Message-Id: <l03130305b34cd633efee@[147.233.159.109]>
In-Reply-To: <371B2F13.D8B446F6@3skel.com>
References: <199904152011.QAA26538@fnur.3skel.com>
	<371ACCB0.6A437C5@TurnAround.com.au>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 28 Apr 1999 18:58:14 +0300
To: Dan Janowski <danj@3skel.com>, J_Shevland@TurnAround.com.au,
	M Simms <grim@argh.demon.co.uk>
From: Herouth Maoz <herouth@oumail.openu.ac.il>
Subject: Re: [SQL] OUTER JOINS
Cc: pgsql-sql@postgreSQL.org

At 16:26 +0300 on 19/04/1999, Dan Janowski wrote:


> Thanks for giving it a stab. Looking at the available
> operators,
> there is no =* or *= ops available. Although I am curious
> where it
> comes from.

To the best of my knowledge, outer joins are not yet supported in
PostgreSQL. As far as I recall, you can achieve the same effect with a
union.

For example, if you have a people table and a pets table, where each
person's pet is joined to the owner by the owner's id colum, you'll
theoretically do something like:

  SELECT surname, firstname, pet_name
  FROM people, pets
  WHERE id *= owner_id
  ORDER BY surname, firstname;

So instead, you'll have to do something like:

  SELECT surname, firstname, pet_name
  FROM people, pets
  WHERE id = owner_id
  UNION
  SELECT surname, firstname, null
  FROM people
  WHERE not exist (
     SELECT 1
     FROM pets
     WHERE owner_id = id
  )
  ORDER BY surname, firstname;

Excuse me if I goofed something syntactically, I don't have time to
actually create the tables and test the above. Note, however, that a UNION
removes duplicates.

Herouth

--
Herouth Maoz, Internet developer.
Open University of Israel - Telem project
http://telem.openu.ac.il/~herutma