Re: abstraction of table relationships

Jan Hidders wrote:
> Marshall wrote:
> > [...] He gave the example of Books and Authors.
> > If they have a many-to-many relationship, he said it would
> > be better if one could query all books by a given author
> > in a simpler way, and have the system find the relationship
> > automatically. So you could, say, simply:
> >
> > SELECT * from Books where AuthorId = 'Jane Austen';
> >
> > even if Author weren't a field of Books. The system
> > would look at the schema for Books, and locate the
> > "path" in the system whereby it could uniquely attribute
> > an Author.
>
> Oh my God! Do we really have to debate the Universal Relation Model
> again? :-)
>
> -- Jan Hidders

Speaking of which, Ullman presented at 2006 SIGMOD and talked a little about how the concept is reinvented so often. http://www-db.stanford.edu/~ullman/. Click the "slides" link under 2006 SIGMOD Talk section.

His common theme pounds on practical versus theory and what the two can take from one another. The Universal Relation is one of the prime examples.

• Dan