Date: Thu, 17 Jun 2004 12:54:22 -0500
From: Bruno Wolff III <bruno@wolff.to>
To: Interzone <lists@interzone.gr>
Cc: pgsql-sql@postgresql.org
Subject: Re: Optimal query suggestion needed
Message-ID: <20040617175422.GA7171@wolff.to>
Mail-Followup-To: Interzone <lists@interzone.gr>,
	pgsql-sql@postgresql.org
References: <40D1AEB0.1010604@interzone.gr>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <40D1AEB0.1010604@interzone.gr>
User-Agent: Mutt/1.5.6i

On Thu, Jun 17, 2004 at 14:46:08 +0000,
  Interzone <lists@interzone.gr> wrote:
> I want to create a view that will have:
> from table t0 the elements "code", "address" and "mun"
> from table t1 the elements "code" and "pname"
> from table t2 the total number of elements, and the total number of 
> elements where avail = true, for every value t0_fk (foreign key to t0) 
> and t1_fk (foreigh key to t1).
> 
> After several attempts and changes as the requirements changed, I finaly 
> came up with that :
> 
> select t0.code, t0.address, t0.mun, t1.code as t1code, t1.pname 
> count(t2.code) as t2total, (select count(t2.code) as t2avail from t2 
> where t2.avail = true and t2.t0_fk=t0.code and t2.t1_fk = t1.code) as 
> t2avail from t0, t1, t2 where t2.t0_fk = t0.code and t2.t1_fk=t1.code 
> group by t0.code, t0.address, t0.mun, t1.code, t1.pname

This approach is actually pretty close. I think you just didn't pick a
good way to count the avail = true rows.
I think you can replace the above with:
select t0.code, t0.address, t0.mun, t1.code as t1code, t1.pname 
count(t2.code) as t2total, count(case when t2.avail then 1 else NULL) as 
t2avail from t0, t1, t2 where t2.t0_fk = t0.code and t2.t1_fk=t1.code 
group by t0.code, t0.address, t0.mun, t1.code, t1.pname