Re: Foreign superkey support

Erwin wrote:
> > I have read various classification systems for constraints, and they
> > all seem unnecessary to me. [...] For any
> > constraint, we have a set of relations that it is quantified over.
>
> I do think type constraints are different.

I'm not 100% sure I understand the term, but yes, I think they are different. Especially if they can be checked statically; constraints in general cannot.

> I also think that what I call "Tuple constraints" are interesting to be
> considered separately, I mean expressions that are well-defined "over"
> one single tuple, i.e. the only "parameters" in the expression are
> references to attributes of that tuple. From a certain perspective,
> they might even be regarded as type constraints for some given tuple
> type (and thus be "different" in the same sense as type constraints for
> scalar types are "different").

To me, these are better thought of as database constraints universally quantified over a single relation. (That is, "a > b" is just shorthand for "forall R(a,b), a > b".)

> All the others, I think, are indeed basically just database constraints.

Agreed.

Marshall Received on Wed Aug 09 2006 - 16:32:27 CEST