Received: from sss2.sss.pgh.pa.us (sss.pgh.pa.us [209.114.166.2]) by hub.org (8.10.1/8.10.1) with ESMTP id e58EcIT47472 for ; Thu, 8 Jun 2000 10:38:18 -0400 (EDT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss2.sss.pgh.pa.us (8.9.3/8.9.3) with ESMTP id KAA15876; Thu, 8 Jun 2000 10:38:14 -0400 (EDT) To: "Cary O'Brien" cc: mikeo@spectrumtelecorp.com (mikeo), pgsql-general@hub.org Subject: Re: Re: [SQL] oracle rownum equivalent? In-reply-to: <200006081245.IAA05207@saltmine.radix.net> References: <200006081245.IAA05207@saltmine.radix.net> Comments: In-reply-to "Cary O'Brien" message dated "Thu, 08 Jun 2000 08:45:32 -0400" Date: Thu, 08 Jun 2000 10:38:14 -0400 Message-ID: <15872.960475094@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200006/239 "Cary O'Brien" writes: > What I meant to say, and failed, was that in Oracle, rowid > "psuedo-column" are hex encoded strings containing the block, row, and > file where the tuple is stored. So lookup on unindexed rowid *should* > be fast, although this is not spelled out in my ancient Oracle book. Oh, in that case it's pretty much like our CTID column. For example: regression=# select ctid,* from int8_tbl where ctid = '(0,3)'; ctid | q1 | q2 -------+------------------+----- (0,3) | 4567890123456789 | 123 (1 row) There is some code in 7.0 that tries to implement this in a fast way, but I don't know how well it works... regards, tom lane