Re: 1GB Tables as Classes, or Tables as Types, and all that refuted

From: Dawn M. Wolthuis <dwolt_at_tincat-group.comREMOVE>
Date: Mon, 13 Dec 2004 15:31:33 -0600
Message-ID: <cpl1nm$5ap$>

"Alfredo Novoa" <> wrote in message
> On Sun, 12 Dec 2004 10:41:09 +0100, "Ja Lar" <ingen_at_mail.her> wrote:
>>> You are playing with words again. Graph theory is not better or worse
>>> than predicate logic or differential calculus, but the Relational
>>> Model is dramatically superior to the graph based data models.
>>Don't you remotely see that you are "playing with words" yourself.
> No, I am not.
>>You are giving no arguments what so ever, but "opininons" and adjectives
>>"horrendous complex and inflexibility", "severe inherent", "from 1st
>>century" (well, that's younger than logic btw), "dramatically superior"...
> They are well known facts that everybody that works with databases
> should know.
>>Repeating dislikes with (new) adjectives does not prove any point (except
>>perhaps that there is no proof).
> They were proven decades ago. They were inquestionable facts when I
> studied database theory in the university.

As best I can tell when reviewing the history, the word "proven" is FAR too strong a word. This has been unquestioned for too long or at least by far too few people in the academy. That, I believe, will change within the coming decade. I didn't question it either, until I was trying to figure out why relational theory seemingly did not improve either s/w development flexibility, quality of data, or maintainbility on the whole (there are definitely anecdotes of some improvements). When I took a peek at this supposed "proof" for relational theory, the emperor seemed a bit on the naked side. The fact that some still think the emperor is clothed doesn't necessary mean he is.

> You have a lot of material about this in the oldest volumes of the
> "Relational Database Writtings" serial.
>>You may very well have the correct conclusions (in fact I think you have
>>So, what _is_ the "inflexibility" in XML?
> The well known inflexibility of the Hierarchical Model.

Where's the meat behind this claim. It was once well known that the world was flat. If there is no better arguement than "well known" then that speaks for itself, perhaps?

>>What _are_ the inherent AND severe problems in graph models (we are not
>>talking implentation, remember).
> The navigational and procedural nature of these data management
> approaches among other things.

You are just labeling things and calling them bad. WHAT is bad about graph data models?

>>By _what measure(s)_ are Relations Model dramatically superior to graph
>>based data models (again, not implementation).
> Flexiblility, simplicity, power, and all that you learned when you
> studied database theory.

But that has not been my experience AND I can find no emperical data nor logical arguement to show me otherwise. So, where is the meat of this argument against graph-based data models other than a bunch of name alling -- "procedural, navigational, pointers, complex, Network, Hierarchical, etc" The arguement against graph-based models has no clothes -- am I right?

cheers! --dawn

> Regards
Received on Mon Dec 13 2004 - 22:31:33 CET

Original text of this message