Re: Per-function GUC settings: trickier than it looked

"Florian G. Pflug" <fgp(at)phlo(dot)org> writes:
> It still seems a bit strange that "SET LOCAL" is undone at function-exit,
> if the function has a matching SET-clause. But we need that for backwards-
> compatibility of the secure-search_path workaround, right?

Yeah, I'm afraid we backed ourselves into a corner on that one.

> So, to reiterate, my idea is
> .) Make "SET TRANSACTION" a synonym for "SET LOCAL" at the SQL-Level.
> .) In pl/pgsql, "SET TRANSACTION" sets a new value that is kept after the
> function exits, even if the function has a matching SET-clause.
> .) "SET LOCAL" in pl/pgsql set a new value that is kept if the function
> has no matching SET-clause. If it has one, the value is restored.
> In any case, we emit a warning that "SET LOCAL" is going away.
> .) One day, make "SET LOCAL" in pl/pgsql mean "local to the surrounding
> BEGIN/END block". Independent of any SET-clauses the function
> might or might not have.

I don't think it's a good idea to change SET LOCAL now and plan on
changing it again later ;-). If we really want BEGIN-block-local
SET capability, I'd prefer to think of some new keyword for that.
But I'm not convinced it's interesting --- given the proposed behavior
of function SET-clauses, attaching a SET to your function seems like
it'll cover the need for restoring outer values.

regards, tom lane