- From: Mark Baker <distobj@acm.org>
- Date: Mon, 15 Apr 2002 14:39:22 -0400
- To: xml-dist-app@w3.org
I've identified an issue with the default HTTP binding, regarding how it satisifies R505[1], the requirement that no additional a priori agreement be established. The closing text for issue 54[2], which I provided, is correct that HTTP has some of the necessary features by which uncoordinated parties can coordinate SOAP support. The problem is that the current HTTP binding doesn't use them. As an example, I can write an HTTP conformant app which accepts any HTTP POST content and "saves" that content to a local file. If somebody POSTed a SOAP envelope to this app, they would have the expectation that, at the very least, the processing model would kick in to the point of faulting on unrecognized mustUnderstands. Yet this isn't the case with my app, because it knows nothing of SOAP. This shows that having a HTTP URI and a SOAP processor conforming to our default binding isn't sufficient to satisfy R505. FWIW, SOAP 1.1 addressed this issue with the RFC 2774 binding, which it used to assert that the SOAPAction HTTP header must be understood. I personally believe that this is the cleanest way of solving this problem, though I expect that the WG does not concur. So, one non-2774 way of addressing this issue would be to state that as part of the binding, in order for a SOAP client to know if it's dealing with a SOAP processor, it has to negotiate that support. An example could be; HEAD http://example.org/some-soap-processor HTTP/1.1 response; SupportsSOAP: true or SupportsSOAP: http://www.w3.org/2001/12/soap-envelope or something like that. To avoid requiring a separate round trip before every POST, we could require that this header be used on the POST response to indicate that the resource supports SOAP; anything not returning the header on a 2xx response would be considered an app similar to my example app above (i.e. not SOAP). [1] http://www.w3.org/TR/xmlp-reqs/#z505 [2] http://www.w3.org/2000/xp/Group/xmlp-issues#x54 MB -- Mark Baker, Chief Science Officer, Planetfred, Inc. Ottawa, Ontario, CANADA. mbaker@planetfred.com http://www.markbaker.ca http://www.planetfred.com
Received on Monday, 15 April 2002 14:33:00 UTC