Re: Serializing arbitrary RDF

>>>"Seaborne, Andy" said:
> I'm sure this has come up elsewhere before but I'd like to highlight the
> lack of a RDF syntax that can encode arbitrary RDF.
> 
> The system I am working on includes an experimental remote API to stored
> RDF.  As part of this, I need to be able to take in arbitrary RDF from a
> higher layer in the protocol stack (i.e. supplied by another part of the
> system, probably written later) and serialize the RDF graph to send it
> across the network.  N-triple can move any graph structure but it loses
> namespace information and lang tags.  The XML syntax can't cope with
> arbitrary graph structures (shared bNodes) if I've read the spec right.
> 
> Sorry if this is a repeat,

Sorry, it is:
  http://www.w3.org/2000/03/rdf-tracking/#rdfms-qnames-cant-represent-all-uris

and we decided to postpone it
  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Oct/0582.html

as being out of scope, requiring significant syntax changes that
would be beyond the current charter.


About N-Triples and xml:lang tags - this depends on the results of our
current RDF Core WG datatyping discussion (which includes "what is a
literal?").  Personally I expect this will be add this to N-Triples.


With respect to namespaces - they are an XML thing and are not part
of the RDF Model.  Hence, not represented in the graph or N-Triples.

There is nothing to stop you adding statements to the model for your
own application to record such XML things, if you felt they were
useful (which I recognise they are for serialisation).  How about
  <http://example.org/mymodel/> <http://example.org/xmlnamespace> "xmlns:foo=http://example.org/" .
but this can't represent the element-scoped namespaces of XML,
defaults, overriding namespaces in inner elements, ....

Dave