[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [ws-rx] PR33 - Re: [ws-rx] NEW Issue back-channel not defined
|
Paul, In our research through the trail of specs we found that the specific
connection between the ws-addressing anonymous uri and what we have been
calling the backchannel is described in section 5.1.2 of the WS-Addressing SOAP
binding[1]. In that reference says: "When "http://www.w3.org/2005/08/addressing/anonymous"
is specified for the response endpoint and the message is the http://www.w3.org/2003/05/soap/mep/InboundMessage
property of a SOAP request-response MEP [SOAP 1.2 Adjuncts], then any response
MUST be the http://www.w3.org/2003/05/soap/mep/OutboundMessage property of the
same instance of the SOAP request-response MEP [SOAP 1.2 Adjuncts]." WS-Addressing specifies behavior only for that one specific uri. If the uri is not wsa:anon, then section 5.2.2 applies. Section 5.2.2 states: "When
"http://www.w3.org/2005/08/addressing/anonymous" is not specified for
the response endpoint, then any response SHOULD NOT be the
http://www.w3.org/2003/05/soap/mep/OutboundMessage property of the same
instance of the SOAP request-response MEP [SOAP 1.2 Adjuncts]. For instance, a
SOAP 1.2 HTTP binding that supports a one-way MEP could put the reply message
in a separate one-way MEP and a separate HTTP request. N.B. the strike-out above was recently decided by the WS-Addressing WG. My opinion is that the MC uri is not wsa:anon and thus section 5.2.2
applies. In the case of the makeconnection form of anon, the language “SHOULD
NOT” contained in the paragraph above provides suffucuent license that
provides the ability for makeconnection to use the same instance of the SOAP
request-response MEP such that the mime entity contained in the http response
be used for a response (reply or fault). Indeed the ”one-way”
message is a soap request-response otherwise there would be no basis to utilize
the http response mime entity to contain a SOAP body. Thus, I am of the further opinion that there is only one wsa:anon and mc:anon
is not the same. It is up to the implementation to utilize that small
license to connect the request to the response. Further, the behavior we are describing in our use of backchannel is
not well described, in general, in either ws-addressing or in SOAP terms, but
is a characteristic of, possibly, only HTTP. The SOAP 1.2 HTTP binding is
rather non-specific in this regard as well. Backchannel is used in a
number of different ways in the ITU and in the liberty alliance. None of
those usages is what we intend. Not all transports will have behavior of the nature anticipated in
makeConnection, since many other transports go not have the rather unique
characteristic that one endpoint is enabled regarding the ability to make a
connection while the other is not. That is why we chose to describe backchannel as it has been used in the
RM spec to be related to the connection behavior of the underlying transport protocol
rather than a specific WS-Addressing or SOAP related behavior. I am personally of the opinion that complete repair and clarity of all
of the above requires work to several specifications (2-5) and is a windmill
that we best not tilt at this time. Thanks -bob [1] http://www.w3.org/TR/2006/REC-ws-addr-soap-20060509/#id2269957
-----Original Message----- Chris I'm still not sure why *we* are defining backchannel. It seems to me that belongs in WS-A. However, if we were to, I would prefer to refer
to in terms of WS-A: Backchannel: The term Backchannel as used in this specification means a
transport protocol specific facility that allows transmission of a protocol response message on the transport-specific connection indicated
by the use of the WS-Addressing anonymous URI. Paul Christopher B Ferris wrote: > > All, > > Bob and I have noodled on a definition of backchannel that we
believe > addresses the concerns expressed in PR33 > > Proposed resolution: > > Add the following definiton to the glossary section as follows: > > Backchannel: The term Backchannel as used in this specification
means > a transport protocol specific facility that allows transmission of
a > protocol response message on a given connection in the direction > towards the entity that initiated that connection. > > Cheers, > > Christopher Ferris > STSM, Software Group Standards Strategy > email: chrisfer@us.ibm.com > blog: http://www.ibm.com/developerworks/blogs/page/chrisferris > phone: +1 508 377 9295 > > " > 11/07/2006 01:00:26 PM: > > > The WS-RM spec uses the term “back-channel”
several times without > > defining it outside of the use “protocol specific
back-channel” > > > > A cursory scan of rfc 2616 HTTP 1.1 does not define the term. > > It is also not defined in SOAP 1.2 Parts 0-3 > > Where is the term defined? > > That definition should be referenced or a definition provided > > > > Thanks > > -bob -- Paul Fremantle VP/Technology and Partnerships, WSO2 OASIS WS-RX TC Co-chair http://bloglines.com/blog/paulfremantle paul@wso2.com (646) 290 8050 "Oxygenating the Web Service Platform", www.wso2.com |
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]