On 1/25/07, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
"Feng Chen" <fchen(at)covergence(dot)com> writes:
> The problem is that I should not have to and cannot re-load the
> functions every now and then. Why would the type id change and the
> function still references to the old type id thus fails to get the right
> results?
I suspect the problem
is not with the return type at all, but with some table or other type
referenced within the function body.
Although pg_depend handles the dependency of return types of the
functions but the weird thing is that his pg_proc entries show invalid
type OID for return type. Something wrong with his PG instance probably.