OpenFlow mostly agreed with my post on XML and RDF but took issue with me on one point that I think was a misunderstanding.
When I said "the default serialization of RDF as XML should not be the principal way RDF is interchanged", I wasn't against serializing as XML. I meant that a generic RDF to XML serialization isn't necessarily going to result in the optimal XML schema.
XML should be the way RDF is expressed but I don't think a single (or even small handful of) generic mappings is going to give you nearly as nice XML as if you tweaked the mapping for the particular ontology.
OpenFlow suggests "A canonical way of expressing RDF would probably go a long way in minimizing the differences (and flame wars) between RDF and XML" but I don't want to minimize the differences between RDF and XML because I think they serve a very different purpose. I'm trying to reduce the overlap in order to minimize the flame wars.
One advantage of the RDF ontology + mapping + XML schema approach over the RDF ontology + generic XML serialization is people who don't like the generic RDF/XML serialization don't have to use it; they can invent their own XML schema.
Furthermore, we RDFers don't have to lament over every "non-RDF" XML schema developed. We can hope that the developers of individual XML schemas would provide a mapping to an RDF ontology, but if they don't, someone else always can.