[
Lists Home |
Date Index |
Thread Index
]
Micah Dubinko wrote:
> AndrewWatt200:
>
> >how many "root nodes" an XForms processor would see in, or associated
> with,a multinamespace XML document which contains three <xforms:model>
> elements.
>
> An XPath root node can only exist with relation to an XPath data model. The
> XPath spec likes to use the word "document" for this, but it is closer in
> concept to an "infoset". It can be seen through XPath APIs that in-memory
> trees, SAX streams, and other non-document things can be used as a source
> for an XPath data model. The XPath errata have a definitive mapping to
> infoset.
With some persistence over the past two years (e.g.
http://lists.xml.org/archives/xml-dev/200007/msg00945.html) I have insisted
that, taken to the extreme which Micah Dubinko exhibits above, the 'infoset'
view is inimical to the fundamental place of documents in markup. John Cowan has
long and publicly hoped that infosets of information items and documents of
marked-up text might peacefully coexist
http://lists.xml.org/archives/xml-dev/200204/msg01053.html as simultaneous views
of the same reality, but I think that Micah Dubinko's post amply illustrates how
jealous are these two masters, and why we cannot serve both. At the heart of the
issue is the fundamental question of the Text. That will not go away. Should we
insist that the lexical expression is motivated by the fluid preverbal Gestalt
http://lists.xml.org/archives/xml-dev/200205/msg01021.html then we shall quickly
find ourselves making such outrageous assertions as "an XPath root node can only
exist with relation to an XPath data model", as if by definition the parent of
root were not document. How convenient it may be to posit the set of information
items as the progenitor of the text. That conceit solves at a stroke the thorny
problems of concurrent markup, for example, by simply imagining away the premise
of the single document root. It is not, however, XML nor markup in any sense
predicated on an instance text.
Respectfully,
Walter Perry
|
