Date: Tue, 24 Jun 2003 09:33:22 +0200
From: Hubert Lubaczewski <hubert.lubaczewski@eo.pl>
To: "Ryan" <pgsql-sql@seahat.com>
Cc: <pgsql-sql@postgresql.org>
Subject: Re: aggregate question
Message-Id: <20030624093322.2051cbb5.hubert.lubaczewski@eo.pl>
In-Reply-To: <54717.65.102.128.233.1056394792.squirrel@fordparts.com>
References: <54717.65.102.128.233.1056394792.squirrel@fordparts.com>
Organization: EO Networks
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Mon, 23 Jun 2003 13:59:52 -0500 (CDT)
"Ryan" <pgsql-sql@seahat.com> wrote:

>  package_id      | integer | not null default
> I must be a total space case today because I can't hammer out the sql to
> get a listing of all the packages with a count() of the package_log by
> package_id.

if you would just package_id and count it would be simple:
select
	p.package_id,
	count(*) as count
from
	packages p
	left outer join package_log l on p.package_id = l.package_id
;

but, if you want all fields from packages it get trickier.
you can do it in two ways:

select
	p.*,
	(select count(*) from package_log l where l.package_id = p.package_id)
from
	packages p

or:

select
	p.*,
	coalesce(c.count,0)
from
	packages p
	left outer join (select l.package_id, count(*) as count from package_log l) c on p.package_id = c.package_id


should work - but i just wrote it "by hand", and didn't test it. anyway - it should give either working code or idea on how to achieve it.

depesz