Peter Eisentraut <peter_e@gmx.net> writes:
> I'm informed that the last statement of a function that returns void cannot be
> a SELECT. How else is one supposed to call another function which also
> returns void?
Hmm, seems like the best answer is to rejigger the order of the tests in
check_sql_fn_retval. I think that logic was designed before we had an
idea of VOID-returning functions.
regards, tom lane