Re: return setof : alternatives to holder table

On 15/08/10 18:00, zhong ming wu wrote:

> Thanks for any better solution to this

CREATE TYPE

However, you still have to have a special type around just for that
function, and you have to *maintain* it to ensure it always matches the
types/columns of the input tables.

I frequently wish for type inference in PL/PgSQL functions returning
query results, so Pg could essentially create and destroy a type along
with the function, allowing you to reference columns in the functions
results without having to use RETURNS RECORD and all that AS
(column-list) pain.

Of course, I don't want it badly enough to put my time where my mouth is
and try to code it ;-) . I'm not whining about the current situation,
just thinking about ways it could improve further.

--
Craig Ringer