Skip to content

Fix various pandas issues using new debugging skill - #22626

Closed
vyasr wants to merge 31 commits into
NVIDIA:mainfrom
vyasr:fix/more_pandas_improvements
Closed

Fix various pandas issues using new debugging skill#22626
vyasr wants to merge 31 commits into
NVIDIA:mainfrom
vyasr:fix/more_pandas_improvements

Conversation

@vyasr

@vyasr vyasr commented May 21, 2026

Copy link
Copy Markdown
Contributor

Description

Applies #22625 to fix various issues with pandas tests discovered locally.

Checklist

  • I am familiar with the Contributing Guidelines.
  • New or existing tests cover these changes.
  • The documentation is up to date with these changes.

@vyasr vyasr self-assigned this May 21, 2026
@vyasr vyasr added bug Something isn't working non-breaking Non-breaking change labels May 21, 2026
@copy-pr-bot

copy-pr-bot Bot commented May 21, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@github-actions github-actions Bot added Python Affects Python cuDF API. cudf.pandas Issues specific to cudf.pandas labels May 21, 2026
@GPUtester GPUtester moved this to In Progress in cuDF Python May 21, 2026
vyasr added 22 commits May 28, 2026 23:51
GroupBy._grouped passed internal storage keys (self.obj._index_names)
to _from_columns_like_self, which for a nameless MultiIndex produced
[0, 1] instead of the correct public names [None, None]. This caused
GroupBy.apply to return results with wrong index level names, making
the cudf.pandas accelerated test_cython_transform_frame_column fail
when comparing apply() expected values against transform() results.

Use self.obj.index.names (public metadata) instead.

Fixes the following failing pandas tests:
- tests/groupby/transform/test_transform.py::test_cython_transform_frame_column[*-frame_mi-*]
These tests assert that pandas raises NotImplementedError when using
engine='numba' with as_index=False for groupby reductions. cuDF computes
reductions on GPU without numba, so the pandas-specific limitation does
not apply and the operation succeeds. The divergence is intentional.

Skips the following pandas tests:
- tests/groupby/test_numba.py::TestEngine::test_as_index_false_unsupported[*]
Fix pyarrow-backed StringDtype str.match results to match pandas nullable boolean dtype and end-of-string anchor semantics.
Fixes the following failing pandas test:

- tests/strings/test_string_array.py::test_string_array_extract[string[python]]
Fixes the following failing pandas tests:
- tests/reshape/concat/test_append.py::TestAppend::test_append_different_columns_types[Index-MultiIndex2]
- tests/reshape/concat/test_append.py::TestAppend::test_append_different_columns_types[CategoricalIndex-MultiIndex0]
- tests/reshape/concat/test_append.py::TestAppend::test_append_different_columns_types[CategoricalIndex-MultiIndex1]
@vyasr
vyasr force-pushed the fix/more_pandas_improvements branch from a50a5e1 to 69b09cc Compare May 28, 2026 23:57
@copy-pr-bot

copy-pr-bot Bot commented May 28, 2026

Copy link
Copy Markdown

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.

@vyasr

vyasr commented Jun 16, 2026

Copy link
Copy Markdown
Contributor Author

Closing since the contents of this branch have been moved to other PRs.

@vyasr vyasr closed this Jun 16, 2026
@github-project-automation github-project-automation Bot moved this from In Progress to Done in cuDF Python Jun 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working cudf.pandas Issues specific to cudf.pandas non-breaking Non-breaking change Python Affects Python cuDF API.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants