Sergiy Vyshnevetskiy <serg(at)vostok(dot)net> writes:
> Not at all. What's "broken" is the idea of variable as a simple piece of
> memory. It is correct for base types, but not for domains - they may have
> non-empty constructors (in C++ terminology).
That may be, but I'm unwilling to pay the overhead for *every* variable
when most of them won't be domains. I'm inclined to extend PLpgSQL_type
to include a domain indicator and only do it the hard way when we have to.
[ looks at code... ] Actually, I think we already have the flag we
need: look to see if the typinput function is strict.
regards, tom lane