Date: Tue, 28 Aug 2007 09:26:04 -0400
From: Andrew Sullivan <ajs@crankycanuck.ca>
To: pgsql-sql@postgresql.org
Subject: Re: Database normalization
Message-ID: <20070828132604.GB461@phlogiston.dyndns.org>
References: <99b656cb0708210709j5182b73cna63a2d275efa0e84@mail.gmail.com>
	<99b656cb0708280447x4eb7c45axec01c8d0ebc6c82d@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <99b656cb0708280447x4eb7c45axec01c8d0ebc6c82d@mail.gmail.com>
User-Agent: Mutt/1.5.9i

On Tue, Aug 28, 2007 at 12:47:45PM +0100, Sebastian Ritter wrote:
> > The update/message format is exactly the same for both. Should I make two
> > different tables:

> >   one table with extra columns : is_client, client_id, service_id, where
> > either client_id or service_id would be null depending on the is_client
> > boolean?

Is the rest of the data the same?  If so, then one table is right. 
If not, then more than one table.  In either case, I really hate the
idea of two columns, one of which is always null.  But if you're
going to do that, make sure you add a CHECK constraint where !(col1
IS NULL and col2 IS NULL).

A

-- 
Andrew Sullivan  | ajs@crankycanuck.ca
I remember when computers were frustrating because they *did* exactly what 
you told them to.  That actually seems sort of quaint now.
		--J.D. Baldwin