Joshua Allen wrote:
> > I have little belief that a general RDF-in-XHTML solution is necessary
> > or even a good idea, and the discussions I've heard here have not
> We hear you :-)  Some people think it is a fantastic idea, however, and
> they haven't been convinced otherwise.  Is it possible to treat this is
> a general issue, so that we don't have to convince one another?

I think I'm being secularized here. Nothing I've said prohibits anyone
from developing tools that either describe metadata about an XHTML 
document or modifying their XHTML to expedite this process. I've only
been trying to point out that *embedding* generically-described RDF in 
XHTML is at the very least problematic, and IMO more likely practically
> > ad hoc basis. Criticisms of XHTML for not being able to XML-validate
> > inclusions of any well-formed markup are out of order, insofar as *no*
> > XML markup language can do that. It's as I've said a non-sequitor. The
> Well, we could easily validate an XHTML document that had:
> <metadata encoding="N3">
>  .....
> </metadata>
> embedded in it, since N3 is not well-formed XML.  It is only when we
> stick well-formed XML in there that the XML validation tries blindly to
> "validate" it according to the XHTML DTD.  Perhaps I am simple-minded,
> but I am still fascinated by the fact that this block can contain
> anything *except* XML.  This isn't the fault of XHTML; rather it is a
> peculiarity of DTDs.  

No, not a peculiarity of DTDs but of XML. No, you can't embed N3 in any
XML document, as N3 contains characters (like "<") that would cause a
parser to choke. It's not an issue of well-formed content but of content
that can't appear in XML. Java and JavaScript code can't appear in an
XML document either, unless within a CDATA section, or escaped. It may
even be that common HTML browsers allow such content, but XML validation
would flag an error.

The N3 example on the W3C web page would flag an error on line 2 upon
encountering the first "<", as well as every other instance of what are
considered XML markup characters:

1    @prefix dc:
2      <>.
3      @prefix : <>.
5      [ :givenname "Ora"; surname "Lasilla" ] is dc:author of 
6      [ a :Book ; dc:title "Moby Dick" ].
IOW, N3 could be made "XML-safe" by avoiding these characters.

> This also is not a problem specific to RDF,
> because I can use a:
> <script language="some_well_formed_xml_scripting_language">
> ...
> </script>
> and break XHTML validation.  So I hope we can avoid treating this as a
> "RDF should not exist anyway, so the restriction is fine" problem.

I've never advocated that RDF be eliminated. I'm a huge fan of Dublin
Core, Mozilla, and other big users of RDF. Sorry if anyone has this
impression. While it's possible to convert DC into XTM, there's not a
lot of impetus to do so. DC is a perfectly decent design, IMO. 

> > only solution possible would be to CDATA section the whole thing, but
> > then XML processors would treat it as CDATA rather than something with
> > a notation of RDF. This would not be useful. External links to RDF is
> Is this really a problem?  Maybe this is the best way to do it?  Does
> anyone have any objections to using:
> <metadata encoding="rdf">
> <!CDATA[
>  <rdf:rdf ..namespace declarations..>
>   <rdf:Description about="" ... />
>   <rdf:Description about="#someid" ... />
>  </rdf:rdf>
> ]]>
> </metadata>

Well, this works to avoid the validation problem (so long as the CDATA
content never includes "]]>") but doesn't *quite* solve the puzzle,
because the CDATA node is treated as an opaque block. The content is
never processed or validated. If validation of the RDF isn't necessary
(even according to an RDF schema) then perhaps this is okay. But my
solution would be to add a notation identifier to the <metadata> element
so that it would be identified *as* RDF and the node would then be 
sent off to an RDF processor. 

[I've always considered it a bit sad and ironic that notations have been
summarily ignored in XML, when they're a perfect solution for this type
of thing. You hang a notation-based processor onto the XML validation,
and nodes identified as being of that notation are sent to the processor.]


Murray Altheim, SGML/XML Grease Monkey     <mailto:altheim&#64;>
XML Technology Center
Sun Microsystems, 1601 Willow Rd., MS UMPK17-102, Menlo Park, CA 94025

      the wood louse sits on a splinter and sings to the rising sap
      ain't it awful how winter lingers in springtimes lap -- archy

Received on Tuesday, 17 April 2001 16:44:58 UTC