Re: A Normalization Question

"Neo" <neo55592_at_hotmail.com> wrote in message news:4b45d3ad.0407201651.792cdd6c_at_posting.google.com...
> > > The string 'brown' is a value in the context of a tuple and attribute,
> > > not by itself. Here we are discussing the string 'brown' by itself.
> >
> > No, that's exactly what I meant: the string "brown" is a value, all by itself.
>
> Please show that "brown" is a value in RM without context to tuple and
> attribute.

The definition of value has nothing to do with tuples and attributes. A value is simple a member of a set. In this case, the set is "list of characters."

> Also see C.J. Dates "Intro to Db Sys", 6th Ed, Chapter 19, Section 5,
> p560 Further Topics / Relation-Valued Attributes, where he discusses
> the concept of values being a relation: "as we explained that such
> 'scalars' can have a structure of arbitrary complexity when viewed at
> a lower level of abstraction and we mentioned that they might even be
> relations. In other words, we can have domains, and therefore
> attributes, that are relation-valued, and thus have relations that
> contain other relations inside themseleves and so on recursively to
> any levels. Not only does this concept appear in TM also, it has been
> implemented.

I don't see the relevance. This excerpt seems to simply be advocating for relation-valued attributes, which I am on the record as being in favor of. (I think you left off the close-quote; the last sentence is yours, I'd guess.)

Marshall Received on Wed Jul 21 2004 - 03:58:53 CEST