Since I have now seen two other people complain about this, I will join
the fray.
I am running Solaris 2.6, Postgresql v6.3.2.
I have also gotten the same notice on FreeBSD 2.1.<something, I forget,
but not the threaded one -- this was for a client a while back) running
v6.3.1
Well, the "bug" only happens if one forces the complete build to use
dynamic libraries (.so instead of .a).
And, no, I do NOT accept the Microsoft-type solution: "don't use
dynamic libraries."
It doesn't seem to have any effect on operation other than being
annoying....
Other info: The freeBSD was built with gcc 2.7.<something>; the Solaris
2.6 was built with gcc 2.8.1. All the auxiliary programs (readline,
history, etc.) are the latest versions (I'm too lazy to look right now)
and linked in as dynamic, also.
A bogus value is being passed somewhere, and the function it winds up in
just doesn't know what to do with it. I didn't have the time to chase
down the path it was using to get there because I am not that familiar
with the internals.
This message appears to show up in anything related to a trigger --
whether it's a restriction in the schema definition or a manually
created trigger.
cat