Re: Controlling changes in plpgsql variable resolution

"David E. Wheeler" <david@kineticode.com> writes:
> On Oct 19, 2009, at 12:05 PM, Tom Lane wrote:
>> Where exactly would you put the modifier, and why is that better than
>> the existing #option convention?

> CREATE OR REPLACE FUNCTION foo()
> RETURNS BOOLEAN
> LANGUAGE plpgsql WITH opt1, opt2
> AS $$...$$;

> That is, the specification of options is made outside of the language  
> in question.

I don't think I particularly care for this.  It's inventing a global
mechanism to cover a problem that we currently have one instance of
in one PL.  That's a mighty thin justification.  Also, I tend to think
that you should have several instances of a problem before you venture
to design a global solution --- else your one-size-fits-all solution
might turn out to be a lot less general than you thought.
        regards, tom lane