API: dont infer freq in DTA/TDA arithmetic ops #33487
Conversation
| @@ -865,7 +865,7 @@ def test_dt64arr_add_sub_td64ndarray(self, tz_naive_fixture, box_with_array): | |||
| tdi = pd.TimedeltaIndex(["-1 Day", "-1 Day", "-1 Day"]) | |||
| tdarr = tdi.values | |||
|
|
|||
| expected = pd.date_range("2015-12-31", periods=3, tz=tz) | |||
| expected = pd.date_range("2015-12-31", periods=3, tz=tz)._with_freq(None) | |||
There was a problem hiding this comment.
can you not just pass freq='None' ?
| for result in [idx - delta, np.subtract(idx, delta)]: | ||
| assert isinstance(result, DatetimeIndex) | ||
| tm.assert_index_equal(result, exp) | ||
| assert result.freq == "2D" | ||
|
|
||
| # When adding/subtracting an ndarray (which has no .freq), the result | ||
| # does not infer freq |
There was a problem hiding this comment.
i think u need to pass freq=None when constructing th index rather than using a private method
There was a problem hiding this comment.
yah, first we need to fix the DTI constructor so that if you explicitly pass freq=None it wont override the arg's freq, ATM we get:
>>> dti = pd.date_range("2016-01-01", periods=3)
>>> pd.DatetimeIndex(dti, freq=None).freq
<Day>
There was a problem hiding this comment.
yeah i guess freq should be a no_default arg s can handle None
There was a problem hiding this comment.
yah. ill make a PR for that in the AM
There was a problem hiding this comment.
working on the freq=no_default branch and its hairier than anticipated. if we can push this through in the interim that'll help me keep momentum on the "check freq in assert_index_equal" project
There was a problem hiding this comment.
ok sure, yeah like to avoid using _with_freq in tests at all
freq inference is pure overhead for the Series/DataFrame op because freq gets discarded.