Re: Hierarchical Model and its Relevance in the Relational Model

From: Derek Asirvadem <derek.asirvadem_at_gmail.com>
Date: Mon, 2 Feb 2015 06:14:12 -0800 (PST)
Message-ID: <7c3de54e-cf64-4695-8255-453f880fc882_at_googlegroups.com>


> On Monday, 2 February 2015 22:46:19 UTC+11, Derek Asirvadem wrote:

> My practised opinion is, after 39 years in the industry, 36 in databases, 30 years pure Relational, is that all data in a database, is related, to all data, in that database. Obviously, not limited to "by referential constraint", related in various ways. At the very least, it is all related simply because the database is an recovery unit. One step up from that, that you will understand, is the referential constraint, so that is the limit for you guys. Codd taught me that their are myriad other relations, and applying the RM, I implement some of them. But that is about thirty years ahead of you guys, and I will not try to explain any of it here. We are stilt, heh heh, arguing about what a relation is, and whether hierarchies are part of the RM.

In case it needs to be said, I do not mean "Codd taught me" literally, in person. I mean through his papers and articles, through the RM. The more I applied the RM, the more I understood it, that is applied more than I previously thought. So I applied it more the next time, and in doing so, I learned that it applied in even more ways.

In that sense, Codd is still very much alive, much like my dead father is alive to me internally. His work is a living work, if you interact with it and respect as the Commandments for Relational Databases, you get way more out of it than if you argued and fought. That might take you ten or twenty years of incorrect implementations, as some of you have mentioned. Genuine disciples of Codd never had that problem.

Of course I implement 100% of all data controls in the db, using RDBMS facilities, declaratively. But there is much more to data integrity than that. There is a whole level of what I call Logical Integrity. To portray that, to document that, I use a set of diagrams, separate to, and entirely bound to, the full IDEF1X data model. Only because IDEF1X doesn't have it. IDEF1X is not like UML, where everyone and his dog has a different set of symbols and notations, so I don't interfere with it, I don't add symbols, I add a separate diagram.

But that is beyond the scope of this thread. I just wanted to clarify the "Codd taught me" statement, in case someone took that literally. Received on Mon Feb 02 2015 - 15:14:12 CET

Original text of this message