Jacques,
Good point. Currently the section 2 Messaging
Model is
implicitly assuming the underlying protocol is Request/Response
type protocol
(Eg. HTTP), which is not good actually, I believe.
Since WS-Reliability should be used with SMTP
also.
So I think this should be clarified. We may need
to add
some more messaging model for one way type
underlying
protocol (Eg. SMTP).
For Section 5, I agree with you.
I appreciate your proposed wording.
Thanks,
Iwasa
----- Original Message -----
Sent: Thursday, January 08, 2004
8:01 AM
Subject: [wsrm] more editorial
comments
Section 2: Messaging Model:
- need to more explicitly state the
assumed properties of this "underlying protocol" . We implicitly assume it is a request-response messaging protocol. We should
state so beforehand. Because SOAP
by itself does not assume this (unless bound to HTTP). An important consequence of this Messaging Model, is
that WS-R 1.1 only covers reliability of business "requests", not of business "Responses".
Section 5: HTTP binding: - need to remind that this spec does not cover
reliability for business messages over HTTP responses (e.g. as may occur with WSDL req-resp
operations).
More generally we should say a word about
the WSDL MEPs that the spec covers: (a) reliability of One-way ops is fully addressed. (b) reliability of Request-response ops is only
covered on the Request side. (and I think we could be more specific here, as only guaranteed
delivery of the response causes problem and would need be adapted, per our past investigation.)
Iwasa, I can propose a wording for
these, if this mail raises no further discussion.
Jacques
|