- From: David Hull <dmh@tibco.com>
- Date: Wed, 29 Mar 2006 13:35:14 -0500
- To: "xml-dist-app@w3.org" <xml-dist-app@w3.org>
- Message-id: <442AD362.7090105@tibco.com>
As per my action item from today's meeting, here are two previous attempts at defining a one-way SOAP MEP. Dave Orchard's [1] dates to December 2004. It came out of the context of adapting protocols like HTTP to one-way use. As such, it includes a sample HTTP binding. We now handle the cases this was aimed at with the SOAP 1.1/HTTP note in WSA and the soon-to-be-put-to-bed request-optional-response SOAP 1.2 MEP (both of which owe much to [1]). Mine (attached as HTML) came along later, just over a year ago. It was intended as a pure one-way MEP. I believe it's more appropriate to the problem we're solving now. In particular * It defines the one way MEP as consisting of "exactly one SOAP message" as opposed to "one SOAP message and one optional binding specific message" * It defines a state machine with three states (init, success and fail) for the sender and receiver, as opposed to five (init, requesting/receiving, sending+receiving/receiving+sending, success, fail). * It does not define an OutboundMessage property (only InboundMessage). * It defines ImmediateDestination but not ImmediateSender. The receiver of a message will not generally have the sender's identity available. If ImmediateSender is defined, it should be optional. * It "makes no claims about the disposition or handling of SOAP faults generated by the either SOAP node" as opposed to having faults generated while the receiver is "Receiving" made available in the OutboundMessage property. I'm sure this will have to be tweaked to meet our needs, but it seems like a better starting point, as it is aimed more directly at the "pure one-way" case. [1] http://lists.w3.org/Archives/Public/public-ws-addressing/2004Dec/0159.html
Attachments
- text/html attachment: one-way-MEP.htm
Received on Wednesday, 29 March 2006 18:35:30 UTC