Neat

Posted Apr 2, 2026 23:34 UTC (Thu) by calvinowens (subscriber, #100757)
Parent article: The role of LLMs in patch review

My first experience with it was positive, it correctly identified an obvious nesting case I should have caught. It proposed some confusing solutions which I allowed to confuse me, but that's my fault. I've only just started using these tools for review: it has taken some practice, but I feel like I'm quickly getting better at that.

To put a positive spin on it, forcing myself to reject the false positives has frequently been a productive learning exercise for me. But I can certainly understand why somebody under time pressure to finish something could see that as a strict negative.