Re: Domain

From: Marshall Spight <>
Date: Tue, 30 Dec 2003 17:47:40 GMT
Message-ID: <0TiIb.15996$xX.40748_at_attbi_s02>

"Dawn M. Wolthuis" <> wrote in message news:bsqv24$r4r$


> I'd rather we start with Java
> than SQL for defining types since what we need seems to be pretty much there
> already.

Me, too. Java would be a huge step up from SQL with regards to user-defined types. But I'd rather we looked to something more sophisticated, like Haskell, for a new type system.

> I wrote it from a mathematics background, adding in the fact that the set
> must be named in relational databases where it need not be in mathematics.
> I know it is not always how folks perceive domains, but all it really is is
> just a set that limits the possible values for an attribute, right?

Sure. Actually, it doesn't seem like all that useful a word to me, because we already have a word for it: "type." This word is better understood and more commonly used.

> Beware
> of anyone who adds in words like "atomic" into definitions of "domain".
> Unless we (as a profession) can find a useful definition of atomic, I'd like
> to leave such useless jargon out of database discussions altogether.

Agreed. But I think atomic has a well-established meaning: "of or employing nuclear energy." No, wait, it's "indivisible." An atomic operation is one that cannot be broken down into smaller operations; it succeeds entirely or not at all. An atomic value is a value which cannot be broken up into subcomponents.

Marshall Received on Tue Dec 30 2003 - 18:47:40 CET

Original text of this message