Evgeni Burovski
Evgeni Burovski
@j-bowhay no, just an oversight. Would be nice to add it here, if only to avoid this question going forward.
Ugh. In that test, there is `assert_equal(v.shape, (7, 11, 6)`, which seems to hint that the problem is a missed flatten/reshape somewhere. On a side note, it's truly humbling to...
> Actually scratch this completely have made some progress Great! Let me know if you want me to do some debugging (was about to fire up pdb but happy to...
Thanks for the reminder @j-bowhay . Will do when I'm back, will be a couple days though.
@j-bowhay apologies it took me this long to circle back to it! I'm starting to look at this at last, and I think I understand the purpose of this test,...
One more obscure glitch masking one more problem in RGI, cf gh-16780. Might make sense to handle it quickly and rebase this one on top.
https://github.com/j-bowhay/scipy/pull/1 seems to fix failures in current tests, can you take it over @j-bowhay ? Please do check it though, this stuff is confusing so a second pair of eyes...
Okay, git conflicts block the CI. Let me know if you want a hand resolving them. Now or once you add stuff already in your pipeline.
> What needs to be added to the document for RGI.__call__? I did not manage to figure the shape of the result and it's relationship to values.shape and xi.shape without...
The PR LGTM, the CI failure in linalg/test_decomp is unrelated. 27 commits is a bit too much here, needs some squashing. We typically prefer to not squash-merge multi-author PRs, but...