Repository navigation
MNT: upgrading ruff to 0.16.x #20113
Description
A new ruff minor appears
Blog post: https://astral.sh/blog/ruff-v0.16.0
This one is special (to me): of the 316 rule violations we get by upgrading now, 305 are for too-many-positional-arguments, a "rule" I've been informally discussing and advising in reviews for years throughout many projects. Crucially, this rule cannot be blinding applied to existing public APIs (without breaking them), and doesn't have as much value for private APIs, but it is extremely precious for new code, so my suggestion is to add local noqa comments (trivially done with ruff itself, e.g. uvx [email protected] check astropy --select PLR0917 --add-noqa) rather than disabling the rule globally.
Other violations should be trivally fixable. I'll keep track of them here.
RUF068
ISC004
- RFC: address
ISC004violations (utils) #20115 - RFC: address
ISC004violations (timeseries) #20116 - TST/BUG: address
ISC004violations (io.fits) #20117 - RFC: address
ISC004violations (io.registry) #20118 - RFC: address
ISC004violations (nddata) #20119
FURB192
RUF036
Activity
pllim commentedon Jul 24, 2026
OK all the linked PRs are merged. Thanks for being on top of this.
So now we just wait for the auto update? Can we close this issue?
neutrinoceros commentedon Jul 24, 2026
Not yet, we'll need to address/ignore violations to
PLR0917too, and I expect that'll be easier to do along with the upgrade.pllim commentedon Jul 24, 2026
As for https://docs.astral.sh/ruff/rules/too-many-positional-arguments/ (PLR0917), I think we should ignore it. Being nitpicky on style is one thing, but it is not ruff's place to tell me how many arguments I can have in my functions.
neutrinoceros commentedon Jul 24, 2026
You can have as many arguments as you want as long as they're keyword-only, which I've been advocating for for years because it's extremely useful for long term maintenance.
pllim commentedon Jul 24, 2026
This is an "old" repo (by software standards). I don't think we should be going around breaking API just because ruff is annoyed. If you want to enable for new code, that is negotiable, but we should ignore for existing code.
neutrinoceros commentedon Jul 24, 2026
That's what I proposed in my description
neutrinoceros commentedon Oct 9, 2026
well this is a new one: ruff 0.17.0 is out and we're still on 0.15.0
Hopefully #20519 can save us from a world of pain