Thinking about what people are saying about XHTML, and, recasting it in
XML/SGML terminology, it sounds like people want to be able to list the
the notations used in a document, and to have this information available
somewhere in the MIME header to figure out whether a document is
acceptable. (I.e. after content negotiation)
Lets invent a language for representing notation dependencies:
() to group alternatives or derived type
| for alternatives versions of the same resource,
where the LHS is preferred
and more specific to the RHS
, for alternative notations useable for the same resource,
where LHS is preferred to
and a subclass of the RHS
+ for additional types,
where the first one is the framework
notation (e.g., text/xhtml or application/rdf )
-> for additional types, pointed to from the document,
which may be overriden by content negotiation
Using this, the notation structure for an XHTML document with lots
of things embedded might be:
(text/xhtml , text/html , text/xml , text/plain )
+ ( application/mathml , text/xml )
+ ( application/ms.vml , text/xml )
+ ( application/x-dublincore , application/rdf , text/xml )
-> ( text/xhtml )
-> ( image/gif )
-> ( image/jpg | image/gif )
-> ( application/pdf | text/xhtml )
(Note that I do not repeat the "," alternatives)
So then the issue becomes, can we derive a useful signature from these,
to be used in a media type name?
If not, then is some language like this appropriate, presumably to be
used as a parameter?
If it weren't for XHTML, I think this kind of thing would be better kept
out of MIME and made part of the W3C fragments or packaging work.
Rick Jelliffe