Message-ID: <001601c3d32c$dd0267d0$7201a8c0@mst1x5r347kymb>
From: "Iain" <iain@mst.co.jp>
To: "Tom Lane" <tgl@sss.pgh.pa.us>
Cc: <pgsql-sql@postgresql.org>
References: <013d01c3ca89$7307d7d0$7201a8c0@mst1x5r347kymb>
	<20597.1072369941@sss.pgh.pa.us>
Subject: Re: not in vs not exists - vastly diferent performance 
Date: Mon, 5 Jan 2004 10:40:13 +0900
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-2022-jp"
Content-Transfer-Encoding: 7bit

Hi Tom,

> You didn't say what PG version you are using, but I'd venture to bet
> it is pre-7.4.

Sorry I didn't give the version number, here is the output from select
version():

PostgreSQL 7.4 on i686-pc-linux-gnu, compiled by GCC 2.95.3

This is the first time that we have compiled the system from source,
previously we used rpm, but it seems very straight forward. I don't know if
there may have been something we did wrong there.

I understand that every DB environment is different, but it sounds like for
the example I gave, you expected that the NOT IN optimization would have
chosen a better plan. As to why it didn't I have no idea at this stage. If
there is anything you want me to do to test it, for example on the
regression test DB, then let me know.

regards
Iain