Re: Access as a RDBMS--why the multiple relationships?

From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Fri, 14 Dec 2007 17:46:47 -0400
Message-ID: <4762f9cb$0$5297$9a566e8b_at_news.aliant.net>


raylopez99 wrote:

> OK, this is about my fifth day doing databases, and I've read about
> 1NF, 2NF, 3NF here: http://www.utexas.edu/its-archive/windows/database/datamodeling/rm/rm7.html,
> and I have a crappy book by a guy named Louis Davidson (APress), full
> of typos, bloated, and light on theory, but here's my question
> concerning Microsoft Access.
>
> I notice in Microsoft Access the relationship chart can, unlike most
> textbooks, have TWO, not just one, relationship arrows between
> tables. But I think (and I just want confirmation of this) that one
> of these two relationship arrows is bogus, and more like a query than
> a true relationship.
>
> Here goes:
>
> relationship arrow one (sorry I can't do the SQL statement thing from
> my head, like some of you can, so I'll use prose instead):
>
> entity TABLE A has Primary Key PKA that migrates to entity TABLE B as
> a (manditory, and non-exclusive, but doesn't matter) Foreign Key FKB.
> In the "relationship diagram" (Access likes to use diagrams), this is
> properly shows as a 1 to infinity symbol, no problem, this is also in
> my textbook.
>
> But I also using Access can set up ANOTHER SECOND?! relationship
> comprising a non-primary, non-key, non-unique field in TABLE A with a
> non-unique field in TABLE B. In the relationship diagram this shows
> up as "indeterminite" (no infinite or numbered symbols) and further
> you cannot check the box for "enforce referential integrity" because
> you get the error message (which makes sense) of "No unique index
> found for the referenced field of the primary table" (since the
> referenced field is not a primary key, or any kind of key).
>
> I just want confirmation that the SECOND relationship above is simply
> a bogus construct of Access, akin to a query constraint of some sort,
> and not really a 'relationship' as defined by RDBMS theory.
>
> RL

RDBMS doesn't really define 'relationship'. It's true Codd mentioned the name in an early paper, but I doubt his use of the word in any way resembles the crap that MSFT has been feeding you.

Relational is not named for referential integrity. It is named for relations as in those things you may have studied in a math course on functions and relations. Received on Fri Dec 14 2007 - 22:46:47 CET

Original text of this message