Re: DB2

Thank you for the reference Jan.  Joe is right, this is about opinions, and
frankly Codd misses the mark by a very long ways in my opinion.  

Codd is attempting to deal with far more than database relationships.   Codd
is using his rules to both define the relationships and to also make
judgements about them.  There seems some interesting concepts, including the
comments about the backdoor changes, but aren't those sorts of abilities
just an attribute?  The rightness and wrongness of that attribute is a
judgement, not a rule.
  
For a comparison, lets say it like this:  My mail is mail.  My own opinion
is that most of my mail is junk, not mail, but so what.  That's just my
judgement  -  it is still mail.   Same with relational data bases.  There
may be opinions about the quality of a relational data base, but they can't
be defined out of existence by Dr. Codd.
 
These rules seem like a simple attempt by a college professor to show his
students good design technique.  His discussion of a primary key is trivial
and probably harmful.  All he really should care about is whether or not the
ecosystem can return any unique piece(s) of data at anytime.  What good is a
primary key of Employee ID when you have no way of discovering the employee
s ID?   

The idea of defining nulls is nice too, but why is that worthy of a rule?  
I am sorry, but my own opinion is that Codd's work is both primitive and
crude.  Essential at the time, and contributed to our understanding, but no
longer useful as a guideline for design.  
 
---------------------------------------------------------
Booth Martin   http://www.MartinVT.com
Booth@xxxxxxxxxxxx
---------------------------------------------------------
 
-------Original Message-------
 
From: Midrange Systems Technical Discussion
Date: Thursday, March 13, 2003 14:55:23
To: Midrange Systems Technical Discussion
Subject: Re: DB2
 
Leif Svalgaard wrote
It was not relational at the time and still isn't. The relational stuff is
built on top of a basic ISAM-type database allowing split-key
alternate indices (A.K.A. logical files).

Reply:

I have always maintained that the DB on the AS/400 is only relational 
insofar as:
The design of the database and
The application that maintains it' s integrity.

If a database does not have the ability to truly ensure it's own 
referential integrity, is it a true relational database?
There are some rules with which it should comply. Does DB/2, or is it 
only now starting (attempting) to comply?

The following is interesting:
http://www.frick-cpa.com/ss7/Theory_RelationalDB.asp

Cheers.

Jan.