Re: ISSUE-137 (including XML includes)

From: "Alan Ruttenberg" <alanruttenberg@gmail.com>
Subject: Re: ISSUE-137 (including XML includes)
Date: Fri, 12 Sep 2008 06:11:59 -0400

> Hi Peter,
> 
> Thanks for the summary of the problem, which addresses my
> understanding of the issue. The idea of having an inclusion mechanism
> in addition to imports seems like a reasonable idea and I would
> support such a move in general. However the proposal to link it to XML
> raises issues - there are different serializations of RDF and it is
> preferable to have a solution that is independent of a particular
> serialization, as others have pointed out in different discussions
> earlier in the working group. In addition the use of the general power
> of XInclude and XPointer seems like substantially more than what is
> needed here, and may impose implementation burden as the technology is
> not afaik, not widely deployed.

One benefit of using XML include is that it is a pre-existing W3C
recommendation.  It may be that XML include is not widely deployed, but
doesn't its existence indicate that we should use it if at all possible?

> I wonder if there might be a way to use a similar mechanism but
> independent of the particular RDF serialization. 

I'm not sure why.  Isn't RDF/XML *the* RDF serialization?

> For example, could it
> make sense to have an directive (ontology property in OWL)
> owl:includesRDF and specify it's behavior only as part of the RDF
> parsing?

This seems a bit strange in that it would be a piece of OWL syntax that
only works in a particular serialization.  

If it turns out that XML includes is not suitable, then I think that it
would be better to employ something like the core syntax terms from the
RDF/XML grammar.  This would be a bit of RDF/XML syntax that required
the inclusion, something like:

  ...
  <rdf:RDF ... include="http://example.org/o1.rdf">
  ...
  </rdf:RDF

The beauty of XML include is that it works somewhat this *and* that it
happens underneath the RDF.  The non-beauty of XML include is that it is
more general than required and the include wanted is not a simple one.

> -Alan

peter

Received on Friday, 12 September 2008 10:49:17 UTC