We have a PR https://github.com/apache/spark/pull/54840 aiming to replace
black with ruff as the python formatter.

There are a few reasons to do that:
1. ruff is significantly faster than black - 30x according to their blog,
100x on spark repo per @LuciferYang's test. This makes the pre-commit hook
more pleasant.
2. Since we already use ruff for linting, using it for formatting will
reduce our development tool dependencies.
3. ruff has a slightly better taste than black in my opinion and is more
customizable (if anyone's interested, feel free to check the PR to view the
diff).

This is also a good timing to upgrade because we just upgraded black
version which caused a large refactoring. If we do this now, the two code
refactoring will feel like one to most people, less interruption.

Feedbacks are welcome!

Tian

Reply via email to