X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 02719D1B500 for ; Thu, 30 Oct 2003 04:20:32 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68336-05 for ; Thu, 30 Oct 2003 00:20:00 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 3998FD1B4ED for ; Thu, 30 Oct 2003 00:20:00 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id F194436B2F; Wed, 29 Oct 2003 23:19:58 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1AF4Hu-0001tu-00; Wed, 29 Oct 2003 23:19:58 -0500 To: Tom Lane Cc: mallah@trade-india.com, pgsql-performance@postgresql.org, jesse@fsck.com Subject: Re: PostgreSQL 7.4beta5 vs MySQL 4.0.16 with RT(DBIx::SearchBuilder) References: <33280.203.145.130.142.1067455363.squirrel@mail.trade-india.com> <33288.203.145.130.142.1067456744.squirrel@mail.trade-india.com> <23780.1067469787@sss.pgh.pa.us> In-Reply-To: <23780.1067469787@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 29 Oct 2003 23:19:58 -0500 Message-ID: <877k2nfjbl.fsf@stark.dyndns.tv> Lines: 24 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200310/744 X-Sequence-Number: 4492 Tom Lane writes: > (I'd argue that the SQL generator is broken anyway ;-) if it generates > such horrible conditions as that. Or maybe the real problem is that > the database schema is a mess and needs rethinking.) I had the same reaction when I first saw those queries. But I think the problem with the RT schema is that it needs to implement an ACL system that satisfies lots of different usage models. Some people that use it want tickets to be accessible implicitly by the opener like a bug tracking system, others want the tickets to be internal only like a network trouble ticketing system. Some people want to restrict specific operations at a fine-grain, others want to be have more sweeping acls. I've tried doing ACL systems before and they always turned into messes long before that point. I always end up pushing back and trying to force the client to make up his or her mind of exactly what he or she needs before my head explodes . If there's a nice general model for ACLs that can include completely different usage models I've never found it. -- greg