BUG: concat(axis=1) sorts by default for DatetimeIndex

[lukemanley]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

a = pd.Series(1, index=pd.to_datetime(['2000-01-01', '2000-01-03']))
b = pd.Series(2, index=pd.to_datetime(['2000-01-02', '2000-01-04']))

pd.concat([a, b], axis=1)

Issue Description

concat documentation states concat will not sort by default. I believe the default sort behavior was deprecated some time ago. However, there is still a sort=True behavior for DatetimeIndex:

              0    1
2000-01-01  1.0  NaN
2000-01-02  NaN  2.0
2000-01-03  1.0  NaN
2000-01-04  NaN  2.0

The source of this behavior is a hard-coded special-casing in pandas.core.indexes.api.union_indexes. Making this not sort by default seems to break at least one test (test_concat_datetime_timezone) and possibly more.

Expected Behavior

Per the docs, I would expect no sort by default. But that might surprise users if changed. I suppose this either needs to be documented (if intentional) or deprecated (if not intentional).

Installed Versions

.

[kshitiz305]

Hi I am a first time contributor would like to know if I shall update the docs or shall make changes to sort the bug

[lukemanley]

Hi I am a first time contributor would like to know if I shall update the docs or shall make changes to sort the bug

Hey @kshitiz305, I think this needs a bit more discussion before determining which path to take.

[lukemanley]

@phofl - just noticed you had opened #51683 for this issue. Any particular reason for closing it?

[phofl]

That kind of fell off my radar back then. You can continue if you are interested, the fix itself should be good