2008-04-20 (日) の 05:37 -0400 に Glen Mazza さんは書きました: > 2008-04-19 Willem Jiang wrote: > > Hi Glen, > > > > My comments are in the mail. > > > > Glen Mazza wrote: > > > Hello again, I have a few more questions. I'm using a CXF-based web > > > service provider running on Tomcat, using CXF's JAX-WS front end. Its > > > WSDL is located on my machine as follows: > > > http://localhost:8080/doubleit/services/doubleit?wsdl. > > > > > > Questions about the CXF component page[1]: > > > > > > 1.) Using this type of URI: "cxf://address?options", which of the > > > options--wsdlURL, serviceClass, serviceName, portName--do I have to > > > provide? It seems strange to need to provide any of these options, > > > because the first can be obtained via "?wsdl" by default, the second is > > > not necessarily known by the client, and the third and fourth are > > > readable from the WSDL--perhaps these options are needed only by CXF's > > > simple frontend? > > > > > Yes , the endpoint's address can get from the WSDL , but if the WSDL has > > more than two ServiceNames or PortNames, you need to specify the right > > one that you want. > > The URI "cxf://address?options" is common pattern for cxf endpoint ( not > > only the CXF provider but also CXF consumer), and camel-cxf component > > can't know which frontend will be used if it do not check the annotation > > from the serviceClass, so the serviceClass and address in required > > parameter for the camel-cxf component, the others parameters are optional.
I need a little bit more clarification here. If "cxf://address?options" is being used for a CXF consumer (not the provider), is the only reason why the serviceClass is needed is to determine whether the simple or jaxws frontend is being used? If so, I would like to say that Camel should default to the jaxws frontend if no serviceClass is given. The reason why I think this is because if you're using a CXF consumer you won't always have access to the SEI implementation class of the web service provider, so an ability to configure a CXF consumer without providing the SEI impl. would be beneficial I think. Thanks, Glen
