xml-dev - RE: XSet, an XML Property Set, was: re: Why the Infoset?

[Jonathan Borden <jborden@mediaone.net>]

Bullard, Claude L (Len) wrote:
>
> 1. "I don't understand groves."  Asserted many times
> and may be the fault of the authors.  On the other
> hand, explained many times on the list and apparently
> not that difficult to grasp.  Per Eliot Kimber's examples
> presented in threads here, not that hard to apply either.
> I've asked for and gotten that explanation from Eliot
> and Prescod and frankly, it looks easy to me once
> one gets past the initial concept of node/property
> sets and the awful names they chose for the abstractions.

I agree, and one should read http://www.prescod.net/groves/shorttut/ prior
to developing a strong opinion in either direction. On the other hand there
is a difference between an idea and an implementation. I have a strong idea
about what this all looks like in the Directed Graph within my head, but the
real question is whether one could get there by reading the ISO specs. In
teaching at the post graduate level, I find that it is alot easier to give a
really complicated lecture than it is to give a really simple lecture. I'm
perhaps more guilty than the next at getting too complicated, but it is
really really hard to talk to a somewhat general audience about a technical
subject without getting caught up in too much jargon (and advanced
DTD'isms). What we need is a simple but normative specification. At the very
least it needs to be understandable by 95% of the people subscribed to(or
perhaps participating on) this list.

> 2. ...
is the argument for the Infoset.

> 3. "We can do a better job now."  Can we?  Will we?
> Or will it be another long tedious committee experience

... groan, I was hoping for perhaps another appendix to infoset defining the
subset (in the 'grove plan' sense) that gets us from full-fidelity to
current infoset. Are you are arguing that XSet ought to be specified exactly
as an XML Property Set?

>
> Third, if the groves really are too hard, then do
> with them what was done with XML, make that subset
> a high priority and get the folks who wrote the
> originals to help you do it.

In reality I think that XML Namespaces are fundamental enough to XML, as
distinct from SGML, to require an XML Property Set as opposed to defining
XML as a subset of the SGML Property Set (to speak formally). To SGML XML
namespaces look like attributes, but they are full fledged properties. No?

>
> John Cowan and others sweated blood for the InfoSet
> and a lot of the rest of us depend on it.  Let's be
> sure it isn't sufficient.

Sufficient for what? Sufficient for the in-scope task of the XML Infoset WG.
Not sufficient as a 'full fidelity' abstract description a.k.a. XML Property
Set. Again, what I am saying is:

(a) XML Property Set + (b) XML Infoset 'grove plan' -> (c) XML Infoset

What is up for discussion (here is as good a place as any?) is whether (a)
and (b) are to be specified as an ISO Property Sets/HyTime or by another
mechanism such as RDFS.  We need something that both works and is
understandable.

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org