Re: Nested Sets vs. Nested Intervals

From: Alexandr Savinov <spam_at_conceptoriented.com>
Date: Thu, 10 Nov 2005 10:14:48 +0100
Message-ID: <43730f94$1@news.fhg.de>

Mikito Harakiri schrieb:
> amado.alves_at_netcabo.pt wrote:
>

>>Mikito, ..., the join operation per
>>se does not give you a graph, ...

>
>
> Take a set of all tables as a set of vertices. Take a set of all
> foreign key integrity constraints as a set of edges. Do you challenge
> the idea that I just defined a graph?

Yes, but in such a formulation it is not RM! It is a graph model implemented via RM.

Nobody argues that many things can be implemented via RM including graph-based models. I could even imagine that someone could implement OOP or AOP techniques using RM. Or, vice versa, if I implement a behaviour of RM using Turing machine then it does not mean that Turing has invented RM.

The question is (as far as I understand it):

Is it more appropriate to view the (data) world as a set of relations (with some operations, constraints etc.) or as a graph?

-- 
http://conceptoriented.com

Received on Thu Nov 10 2005 - 03:14:48 CST