Re: Are we losing out because of grammars?

["Clark C. Evans" <cce@clarkevans.com>]

On Sun, 28 Jan 2001, Rick Jelliffe wrote:
> If we know XML documents need to be graphs, why are we working as if they
> are trees? Why do we have schema languages that enforce the treeness of the
> syntax rather than provide the layer to free us from it?

This is very interesting.  Most information models for XML texts
are based on a tree or the hedge model.  The support for XML as
a representation for general graphs just isn't "core". 

Certainly there are "layered" mechanisms for building graphs,
SOAP uses one method, we have the HTML href mechanism, then
we have xlink/xinclude/xpointer trio.  All of these are 
slightly different, no?  Is this what we want?

I've been doing some practical XML work lately, building a
user level application which uses XML/XSLT and I must say,
I've come up with my own style of "linking", much like the 
HTML href mechanism only I use elements instead of attributes
since I have a personal grudge against attributes.

Actually... I've found that I can't do anything without
some sort of linking.  Even the most trivial timesheet
example is a graph structure, else you have de-normalized
the information.  Certainly if you do any relational database
interaction you are definately in the world of graphs,
partial information, and network databases.

Perhaps Rick is on to something here.  Are we modeling the "parts"
without looking at the "whole"?  Can one truely have a 
schema for a single document type?  Or must a schema necessarly
model a set of inter-connected document types.

By far the most interesting (and practical) thread in months,

Clark