Re: MV Keys

Marshall Spight wrote:
> Jon Heggland wrote:
> >
> > How do you define "repeating group" as opposed to "nested relation",
> > then?
>
> A repeating group is a set of attributes, all doing the same thing,
> but repeated to allow higher cardinality. Nested relation/RVA
> is a single atribute whose type is relation. An example of a
> repeating group is allowing up to three phone numbers for a
> customer by having three attributes, PhoneNumber1,
> PhoneNumber2, and PhoneNumber3. This approach has
> significant disadvantages.

I'm catching up, but I don't think this is what Codd meant by the phrase "repeating group." We used to use "repeating group" to refer to data defined with an OCCURS clause in COBOL, for example. I do think you could claim there to be a difference in that a repeating group can be multiple attributes, such as PhoneType & PhoneNumber that is then repeated, while nested relations would be a single attribute with a value that is a relation of those PhoneTypes and PhoneNumbers. But from a practical standpoint, I don't see a difference provided that each has a means of naming the nested relation as well as it's component attributes.

> (I was under the impression this usage was quite standardized;
> yes? No?)

No

> > I assume "nested relation" and "relation-valued attribute" is the
> > same thing.
>
> Yes; that's how I'm using the terms. For some reason I prefer
> the "nested" term to the RVA term.

I don't know, but suspect your reason for your choice is related to your design for an implementation.

Cheers! --dawn Received on Fri Mar 03 2006 - 06:09:34 CET