Skip site navigation (1)
Skip section navigation (2)
Peripheral Links
Header And Logo
|
Site Navigation
Re: Bad plan for join to aggregate of join.
- From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
- To: mischa(at)ca(dot)sophos(dot)com
- Cc: pgsql-performance(at)postgresql(dot)org
- Subject: Re: Bad plan for join to aggregate of join.
- Date: Tue, 12 Sep 2006 19:42:30 -0400
- Message-id: <3215(dot)1158104550(at)sss(dot)pgh(dot)pa(dot)us>
Mischa Sandberg <mischa(at)ca(dot)sophos(dot)com> writes: > can PG see that a join on an grouped-by field > can be pushed down into the query as an indexable filter? No. The GROUP BY serves as a partial optimization fence. If you're concerned about the speed of this query, I recommend making a different view in which 'message' is joined inside the GROUP BY. regards, tom lane
- Follow-Ups:
- Re: Bad plan for join to aggregate of join.
- From: Mischa Sandberg
- Re: Bad plan for join to aggregate of join.
- References:
- Abysmal hash join
- From: Florian Weimer
- Bad plan for join to aggregate of join.
- From: Mischa Sandberg
- Abysmal hash join
- Prev by Date: Re: Poor performance on seq scan
- Next by Date: Re: Poor performance on seq scan
- Previous by thread: Bad plan for join to aggregate of join.
- Next by thread: Re: Bad plan for join to aggregate of join.
- Indexes: