This data as json
| sha | message | author_date | committer_date | raw_author | raw_committer | repo | author | committer |
|---|---|---|---|---|---|---|---|---|
| db2d414c6af236be2df4f5ca353e154b2f4cf885 | `pandas=2.0` support (#7724) * unpin pandas in the package metadata * unpin pandas in the ci environments [skip-rtd] * create the input arrays in the parametrization * split `test_sel_float` into variants * skip the float16 variant if pandas>=2.0 is installed * [skip-rtd] * add tests for `days_in_month` and its alias * make sure the name and dtype match the expected * actually verify that the dtype stays the same * apply the dtype for non-dask * always use `int32` to follow `pandas=2.0` * back to `int64` * same for the test * final undo of `int64` → `int32` * update the comment to make more sense * simplify the conversion of the expected data * change back to the old condition * cast float16 to float64 when creating indexes (but warn anyways) * convert float16 to float64 when selecting using arrays (all except 0d) * move the float16 variant to a separate test this allows us to check for expected warnings * explicitly type the kwargs as a mapping of str → str * reword the warning message * restore the pin * [skip-ci] [skip-rtd] * rerun to make sure we don't introduce failures with `pandas<2` * changelog | 2023-04-12T13:04:11Z | 2023-04-12T13:04:11Z | f48b8f0d94dbaa4588e3a965a600d38e311eab17 | cd792325681cbad9f663f2879d8b69f1edbb678f | 13221727 | 14808389 | 19864447 |