API: Treat 1D vectors as columns in some cases for arith/cmp ops

[jbrockmendel]

Definitely needs discussion.

There are a bunch of GH issues (will track down the list in a bit) that boil down to users expecting DataFrame+(Series|ndarray|listlike) ops to operate column-by-column. This PR implements that behavior for a subset of cases. In particular:

• other is a Series: indexes match exactly (and columns do not)
• other is a list/tuple/np.ndarray: length matches df.index (and not df.columns)

This means that things like df == df['A'] will now work.

• closes #xxxx
• tests added / passed
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[pep8speaks]

Hello @jbrockmendel! Thanks for updating the PR.

• There are no PEP8 issues in the file pandas/core/frame.py !

• There are no PEP8 issues in the file pandas/core/ops.py !

• There are no PEP8 issues in the file pandas/tests/arithmetic/conftest.py !

• There are no PEP8 issues in the file pandas/tests/arithmetic/test_datetime64.py !

• There are no PEP8 issues in the file pandas/tests/arithmetic/test_period.py !

• There are no PEP8 issues in the file pandas/tests/arithmetic/test_timedelta64.py !

• There are no PEP8 issues in the file pandas/tests/frame/test_operators.py !

Comment last updated on October 24, 2018 at 02:56 Hours UTC

[jreback]

this is really confusing

we have .eq for this exact reason

pls justify why this should change

[jbrockmendel]

pls justify why this should change

I'm not 100% sure that it should. But it is only now feasible following #23132 and it does come up often enough to merit discussion. I'll try to fill out the list of Issues over the next couple days.

[jorisvandenbossche]

Can you first open an issue for this to propose a specific behaviour and discuss it?