Message-ID: <009301c4072a$19b28fd0$7201a8c0@mst1x5r347kymb>
From: "Iain" <iain@mst.co.jp>
To: "Enver ALTIN" <enver.altin@frontsite.com.tr>
Cc: <pgsql-sql@postgresql.org>
References: <1078937297.29628.12.camel@enver.office.frontsite.com.tr>
	<20040310190250.GA6629@wolff.to>
Subject: Re: randomized order in select?
Date: Thu, 11 Mar 2004 14:31:26 +0900
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit

If you have a lot of tips, you could create a unique indexed tip number
column. Select the highest tip number using:

select tip_number from tips order by tip_number desc limit 1;

Then generate a random number and select using that tip_number.

Of course, you would have to allow for the possibility of missing tip
numbers, by repeating the random number generation/read sequence until you
find something. Since the tip_number isn't the PK of the table, you can
regenerate the tip numbers to eliminate holes from deletions any time you
like. Just reset the sequence to 1 and update all rows with the
nextval(tipnumber_seq).

Sounds like a lot of work to me though...