To: "Ozer, Pam" <pozer@automotive.com>
cc: pgsql-sql@postgresql.org
Subject: Re: Random sort with distinct 
In-reply-to: 
 <216FFB77CBFAEE4B8EE4DF0A939FF1D10182C8@mail-001.corp.automotive.com>
References: 
 <216FFB77CBFAEE4B8EE4DF0A939FF1D10182C8@mail-001.corp.automotive.com>
Comments: In-reply-to "Ozer, Pam" <pozer@automotive.com>
	message dated "Fri, 01 Oct 2010 13:51:38 -0700"
Date: Sat, 02 Oct 2010 09:52:50 -0400
Message-ID: <5070.1286027570@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

"Ozer, Pam" <pozer@automotive.com> writes:
> Select Distinct VehicleMake, VehicleModel
> From VehicleYearMakeModelTrim
> Order by random()
> Limit 10;

> I don't want to bring back the random number I just want the sort order
> to be random. How can I sort randomly?  This query breaks because
> random() is not in the select.

Well, yeah: the implication of the ORDER BY is that a new random value
is to be computed for each row of VehicleYearMakeModelTrim.  After you
combine rows with DISTINCT it's not clear which of those values should
be used to sort a grouped row.

You need to put the DISTINCT and the ORDER BY in separate query levels,
like this:

select * from
  (Select Distinct VehicleMake, VehicleModel
   From VehicleYearMakeModelTrim) ss
Order by random()
Limit 10;

			regards, tom lane