Re: MultiValue Databases

From: Jon Heggland <heggland_at_idi.ntnu.no>
Date: Tue, 7 Jun 2005 19:04:09 +0200
Message-ID: <MPG.1d0ff1ea4133bf40989687_at_news.ntnu.no>


In article <1118158988.399655.126370_at_g47g2000cwa.googlegroups.com>, neo55592_at_hotmail.com says...
> Could you explain why RM is more
> general/superior when it exposes more of the underlying data model and
> implementation details (ie TABLES containers, ID fields, data types,
> etc).

Sorry about barging in without having read the rest of the thread, but...

Relvars (tables) and data types (domains) are not implementation details. They are integral to the semantics of the RM, just as nouns and sentences are essential to natural language.

The ID fields are not needed. In fact, a single relvar is sufficient:

CREATE TABLE person_phone (char name, char number);

---unless you want to record more information about persons and phone numbers, or want to support several persons with the same name.

> Since I am obviously no RM expert, could you help write a query
> to find a person with phone numbers 111-1111 and 222-2222. I started
> the query, but I could not complete it.

I may very well be wrong here, but isn't that just a simple division?

person_phone DIVIDEBY { { number '111-1111' }, { '222-2222' } }

SQL doesn't support it, though (afaik (perhaps D4 does?)). Blame SQL, not the RM.

-- 
Jon
Received on Tue Jun 07 2005 - 19:04:09 CEST

Original text of this message