Post

Friggin' AI coding...I really don't want to like it but... It just tested >10k queries against two builds to find and understand where they diverge. Sure, I could have coded up a way to that diff queries. But understanding why they're different would have taken eyeball time.