Should ruff check be renamed to ruff lint?

[thernstig]

This is a wild idea, but since Ruff is still in its 0.X semver phase, maybe not a bad idea after all.

Should ruff check be renamed to ruff lint?

Since ruff format was introduced, the term check is harder to to understand what it does. In addition the website goes about mentioning the terms format and lint. As evident by this:

It adds more "confusion" since you have ruff format --check. One could think it does both the format and the normal ruff check.

[zanieb]

Thanks for raising.

We're interested in making ruff check a unified command that checks for both lint and format errors #8232. We might want to separately introduce a ruff lint command though — this seems as good a place as any for discussion on that.

[T-256]

Like ruff format that applies modifications by default, I suggest ruff lint do the same.
The current behavior of --fix would be default for ruff lint and can be disable by --check. (like ruff format we have today)

[parafoxia]

I disagree that a ruff lint command should fix code by default. Ruff (afaik) is the only Python linter with the capability to fix issues, so may be confusing for people switching from something else. Linters in other languages, such as Cargo, tend not to fix by default either, so I don't believe there's a precedent for this. Additionally, the linter fixes aren't 100% safe to apply (see #8556 as an example).

Whether a unified ruff check should run something equivalent to ruff lint --fix && ruff format is another question though.

[thernstig]

Linters I am aware of has a --fix for linting faults. Or e.g. Prettier (a formatter only) has --write. I think code transformations (which linters are) should not do fixes by default as @parafoxia mentions, due to the dangers of it. When users specifically has to add a --fix, they can check the changes made and are aware they are being made.

[tmeiczin]

I agree, a ruff lint should identify, but not fix issues by default. There are times when the linter does not have enough context to determine if a change is appropriate and can potentially introduce a bug. In regard to formatting, I think ruff format is explicit in that I'm asking it to format my code and should do so by default. I like the idea of ruff check identifying both lint and format, but not fixing and I kinda find --fix to be awkward. I would probably prefer a ruff fix command, but it's a balance between number of commands and options per command.

[Avasam]

My thoughts:
ruff lint: only lints by default
ruff format: formats by default
ruff check: shorthand for ruff lint && ruff format --check
ruff fix: shorthand for ruff lint --fix && ruff format

[MichaReiser]

I'll close this in favor of #8232 where potentially introducing a ruff lint command will come up.