- From: David Hull <dmh@tibco.com>
- Date: Thu, 17 Mar 2005 07:55:17 -0500
- To: "public-ws-addressing@w3.org" <public-ws-addressing@w3.org>
- Message-id: <42397E35.6020208@tibco.com>
This is mostly like what I've mentioned before, but it puts the reply endpoint on the wire as <wsa:ReplyTo> and likewise for [fault endpoint]. * Instead of [reply endpoint] and [fault endpoint] MAPs, we have an [endpoints] MAP (or whatever we want to call it). This is a set of /(qname, /EPR) pairs (not (IRI, qname)). We pre-define wsa:ReplyTo and wsa:FaultTo, which we tie to the usual semantics. * Instead of ReplyEndpoint and FaultEndpoint properties in the SOAP 1.2 module, we have an Endpoints (or whatever) property with the same value as the [endpoints] MAP. I believe the same approach used to back-port the current 1.2 module to SOAP 1.1 will work here, mutatis mutandis. * The Endpoints SOAP property is mapped to one header per pair, using the qname element of the pair as the header name. So the wsa:ReplyTo entry becomes the wsa:ReplyTo header, the wsa:FaultTo entry becomes the wsa:FaultTo header, and the alt:AlternateReplyTo entry becomes the alt:AlternateReplyTo header. * Headers other than the pre-defined wsa:ReplyTo and wsa:FaultTo carry an attribute identifying them as endpoints. This allows the receiver (and intermediaries) to reconstitute the [endpoints]/Endpoints properties without any special knowledge.
Received on Thursday, 17 March 2005 12:55:52 UTC