PEP: pandas/core round 3 (generic, groupby, index)

[rockg]

yapf with manual modifications for error strings and bad yapf formatting.

[rockg]

No github diffs again!

There are some pieces of code that I think can be removed but want to confirm:
index.py i/j are not used anywhere.
groupby.py how can this work? agg_func is not defined.

[jreback]

index i/j ok to remove

[jreback]

yeh that looks not tested, it for a transform on a Panel or greater. I think just raise NotImplemented (if the ndim >= 3) & remove the 2nd clause (where you noted). If nothing breaks in the test suite then ok!

[rockg]

I assume it should be transform_func(result[:, :, i], chunk, comp_ids, accum)

[rockg]

k, I can do that.

[jorisvandenbossche]

@rockg I would keep style changes and code clean-up (as in removal of unused code or a change of actual code) strictly separated (at least in a separate commit). Just to be sure, in case we overlooked something, then it is not burried in a gigantic commit for which you can't see the diff.

[rockg]

The main goal of this exercise is to make flake8 pass and cleanup some style were necessary. What I cleaned out were unused variables reported by flake8 and, as a result, I don't see a problem including that here.

[wesm]

@rockg I've been putting # noqa on these lines and adding a TODO comment to come back and investigate

[wesm]

This is especially important in unit tests where I suspect that there are some test cases that may have skipped some intended checks

[rockg]

Okay, that seems reasonable.

[jreback]

ok, rebasing core/groupby.py is going to be painful for #11841 but go ahead (my touches are mostly reorg anyhow)

[wesm]

Is this good to merge?

[wesm]

I'll leave this to @jreback with the pending rebase

[jreback]

if u want to remove groupby.py from this would be great (I'll catch that in resample fixes)

otherwise lgtm

[wesm]

thanks! we're almost there =)