Re: anonymous composite types for Table Functions (aka SRFs)

Joe Conway <mail(at)joeconway(dot)com> writes:
> Actually, I was just looking at this and remembering that I wanted to
> disallow a coldeflist for non-RECORD return types. Do you prefer to
> allow it (but properly apply the alias and enforce the type)?

We could do that too; what I was unhappy about was that the system
took the syntax and then didn't apply the type checking that the syntax
seems to imply. I'd prefer to do the type checking ... but I don't
want to expend a heckuva lot of code on it, so maybe erroring out
is the better answer.

regards, tom lane