BUG: DatetimeArray._from_sequence accepting bool dtype #32668
Conversation
| with pytest.raises( | ||
| ValueError, match="The dtype of 'values' is incorrect.*bool" | ||
| ): | ||
| DatetimeArray(np.array([1, 2, 3], dtype="bool")) | ||
| DatetimeArray(arr) |
There was a problem hiding this comment.
why is this a ValueError when DatetimeArray._from_sequence(arr) is a TypeError? Is is also worth considering making error messages consistent?
There was a problem hiding this comment.
originally i made it type error, then changed for consistenty with... something, but i dont remember what. will try to figure that out and report bakc
jreback
added
Dtype Conversions
|
looks fine. need a whatsnew as a bug fix for pd.to_datetime erroneously accepting bools |
| try: | ||
| arg, _ = maybe_convert_dtype(arg, copy=False) | ||
| except TypeError: | ||
| if errors == "coerce": |
There was a problem hiding this comment.
actually do we have coverage for all of these paths? i seem to recall we using similar code elsewhere?
There was a problem hiding this comment.
yes, we have tests in tests.tools.test_to_datetime that hit all of these cases
There was a problem hiding this comment.
kk
also maybe able to de-duplicate some of this code
|
can you rebase your PRs again, now that azure if running. (likely not a big deal, but just to make sure) |
|
thanks |
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diffIdentified when making the change this makes in Series.