Feeling Stupid about Invalid XML for RDF

All right, I give up.

There is a practice, exhibited in the OWL Reference Appendix B OWL RDF Schema 
<http://www.w3.org/TR/2003/CR-owl-ref-20030818/#appB> that I don't understand.  It is also 
employed in the file that is linked from that section.

I saw this in the current working draft of the RDF Primer, and I thought it was simply a mistake
or over-simplification in the examples.

The practice I am referring to is the use of 

1.	An XML 1.0 document declaration, combined with
2.	A Document Type Declaration that provides only an internal DTD for the purpose of making
some entity definitions.  There are no element markup declarations, including for the root
identified in the Document Type Declaration.

I see in the XML 1.0 section on Conformance, <http://www.w3.org/TR/REC-xml#sec-conformance>, 
that a non-validating processor would certainly accept this practice.  

Is that the deal?  RDF is an application that *REQUIRES* the use of a non-validating XML 
processor?  Is this said anywhere?  I haven't found it.

-- Dennis

Validation Experience with the OWL RDF Schema:
----------------------------------------------

I thought that presence of a Document Type Declaration constituted an assertion that the XML 
document be [DTD] valid, even though the processor might not validate the DTD.  Of course, so 
does my validating XML editor plug-in (in jEdit) and the W3C HTML Validator (that also validates 
XML).  Silly us.

The XSV 2.5-2 Validator is more tolerant.  It simply notes that there is no markup declaration 
for the root element and then proceeds to conduct "lax" validation.  Then it produces a long list 
of reports on how it can't find schema definitions that it can use. I didn't research that one 
any further.

Internet Explorer 6.0 SP1, surprisingly, loads and displays the OWL RDF Schema without complaint.  
Considering that it will refuse to present an invalid XML document under other conditions (and 
that includes some cases of valid XML documents that it pukes on), that's remarkable.  Is there 
some out-of-band agreement that I don't know about? I wonder.


-- dh

Dennis E. Hamilton
------------------
AIIM DMware Technical Coordinator
mailto:Dennis.Hamilton@acm.org | gsm:+1-206.779.9430
http://DMware.info
   ODMA Support: http://ODMA.info
OpenPGP public key fingerprint BFE5 EFB8 CB51 8781 5274  C056 D80D 0C3F A393 27EC

Received on Wednesday, 1 October 2003 18:31:10 UTC