Re: Database design

Mark Johnson wrote:
>
> A relation is a set of unsorted, unordered unique tuples,

There's some signficant redundancy in there:

set => unsorted
unsorted => unordered
unordered => unsorted
set => unique

How about we refer to the classical definition:

"a relation is a subset of the cross product of two sets."

This is the definition of a binary relation, which is the kind that gets the most ink in the math books. This defintion is easily extended to n-ary relations:

"an n-ary relation is a subset of the cross product of n sets."

> which may only contain data - not text.

Text is a kind of data.

 > But which leads back to the question. If a sort column/attribute is
> included, is it definitionally destroyed by saying it doesn't sort, or
> that whatever is sorted is not a, relation?

No. This never happens. By definition.

An ordered set is something else: it is pair. The pair is (relation, ordering relation.)

Just for fun, consider a cardinality-one relation. Is it in order? Yes, it is. For every possible traversal strategy for the relation, and for every possible ordering relation, it is in order. The same is true for a cardinality-zero relation. And yet we still do not call it an ordered set.

Consider a cardinality-two relation. For every order relation that could exist and for which the relation is in order, there exists another ordering relation for which the set is not in order.

Again, I suggest reviewing some basic texts. Newsgroups are not a good way to learn the basics; a book is better for that. I suggest Schaum's Outlines: the "Set Theory and Related Topics" is a good reference.

> Is it just semantics, as the term is commonly used?

I have no idea what you mean by "just" semantics. Semantics is the foundation; without semantics there is nothing.

Marshall Received on Wed Feb 22 2006 - 05:05:22 CET