BUG: lib.infer_dtype with incompatible intervals #41749
Conversation
| cdef: | ||
| IntervalValidator validator = IntervalValidator(len(values), | ||
| skipna=True) | ||
| return validator.validate(values) |
There was a problem hiding this comment.
can you remove the IntervalValidator code?
There was a problem hiding this comment.
sure, will add that to the upcoming misc branch
jreback
added
Interval
| return False | ||
| else: | ||
| raise ValueError(val) | ||
| elif util.is_nan(val) or val is None: |
There was a problem hiding this comment.
couldn't this be a float nan or NA? (or None)? i am not sure we really specify
There was a problem hiding this comment.
this matches what we do in corresponding libinterval function
There was a problem hiding this comment.
got it
ok i think we should open an issue about this (eg letting the missing value be NA) or NaT if it's a datetimelike in a new issue as this is a bit inconsistent with what we allow for missing values in other dtypes
but out of scope for this PR
pandas/_libs/lib.pyx
Outdated
| return False | ||
|
|
||
| if closed is None: | ||
| # we saw all-NaTs, no actual Intervals |
There was a problem hiding this comment.
yah, will update in the follow-on brnach
Nothing user-facing.