RE: SGML output from XSL?

Subject: RE: SGML output from XSL?
From: "Borden, Jonathan" <jborden@xxxxxxxxxxxx>
Date: Mon, 8 Feb 1999 23:04:39 -0500
Paul Prescod wrote:
> Chris Lilley wrote:
> >
> > > "Borden, Jonathan" wrote:
> > > >
> > > > XSL can output HTML because HTML can be made well-formed.
> > >
> > > Not really. HTML's idea of a well-formed empty element is
> different from
> > > XML/XSL's.
> >
> > HTML Voyagers idea of a well formed empty element is *exactly the same*
> > as XMLs, since the former is written in the latter.
> True, but from context I know that Didier's message was not about Voyager.
> It was about HTML-as-it-is-standardized. That's the HTML discussed in the
> XSL specification.
> > If people just want to spit out random < and > all over the place and
> > not have any clue or any way of knowing whether the result is well
> > formed or not, just use perl or awk or something.
> That's not what we are discussing. We are discussing the possibility
> (described in the XSL specification) of generating data that is
> well-formed according to some syntax other than XML (i.e. legacy HTML,
> comma delimited file, etc.).
> "If an implementation wishes to use something in the result tree or
> stylesheet to control the output of a non-XML representation of the result
> tree, it should use the result namespace."
> This seems reasonable to me.

	Under this spec, there is no mechanism for enforcing an SGML DTD assignment
on the result-ns and hence it is impossible to determine whether start and
end tags are optional. The issue, I hypothesize, is not whether the output
tree is XML, rather whether, when serialized the output tree be well-formed
to the extent that this term is applicable to documents described by SGML
DTD's. This is to say that optional start and end tags will be present.

	Binary goo, on the other hand, can be produced by the output of a FOT
back-end (which is outside the XSL spec).

	The specification is silently ambiguous about how non-XML result-ns are to
be produced. Under the current document, the XSL implementation has leeway
to do as it sees fit, since there is no meaning attached to the namespace
URI (and this namespace *is* specified to be an XML namespace).

	This needs clarification.

Jonathan Borden

 XSL-List info and archive:

Current Thread