Message-ID: <46C0B994.6090906@obviously.com>
Date: Mon, 13 Aug 2007 13:05:40 -0700
From: Bryce Nesbitt <bryce1@obviously.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US;
	rv:1.8.1.4) Gecko/20061107 SUSE/1.1.2-1.1 SeaMonkey/1.1.2
MIME-Version: 1.0
To: pgsql-sql@postgresql.org
Subject: Getting pk of the most recent row, in a group by
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

I've got a table of "coupons" which have an expiration date.  For each
type of coupon, I'd like to get the primary key of the coupon which will
expire first. 

# create table coupon
(
        coupon_id serial primary key,
        type varchar(255),
        expires date
);
insert into coupon values(DEFAULT,'free','2007-01-01');
insert into coupon values(DEFAULT,'free','2007-01-01');
insert into coupon values(DEFAULT,'free','2007-06-01');
insert into coupon values(DEFAULT,'free','2007-06-01');
insert into coupon values(DEFAULT,'50%','2008-06-01');
insert into coupon values(DEFAULT,'50%','2008-06-02');
insert into coupon values(DEFAULT,'50%','2008-06-03');

The desired query would look like:

# select coupon_id,type,expires from coupon where type='free' order by
expires limit 1;
 coupon_id | type |  expires  
-----------+------+------------
         1 | free | 2007-01-01


But be grouped by type:

# select type,min(expires),count(*) from coupon group by type;
 type |    min     | count
------+------------+-------
 free | 2007-01-01 |     4    ; pk=1
 50%  | 2008-06-01 |     3    ; pk=5

In the second example, is it possible to get the primary key of the row
with the minimum expires time?

-- 
----
Visit http://www.obviously.com/