Support for docformatter rules

[cpcloud]

We just moved our linting setup over to ruff and it's great!

It would be amazing if we could remove our docformatter dependency as well in favor of ruff.

[cpcloud]

It's possible we can use the ruff pydocstyle support for this too, which may obviate the need for docformatter support.

[charliermarsh]

Awesome :)

Yeah I've used docformatter in the past, it's great! I wonder how much of its functionality is covered by our pydocstyle support with autofix enabled. I'd need to do a full audit... But I'd definitely like Ruff to be a full replacement for docformatter in that sense.

[charliermarsh]

(Labeling as question, to figure out what Ruff is missing here, but it's not necessarily aimed at you.)

[cpcloud]

In the meantime, I've flipped on the D switch and it appears to cover what we're currently using docformatter for, which was effectively a less-strict pydocstyle with the ability to actually fix things instead of only reporting them.

Since ruff covers both linting and fixing I think at least for ibis we no longer need docformatter support.

[charliermarsh]

Okay cool! (Note that some of the D rules are conflicting, like "No blank line before class" vs. "One blank line before class", since we support a couple different docstring conventions: https://github.com/charliermarsh/ruff#does-ruff-support-numpy--or-google-style-docstrings)

[cpcloud]

Yeah, the list of ignores I'm using for that category of lints is ... long.

[charliermarsh]

It might be easier to explicitly enable via --select if you want? I can post the list here if you post your ignores (and I could add it to the docs).

[cpcloud]

It's actually not too bad:

ignore = [
  "E501",
  "PGH003",
  "RET504",
  "RET505",
  "RET506",
  "RET507",
  "RET508",
  "D100", # public module
  "D101", # public class
  "D102", # public method
  "D103", # public function
  "D105", # magic methods
  "D107", # init
  "D202", # blank lines after function docstring
  "D203", # blank line before class docstring
  "D205", # blank line between summary line and description
  "D213", # Multi-line docstring summary should start at the second line
  "D402", # First line should not be the function's signature
  "D417", # Missing argument descriptions
]

[cpcloud]

And then of course this bit:

[tool.ruff.per-file-ignores]
"*test*.py" = ["D"] # ignore all docstring lints in tests

[charliermarsh]

Ah okay yeah, in that case it's more concise to specify the ignores than the selects, which would be like:

select = [
  "D104",
  "D106",
  "D200",
  "D201",
  "D204",
  "D206",
  "D207",
  "D208",
  "D209",
  "D210",
  "D211",
  "D212",
  "D214",
  "D215",
  "D300",
  "D301",
  "D400",
  "D403",
  "D404",
  "D405",
  "D406",
  "D407",
  "D408",
  "D409",
  "D410",
  "D411",
  "D412",
  "D413",
  "D414",
  "D415",
  "D416",
  "D418",
  "D419",
]

[charliermarsh]

Closing for now since I think we actually have good coverage here.