From: MURATA Makoto
<murata(_at_)apsdc(_dot_)ksp(_dot_)fujixerox(_dot_)co(_dot_)jp>
There *is* an agreement that generic XML processors cannot do anything
useful.
I would like to make sure that we go forward (not backward). I also would
like to make sure that people's time and effort are well spent. Thus, I
do
not want to reopen this issue.
I think Murata-san means "(fallback to) generic processors (on
application-specific
documents) cannot do anything useful", rather than that text/xml and
application/xml are useless, which is what Simon seems to think he means.
I thought participants have agreed that
* there are generic XML documents which contain enough information in them
to be useful (e.g. with a stylesheet), and that these can use text/xml or
application/xml;
* there are application-specific XML documents which need their own MIME
type: e.g., application/*;
* fallback from application-specific XML to generic XML and from generic XML
to text has not been proved to be particularly useful, except for debugging;
* MIME parameters are not particularly useful at the current state of
technology (they may be useful for negotiation during requests, but the
current and new generation of browsers don't use them for dispatching
received documents);
* people are seem open to change there minds if there are convincing cases
shown and if technology changes.
Finally, I cannot see why people are scared of having a million new XML MIME
types under application/* or audio/*; or the other places.
Perhaps what is needed is an XML MIME registration tree, rather than a new
type.
Perhaps it could even be an automated system: you send the desired MIME
content-type and a minimal RFC to describe it (e.g., the DTD), and the
system allocates and confirms MIME content-type, under the XML registration
tree. That would be better than x- (no name collisions) and easier than
getting a content-type registered under the IETF registration tree (because
using XML answers some of the mandated questions about encodings, and
because the types would be allocated on demand; a subsequent administrative
review would disallow bogus requests).
I don't think that new XML-based media types whould require the same
scrutiny or comment as text/xml and application/xml; they are just
subclasses of them. IETF has a good role to play for text/xml and
application/xml comments, for the XML class of documents. But having
established that text/xml and application/xml are OK, I think creators of
XML document types shouldn't have to duplicate the effort involved in
getting an IETF MIME type registered (that effort is small and efficient,
but as Murata-san has found out, is not nothing).
This would give, for example, application/xml.ddml , which is useable by
existing browsers, and allows a program to fall back to generic XML (if it
knows that all application/xml.* documents are also application/xml).
(Refer http://www.freenic.net/rfcs/rfc2000/rfc2048.txt)
Rick Jelliffe