Re: Relational Divsion

From: Rich Ryan <rryan_at_cshore.com>
Date: Thu, 01 Jun 2006 21:59:38 GMT
Message-ID: <erJfg.43977$Lm5.26532_at_newssvr12.news.prodigy.com>


"-CELKO-" <jcelko212_at_earthlink.net> wrote in message news:1149159270.208770.95870_at_i40g2000cwc.googlegroups.com...
>
> Relational division comes in several flavors -- Codd's, Todd's and
> Romley's are the major ones. Basically, you can do it as exact division
> or division with a remainder. Chris Date's version in his text book
> says that a division by an empty set returns all of the divisor; my
> version returns none of them (which is Date's preferred answer!).
> Romley's division classifies the relationship as "none match", "some
> match", "match exactly", and "match with a remainder", making it the
> best choice for reporting.
>
> I cover them in detail in SQL FOR SMARTIES and give Standard SQL code.
>

I have your SQL FOR SMARTIES, however it is unaccessable because of interior painting we're doing. I did find an article by you on DBAZine.com which helped. Also, Terry Halpin gives a good explanation in his Information Modeling and Relational Databases.

Speaking of Halpin, I used ORM on several projects. It was/is a great modeling method. A bit verbose(every attribute is modeled as a relationship), but the ability to define intra-predicate constraints made the effort worthwhile. As a methodologist at KnowledgeWare, I came up with a product that allowed you to describe the data side by stating facts, much like ORM. We even had a prototype software product to support it. Alas, ERD was and still is the mainstream technique. Oh yes, and Halpin is much smarter than I and did it right.

Rich Received on Thu Jun 01 2006 - 23:59:38 CEST

Original text of this message