ietf-822
[Top] [All Lists]

Re: Finishing the XML-tagging discussion

2000-04-21 02:13:51
I have been reading the XML tagging discussion, here is a
short summary of my understanding of the problem:

The discussion relates to how a specialized application,
such as foobar, should label its MIME content-type, when
foobar uses XML as a basis for its encoding format.

Requirements:

(1) The normal MIME format should be used, with a content-type
    such as application/foobar.

(2) There is a need to indicate in the header, that foobar
    actually is based on XML.

(3) Body parts in the application/foobar format should be handled
    by a special foobar application handler. If, however, no
    foobar application handler is available, the applications
    should alternatively be handled by a generalized xml
    processor.

Possible solutions:

(a) Use content-type such as application/foobar-xml, where
    "-xml" tells the mailer, that if no special "foobar-xml"
    handler is available, then a general-purpose application/xml
    handler should be used.

(b) Use content-type such as application/foobar, and add a
    parameter to this content-type such as
    Alternative-Content-Type: application/xml.

(c) A variant of (b), where the added header is named
    Super-Content-Type: application/xml.

Pros and cons of the different solutions:

Solution                Pros                    Cons

(a) application/        No change to            Not neat to require
foobar-xml              MIME specs.             mailer to parse the
                                                subtype string.

(b) Alternative-        Easy to implement,      Some mailers which
Content-Type parameter  neat.                   do not know of this
                                                parameter may be
                                                confused.

(c) Super-              Gives MIME new          Should be carefully
Content-Type parameter  powerful capability     thought-through
                        with object-oriented    to fit other possible
                        structure of content-   uses, what about
                        types.                  several levels of
                                                superclassing for example?

My conclusion is that (b) seems to be the best alternative, unless
we want to really think through an object-oriented structuring of
subtypes.
-- 
Jacob Palme <jpalme(_at_)dsv(_dot_)su(_dot_)se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/jpalme/

<Prev in Thread] Current Thread [Next in Thread>