Hello, I'm looking for some pointers on how people have made effective use of WS-BF in specifications, and what is the general best practise. 1. How do people wrap SOAPFaults? Do they define a WrappedSoapFault extension of BaseFault? Do they extract the unofficial extra fields, like the Axis HttpFaultCode and stack trace? 2. What is the i18n policy on descriptio? Is it assumed to be in en-us or whatever the implementor chooses. Given that <description> is allowed multiple times, a language attribute would let me return messages in different languages. Does anyone do that in conjuction with the appropriate xml:lang attributes? 3. How do people handle native language exceptions? Have the framework generate a SOAPFault and wrap it, or redo all the work themselves? I'm planning on having a fairly generic 'DeploymentFault' with extra info that I want for diagnosis (hostname, process identifier, stack trace), and under there a few more faults (WrappedSoapFault, LanguageFault) with extra data. All my operations will declare DeploymentFault as their sole return type; the Java equivalent of 'throws Exception'. That way future implementations get scope to add more stuff. For specific faults which do not need to add new fields, I am not going to type them, but just use the BaseFault/ErrorCode element with different QNames for each fault. Thoughts? -Steve