RE: Enlightenment via avoiding the T-word

[Michael Brennan <Michael_Brennan@allegis.com>]

> From: Mary Holstege [mailto:holstege@mathling.com]

<snip/>

> I agree with the vision completely. What I don't grasp is why 
> you think
> rejecting the PSVI is in consort with this vision. The PSVI 
> is a first 
> attempt to define a particular domain of application in 
> exactly the terms you
> ask for: as an extended infoset that carefully avoids messing 
> with the core
> Infoset. 

But it takes that extended infoset set and presents it as the one and only
true form of the document. That's the problem. That and the fact that XML
Schema conflates too many roles and purposes into one, and pushes them down
into the foundations of XML. As Ronald Bourrett stated, "Factoring,
factoring, factoring..." [1]. I'd prefer to see a more factored approach. As
one example of this, check out the work the RELAX NG committe is doing for
DTD compatibility [2]. They are providing a layered, well-factored approach.
I could see similar approaches using annotations in a schema that provide
additional metadata suited to processing in other application domains. One
can include annotations intended to provide hints to data-binding tools, for
example. 

I'd also like to see a more pluralistic vision, the recognition that there
is not just one correct approach to producing a PSVI, and any such PSVI is
derivative. It is not the true form of a document instance, but rather an
infoset that is the result of some processing to suit a particular
application domain. As part of that pluralistic vision, I'd like to see
allowance for entirely different metadata vocabularies. I don't just mean
different schema languages, here, but actually different vocabularies for
the constructs that go into a PSVI.

I think perhaps I need to go study these Grove Plans and Property Sets that
others keep mentioning on this list. [3,4] Maybe that offers the right path
to achieving what people are trying to achieve with the PSVI.  

[1] http://lists.xml.org/archives/xml-dev/200108/msg01131.html
[2] http://www.oasis-open.org/committees/relax-ng/annotate-20010810.html
[3] http://lists.xml.org/archives/xml-dev/200108/msg01133.html
[4] http://lists.xml.org/archives/xml-dev/200107/msg00448.html