Message-ID: <3928F748.3F51EA87@tm.ee>
Date: Mon, 22 May 2000 12:00:56 +0300
From: Hannu Krosing <hannu@tm.ee>
Organization: Trust-O-Matic =?iso-8859-1?Q?O=DC?=
MIME-Version: 1.0
To: Chris Bitmead <chris@bitmead.com>
Cc: Chris Bitmead <chrisb@nimrod.itg.telstra.com.au>,
	PostgreSQL Development <pgsql-hackers@postgresql.org>
Subject: Re: Thus spoke SQL3 (on OO)
References: <Pine.LNX.4.21.0005201507330.423-100000@localhost.localdomain>
	<00052104000003.14806@comptechnews> <392830DA.B5546EF7@bitmead.com>
	<00052108053904.14806@comptechnews>
	<39286B3F.F4594E40@bitmead.com> <39283008.ED83E94@tm.ee>
	<39287873.D023D7E5@nimrod.itg.telecom.com.au>
	<3928F01B.B0C6D563@tm.ee> <39298E29.16B280CC@bitmead.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Chris Bitmead wrote:
> 
> Hannu Krosing wrote:
> >
> > Chris Bitmead wrote:
> > >
> > > Hannu Krosing wrote:
> > >
> > > > Chris, what is your position on having a single primary key for all
> > > > inherited columns ?
> > >
> > > What is the significance of a primary key compared to any old unique
> > > key?
> >
> > I don't know ;) Some theorists seem to think it important, and PG allows
> > only one PK per table.
> >
> > I just meant that primary key (as well as any other uniqe key) should be
> > inherited from parent table
> 
> What object theory would say is that oid uniquely identifies an object.
> Other unique keys should usually be inherited.

it would be hard to define RI by just saying that some field references "an
OID",
often you want to be able do define something more specific.

It would be too much for most users to require that all primary and foreign
keys 
must be of type OID.

It about flexibility, much much like the situation with SERIAL vs.
INT DEFAULT NEXTVAL('SOME_SEQUENCE')

------------
Hannu