Comment by buggymcbugfix

2 months ago

Thanks for the nice article. The SQLite docs that explain this bug are emphatic about how utterly rare, even irreproducible it is. How, then, was it found, one wonders?

We found it at Tailscale and bought an enterprise support contract from SQLite to debug it for us. Worth every penny.

We should probably blog about it.

My guess is that they noticed something was not quite right when inspecting the code and then tested their hypothesis manually to see if it was actually a real bug.

Something similar happened to me when I was working on some code that should clear the state and, by inspection, I noticed some field was not properly cleared under specific circumstances. Then I tested the hypothesis and indeed found a bug. Sometimes intuition and guessing is what it takes.