Wouldn't it be nice if it were expressible/expressed as a property
set? That's how SGML's data model is expressed. Also HyTime's. It's
very likely that XML's data model can be expressed as a true subset
("grove plan") of the SGML Property Set. However, this would probably
not be the friendliest possible way to express XML's property set. I
suspect that the friendliest possible XML property set would use XML
(e.g. DOM) terminology wherever possible.
Work done by Fujitsu Labs has shown how XLink is expressible as a
property set. Unsurprisingly, it turns out to be about the same as
the relevant portions of the HyTime property set, except that all the
names have been changed to correspond to XLink terminology.
Property sets have some pretty attractive characteristics, and the
"grove object model" which they serve as schemas was originally
devised to describe the formal characteristics of SGML syntax. It's
very neutral, very standard, very pure, and as simple as it can be.
Moreover, property sets are expressible as XML documents; the DTD
for property set documents already exists.
Property sets describe classes of nodes, and the properties of each
class of node, as such nodes are output by a parser for a given
notation. A property set does not describe any methods, so it can
form an excellent all-purpose foundation for methods and applications.
Since every property of every syntactic construct is assigned a name
in a property set, the names of properties readily form a natural
basis for query languages, too.
Having a property set for XML would set the stage for XML to become
the language of documents that integrate information expressed in all
other notations, because they can pretty much all have property sets,
too.
-Steve
-- Steven R. Newcomb, President, TechnoTeacher, Inc. srn@techno.com http://www.techno.com ftp.techno.comvoice: +1 972 231 4098 (at ISOGEN: +1 214 953 0004 x137) fax +1 972 994 0087 (at ISOGEN: +1 214 953 3152)
3615 Tanner Lane Richardson, Texas 75082-2618 USA