Hi Balaji, For some more info/documentation on the feature Eoghan mentions below see:
http://cwiki.apache.org/confluence/display/CXF20DOC/Multiplexed+Endpoint References A yoko binding would negate the need for WS-A by using a default servant approach. Gary. > -----Original Message----- > From: Glynn, Eoghan > Sent: 23 April 2007 09:54 > To: 'yoko-dev@incubator.apache.org' > Subject: RE: WS-Addressing support in Yoko > > > > Hi Balaji, > > I believe Gary is planning to write up a how-to on the CXF wiki. > > But in the meantime, you could have a look at the CXF code, > specifcally: > > - MultiplexDestination.java: basic API for generation of > fine-grained EPRs adorned with uniqueID > > - AbstractMultiplexDestination.java: example > transport-neutral mechanism > > - AbstractHTTPDestination.java: example transport-specific > mechanism for HTTP > > - > rt/frontend/jaxws/src/main/java/org/apache/cxf/jaxws/ServiceIm > pl.java: JAX-WS 2.1 style extension to Service.getPort() to > allow for binding on application-supplied EPR > > - > systests/src/test/java/org/apache/cxf/systest/factory_pattern: > system test tying it all together > > All available in the CXF repo from revision 530804. > > Cheers, > Eoghan > > > -----Original Message----- > > From: Mosur Ravi, Balaji [mailto:[EMAIL PROTECTED] > > Sent: 20 April 2007 15:45 > > To: yoko-dev@incubator.apache.org > > Subject: RE: WS-Addressing support in Yoko > > > > Thanks Eoghan... > > > > I would like to encode the WS-A properties in the > ServiceContext, so > > that we can take advantage of the transport neutral way of > doing this. > > But I would also like to understand the other transport > specific way > > of passing in the information. > > > > Is there a wiki where these things are documented? Or can > you point me > > to a piece of code that I can take a look at? > > > > Thanks > > > > Balaji > > > > -----Original Message----- > > From: Glynn, Eoghan [mailto:[EMAIL PROTECTED] > > Sent: Friday, April 20, 2007 10:21 AM > > To: yoko-dev@incubator.apache.org > > Subject: RE: WS-Addressing support in Yoko > > > > > > > > > > > -----Original Message----- > > > From: Mosur Ravi, Balaji [mailto:[EMAIL PROTECTED] > > > Sent: 20 April 2007 14:57 > > > To: yoko-dev@incubator.apache.org > > > Subject: RE: WS-Addressing support in Yoko > > > > > > Hi, > > > > > > Currently, the ws-addressing support in yoko is limited. > It is just > > > using EndpointReferenceType. But we might think of > extending it to > > > include the other WS-A properties. Can you describe > briefly what you > > > mean by wsam:Addressing policy? > > > > <wsam:Addressing> is a WS-Policy that can be used to enable > > WS-Addressing for a particular subject in the WSDL (e.g. a port, or > > service, or binding). > > > > It can also be used to assert various requirements and capabilities > > related to WS-A, e.g. that the presence of WS-A properties > in incoming > > messages is mandatory, or that non-anonymous responses are > supported. > > > > CXF can consume these policy assertions and auto-magically > include the > > corresponding interceptors into the client-side interceptor chain. > > > > Now if a server was to expose *both* SOAP and CORBA > bindings via the > > same <wsdl:service>, then asserting the policy at the service level > > would obviously be inappropriate as Yoko couldn't match the > > requirement with the corresponding capability. Hence it would only > > make sense to attach the Addressing policy to the SOAP > binding. Which > > is fine. > > > > My main interest in this currently is figuring out which policies > > would realistically tend to be asserted for a specific binding, but > > would also have a "logical" component that could stand on > its own if > > the binding were by-passed completely, as would be the case for an > > optimized colocated dispatch. Addressing seemingly falls into this > > category. > > > > Would it make sense to extend Yoko to encode WS-A properties in > > IOP::ServiceContexts? Well one obvious use-case is a > default-servant > > type pattern where the individual fine-grained targets are > > distinguished via a uniqueID or marker passed via WS-A reference > > parameters. This is the default transport-neutral way of > doing this in > > CXF (which I just committed today on behalf of Gary Tully). > > > > However CXF also allows a transport-specific alternative > mechanism to > > be supported, e.g. buring the uniqueID into a URI for HTTP. And for > > CORBA, the obvious approach would be to burn it into the object key. > > So I guess > > the point is that support for WS-A reference parameters wouldn't be > > required to make the CXF version of the default-servant > pattern work > > for Yoko. > > > > > Also, what type of policy assertions are you planning for > the soap > > > cases? > > > > Other examples that typically only make sense in the SOAP > case would > > be policies to enable WS-RM (currently supported by CXF), MTOM (not > > yet supported but easy to do) and WS-SecurityPolicy (not > yet supported > > but harder to do, partly due the way WS-Security is implemented in > > CXF, partly due to its intrinsic complexity). > > > > Cheers, > > Eoghan > > > > > > > - Balaji > > > > > > > > > > > > -----Original Message----- > > > From: Glynn, Eoghan [mailto:[EMAIL PROTECTED] > > > Sent: Friday, April 20, 2007 9:36 AM > > > To: yoko-dev@incubator.apache.org > > > Subject: WS-Addressing support in Yoko > > > > > > > > > > > > Folks, > > > > > > I'm curious about the extent of WS-Addressing support in > Yoko. This > > > relates to a scenario in CXF whereby a policy might be usefully > > > asserted in the WSDL for one binding but not another. The > canonical > > > example I had in mind was the <wsam:Addressing> policy being > > > asserted for the SOAP binding but not the CORBA binding. The > > > assumption being that there's no standard way of encoding > the WS-A > > > message addressing properties (wsa:To, wsa:ReplyTo, > wsa:MessageID, > > > wsa:isReferenceParamater etc.) as part of a CORBA payload, so it > > > would only make sense to aggregate these properties in the SOAP > > > case. > > > > > > However I was interested to see that Yoko includes a > > > bank_ws_addressing demo. A quick look at the code > suggests that the > > > extent of WS-A support is really just the usage of the > > > EndpointReferenceType, and not the encoding of the WS-A > properties > > > on the wire. I'd be happy to be corrected on this point > though, for > > > example if the WS-A properties were encoded by Yoko as as an > > > IOP::ServiceContext or some-such. > > > > > > Cheers, > > > Eoghan Glynn > > > [CXF committer] > > > > > > > > >
