Re: eXtreme eXtensibility

Thanks for your excellent comments Eric!

I think that your main objection is that I am trying to use XML Schemas
for doing something that RDF is much better suited for, so why not
simply use RDF?  Excellent objection!  Let me try to respond:

Wouldn't it be the best of both worlds if an instance document could be
processed both by XML Schema tools AND by RDF tools?  Then you get the
strong type checking and the plethora of tools that XML Schemas
provides, PLUS, you get the inductive/deductive capabilities that RDF
provides.  This is why it makes good sense to use this design pattern. 
I have updated my write-up to address this very issue.  See the section
titled "Unification of XML Schemas and RDF".

Also, a colleague pointed out that there is a potential problem with the
design pattern when doing aggregation.  I have added a discussion on
Aggregation and the problem to the write-up as well.  I am eager to hear
your (everyone's) thoughts on this problem and the solutions I pose. 
Cheers!  /Roger

Eric Jain wrote:
> 
> > I have written a paper[1] which shows a way of designing an XML Schema
> > so that it places no restrictions on the vocabulary that instance
> > documents employ, and which facilitates the growth of data in a highly
> > distributed fashion.
> 
> Very interesting. Nevertheless...
> 
> As you mention, this kind of application is what RDF is meant for. While I'm
> also struggling a bit at the moment how to make effective use of RDF (i.e.
> how to efficiently reconstruct conventional objects from a set of RDF
> statements), I'm not sure I see much sense in emulating the kind of
> functionality RDF provides with XML Schema.
> 
> Currently, the main benefits of providing an XML Schema along with our data
> are:
> 
> 1) Allow simple code generation a la JAXP, so people don't have to bother
> with SAX/DOM.
> 
> 2) Allow access to certain data through web services.
> 
> As far as I can see, the kind of open schemas you describe wouldn't work
> very well with either use case, so I'll rather use XML Schema in the
> classical way, and provide a separate RDF view of our data.
> 
> --
> Eric Jain

Received on Saturday, 25 January 2003 10:49:55 UTC