"Dan \"Heron\" Myers" <heron(at)xnapid(dot)com> writes:
> One case that fails is essentially copied from the V1 section in the
> documentation:
Well, there's nothing wrong with that C code, so the problem is
someplace else.
Did you remember to declare the function STRICT? If not, and if
there are any nulls in your test table, a crash would be expected;
there's nothing in this function that's guarding against a null
pointer dereference.
regards, tom lane