TYP: fix ignores #40412
TYP: fix ignores #40412
Conversation
can you confirm that you have numpy 1.20.1 in those envs. |
| arr = com.asarray_tuplesafe( | ||
| data, dtype=object # type: ignore[arg-type] | ||
| ) | ||
| arr = com.asarray_tuplesafe(data, dtype=np.dtype("object")) |
There was a problem hiding this comment.
do not change but I think dtype should be the string version to satisfy the typing in asarray_tuplesafe since not sure if there are perf issues that will accumulate creating a dtype object unnecessarily. but there does look like there may be an issue in asarray_tuplesafe since "object" in [np.object_, object] is False so the type annotations for asarray_tuplesafe maybe incorrect.
side note:
I think that we want to import the numpy type definitions for dtype in pandas._typing
in asarray_tuplesafe we use NpDtype from pandas._typing, which is not as permissive as the numpy alias, or what is actually accepted by numpy for dtype arguments (with the exception of object which is a false positive but is a special case since it creates issues with static typing)
There was a problem hiding this comment.
since not sure if there are perf issues that will accumulate creating a dtype object unnecessarily
my thought is that these will have to be converted to np.dtype objects somewhere in the call stack, so should be roughly perf neutral. no actual measurements on that though.
| # error: Incompatible types in assignment (expression has type | ||
| # "str", variable has type "Union[dtype[Any], ExtensionDtype]") | ||
| dtype = "float64" # type: ignore[assignment] | ||
| dtype = np.dtype("float64") |
There was a problem hiding this comment.
dtype is passed to astype which accepts a string? i'm not convinced this is the best approach to removing the ignores. maybe could add a type variable declaration to dtype = find_common_type(... to avoid the Incompatible types in assignment... error.
There was a problem hiding this comment.
Thanks @jbrockmendel. In general, I prefer not to change code to appease mypy but i'll defer to you on this assessment.
that fixed it, thanks |
cc @simonjayhawkins
Locally when I run mypy on master I get 427 errors mostly "unused 'type: ignore' comment". Same result when i switch from py39 to py38, or from OSX to Ubuntu. mypy==0.812 in all cases. Any guesses what config the CI has different?