Re: Do Data Models Need to built on a Mathematical Concept?

Paul Vernon wrote:
> "Paul Vernon" <paul.vernon_at_ukk.ibmm.comm> wrote in message
> news:b8eg1v$1pbo$1_at_gazette.almaden.ibm.com...
> [snip]
>

>>His other question "How do we support multiple views of a system while
>>retaining the advantages of hierarchical structuring?", is, if you will

>>being partially answer by the more sensible efforts to integrate XML with

>>databases. Specifically I'm thinking of XQuery and XTables, that look much
>>more respectable than you might think would come out of such an effort. (See

You'll excuse me if what I read in the article triggered my recollection of this formidable saying of Dijkstra:

"In the software business there are many enterprises for which it is not clear that science can help them; that science should try is not clear either."

XML is doomed to be such an enterprise.

I find many of the (unspecified) premises of the article rather not happening in practice, if I'm ever going to see them I'll probably consider them bad engineering, and of course those premises do raise a hell of a lot of problems, that the proposed approach might or might not address.

The quality of ad-hoc solutions to ad-hockery kind of problems is certainly not going to be great, and frankly speaking reading the article and seeing some of the unnecessary complexities in their code samples, one has to wonder how have we got here ? Where do we depart them and what is such a serious problem that requires such a complex of a solution for moving a bunch of bytes around the computer ?

Well, the point that the authors departed from is that you have to have XML in the database, or to "integrate" XML whatever that means. Yes, of course, then you have problems, lots of them.

Well, those are not the problems of my users. The problems of my users are orthogonal to whether I store native XML in relational databases (which is something that I refrain from qualifying for the sake of civility). So the article has 0 value for me, and I consider that IBM will likely spend a hell of a lot more money for no benefit of their users. Reminds me of the sheer engineering effort of Oracle to put first objects in the database as a PL/SQL extension, and then make room for them in tables, of course I would regard any usage of those features a serious engineering mistake.

Of course, I can imagine a reply, but you have to deal with XML because it is industry standard, and other systems will want XML and so on so forth, and even if XML is technically less than desirable, the sheer force of the whole industry being behind it makes it a better engineering solution to virtually all its competitors, byt the amount of tools put at your disposal, your needs fo integration, etc, etc, etc

Yes. I will admit XML is a neccessary evil for the time being, but I won't make a deal with the devil for the sake of it :) And the forces of industry might be great, but they are petty nothing compared to the daunting force of Mathematics.

That's what made COBOL obsolete, C++ obsolete, and will make Java , UML, C# and XML onsolete soon. That's what makes Lisp still alive with a design dating back to 1958, that's what makes Dijkstra's Discipline Of Programming be still a book in print. You can have all your industry behind you, that's easy part, the hard part in our business is to deal with Mathematics.

Best regards,
Costin Cozianu Received on Sun Apr 27 2003 - 19:31:38 CEST