[Dev] Use pre-commit-hook instead of "archery lint" for all lints

[kou]

Describe the enhancement requested

We use archery lint ... to run linters. It's convenient because we don't need to know linters details but we need to install archery[lint] and we need to write Python code to add a new linter. They're a bit annoying.

How about using pre-commit https://pre-commit.com/ for all linters? If we use pre-commit:

• We don't need to write Python code for new linter and changing linter behavior (We need to write YAML instead)
• C++: We don't need to install clang-format-14 separately (pre-commit can do it automatically)
• C++: We don't need to run CMake to run clang-format-14
• We can get easy to read lint failure messages (archery docker run ubuntu-lint outputs many unrelated messages)

Note that we already have some pre-commit configuration: https://github.com/apache/arrow/blob/main/.pre-commit-config.yaml

Sub issues:

• 40448

• 40544

• 40586

• 35941

• 40854

• 43979

Component(s)

Developer Tools

[lidavidm]

+1 for pre-commit, which is at least reasonably common across projects these days IMO

[vibhatha]

+1 this is a great initiative.