3 ms·
That's exactly how I treat reviewing AI generated PRs at work. Except... the next day they are still there and I have to push myself a bit further through the
by fallingbananna 17d ago
That's exactly how I treat reviewing AI generated PRs at work.
Except... the next day they are still there and I have to push myself a bit further through the wall of changes before closing the tab again.
- RobKohr 17d agoIf it takes more than 20 min to review a pr, ask the dev to come on a call and explain what the code does.
- pc86 17d agoIt must be nice to work on such small-scale systems that you can reasonably expect PRs to be this quick.
- gedy 17d agoIt's a cultural thing, but you can do incremental PRs towards a large goal. Giant PRs that are expected to be reviewed never really seemed worth it imho.
- pc86 16d agoI don't think a PR needs to be tens of thousands of lines for a review to take more than 20 minutes. I've worked on projects where you can't reasonably get a branch set up and running somewhere to test it within 20 minutes if you need to set up dependencies, peripherals, external systems, etc.
- zamadatix 17d agoI don't know it has as much to do with scale of the system vs the general architecture. E.g. the system I primarily work with these days has millions of lines but most PRs are for a small changes which are well contained in scope by the overall architecture.
- 8note 17d agoreasonably you should already know what the code and project is supposed to do before they even start writing code, so you can get directional feedback in. then you are maybe reviewing 1 out of 7 PRs that implement the agreed upon change