feedback on rdfcore syntax changes: pls allow unqualified about=, etc

From #rdfig discussion:

http://ilrt.org/discovery/chatlogs/rdfig/2002-10-22#T20-16-43

[09:10] * danbri realises that all Adobe XMP RDF isn't RDF according to
current RDFCore RDF/XML syntax
[09:10] <danbri> they use unqualified 'about="..."'
[09:10] <dajobe> yeah
[09:11] <dajobe> but if MikeD gets the namespace change, it'll all be
broken
[09:11] <danbri> How would you feel about revisiting that decision in the
light of implementor feedback?
[09:11] <danbri> I don't want a namespace change...
[09:11] <dajobe> I'll wait for the feedback
[09:11] <dajobe> there's been too much predicting of potential complaints
[09:11] <DanCon> er... we have the feedback; danbri just checked their
shipped product.
[09:12] * DanCon will send it to rdf-comments if that's easier for dajobe
[09:12] <danbri> I'll do it.


I believe the rdfcore decision on about= vs rdf:about=, namely to
dissallow the former, goes against the chartered commitment to backwards
compatibility.

http://www.w3.org/2000/10/rdf-tests/rdfcore/rdf-ns-prefix-confusion/error0001.rdf
<!-- Test about - MUST FAIL -->

Currently we say that docs that use the unqualified 'about=' idiom are not
RDF/XML documents. This includes some examples from the M+S
RECommendation, as well as the implementation by Adobe in their XMP
toolkit (and hence a great many PDFs and other files contain
not-quite-RDF).

We could instead take the line that about= and rdf:about= are specified by
the RDF/XML syntax to be functionally equivalent, even though they are not
associated (by the XML Namespace machinery) with a common namespace URI.
(same goes for rdf:ID and other syntactic gizmos).

See http://www.w3.org/2001/sw/Europe/200206/imagemeta/extract/extract for
an online tool that extracts RDF/XML from XMP documents, eg see innards of
http://www.adobe.com/products/framemaker/pdfs/idn2_vs_pm7_vs_fm7_ue.pdf

<rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'
 xmlns:iX='http://ns.adobe.com/iX/1.0/'>
 <rdf:Description about=''
  xmlns='http://ns.adobe.com/pdf/1.3/'
  xmlns:pdf='http://ns.adobe.com/pdf/1.3/'>
  <pdf:CreationDate>2002-05-16T10:35:48Z</pdf:CreationDate>
  <pdf:Producer>Acrobat Distiller 4.05 for Macintosh</pdf:Producer>
  <pdf:ModDate>2002-05-22T17:22:24-07:00</pdf:ModDate>
 </rdf:Description>

 <rdf:Description about=''
  xmlns='http://ns.adobe.com/xap/1.0/'
  xmlns:xap='http://ns.adobe.com/xap/1.0/'>
  <xap:CreateDate>2002-05-16T10:35:48Z</xap:CreateDate>
  <xap:ModifyDate>2002-05-22T17:22:24-07:00</xap:ModifyDate>
  <xap:MetadataDate>2002-05-22T17:22:24-07:00</xap:MetadataDate>
 </rdf:Description>
</rdf:RDF>


According to M+S '99 REC (which has about='' examples) this is OK.
According to RDF Core, it isn't. The new RDF syntax spec doesn't make
clear why such documents are no longer considered RDF, only that they are
not. Perhaps there is a case based on parser complexity, efficiency etc.,
but I've not yet seen it made strongly enough to justify the backwards
compatibility hit.

Dan


-- 
mailto:danbri@w3.org
http://www.w3.org/People/DanBri/

Received on Tuesday, 22 October 2002 16:28:07 UTC