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 032AE5E37D0 for ; Wed, 18 Aug 2004 18:47:49 -0300 (ADT) 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 69336-05 for ; Wed, 18 Aug 2004 21:47:53 +0000 (GMT) Received: from mproxy.gmail.com (rproxy.gmail.com [64.233.170.198]) by svr1.postgresql.org (Postfix) with ESMTP id B40955E46E9 for ; Wed, 18 Aug 2004 18:47:41 -0300 (ADT) Received: by mproxy.gmail.com with SMTP id 76so172676rnl for ; Wed, 18 Aug 2004 14:47:50 -0700 (PDT) Received: by 10.38.164.57 with SMTP id m57mr236744rne; Wed, 18 Aug 2004 14:47:50 -0700 (PDT) Message-ID: <19ab0ccd04081814473ca44daf@mail.gmail.com> Date: Wed, 18 Aug 2004 17:47:50 -0400 From: Aaron Birkland Reply-To: apb18@cornell.edu To: Tom Lane Subject: Re: Array types and loading Cc: pgsql-performance@postgresql.org In-Reply-To: <4152.1092865161@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit References: <19ab0ccd040818123936c1cdd3@mail.gmail.com> <4152.1092865161@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200408/281 X-Sequence-Number: 7917 You got it.. 7.3 (should have mentioned that). We're planning to upgrade to 8.0 anyway in the future, so it's good to know. Thanks! -Aaron On Wed, 18 Aug 2004 17:39:21 -0400, Tom Lane wrote: > Aaron Birkland writes: > > In a nutshell, it looks like whenever COPY is invoked, and when a user > > defined type is used in an array, then stat64() will be called for > > each row accessed on the shared library relevant for the user defined > > type. > > Let me guess ... PG 7.3 or older? > > 7.4 should avoid the problem because array_in() caches function lookup > information for the element type's input function across multiple calls. > > In 8.0 there's also a cache at the fmgr_info() level to eliminate > repeated searches for a dynamically loaded function. > > regards, tom lane >