- From: Noah Mendelsohn <noah_mendelsohn@us.ibm.com>
- Date: Tue, 15 Jan 2002 11:42:49 -0500
- To: skw@hplb.hpl.hp.com
- Cc: "gtn" <gtn@rbii.com>, "www-tag" <www-tag@w3.org>, "xml-dist-app" <xml-dist-app@w3.org>
Ironically, a primary counterexample to the proposition that the root element namespace determines the "type" of a document is SOAP itself. If I send a purchase order wrapped in a SOAP envelope, is it most useful to consider this as "a SOAP document that by the way contains a purchase order" or as "a purchase order that happens to be wrapped in a SOAP envelope"? I would argue that there are many cases in which the latter represents a more useful view than the former, though both are valid in principle. Consider a message queuing system that receives different types of requests, including purchase orders, through a variety of protocols, one of which is SOAP. In such a system, it may be as useful to type the document as a purchase order rather than as a SOAP message. Perhaps the most useful labeling would be as "a SOAP purchase order". In other words, a label that potentially draws on a range of information salted through the document to determine its nature. Unfortunately, the current system of MIME types probably does not scale to this sort of usage---URI names would be a better base on which to build. That being the case, I don't think we should go too far in limiting the mime types that can be applied to a SOAP message. It may indeed be appropriate to choose a type such as xml+soap or whatever as a standard to promote interoperability in the case where "xmlness" and "soapness" is the most useful thing to convey in the MIME type. ------------------------------------------------------------------------ Noah Mendelsohn Voice: 1-617-693-4036 Lotus Development Corp. Fax: 1-617-693-8676 One Rogers Street Cambridge, MA 02142 ------------------------------------------------------------------------
Received on Tuesday, 15 January 2002 11:43:56 UTC