On Mon, 2023-03-13 at 11:22 +0100, Sebastian Berg wrote: > Hi all, > > Without concerns voiced, I will probbaly merge the PR: > > https://github.com/numpy/numpy/pull/23240
I put this PR in, if it turns out there are issues or general dislike, please let us know and we can reconsider. - Sebastian > > soon. It proposes to dispatch also on the `where=` argument to > ufuncs > for __array_ufunc__. This allows a use-case of an "uncertainty" > boolean array (which is not really boolean). > I guess there may be other use-cases, although I am not sure. It > thus > may be a bit narrow in scope. > > The main effect of this should be that implementors of > __array_ufunc__ > may have to add logic to check for `where` in the kwargs > (eventually), > otherwise a user passing their array type as `where=` might run into > issues (for some types). > > That is slightly tedious, but I doubt it causes much issues in > practice > (where isn't used much, and for most types it doesn't matter much > either way, except for rejecting weirder inputs). > > Cheers, > > Sebastian > > _______________________________________________ > NumPy-Discussion mailing list -- numpy-discussion@python.org > To unsubscribe send an email to numpy-discussion-le...@python.org > https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ > Member address: sebast...@sipsolutions.net > _______________________________________________ NumPy-Discussion mailing list -- numpy-discussion@python.org To unsubscribe send an email to numpy-discussion-le...@python.org https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ Member address: arch...@mail-archive.com
