- From: Williams, Stuart <skw@hplb.hpl.hp.com>
- Date: Fri, 19 Oct 2001 14:47:16 +0100
- To: "'Glyn Normington'" <glyn_normington@uk.ibm.com>
- Cc: xml-dist-app@w3.org
Hi Glyn, > -----Original Message----- > From: Glyn Normington [mailto:glyn_normington@uk.ibm.com] > Sent: 19 October 2001 14:13 > To: Williams, Stuart > Cc: xml-dist-app@w3.org > Subject: RE: SOAP Binding Framework, HTTP Binding, MEP documents > > > Stuart, > > That was a helpful reply - thank you! The mystery of QoS > should start to unravel when features are further defined. > > I am trying to discern the relationship between SOAP, MEPs, > features, and transport bindings. Would it be fair to say > that a transport binding may implement SOAP with a > particular set of MEPs and a particular set of features? Basically yes... but I would try to maintain a careful distinction between transport meps (provided to SOAP by bindings) and application meps (provided by SOAP to the things that use SOAP). > If this is the case, I wonder if multi-hops scenarios with more than one > transport binding involved effectively result in only the MEPs and features > common to all of the involved transport bindings being implemented? I've a couple of thoughts on this... Firstly, I think that application meps can be synthesised within the SOAP layer from whatever raw materials the underlying binding support. So for example, a simplistic UDP binding might support a one-way, lossy, unsequenced... transport message exchange. SOAP could still provide a request-response pattern to an Application, however, the machinery within SOAP layer would have to add headers to the message to do the necessary message correlation and duplicate detection - yes arms are flapping a bit here... but you could see with some repetoire of application meps and some repetoire of transport meps, we could get to describing how to synthesise particular application patterns from what the transport bindings are able to offer. Regarding features... for features that in some sense extended across multiple hops, I think the description of that feature needs to describe the behaviour required at a SOAP intermediary to correctly relay the feature across multiple hops. > > Glyn > Hope this helps, Regards Stuart
Received on Friday, 19 October 2001 09:49:26 UTC