On Fri, 2003-01-10 at 20:28, Tom Lane wrote:
> The core dump appears to occur when the SELECT INTO fails to retrieve
> a row, leaving "finalrec" undefined.
Thanks very much for your help, Tom.
> Clearly, RETURN NEXT with an undefined record variable shouldn't dump
> core, but what should it do? Raise an error, or perhaps be a no-op?
I'd vote for making it a no-op. Raising an error is too severe for a
fairly routine occurence, IMHO. If we make it a no-op, it's consistent
with how I understand a SELECT INTO of 0 rows -- it doesn't produce an
"undefined value", but an "empty result set" (like the difference
between "" and a NULL pointer).
Cheers,
Neil
--
Neil Conway <neilc@samurai.com> || PGP Key ID: DB3C29FC