RE: RDF Issue: mime-types-for-rdf-docs

Hi Garret,

>The adoption of the "application/rdf+xml" MIME type makes sense for general
>RDF, just as "text/xml" makes sense for general XML.

Strongly agreed.

I'm glad action is happening on this, as it is an issue that could have a
detrimental effect on the adoption of RDF beyond first appearances. Recently
I've been looking at a lot of SVG related material, and a big stumbling
block for a lot of people seems to have adding the mime type to the server
config, so that the browser (usually IE with the Adobe plugin) understands
it.

>There have been later recommendations that "+xml" be appended to MIME types
>for specific applications of XML (as evident in the new RDF MIME type above
>when RDF is serialized as XML). Has there been any thought of making a
>similar recommendation ("application/...+rdf+xml") for specific
>applications of RDF? For instance, this would
>allow "application/pics+rdf+xml", "application/xpackage+rdf+xml",
>and "application/annotea+rdf+xml".

The problem with SVG above is one reason I'd use against this suggestion,
but there's a deeper one.

I may be completely off the mark here (I'm certainly no Marcel Marceau), but
I think perhaps MIME stumbles a bit around this point - text/xxx image/xx
application/xxx seems a bit too much like a short-term shortcut for
application developers, a hangover from the days before standard formats
could better describe themselves. Pandering to today's (HTML+) textual
document browsers, even.

I would have thought that any system that gets application/rdf+xml shouldn't
have trouble deciding whether it's pics, xpackage or annotea, *after* the
header, in an intermediate routing layer. Going further, RDF can pretty well
contain information about anything, or at least from any namespace. To try
and prescribe a specific receiver for the data when fed over http (or
whatever) whould strike me as a lost cause - a generic RDF reader/parser
with routing rules governed by local preferences would strike me as a much
more promising approach. The exact same data received by my 'environmental
concerns' agent would presumably need different handling that received by my
'teak commodities' agent.

Incidentally, such a post-header, pre-application routing layer would be a
way to separate embedded or linked RDF as well. I'm sure it's been done
already...

Cheers,
Danny.

---
Danny Ayers
<stuff> http://www.isacat.net </stuff>