REF: dont _try_cast for user-defined functions #29698
Conversation
|
Can you add test(s) for this? |
This is turning out to be surprisingly difficult. Which suggests its all-the-more worthwhile |
So far the answer is "no". i haven't been able to find a case that fails in master but works in this PR. |
|
How did you come across the behavior in the whatsnew? |
Thats the whatsnew note I forgot to include in #29641. |
|
Ah OK. So not necessarily tied to the rest of the PR? |
Right. ditto for the comment in core.apply |
doc/source/whatsnew/v1.0.0.rst
Outdated
| @@ -439,6 +439,7 @@ Groupby/resample/rolling | |||
| - Bug in :meth:`DataFrameGroupby.agg` not able to use lambda function with named aggregation (:issue:`27519`) | |||
| - Bug in :meth:`DataFrame.groupby` losing column name information when grouping by a categorical column (:issue:`28787`) | |||
| - Bug in :meth:`DataFrameGroupBy.rolling().quantile()` ignoring ``interpolation`` keyword argument (:issue:`28779`) | |||
| - Bug in :meth:`DataFrameGroupBy.agg` with timezone-aware datetime64 column incorrect casting results to the original dtype (:issue:`29641`) | |||
There was a problem hiding this comment.
can you add a sub-section in api-changes. this is a non-trivial change i think.
There was a problem hiding this comment.
sure, but since this note is orthogonal to this PR, im going to do that in a separate PR. removing this note for now.
|
thanks |
Also add a whatsnew note for #29641 and a comment in core.apply that ive been meaning to get in