BUG: Slicing of Index with strings does not raise anymore when keys are not in Index

[phofl]

The following

df2 = pd.DataFrame({"A": [1, 2, 3]}, index=['a', 'c', 'b'])
df2.loc['d':'e']

Should raise a KeyError. This worked in the past (see #18531) but does not anymore. Question is if we should raise in slice_locs if start and end are equal or if we should do this after calling the function.

cc @jreback

[jorisvandenbossche]

This still raises a KeyError for me. With what version do you see something else? And can you show the output?

[phofl]

Sorry, i applied the loc to the DatetimeIndex, which did not raise on master (as expected). Mixed this up... You are right this still raises

[phofl]

cc @jorisvandenbossche

Hm used the wrong example. The above raises, but when the index is monotonic, it does not raise.

df2 = pd.DataFrame({"A": [1, 2, 3]}, index=['a', 'b', 'c'])
df2.loc['d':'e']

This does not raise for me on master

[jorisvandenbossche]

As far as I know, that is the expected result, because it is a monotonic index, the exact labels in the slice don't need to be present in the index