CLN: assorted, mostly typing #29419
CLN: assorted, mostly typing #29419
Conversation
gfyoung
added
Clean
Internals
| @@ -971,7 +972,7 @@ def _cython_agg_blocks(self, how, alt=None, numeric_only=True, min_count=-1): | |||
| locs = block.mgr_locs.as_array | |||
| try: | |||
| result, _ = self.grouper.aggregate( | |||
| block.values, how, axis=agg_axis, min_count=min_count | |||
| block.values, how, axis=1, min_count=min_count | |||
There was a problem hiding this comment.
Interesting that we had essentially hardcoded this value all this time...
There was a problem hiding this comment.
I think all of these changes look okay, but I also feel that this PR covers a little too much ground and could have benefited from being broken up into smaller parts (e.g. typing, Cython refactoring).
I somewhat on the fence, so would be good to get some other eyes to look at this.
Totally reasonable. If others find this a deal-breaker I'll split this up. Otherwise I'll apply this advice in future CLN PRs |
|
this is ok, but generally yes would like to split largish changes. note to @WillAyd this might conflict with your groupby PR? |
diff_2d no longer needs to be in the pxi.in file, so moved it to the pyx
A couple of recently-identified bugs in the groupby code are caused by passing incorrect types, so im getting more motivated to add annotations in/around the affected code.