dns media type registration tree

[Mark Baker]

Hi,

I wanted to express my *strong* support for the recently published draft
about the "dns" media type registration tree[1].  I think this is long
overdue, and I thank Mark for devising such a straightforward mechanism
for decentralizing the media type assignment process (you don't want to
know what I was going to propose 8-).

I would like to propose one substantive change to the document though.

As I see it, there are two objectives that we'd really like to achieve;
a decentralized namespace, and dereferenceable identifiers (for all the
reasons in [2]).  Mark's draft ably addresses the former, but doesn't
seem to treat the latter as an objective, although the "doc" parameter
is a step in that direction.  I think a fairly small tweak can give us
both.

What I propose is that the doc parameter be replaced by a "base"
parameter whose value is a URI that is intended to permit the media
type name itself to be used as a relative URI.  For example, if we had
the media type "text/dns.foo.example.com", with a base parameter value
of "http://example.com/mtype/" then the authoritative URI for the media
type would be "http://example.com/mtype/text/dns.foo.example.com"
(see the discussion[3] about the relative URI approach by the W3C TAG).
The document should be clear that the resultant URI is the authoritative
URI for the media type.  It would still be optional, though I'd like to
say that we recommend that it SHOULD be used, and SHOULD be
dereferenceable using currently ubiquitous technologies.

Thoughts?

 [1] http://www.ietf.org/internet-drafts/draft-nottingham-dns-media-tree-00.txt
 [2] http://www.ietf.org/internet-drafts/draft-connolly-w3c-accessible-registries-00.txt
 [3] http://www.w3.org/2003/12/15-tag-summary.html#uriMediaType-9

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca