Received: from localhost (mx1.hub.org [200.46.208.251]) by postgresql.org (Postfix) with ESMTP id C153C9FA594 for ; Fri, 13 Oct 2006 14:45:29 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.208.251]) (amavisd-new, port 10024) with ESMTP id 29563-05 for ; Fri, 13 Oct 2006 14:45:05 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey- Received: from servata.com (ip-216-152-249-241.servata.com [216.152.249.241]) by postgresql.org (Postfix) with ESMTP id 936EA9FB4F8 for ; Fri, 13 Oct 2006 14:45:05 -0300 (ADT) Received: from [209.162.219.253] (helo=dogma.v10.wvs) by servata.com with esmtpa (Exim 4.50) id 1GYR5M-0003lL-SV; Fri, 13 Oct 2006 10:44:41 -0700 Subject: Re: [PERFORM] Hints proposal From: Jeff Davis To: "Jim C. Nasby" Cc: Bucky Jordan , Alvaro Herrera , josh@agliodbs.com, pgsql-hackers@postgresql.org In-Reply-To: <20061013173024.GE28647@nasby.net> References: <78ED28FACE63744386D68D8A9D1CF5D4209CBD@MAIL.corp.lumeta.com> <1160760211.31966.231.camel@dogma.v10.wvs> <20061013173024.GE28647@nasby.net> Content-Type: text/plain Date: Fri, 13 Oct 2006 10:44:41 -0700 Message-Id: <1160761481.31966.242.camel@dogma.v10.wvs> Mime-Version: 1.0 X-Mailer: Evolution 2.0.4 (2.0.4-7) Content-Transfer-Encoding: 7bit X-Virus-Scanned: Maia Mailguard 1.0.1 X-Spam-Status: No, hits=0.234 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, SPF_HELO_PASS X-Spam-Level: X-Archive-Number: 200610/774 X-Sequence-Number: 92671 On Fri, 2006-10-13 at 12:30 -0500, Jim C. Nasby wrote: > On Fri, Oct 13, 2006 at 10:23:31AM -0700, Jeff Davis wrote: > > Right. And I think the sane ideas are along the lines of estimate & cost > > corrections (like Tom is saying). > > Let me ask this... how long do you (and others) want to wait for those? > It's great that the planner is continually improving, but it also > appears that there's still a long road ahead. Having a dune-buggy to get > to your destination ahead of the road might not be a bad idea... :) Fair enough. I can wait indefinitely right now, because I don't have any serious problems with the planner as-is. I am trying to empathize with people who are desperate to force plans sometimes. Your original proposal included hints in the client queries. I suggested that regexes on the server can accomplish the same goal while avoiding a serious drawback. Don't you think some kind of server matching rule is better? I think an idea to get the hints into the server, regardless of the types of hints you want to use, make it more likely to be accepted. Don't you think that's a better road to take? Regards, Jeff Davis