X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 843FC539A6 for ; Sun, 3 Apr 2005 09:01:28 +0100 (BST) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94829-03 for ; Sun, 3 Apr 2005 08:01:16 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id C1B01538D3 for ; Sun, 3 Apr 2005 09:01:15 +0100 (BST) Received: (qmail 14035 invoked from network); 3 Apr 2005 10:01:33 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 3 Apr 2005 10:01:33 +0200 To: "Hannes Dorbath" , pgsql-performance@postgresql.org Subject: Re: Query Optimizer Failure / Possible Bug References: Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Sun, 03 Apr 2005 10:01:13 +0200 In-Reply-To: User-Agent: Opera M2(BETA2)/8.0 (Linux, build 987) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.003 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200504/39 X-Sequence-Number: 11436 Noticed this problem,too. You can always make the calculation you want done once inside a set returning function so it'll behave like a table, but that's ugly. On Mon, 28 Mar 2005 16:14:44 +0200, Hannes Dorbath wrote: > hm, a few days and not a single reply :| > > any more information needed? test data? simplified test case? anything? > > > thanks > > > Hannes Dorbath wrote: >> The query and the corresponding EXPLAIN is at >> http://hannes.imos.net/query.txt >> I'd like to use the column q.replaced_serials for multiple calculations >> in the SELECT clause, but every time it is referenced there in some way >> the whole query in the FROM clause returning q is executed again. >> This doesn't make sense to me at all and eats performance. >> If this wasn't clear enough, for every >> q.replaced_serials AS some_column >> in the SELECT clause there is new block of >> --------------------------------------------------------------- >> -> Aggregate (cost=884.23..884.23 rows=1 width=0) >> -> Nested Loop (cost=0.00..884.23 rows=1 width=0) >> -> Index Scan using ix_rma_ticket_serials_replace on >> rma_ticket_serials rts (cost=0.00..122.35 >> rows=190 width=4) >> Index Cond: ("replace" = false) >> -> Index Scan using pk_serials on serials s >> (cost=0.00..3.51 rows=1 width=4) >> Index Cond: (s.serial_id = "outer".serial_id) >> Filter: ((article_no = $0) AND (delivery_id = $1)) >> --------------------------------------------------------------- >> in the EXPLAIN result. >> For those who wonder why I do this FROM (SELECT...). I was searching >> for >> a way to use the result of an subselect for multiple calculations in the >> SELECT clause and return that calculation results as individual columns. >> I tested a bit further and found out that PG behaves the same in case q >> is a view. This makes me wonder how efficient the optimizer can work >> with views - or even worse - nested views. >> Tested and reproduced on PG 7.4.1 linux and 8.0.0 win32. >> Thanks in advance, >> Hannes Dorbath >