Re: One-To-One Relationships

From: David Cressey <>
Date: Sun, 02 Dec 2007 18:33:30 GMT
Message-ID: <_TC4j.5085$VJ.3384_at_trndny05>

"Marshall" <> wrote in message
> On Dec 2, 2:46 am, Jan Hidders <> wrote:
> >
> > Actually the problem is with OO is the same as with the RM, but at the
> > other side of the spectrum. In OO the notion of entity is assumed to
> > be central, in RM the notion of relationship is. Both are wrong.
> This is a very provocative statement. I wish I understood it!
> I understand what you are saying about OO. The way I
> used to hearing OO discussed is vs. FP.
> In OO the object is the organizational focus, and
> functions are organized around this. In FP, functions
> are the organizing focus, and structures are
> organized around functions. Thus in OO it is
> easy to add new datatypes but hard to add new
> functions to datatypes, whereas in FP it is easy
> to add new functions but hard to add new
> datatypes to functions.
> However neither issue seems to apply to the RM.
> The RM, it seems to me, uses a single abstraction
> to describe both entities and relationships. So your
> statement that the RM favors relationships over entities
> is hard for me to make sense of. Please elaborate.
> Marshall

If we are going to bring OO into the discussion we really ought to distinguish between OOA on the one hand and OOD (plus OOP) on the other. I once read a book on OOA by Peter Coad. It seemed to me that ER modeling fit quite well into OOA modeling. I've never got a chance to put OOA into practice, but it seems to me that it would to what ER does only more so, because it deals with both state and behavior. ER deals with state (data) but not with behavior.

However, just because there is a good fit between OOA and ER does not mean that there is necessarily a good fit between OOD and database design (whether relational or not). Received on Sun Dec 02 2007 - 19:33:30 CET

Original text of this message