http://www.w3.org/2000/03/rdf-tracking/#rdfms-empty-property-elements [was ...]

jos.deroo.jd@belgium.agfa.com wrote:

[...]
> Well I was actually meaning that we *could*
> have an error1.n3 file in the testcase suite
> and that that file *could* contain no triples.

I see what you mean.  I think we decided to avoid trying to specify the
behaviour of processors when they encounter incorrect RDF.  I'd suggest
that a process where we specify the triples which correct RDF/XML
represents, and we have test cases which illustrate bad RDF/XML but
don't specify what triples should be output for bad RDF is consistent
with that.

> The broader issue here is when you have a
> "nesting" (to use Pat's new term) supposed
> to contain triples and there is a
> 'syntactic problem' then WHAT is going to be
> in e.g. an engine's memory?
> Craching is definitely out of question and
> exception reporting/logging is obvious, but
> what is the exception handling supposed to be.

I suggest that is not the concern of this WG.

Brian