It is my understanding that MIME is a transport layer, not an
application layer. It is used to tranport multimedia objects through
an environment that normally does not allow such transference (like
email) and to allow easy classification of said objects (read:
content-type, etc).
I prefer to call MIME a framework. It isn't a transport per se since all
it does is specify something you could put in a file. It also isn't
a data format in any real sense since all it does is encapsulate and
label other data formats. All of our "specification" of data formats
consists of pointers to other documents.
MIME itself does not contain a linking language, but that is something
that can be built on top of it.
Exactly. The linking language has to be part of some specific data format,
which puts it more or less out of scope of MIME proper. The primary reason for
this is that the designers of MIME, myself included, realized that previous
efforts to come up with multimedia mechanisms on the Internet had screwed up by
trying to specify everything down to the bit. No single group of people has
sufficient expertise to deal with this problem comprehensively for all types of
data, and even if they did whatever they did would be obsolete before they
finished doing it. And even if they were able to stop the clock somehow,
there's no way for a single set of formats to be politically acceptable to
everyone.
I think further details would be inappropriate for the pem-dev mailing
list, and should be taken offline.
Probably true. I would suggest either the
ietf-822(_at_)dimacs(_dot_)rutgers(_dot_)edu list or
the SGML list for further discussion.
Ned