Re: [syndication] Re: Thoughts, questions, and issues.

[Gabe Beged-Dov <begeddov@jfinity.com>]

Ken MacLeod wrote:
> <snip/> 
> Outside of RSS1.0 core (but lurking in section 7. Examples, 3rd
> example) is the implication that RDF statements are allowed as the
> values of some properties (see usage of <dc:subject> in that 3rd
> example).  This usage is problematic, and no one has yet to explain
> how RSS tools should even begin supporting that, or if that was really
> intentional or just a brainfart on someone's part.

One of the problems people have had with RDF is the multiple ways that
an RDF graph can be encoded in XML based on the rules in the RDF
specification. The RSS1.0 proposal uses a severely constrained subset
of the RDF XML syntax. 

We have not published guidelines for what the specific subset is other
than the fact that all the syntax used in both the core spec and the
modules uses a _single_ XML syntax for any RDF processable stuff. This
syntax is documented using DTD syntax in the  "Model:" subheading of
each element description.

The example you mention (I've excerpted the fragment below) seems to
be a little off since the subject element is defined to have PCDATA
(i.e. plain-text) content in the dublin core module [1] while it is
defined to have a single rdf:Description element as its content in the
taxonomy module [2].

The example shows the subject element as being from the dublin core
module yet uses the content model of the subject element in the
taxonomy module. I'm guessing that this is an editing snafu or an open
issue that we need to work on.

Either way, you can process the document totally in the XML syntactic
domain without any RDF awareness. If you see a reason that it can't be
processed unambiguously using straight XML tools, please let us know.
Other than the fact that you may find the RDF syntactic sugar
aestetically unpleasing, it shouldn't impact your ability to process
RSS10 using RDF unaware tools. This should be true both on the
producer and consumer sides of things.

Module authors will always document the content model for the one true
XML syntax of thier module. If the author decides to use RDF, they
will have to choose the one RDF/XML syntax they want to use. 

Module authors who don't use RDF have to do one thing in order to
allow RDF processors to consume RSS10 documents that contain thier
modules. This is to add a single attribute to each of thier top-level
module elements that says that they aren't intending for this element
to be RDF processable. This magic attribute is
'rdf:parseType="Literal"'. This really doesn't seem like too much of a
mechanical overhead. 

Gabe


<item rdf:about="http://c.moreover.com/click/here.pl?r123";
position="1">
  <!-- ... -->
  <dc:subject>
    <rdf:Description
resource="http://meerkat.oreillynet.com/?c=cat23";>
       <rdf:label>Data: XML</rdf:label>
    </rdf:Description>
  </dc:subject>
</item>



[1] http://www.egroups.com/files/rss-dev/Modules/Proposed/mod_dc.html
[2]
http://www.egroups.com/files/rss-dev/Modules/Proposed/mod_taxonomy.html
-- 
--------------------------- 
http://www.jfinity.com/gabe