RE: MIME types vs. DOCTYPE (was RE: ANNOUNCE: New XHTML WD)

David Megginson wrote:
>
>
>  Isn't it easier to identify the resource type externally so
> that it can be handed directly to the correct processor?
>
>
	Assuming that HTML is defined in XML, then isn't the correct processor the
XML processor? text/xml correctly identifies the content-type. If you make
an exception for the specific XHTML DTD then why not for every DTD! The
argument that text/xhtml for content negotiation is a shaky one because the
problem of content negotiation is a well known problem for HTTP. Proposed
solutions include RFC 2295. A better solution is to employ specific
request/response headers e.g.

Content-Type: text/xml
Content-Document-Type: http://www.w3.org/html50.dtd

or,

Content-Type: text/xml; document-type=http://www.w3.org/html50.dtd;
charset=us

The problem with content-type proliferation is that lots of software depends
on known content-types. For example, how can you programmatically tell if a
MIME message body contains XML? Parse it and if it succeeds then TRUE?

Its alot easier to add a new header recognized by new UAs than it is to
modify legacy and currently working code.

Jonathan Borden
http://jabr.ne.mediaone.net

Received on Wednesday, 24 February 1999 20:04:30 UTC