BUG: eval and query not working with ea dtypes #50764
Conversation
mroeschke
added
expressions
|
cc @mroeschke could you have a look? Remaining failures were only because pyarrow was not installed in the pipelines in question |
pandas/core/computation/common.py
Outdated
| non_ea_dtypes.append(arr_or_dtype) | ||
|
|
||
| if non_ea_dtypes: | ||
| np_dtype = np.result_type(*non_ea_dtypes) |
There was a problem hiding this comment.
Could this hit the except ValueError: as above?
There was a problem hiding this comment.
Not totally sure, we don't have a test hitting the ValueError above, but added it down here as well
| for elem in parsed_expr.terms.operand_types | ||
| ) | ||
| ): | ||
| engine = "python" |
There was a problem hiding this comment.
Does this mean if a user is specifying engine="numexpr", this switches the engine to python?
If so, I think it would be better to explicitly raise a NotImplementedError and tell the user to switch the engine manually.
There was a problem hiding this comment.
Yep, would you be ok with a RuntimeWarning? Since numexpr is the default, raising seems a bit noisy maybe?
There was a problem hiding this comment.
Well the issues linked already raise an exception currently, so a NotImplementedError would be similar but more explicit. Unless there's a case I'm not thinking of.
There was a problem hiding this comment.
I am more concerned with people using our nullable option who are switching over from NumPy. Would be annoying if query/eval stop working then.
But you are correct, an appropriate NotImplementedError would be an improvement
There was a problem hiding this comment.
That's fair. Yeah a RuntimeWarning warning & encouraging users to switch to the python engine would be okay
|
ci is green now |
|
Thanks @phofl |
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.This is really ugly, not sure if we want to do this. But could not find another way to make this work