(bcc: http-wg because this is about the ietf-types proposal for RDF types) I think it would be good to separate out the discussion of HTTP & browser behavior from the MIME registrations of RDF representations. The simple solution for Media Types for RDF languages is -- if you really want to use text/*, then always use a charset parameter, even if you'd rather not. Don't try to change MIME behavior or update the default or otherwise mess with it. You can say, using current MIME, what you want to say. Perhaps the default to US-ASCII is annoying, the content-type string is longer than you would like, but changing MIME is unnecessary. Secondly, don't use use the '+' syntax, as in: text/rdf+n3, text/rdf+turtle since it doesn't fit into the paradigm established for +xml; this was noted during the discussion of registration of +zip types. For an alternative, I imagine there are lots of possibilities. Personally, I prefer being more explicit, e.g., Label N3 documents as: text/w3c.rdf.n3;charset=utf-8 label Turtle files as: text/w3c.rdf.turtle;charset=utf-8 Larry