"Alex Hunsaker" <badalex(at)gmail(dot)com> writes:
> I can turn that off and only use DBI->prepare() as a test. Or heck
> just cut DBI->prepare() out and just quote everything and send them
> through using DBI->do() instead. That is if you think that could be
> the culprit.
The path of control that's dying cannot be entered at all unless you're
using a prepared statement, so cutting out DBI->prepare should make the
problem go away.
Now personally, I am much more interested in *reproducing* the problem
than merely dodging it. I can understand if you just need a workaround
yesterday, but please see if you can get a reproducible test case ...
regards, tom lane