API: Index.__getitem__(True) returns 2D ndarray

[jbrockmendel]

• 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 master branch of pandas.

Reproducible Example

Not clear if this is intentional:


import numpy as np
import pandas as pd

arr = np.random.randn(4)
idx = pd.Index(arr)

idx[True]

>>> idx[True]
array([[ 1.39365849, -0.03655461,  0.17112114, -1.4612337 ]])

Issue Description

For other cases where indexing on index._data returns 2D (e.g. idx[:, None]) we issue a FutureWarning, but this goes through a different code path.

Expected Behavior

I don't have an opinion here, just think whatever we do should be intentional

Installed Versions

Replace this line with the output of pd.show_versions()

[jorisvandenbossche]

To be explicit for other readers, this is following the numpy behaviour:

In [2]: arr = np.array([1, 2, 3])

In [3]: arr[True]
Out[3]: array([[1, 2, 3]])

(I don't fully understand the rationale of the numpy behaviour here, though ..)

Personally I think we should be consistent with the other cases that currently returns a 2D array -> deprecate this and disallow it in the future.