Re: domain theory and databases

neo55592_at_hotmail.com (Neo) wrote in message news:<4b45d3ad.0305160838.3c89a682_at_posting.google.com>...
> The definition of a mathematical domain at the above site seems
> slightly different than the definition of domain in the relational
> data model(rdm). In rdm, a domain seems to be a set of "scalar" values
> that are in no particular order. In constrast, the mathematical domain
> seems to be a set of ordered values. If the values represent pig,
> banana, book, etc, how could they be logically ordered. It is true
> they could be ordered by their name, but they would be ordered
> differently in different languages thus the imposed order seems to be
> arbitrary. In the general/natural case, aren't the values of a domain
> unordered?

Yes, that does seem strange. I can't see where the relational model would require domains to have an ordering. Clearly there would always be an ordering by looking at the physical representation but, as you say, not at the logical level. For example a domain of pictures or videos or sounds.

Maybe the term "domain" is just overloaded so its mathematical meaning is different from its relational meaning.

In fact, "domain" has another mathematical meaning as well: http://www.wikipedia.org/wiki/Domain_(function) If f is a function mapping the set A to the set B, A is called the "domain" of the function f.

I suppose this is closer to the relational meaning as a relation can be thought of as a truth-valued function with "domain" (in this sense) being the cross product of the domains (in the relational sense) of the columns.

Paul. Received on Mon May 19 2003 - 11:17:05 CEST