Re: Question about Sql SELECT and optimizer

On Thu, Mar 02, 2006 at 11:19:32AM -0500, Chaz. wrote:
> I am trying to understand something I have seen happen. I had a select
> that looked like:
>
> select f(A) from A, B, C where g(A)
>
> Where f(A) is the select that only depends on table A;
> g(A) is the where part that only depends on table A.
>
> What I saw happen was the optimizer will waste a lot of time (seconds!)
> bringing in table B and C. I was wondering why doesn't the optimizer
> drop references to tables B and C since they aren't used any where?

The above query does a cross join. Even though you're not using
values from B and C they're still contributing rows to the result
set.

test=> SELECT * FROM a;
aid
-----
a1
a2
(2 rows)

test=> SELECT * FROM b;
bid
-----
b1
b2
(2 rows)

test=> SELECT * FROM c;
cid
-----
c1
c2
(2 rows)

test=> SELECT a.*, b.*, c.* FROM a, b, c WHERE a.aid = 'a1';
aid | bid | cid
-----+-----+-----
a1 | b1 | c1
a1 | b2 | c1
a1 | b1 | c2
a1 | b2 | c2
(4 rows)

test=> SELECT a.*, b.* FROM a, b, c WHERE a.aid = 'a1';
aid | bid
-----+-----
a1 | b1
a1 | b2
a1 | b1
a1 | b2
(4 rows)

test=> SELECT a.* FROM a, b, c WHERE a.aid = 'a1';
aid
-----
a1
a1
a1
a1
(4 rows)

--
Michael Fuhr