[regrep] Re: Core Components Specifications

[David RR Webber - XMLGlobal <Gnosis_@compuserve.com>]

Message text written by Philip Goatly
>XML in the traditional hierarchical for is also inadequte in that it has
difficulty in representing many to many relationships, this
is I guess because it derives from a document layout model which is very
hierarchical and does not need many to many 
relationships. The first databases, which were hierarchical had the same
problem, I know I was there. ( my first DB in 1974).
XML is evolving away from the 'pure' hierarchical form allowing linkage via
data values, as opposed to a hardwired structure.
This is hardly a surprise since that is what happened with databases.
Behold their is nothing new under the sun.

Cheers, Phil<

>>>>>>>>>>

Phil,

The same comment applies to XML - its merely a piece of clay - you get to
shape it.

Notice the Registry information model provides for associations and
classifications.
Of course there extend the basic heirarchical model and provide the ability
to capture
complex interrelationships and then exploit these programmatically.

Its this fusion of the CCR/ CRI and the classifications and associations in
the
Registry that I'm talking about - and being able to assemble content in a 
neutral way.  The other missing piece is context - and the CCR team is
already
working on that and planning to produce a draft in Seattle.

Ultimately people will be able to do all this in UML drawing little boxes
and arrows
but first we have to establish the underpinning.   Otherwise you have a
very nice 
looking car - but no engine under the hood!

Thanks, DW.