Add missing meta kwarg to gufunc class #7423
Conversation
|
Reposting #7384 (comment) here to provide a brief explanation on what this PR is attempting to resolve:
|
There was a problem hiding this comment.
Thank you for writing this up @pentschev and sorry I didn't review it sooner. It seems to me that this is a clear improvement and if we want to finesse how other kwargs are handled in the future we can reasonably do that as well. Having the meta as a kwarg is a good escape hatch.
| @@ -560,6 +560,9 @@ class gufunc: | |||
| dimensions are to consist only of one chunk. | |||
| Warning: enabling this can increase memory usage significantly. | |||
| Defaults to ``False``. | |||
| meta: Optional, tuple, keyword only | |||
| tuple of empty ndarrays describing the shape and dtype of the output of the gufunc. | |||
There was a problem hiding this comment.
| tuple of empty ndarrays describing the shape and dtype of the output of the gufunc. | |
| tuple of empty ndarrays describing the type of array and dtype of the output of the gufunc. |
| @@ -602,6 +616,7 @@ def __init__(self, pyfunc, **kwargs): | |||
| self.output_sizes = kwargs.pop("output_sizes", None) | |||
| self.output_dtypes = kwargs.pop("output_dtypes", None) | |||
| self.allow_rechunk = kwargs.pop("allow_rechunk", False) | |||
| self.meta = kwargs.pop("meta", None) | |||
There was a problem hiding this comment.
Should we check, and throw a warning in case of, mismatch between output_dtypes and the dtypes as given in meta?
|
@jsignell Oops, I had still noticed two small improvements to this PR... Sorry, for not giving input sooner, had other stuff on my plate. Just too late I guess. |
|
No worries, you are welcome to open another PR with these changes if you like :) |
black dask/flake8 dask