Stephen Frost <sfrost@snowman.net> writes:
> The other idea which comes to mind is- could we try to actually resolve
> what the 'right' answer is here, instead of setting a special value and
> then having to detect and fix it later?
No, absolutely not. Read the NOTES at the head of gram.y. Or if you
need it spelled out: when we run the bison parser, we may not be inside a
transaction at all, and even if we are, we aren't necessarily going to
be seeing the same current user that will apply when the parsetree is
ultimately executed. (Consider a query inside a plpgsql function, which
might be called under multiple userids over the course of a session.)
I think Alvaro's suggestion is a perfectly appropriate solution.
regards, tom lane