Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Yep, and he couldn't reproduce it either, and on a different platform.
> I think that indicates we do have a problem in there, it just doesn't
> show very often.
I agree, this looks a lot like a low-probability bug. But how to attack
it when we can't reproduce it with even small probability? We need the
reporters to try to figure out what environment made it happen for them.
I can chase a bug if I can make it happen one-time-in-ten, or even
one-time-in-a-hundred, but I can't do much with a bug that I've only
heard secondhand reports of.
regards, tom lane