- From: Marc Hadley <marc.hadley@sun.com>
- Date: Wed, 30 Jan 2002 10:12:51 +0000
- To: "Williams, Stuart" <skw@hplb.hpl.hp.com>
- CC: XML Protocol Discussion <xml-dist-app@w3.org>
Hi Stuart, My comments interspersed below. Williams, Stuart wrote: > >I find myself regarding the primary purpose of a SOAP protocol binding is to >(faithfully) transfer the infoset of a SOAP Envelope (and its contents) >between adjacent SOAP nodes along a SOAP message path. > I agree, and in my view the primary purpose of a binding *specification* is to describe how that is done using the underlying protocol mechanisms. > >In addition the binding may also convey the value of various properties >associated with features that the binding supports between adjacent SOAP >nodes in a compliance with the requirements of the relevant feature spec. >and in a manner specified in the binding specification. > Yup. > >As a consequence I find it difficult to usefully conceive a specification >that describes how a SOAP protocol binding provides (implements) what you >call a SOAP MEP (presumably intended to cover multi-hop cases in general) >except for two special cases: > >1) The end-2-end message exchange occurs over a single hop on a single >binding. > >2) An end-2-end message exchange that occurs over multipe hops using >multiple binding instances conforming to the same binding specification. > >I conceived of Transport-MEPs as patterns of exchange of SOAP messages over >a single binding instance between adjacent SOAP nodes. I think these are >exactly what bindings implement. > I agree and just to be clear I wasn't advocating that bindings should support split protocol MEPs or multihop scenarios. However I do think we should be clear about our goals and discussions during TBTF conf calls do tend to drift into these areas. Also note that there are currently sections in the WD that discuss MEP operation through intermediaries... > >SOAP Message exchanges have larger scope (in the presense of intermediaries) >and can occur over multiple different protocol bindings. > >I think that SOAP-MEPs (as you name them) can be defined by extension of >Transport-MEPs... eg. a SOAP Request-Response MEP over multiple hops arises >by extension of Request-Response Transport-MEP. However, simple extension of >a TMEP is not the only way to provide a semantically equivalent SOAP-MEP. >So, to take your split binding example, a SOAP Request-Response MEP could be >provided by the use of an underlying one-way transport MEP in conjunction >with a request/response corelation module defined for use inconjunction with >a two one-way transport message exchanges. > Again I agree, but if we want to do this then we need to be clearer in the spec, at the moment I think it mixes SOAP-MEPs and Transport-MEPs up. > >A way out of this 'fix' may be to regard MEP's as SOAP MEPs (as you describe >them) which include a description of the MEP is relayed through an >intermediary... and to regard the role of a SOAP binding to describe the >operation of that MEP over a single hop between adjacent SOAP nodes.... I >think I almost like this :-) It would leave all the intermediary behaviour >specified as part of the MEP description, and not the binding... which would >focus on the transfer of message infoset and feature properties between >adjacent SOAP nodes. > I like your idea. The alternative, I think, is to concentrate on Transport MEPs and eliminate intermediaries from our discussion of bindings. >Maybe I come full circle and contradicted myself... comments? > I think that maybe we just agreed all along :-) Marc. >>-----Original Message----- >>From: Marc Hadley [mailto:marc.hadley@sun.com] >>Sent: 29 January 2002 17:55 >>To: XML Protocol Discussion >>Subject: TBTF: SOAP MEP vs TMEP >> >> >>Following on from our discussion on the TBTF call today I would like to >>raise a point that I alluded to in a previous message [1]. I think that >>current text does not draw enough distinction between SOAP message >>exchange patterns and transport message exchange patterns and as a >>result is rather confusing. >> >>In order to address the confusion I would propose that when we refer to >>MEPs in the specification we refer to SOAP MEPs, i.e. patterns of >>exchange of SOAP messages rather than underlying protocol exchanges. We >>would define an open ended set of such MEPs: outbound one-way, inbound >>one-way, request-response, ... The purpose of a binding specification >>would be to describe how an underlying protocol is used to support a set >>of these SOAP MEPs plus any other features desired. Our HTTP binding >>specification would describe how HTTP is used to provided the SOAP MEPs >>that we decide we want to support. >> >>We also need to decide a number of issues including: >> >>(i) Can a SOAP MEP be split over multiple bindings ? E.g. request >>response with request via HTTP, response via SMTP. >>(ii) Are SOAP MEPs single hop or multi-hop in nature ? >> >>and what bearing the answers to these questions have on bindings >>"supporting" a given MEP. >> >>Regards, >>Marc. >> >>[1] http://lists.w3.org/Archives/Public/xml-dist-app/2002Jan/0194.html >>
Received on Wednesday, 30 January 2002 05:13:09 UTC