Consider:
CREATE VIEW_X AS
SELECT
UNION ALL
SELECT
UNION ALL
SELECT ;
versus
CREATE VIEW_A AS
SELECT ;
CREATE VIEW_B AS
SELECT ;
CREATE VIEW_C AS
SELECT ;
where , and are each somewhat complex
with several joins, but utilizing different tables for each of A, B
and C.
Performance on
SELE
[[ Oops, still not used to gmail. Accidentally posted this straight to
Tom and not the list]]
> There are a bunch of special cases where it can't do that, though.
> Look into src/backend/optimizer/path/allpaths.c, particularly
> subquery_is_pushdown_safe:
My case probably fits the 'special case'
On 11/3/07, Tom Lane wrote:
> "Jeff Larsen" <[EMAIL PROTECTED]> writes:
> > My case probably fits the 'special case' description. Not all the
> > columns in each subquery matched up, so there were NULL::text
> > placeholders in some spots in the SE
On Nov 23, 2007 7:29 AM, Dean Rasheed <[EMAIL PROTECTED]> wrote:
> I am having a performance problem trying to query a view which is a
> UNION ALL of 2 tables. I have narrowed the problem down to my use of
> DOMAINS in the underlying table. So in the test-case below, when the
> column "a" is of dom
