* Tom Lane (tgl@sss.pgh.pa.us) wrote:
> Stephen Frost <sfrost@snowman.net> writes:
> > * Tom Lane (tgl@sss.pgh.pa.us) wrote:
> >> Quite aside from any security risks, that means that running "make
> >> installcheck" twice in a row fails. Please fix.
>
> > Right, will do, though one kind of requires the other (we can't drop the
> > only user we know how to connect as which is a superuser...). I'll
> > figure out a way to make it work though.
>
> Instead of physically reconnecting, could you do SET ROLE or SET SESSION
> AUTHORIZATION? I think that's what we do in the core tests.
Unfortunately, neither of those reset the role GUCs which are set with
ALTER ROLE.
> > I'm at least somewhat suspicious it has to do with loading the library,
> > since we (pretty clearly..) didn't test that much as it's not really
> > intended.
>
> There are several more crashes in the BF now. They're not at initial
> library load AFAICS. Hard to tell if it's platform-specific or just
> randomly fails sometimes. Have you tried valgrind to see if there's
> uninitialized-memory touches?
Will give that a shot and see what it says.
Thanks!
Stephen