Jeff Davis <pgsql@j-davis.com> writes:
> I don't have a strong opinion, but allowing zero-argument variadic
> function calls -- and therefore causing foo(variadic int[]) and
> foo(variadic text[]) to conflict -- makes more sense than requiring one
> argument.
I hadn't even thought about that point, but the idea that those two
would conflict bothers me quite a lot. Not least because there's no
reasonable way to enforce it with the existing unique indexes on pg_proc.
I think you'd have to leave the variadic argument out of proargtypes
altogether, and that seems mad.
regards, tom lane