[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Proposed Suffix for XML-based media types

At 09:28 AM 7/20/99 -0400, Keith Moore wrote:
>[Steven Pemberton wrote:]
>> Now when we need an extra layer of fallback, it is obvious how to do
>> it:
>>         text/xml/xhtml
>sorry, it's not going to happen.  the MIME content-type syntax is 
>well-established; changing it would break a huge amount of deployed software.

On this count, I've surrendered long ago and agree with Keith, though
fortunately that's not what I'm proposing here.

>furthermore a two-level type system was an explicit design decision
>of the 822ext working group.  this decision was made for good and valid 
>reasons, and nothing which has happened since has demonstrated that 
>such a decision was shortsighted.  in particular:

In this case, I think XML - and other continuing developments around the
concept of generic formats - have made a significant case for the creation
of formats that can be used with multiple vocabularies.  At this point,
especially with the rapid growth of XML and supporting generic XML
standards (XPointer, XLink, XSL, CSS, and others), I have no qualms about
saying that generic XML processing is both useful and important to
meaningful information exchange.
>nobody has demonstrated a need for an extra level of type classification. 
>nobody has demonstrated a need for an extra level of fallback behavior.  
>if anything, experience with the existing two-level type system 
>indicates that fallback behaviors are of dubious value.

I would hope you explore sections 2 and 4 of the full proposal.  XML was
designed explicitly to be a solid foundation for many document types, in
effect a 'fallback' of very sound value.  

At the time RFC 822 was written, I think they pretty much did the right
thing, and I'm willing to live as close to their rules as possible (which
is why the proposal uses a suffix, not an extra level with /).  However, I
think we should seriously consider the fact that XML has changed the rules
significantly, in a way which appears to be gathering widespread support
and generating new media types at a feverish pace.

Simon St.Laurent
XML: A Primer / Building XML Applications
Inside XML DTDs: Scientific and Technical
Sharing Bandwidth / Cookies