flake8's (W503 and) W504

[jamesbraza]

flake8 has two rules W503 and W504 that don't seem to be in ruff==0.0.263.

• W503 shouldn't exist PyCQA/flake8#494 mentions that W503 shouldn't exist
• https://stackoverflow.com/questions/73568036/flake8-line-break-before-binary-operator-how-to-fix-it discusses W503 and W504

Any thoughts on ruff implementing W504 with autofix?

[charliermarsh]

I think we're unlikely to implement W503 and W504, since we're likely not going to implement the pycodestyle rules that are ignored-by-default (due to being non-unanimously accepted, and non-violations of PEP8, per the pycodestyle docs).

W503 is a little more problematic, since it conflicts with Black. W504 might not conflict with Black, but I'm still hesitant to support it given that it's disabled-by-default in pycodestyle and Flake8.

[charliermarsh]

(As always, totally open to further discussion -- I just tend to close issues by-default when answering a question, to try and manage the issue queue responsibly.)

[jamesbraza]

Oh sure, this conclusion is fine with me, thanks! I realize I have been making a lot of issues here, but it's okay to close them.

For me to remove flake8, I want to make sure the rule coverage is the same. With W504 dispatched (per this issue), the last outstanding items are:

• doctests = True: Add flag for syntax checking of doctests as in flake8 --doctests #3542
• flake8-requirements: Request: adding flake8-requirements #4100

[ccostello97]

@charliermarsh I'd like to propose this be re-opened. Black does take the stance of W504 against line breaks after binary operators, which also aligns with the PEP-8 recommendation.

To keep my code in check with PEP-8, I have to use ruff with flake8 --select=W504. It would certainly be nice to have this encapsulated in ruff.

[Avasam]

FWIW, this would also bring Ruff autofixes closer to parity with autopep8, even if ignored/disabled by default #9057 (comment)