OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] XML and the Relational Model (was Re: [xml-dev] A standar

[ Lists Home | Date Index | Thread Index ]

Hmmm, and I suppose I could respond that Ada was never fully appreciated 
either, even unto this very day. But that would also be off topic. Few 
scientists who play by all the rules are ever appreciated, even within 
their own communities, in my opinion.

Not to give away the store here, but I like doing XML docs with dbms 
structures in systems like Oracle, thereby gaining I think the best of both 
worlds, relational and XML, even if the doc is like gigabytes in size. Nor 
does such an approach require a heavy metal solution, or TeraText or other 
bleeding edge product sets. I just am not going to advocate, as far as my 
own work goes, what I consider to be a poor practice, or a failure prone 
practice - namely Pure XML.

My point is that while doing a database as rows and columns of data in a 
spreadsheet (like Excel) within a document (like Word) within files across 
directories is just not the best approach, even if it does initially run in 
like seconds. The maintenance is murder.

Various math proofs point this out, from both the Relational Model (dbms, 
RDBMS) side and from the Heiarchichal Data Model (XML) side. Various well 
known business cases also exist, in casual journalism as well as academic 
journals and academic text books, showing the problems with this kind of 

If the pointy haired boss realized he was authorizing the creation of a 
multi-level spreadsheet data store (with variations), across desks and 
across directories, even he would not do it, having learned from the 
termination of his predecessor for exactly that reason (hundreds or 
thousands of spreadsheets with little or no documentation, over half of 
which are essential to the core business processes). Moral:  Information 
Islands are a Bad Thing in almost all cases.

Now, having said all that it seems important to reiterate what I have said 
before - that XML is the best tool for the job in some cases, and a dbms is 
the best tool for the job in some cases, and neither is __always__ the best 
or "only" tool that should be considered.

It seems to me that the information island thing is happening, is happening 
alot, and that there will be hell to pay (but not by me, thank you very 
much). The pointy haired boss dude is going to let it land on your desk, or 
mine, if he can... and I Do Not Want It.


At 10:51 AM 8/25/2003 -0700, Mike Champion wrote:
>--- Jonathan Robie <jonathan.robie@datadirect.com>
> > >Reconciliation?  You mean so that RM is the same as
> > a Tree Model, and vice
> > >versa? I suppose a translation or interpretive
> > method can be done, but
> > >that does not make RM native to the Tree Model, or
> > vice versa.
> >
> > Of course not. Every model is a universe unto
> > itself. Computer Science
> > allows more than one model. Mappings among models
> > are useful.
>Yup. And new models can come along that transcend the
>competing models into special cases of a more abstract
>model.  I guess the canonical example would be the
>wave vs particle models of light; to the best of my
>very limited understanding, they can be reconciled as
>special cases of a quantum electrodynamical
>thingamabob.  My original point was that people are
>actively looking for tree-relational thingamabobs that
>would do likewise for the RM and XML.
>In the meantime, people gotta do what they gotta do to
>get the job done, I guess. I'd be very surprised to
>find a truly successful pure relational approach to
>the kind of terrabyte-scale, evolving document
>scenario that started this thread.  An ad-hoc
>combination of the salient ideas from the relational
>approach (e.g. normalization, using value-based joins
>rather than hard-coded pointers or entity references
>to compose document components), the XML approach
>(e.g. hierarchical containment as an efficient way to
>bundle tightly related bits of information), and
>practical DBMS engineering techniques (e.g. indexes,
>transaction processing, database distribution and
>replication) seems like the most sensible approach to
>real-world problems.
>Totally Off-Topic: if Marie Curie had "waited to do
>the math" [as someone in one of these threads
>suggested] she might well have avoided an early death
>from cancer, but she might well have avoided the
>immortality she achieved by following her vision.
>Comparison's with Rosalind Franklin seem apt -- she
>did wait for the hard data rather than playing with
>ball and stick models, and died young anyway; it was
>decades before her contributions were recognized.  Not
>too many of us would last long on the payroll after
>assuring the Pointy Haired Boss that although it takes
>hours to do each 100-way join required to perform a
>transaction that was specified to be done in seconds,
>he/she should take comfort from the database
>application's firm grounding in predicate logic and
>set theory :-)
>The xml-dev list is sponsored by XML.org <http://www.xml.org>, an
>initiative of OASIS <http://www.oasis-open.org>
>The list archives are at http://lists.xml.org/archives/xml-dev/
>To subscribe or unsubscribe from this list use the subscription
>manager: <http://lists.xml.org/ob/adm.pl>


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS