Rather than continuing to kick landmines, however much I may dislike their
presence, I'm putting forward something that may settle the political
waters while still providing as much information as seems practical within
the scope of a MIME type.
All XML documents can continue to use application/xml if their designers
choose. Software will be expected to figure out what exactly is contained
in that XML on its own. (I'll leave the fate of text/xml to the rest of you.)
If the developers of a particular XML document type want to register a
specific MIME type for it, they go through the normal IETF registration
process using the top-level headers that already exist. The only
difference is that all document types that use XML as their base format
will be suffixed '-xml'.
This provides for things like:
graphics/svg-xml
application/xpdl-xml
application/ice-xml
model/x3d-xml
This way, applications can know which overall category the information
belongs to (graphics and model are, of course, more meaningful than
application, but that's the breaks), tools like editors, browsers, agents,
and search engines don't have to suffer too mightily guessing which content
they can open and which should be left alone, and we can assign meaningful
(well, sort of) names to XML formats, identifying them more precisely than
plain old vanilla XML.
In cases where content negotiation is required, other protocols can provide
that support. Applications can at least get a broad feeling for the type
of information being transferred from the MIME type, avoiding unnecessary
negotiations for 'simple' cases.
Does this seem possible? Or does it still raise flags?
It seems fine to me -- I would certainly have no problem enforcing it as
reviewer.
Ned