MAINT: Remove the unused keepdim argument from np.ufunc.accumulate #8187
Conversation
numpy/core/src/umath/ufunc_object.c
Outdated
| static char *kwlist2[] = {"array", "indices", "axis", | ||
| "dtype", "out", NULL}; | ||
| static char *reduce_kwlist[] = { | ||
| "array", "axis", "dtype", "out", "keepdims", NULL}; |
There was a problem hiding this comment.
Umm, looks like you may have a hard tab in here. I preferred to previous format to tell the truth, although the alignment was a bit off on the second line.
There was a problem hiding this comment.
That's a set of spaces to align columns, to make visually diffing the arg lists easier.
The previous format was just indented with a bunch of soft tabs until it roughly aligned. I definitely think that having all the list items on one line is an improvement, but could be persuaded to remove that alignment
There was a problem hiding this comment.
Big no-no. The way it was before was correct, and quite standard for C code, just off by a single space. See the doc/C_STYLE_GUIDE.rst.txt.
There was a problem hiding this comment.
There are no hard tabs here:
There was a problem hiding this comment.
Reading that guide, the only mistake I seem to have made is indenting the wrapped line by one "tab" and not two. I don't see anything saying that if a line is wrapped, the first line should be made as long as possible. Which bit specifically are you calling a "no-no"?
There was a problem hiding this comment.
Multiple spaces after the ,. It looks just awful.
There was a problem hiding this comment.
You'd argue that that the alignment of argument names isn't worth the extra spacing then? Happy to remove if so, although I think having it on one line is still desirable. I've added the second "tab" of indentation
There was a problem hiding this comment.
Yes ;) I understand the temptation, but it is a false god. Along the same lines, PEP8 disapproves of aligning on equals signs, etc.
There was a problem hiding this comment.
There might be some exceptions for tabular test data, where having data aligned in columns may be useful, but here is is just a distraction to (my) eye.
numpy/add_newdocs.py
Outdated
| @@ -5919,6 +5919,13 @@ def luf(lamdaexpr, *args, **kwargs): | |||
| First array | |||
| B : array_like | |||
| Second array | |||
| dtype : data-type code, optional | |||
| The type used to represent the intermediate results. Defaults | |||
There was a problem hiding this comment.
I don't think this is correct. I believe what happens is that the arrays are converted to the common type, or dtype if given, the operation done, and then converted to the out type if the out argument is given.
There was a problem hiding this comment.
You can check that that is the case by using various combinations of dtype and out.
There was a problem hiding this comment.
Yep, blindly copied from another ufunc method, without thinking hard enough about whether it applied
There was a problem hiding this comment.
Totally missed the mark with this one - the arguments are just forwarded to the ufunc, including things like where which don't have all too sensible a result
|
Couple of style nits and probably the documentation is incorrect. The actual change looks OK. |
82a9c67 to
93fa698
Compare
|
Ok, should be good for another review |
93fa698 to
aedccb7
Compare
|
OK, this looks good. However, current uses of The temptation is to skip this process as it seems unlikely that the keyword is actually used by anyone, but it is best to follow procedure. |
|
I apologize for not noticing the problem earlier, it just slipped my mind. |
aedccb7 to
ee85ecc
Compare
ee85ecc to
a77d0e0
Compare
|
Ok, broken into three PRs:
|
a77d0e0 to
d15c75d
Compare
|
Ok, this one is all set for 1.14 |
|
OK, thanks. |
|
☔ The latest upstream changes (presumably #8302) made this pull request unmergeable. Please resolve the merge conflicts. |
|
☔ The latest upstream changes (presumably #9062) made this pull request unmergeable. Please resolve the merge conflicts. |
d4ea692 to
f44ef18
Compare
|
Rebased |
|
This would seem OK to merge now, so I'll go ahead and do that. Nice to get the code cleaner, even if it is by a little bit at the time! |
No description provided.