Message-ID: <42CEA305.1050904@noao.edu>
Date: Fri, 08 Jul 2005 09:00:05 -0700
From: Steve Wampler <swampler@noao.edu>
User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317)
MIME-Version: 1.0
To: pgsql-sql@postgresql.org
Subject: Re: Make COUNT(*) Faster?
References: <9EF2931D-1097-470D-8ECB-59D00166C96F@apple.com>
	<60vf3mklue.fsf@dba2.int.libertyrms.com>
	<42CE89CA.2070607@noao.edu> <20050708143508.GD22725@wolff.to>
	<42CE96AF.3020903@noao.edu> <22916.1120836900@sss.pgh.pa.us>
In-Reply-To: <22916.1120836900@sss.pgh.pa.us>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Tom Lane wrote:
> Steve Wampler <swampler@noao.edu> writes:
> 
>>So, leave COUNT(*) alone.  But it would be very handy to have a
>>way to get an approximate table size that is more accurate than is
>>provided by a pg_class.reltuples that is only updated on vacuums.
> 
> If you want something cheap, you could use the same technique the
> planner uses nowadays: take RelationGetNumberOfBlocks() (which is
> guaranteed accurate) and multiply by reltuples/relpages.  I don't
> see anyplace where RelationGetNumberOfBlocks is directly exposed to
> users now, but it'd be trivial to code up a couple of C functions to
> provide this functionality.

Yes - this would be an excellent approximation for my needs!  The
solution that Dawid Kuroczko suggested (just call "explain select *
on ..." and parse the result) would be equivalent these days, right?
(I think in the 7.x versions the planner was just using pg_class.reltuples,
which wouldn't have helped.)  If true, I can handle that parsing myself
easily enough without exposing RelationGetNumberOfBlocks.

Thanks (Tom and Dawid)!

-- 
Steve Wampler -- swampler@noao.edu
The gods that smiled on your birth are now laughing out loud.