Simon St. Laurent has announced an updated IETF draft prescribing a MIME media feature tag for XML content, which indicates the relevant namespace or namespaces in the transmitted document. It is meant to add more specific details to content of type text/xml, application/xml, application/XXX+xml, etc. Quoting from the draft (please refer to that document to follow the citations):

" MIME Content-Type identifiers have proven very useful as tools for describing homogeneous information. They do not fare as well at describing content which is unpredictably heterogeneous. XML documents may be homogeneous, but are also frequently heterogeneous. It is not difficult to create, for instance, an XHTML [26] document which also contains RDF [19] metadata, MathML [20] equations, hypertext using XLink [22], XForms [25],and SVG [21] graphics. XSLT [17] stylesheets routinely include information in both the XSLT namespace and in the namespace of the format resulting from proper execution of the stylesheet. "

" This document specifies a Media Feature which identifies the URIs used as XML namespaces [14] in a given XML document. While a list of namespaces cannot tell a recipient application everything about the use of those namespaces and their interactions in a given document, it can provide a baseline set of expectations. "

[...]

" This feature is designed primarily to be used with the XML Media Types defined in RFC 3023 [12]. By providing additional information about the content of the document beyond its overall type, it provides XML applications with a more comprehensive view of information they may (or may not) wish to process, potentially avoiding wasted parsing and processing. "

This specification comes among heavy debate on XML-DEV and the www-tag, the mailing list for the new W3C Technical Architecture Group (TAG) about how to direct processing using in-band or out-of band characteristics of the docuemnt. The discussion has largely split into two camps, one of which is the "document type" camp, which holds that a document should have one primary core of semantics and processing, even though it may contain chunks of XML from other semantic spaces.

The other camp holds that this is impractical and that the only hope there is of communicating the semantics of XML documents to the processors is by examining the "namespace bundle" of XML vocabularies used in the document. The xmlns Media Feature Tag is a handy way to communicate this "namespace bundle" prior to parsing of the XML document, so that the processing can be properly pipelined.

The brave can follow the discussion threads on-line starting at (2) and the various threads from January 2002 on RDDL on XML-DEV.

Related story:

• TAG becomes active