- From: Hugo Haas <hugo@w3.org>
- Date: Thu, 9 Dec 2004 16:24:56 -0800
- To: www-ws-desc@w3.org
- Message-ID: <20041210002456.GA2846@w3.org>
Hi. I am afraid that I am going to reopen an issue (issue 21[1]), but this is with additional information as I was implementing the resolution in Part 3. We took the decision that the HTTP binding supports out-message first MEPs, saying that the service could initiate the HTTP request, if it knew where to do so somehow. Some people at the last F2F pointed out that this "somehow" was an interoperability issue. Actually, I think that it is not "somehow" that this is communicated. The spec clearly says that the HTTP Request URI is built from @location on a binding operation and @address on an endpoint URI. What that means is that the Request URI that a service would use for this request is well-known and communicated in a standard way: in the WSDL document itself. A (pseudo) example: <description> <interface> <operation pattern="out-in"> <output/> <input/> </operation> </interface> <binding type="HTTP"> <operation location="http://myclient.example/"/> </binding> <service> <endpoint address="http://myservice.example/"/> </service> </description> Here, the service does an HTTP request to http://myclient.example/ and gets a reply from the client. The downside is that it is completely static. I have mixed feelings about the usefulness of this, and this changes the text that we agreed to put in the spec, so I think that the WG should consider this (forced) change of plan. Cheers, Hugo 1. http://www.w3.org/2002/ws/desc/4/lc-issues/issues.html#LC21 -- Hugo Haas - W3C mailto:hugo@w3.org - http://www.w3.org/People/Hugo/
Received on Friday, 10 December 2004 00:24:57 UTC