Re: RDDL and MIME types

[Jonathan Borden <jborden@mediaone.net>]

Simon St.Laurent wrote:


> After pondering RDDL and data types yesterday, I'm now pondering RDDL and
> MIME types.
>
> Namespace URIs are identifiers used inside of documents, but it seems
> reasonable that they could become identifiers used earlier in the content
> negotiation process, much as MIME types are today.  Putting RDDL documents
> at the ends of those URIs would give software and human users a chance to
> make a more informed choice about what information they'd be receiving and
> how to process it.
>
> It also struck me that RDDL could be used to describe non-XML formats just
> as easily as XML, making it a plausible replacement (not that it will
> happen in my lifetime) for the MIME type registration process and the RFCs
> which describe MIME types.

Right. Whether the RDDL indirection happens on the client or server is a
matter of software and protocols. It would be straightforward to use the
HTTP extension framework and pass HTTP client headers (either/and)

RDDL-Nature: uri
RDDL-Purpose: uri

From which server side code could extract the matching resource from a RDDL
document and return that resource in the HTTP response.

>
> Maybe I've just thought too hard, but this seems like a reasonable path
> forward for supporting the ever-growing number of XML document types in an
> unconstrained but useful fashion.
>

Now that text/xml and application/xml are defined, we can define document
types by (for example) the namespace URI of the root element (RDDL Nature)

-Jonathan