RE: [ws-rx] RE: i082 - Is there a more detailed proposal?

Chris:

I'd be surprised if my "confusion" was not shared by anyone in the TC about which level this request-response should be understood ... I saw some recent email about keeping its "HTTP sense", while others would (wrongly) "presume a WSDL MEP" as you say.

Now I concur that it is better to not tie CS/CSR to any SOAP MEP or any underlying protocol MEP for that matter.

So removing any possible hint at some MEP would not hurt and would go a long way in avoiding interoperability problems later, I believe.

My proposal to clarify this:

---- formal proposal -----

- L237: remove the end of sentence "in the body of the response message". (its being said enough elsewhere that it should be in the body)

- Everywhere else, simply replace "response message" with "message". (That would avoid such redundancy as "response message in response to...")

- Replace "request message" by "message" everywhere.

- L611: remove the sentence: "Sequence creation uses a CreateSequence, CreateSequenceResponse request-response pattern." which has little meaning in this section anyway (the rest of the paragraph talks of errors incurred by the CreateSequence message only)

--- end proposal ------

More on why the current wording could be misunderstood:

- the wording in L237 implies that the CSR must use the same "response message" as the fault would, which in case of an HTTP binding must be carried over the [SOAP HTTP] response if the CS is on the request, per SOAP 1.1 (6.2).

- With transport bindings out of scope, the terms "response message" and "request message" might be understood as SOAP response and SOAP request. If an HTTP binding is used later, the standard HTTP binding can lead to assuming that a CS/CSR is carried over an HTTP request-response.

Thanks,

Jacques

From: Christopher B Ferris [mailto:chrisfer@us.ibm.com]
Sent: Wednesday, January 18, 2006 3:32 PM
To: ws-rx@lists.oasis-open.org
Subject: Re: [ws-rx] RE: i082 - Is there a more detailed proposal?

Jacques,

I guess I am very much confused by your confusion:-) The RM spec says nothing about the term "response message"
as being in any way related to the response message of a SOAP MEP. There is no mention of SOAP MEPs
in the spec anywhere IIRC. I guess I am surprised to learn that you thought that.

In fact, IMO, there is no, nor should there be any, ties between SOAP MEPs and the CS/CSR request
response. If anything, they might be related to WSDL operations with both an input and an output message.

<aside>
Of course, there are some who will then presume that there is a relationship between the WSDL "MEP" and the SOAP MEP.
That would be a mistake. There is none. In fact, I would argue that there cannot be one. They are orthogonal
to one another, despite the fact that at a trivial level, they align with predominant usage of available Web services
stacks. However, ***this is merely a coincidence***.

Once you toss in WS-Addressing (or some other means of addressing) that coincidence disappears altogether.
</aside>

If you want to change "response message" to "a message sent in response to", I suppose that's fine. However,
I frankly see no need to do so.

Cheers,

Christopher Ferris
STSM, Emerging e-business Industry Architecture
email: chrisfer@us.ibm.com
blog: http://www.ibm.com/developerworks/blogs/dw_blog.jspa?blog=440
phone: +1 508 377 9295

ws-rx message