Stephen Frost <sfrost@snowman.net> writes:
> Working on 8.1 I've recently been annoyed at the need to translate a
> sub-select inside an IN () clause into a fixed list of contents (the
> results of the sub-select, exactly) in order to get better performance.
Better performance than what? Ever since 7.4 we've converted small IN
sub-selects into plans along the lines of
regression=# explain select * from tenk1 where unique1 in (select f1 from int4_tbl);
QUERYPLAN
-----------------------------------------------------------------------------------Nested Loop (cost=1.06..31.20
rows=5width=244) -> HashAggregate (cost=1.06..1.11 rows=5 width=4) -> Seq Scan on int4_tbl (cost=0.00..1.05
rows=5width=4) -> Index Scan using tenk1_unique1 on tenk1 (cost=0.00..6.00 rows=1 width=244) Index Cond:
(tenk1.unique1= "outer".f1)
(5 rows)
which looks OK to me.
regards, tom lane