Re: Does Codd's view of a relational database differ from that ofDate&Darwin?[M.Gittens]
Date: Sat, 2 Jul 2005 17:00:16 +0200
In article <IVhxe.135039$l56.6861917_at_phobos.telenet-ops.be>,
> It's the difference between informally stating that certain
> fields are surrogate keys and waving your hands a bit, and coming up
> with a precise formal theory of what that exactly means.
> > I guess I would be happier if this was presented as a
> > layer on top of the RM instead of as a replacement.
> Well, if that makes you happier then by all means let us present it as
> such. :-)
It would shorten most papers on OODB models down to a few paragraphs, I think. :)
> So where do you get the feeling hat this worldview would be
> somehow restrictive?
For one thing, I don't like being forced to decide whether something is an entity or a relationship. Is marriage an entity or a relationship? What is the difference? I note that this is resolved in ORM, where all the information is represented by relationship types, or fact types in the original terminology---just like in the RM. But then, I consider ER and ORM to be very different, as you know.
> > Anyway, there is one thing that is bugging me when I read articles about
> > object data models like Calvanese/Lenzerini and Van den Bussche /
> > Paredaens. If objects are just identity points, and the properties of
> > objects are other objects, where do actual data values enter into the
> > picture?
> Data values are special objects that have one or more representations
> associated with them by which they are identified. Like LOTs and NOLOTs
> in NIAM.
How do you determine whether a type consists of "objects" or "special objects"? Is there any consensus? What is the decision based on?
What is the point of having this dichotomy? It complicates the model; what are the gains?
In diagrams in the Bussche paper, how is "v12" (or Vertex v12) a different kind of concept than "0" (or Number 0)? How can you tell the difference?
-- JonReceived on Sat Jul 02 2005 - 17:00:16 CEST