[dartos]

Seems like every other AI code analysis tool. Someone excited posts about how it made 1000s of syntactically valid fixes! Of course some are wrong and others are worse than not being there, but surely some are useful! Surely in the coming weeks it’ll improve to make millions of potentially useful tests!

Seems like a waste of time unless you can be sure it’s spitting out valuable fixes.

[dbolgheroni]

Some points: 1) A tool automatically generating thousands or millions of patches won't scale since it can't be reviewed by developers; 2) A tool could automate patch generation but won't automate doing tests to check if it works or of if there are any regressions introduced; 3) https://marc.info/?l=openbsd-tech&m=171817275920057&w=2

[dartos]

I’m afraid that I didn’t translate my sarcasm into text.