BUG: DTA/TDA/PA/Series/Index.view with datetimelike #39788
Conversation
jbrockmendel
changed the title
| if isinstance(dtype, (np.dtype, ExtensionDtype)) and needs_i8_conversion( | ||
| dtype | ||
| ): | ||
| if dtype.kind == "m" and dtype != "m8[ns]": |
There was a problem hiding this comment.
you can handle all m and M types via this branch?
There was a problem hiding this comment.
L757-L759 is for non-nano td64. L754-L765 handles all td64, dt64, dt64tz, and perioddtype
There was a problem hiding this comment.
i think this could use a little more commentary here (similar to what you did on the datetimelike astype), pls catch in a followon
jreback
added
Dtype Conversions
| if dtype is None or dtype is self.dtype: | ||
| return type(self)(self._ndarray, dtype=self.dtype) | ||
|
|
||
| if isinstance(dtype, type): |
There was a problem hiding this comment.
i think make this clear that M8[ns] (or really M8[s]) or whatever will hit here, but a tz-aware / pandas dtype will pass thru. yes after i read it i get it, but in 3 months no-one will remember.
There was a problem hiding this comment.
fleshing out the comment bc its pretty weird. this line isnt catching np.dtype, its catching type, e.g. we get here with dtype=np.ndarray
| if isinstance(dtype, (np.dtype, ExtensionDtype)) and needs_i8_conversion( | ||
| dtype | ||
| ): | ||
| if dtype.kind == "m" and dtype != "m8[ns]": |
There was a problem hiding this comment.
i think this could use a little more commentary here (similar to what you did on the datetimelike astype), pls catch in a followon
This makes it feasible to share some of the SetitemCastingEquivalents tests.