Update handling of breaking around binary operators: W503 vs W504

[IanLee1521]

Based on the discussion that started on python-dev [1] and ended with an update to PEP-8 [2][3], the logic for allowing (preferring really) breaks to occur before rather than after binary operators.

[1] https://mail.python.org/pipermail/python-ideas/2016-April/039774.html
[2] http://bugs.python.org/issue26763
[3] https://hg.python.org/peps/rev/3857909d7956

[hai-ld]

How should we implement this patch, interface-wise? There are two options off the top of my head:

• provide an option to choose the preferred breaking style, with the new style as default
• or warning when inspected files containing both styles

[methane]

My opinion is just stop warning and more permissive about continuation line.
Current pydocstyle is stricter than Guido. There are many cases "this is more readable, but pep8.py doesn't accept it."

See discussion in http://bugs.python.org/issue26763

[warsaw]

I also lean toward not warning about it. It seems like it might be good to warn about mixed styles, but I think ultimately that will be more annoying than helpful.

[sigmavirus24]

#499 is the most straight forward way of solving this. This also allows others who like this rule to re-enable it by selecting it.

[hai-ld]

@sigmavirus24 If I understand it correctly, that means there is no way to:

• ensure consistent breaking
• enforce breaking before binary operator

Is it PyCQA's opinion that above points should be pylint's responsibilities?

[sigmavirus24]

@hai-ld please don't put words in my mouth. The immediate need is to turn this off. In my opinion a new check should be added which is also in the default ignore, otherwise we will cause a lot of problems. This means people will need to enable it by hand. (That said, if we wanted to add the new check in pycodestyle 2.0 we could break it very easily.)

[hai-ld]

@sigmavirus24 I'm sorry if there's any negative connotation in my comments, be assured that it's not intentional (I'm just not a native English speaker). Since you said "#499 is the most straight forward way of solving this", I thought that was the final solution and asked for confirmation.

[sigmavirus24]

The two comments after yours advocated for not warning about this particular style notion from PEP-0008. In that case, the most straight-forward way to solving the problem of warning about this is to disable it. It's not the end of all discussion though.