[
Lists Home |
Date Index |
Thread Index
]
--- lbradshaw@dbex.com wrote:
> I am saying I have not seen a proof that XML
> supports, or can be engineered
> to support, the Relational Model.
I'm not sure what you mean by that. Clearly any XML
schema can be normalized into relations (didn't Codd
prove something to that effect?). I'm less sure about
this, but my reading of C. J. Date's various
discussions of the relationship between the RM and the
OO suggest that one could condider a particular XML
schema a "type" and -- so long as a useful set of
operators were defined on the type -- there's no
difference between the RM's support (or lack of
support) for types such as dates/times and types such
as "invoices" or whatever.
I believe there's a fair amount of work underway to
reconcile the underpinnings of the relational model
and XML's tree model. See for example
http://citeseer.nj.nec.com/jagadish01tax.html and
http://www.eecs.umich.edu/db/timber/files/physical.pdf
Also, FWIW, I made the assertion in a townhall on XML
and databases at XML 2002 that within 5-10 years the
XML model and the RM would be theoretically
reconciled, and some notable luminaries in the field
were nodding their heads :-)
> 2) Data based applications developed using the
> Relational Model, which are
> well engineered and designed, will feature lower
> cost over time with
> greater flexibility.
> 4) Rigorous, scientific proofs exist, and are easily
> found, for adherence
> to the Relational Model (RM).
> 5) I have seen nothing better than RM for improving
> software application
> reliability, flexibility, maintainability and
> lowering software system
> costs overall. If something better exists, as a
> methodology with scientific
> proofs, I would dearly like to see it.
Well, I'll just ask the question that Simon and I have
posed to Fabian Pascal a few times, resulting in a
storm of invective but no real answer: If the RM is
the answer to all data management problems, why do
people keep reinventing and effectively using variants
of the hierarchical model (such as OODBMS and XML DBMS
and the numerous hybrid solutions that the dominant
SQL DBMS vendors offer)? I don't buy the "the industry
is too stupid and fad-obsessed to do the Right Thing"
argument. There are an awful lot of smart people and
very deep R&D pockets at Microsoft, Oracle, and IBM;
if implementing and exploiting the RM "properly" could
give one a competitive advantage over the others, they
would have done this already. Instead, they've all
turned to XML or hybrid SQL-XML-OO approaches.
It would appear that all the "scientific proofs" in
the world don't help one if the techniques they
espouse can't be implemented and used by real humans.
We don't write real software using Turing's
mathematics, although it's important to have that as a
theoretical foundation. I suspect that we'll
eventually conclude something similar about Codd's --
it's wonderful theoretical foundation for databases,
but not a practical tool for ordinary mortals ... and
probably never will be for a large subset of
real-world problems, especially those where the
"types" are complex, poorly understood, rapidly
evolving, etc.
|
