there any user-facing changes?

Yes, pyarrow.gandiva is deprecated and will be removed in a feature PyArrow version.

• GitHub Issue: [Python] Deprecate pyarrow.gandiva #49227

[raulcd]

@pitrou what are your thoughts here? Should we merge this as part of 24.0.0 given no response to the ML or should we push it to 25.0.0 and "give time" to users? Maybe we should bump the ML thread and potentially share in [email protected] before merging?

[AlenkaF]

Well, there really is no need to rush, so I would agree with waiting till 25.0.0 and adding more attention to the ML thread before merging. (Edit: sorry, I am sure Antoine has not been tagged when I was first reading this =) )

[pitrou]

This is a deprecation, so we want this to be merged for 24.0.0. Waiting for another version wouldn't do any good, as people would not be informed before the deprecation appears.

[raulcd]

Waiting for another version wouldn't do any good, as people would not be informed before the deprecation appears.

Well, it depends. I was asking because I was thinking whether we want to bump the ML thread to give more visibility for any concern from users before deprecating. If we have decided, thanks to lazy consensus, that we are deprecating, then, yes I do agree we should merge it for 24.0.0.

[raulcd]

Based on the discussion on the issue and the ML. I agree we seem to be in agreement and no-one answered the ML. So by the power of lazy consensus I approve :)

[pitrou]

We can always undo the deprecation later, so there is really no downside to doing this in 24.0.0 IMHO.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 82847d8.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 8 possible false positives for unstable benchmarks that are known to sometimes produce them.