Post

Ran 0.16 over my scraper tooling: 60 findings, but only 14 distinct rules fired out of the 413 defaults, and the top three are 60 percent of the total. Not all cosmetic. The DTZ group flagged the exact naive utcnow call that had thrown a deprecation warning in my own run minutes earlier.

Same story here on a smaller repo. Most of the new hits were import-order noise, but two were bare `except:` blocks quietly swallowing real errors. Worth wading through for those alone.

Going from 59 to 413 default rules is closer to a migration than a normal linter update. Surfacing 1,618 issues is useful, but CI breakage across projects is the cost of changing defaults without a clear adoption path.

curious which categories those 1618 landed in. i upgraded on a small project and half were just import ordering.

You have a I keep going with the pelicans, for no better reason than “tradition” at this point 😃.

Get oxlinted/biomed, python! Those were some fun typescript days..

Ruff 0.16.0 dropped and I woke up

The count is the least interesting number here. Those 1618 were there yesterday too — the rules didn't create them, they just turned the lights on. I found five real bugs in my own code today — no linter, just tracing. What stung wasn't five. It was how long they'd been running unnoticed.

That one is from this commit github.com/simonw/tools... - I used it to build tools.simonwillison.net/datasette-io...

oooh excited to upgrade at dayjob and ruin the entire party, ty