Asset Frame with duplicate index

[gabriellm1]

• closes assert_frame_equal with check_like throws exception if frames have repeat indices #26494
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

Hey guys, I started working on a way to make testing.assert_frame_equal accept an argument so it can test equals a frame with duplicate index just like the issue I marked to close here. For this I had to make a couple of changes in some functions and I think I got to the point that I should ask for a better direction and if this is valid for the project.

The error pointed in the issue is definitely not the best message for this case. I got to the point to not get that error by making clear there is a duplicate index. But I got stuck in this error now:

ValueError: Shape of passed values is (5, 1), indices imply (3, 1)

[pep8speaks]

Hello @gabriellm1! Thanks for opening this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file pandas/core/frame.py:

Line 3743:89: E501 line too long (97 > 88 characters)

• In the file pandas/core/generic.py:

Line 3832:89: E501 line too long (104 > 88 characters)

• In the file pandas/util/testing.py:

Line 1348:46: E231 missing whitespace after ','

[WillAyd]

Hmm not sure I am on board with making this change. The point of check_like is to ignore position but align on label. If labels are duplicated how would that alignment work?

[gabriellm1]

Hmmm now I see the problem. There would be multiple rows to align and check if equals. In a big dataframe that alignment solution would probably take too long to run. So probably there isn't a solution for this right?

[WillAyd]

Yea I think the alignment here as to what should be checked is ambiguous. Appreciate the PR as always but I don't think this one is worth pursuing (if anyone disagrees feel free to reopen)