When a fault is thrown I'd expect that the fault field in the event would be filled with the exception information. It is doable if user uses custom code in IP but this functionality should be available OOTB.
@Keith - when a fault response is returned by a service, currently the message does not have a fault name (e.g. as defined in a wsdl contract). Therefore in the sample swyd/rtgov app I use an information processor (which usually is used to derived identifiers) to set the fault field in the activity event explicitly. Is there a way that this could be done efficiently without having to use the information processor approach?
Can you elaborate a bit on what you mean by fault "name" in this context? The type described in the service contract for consumer and provider is available on the exchange today. The message name from the WSDL is not something we care about from a runtime POV.
e.g. <wsdl:fault name="loanProcessFault" message="tns:errorMessage" /> I thought that was the case, which is why I used the information processor to set the information. Although I agree it is not information useful for the runtime, it is useful when referring back to the original contract (e.g. wsdl), so is it something that could be made available?
It's possible as an enhancement, but would require looking at the actual message content to identify which fault is actually being sent. We try not to dig into the content unless specifically instructed by the user (e.g. transformation). The core itself isn't aware of WSDL at all actually, since the deployer takes care of parsing the WSDL into a generic contract representation.
Thanks Keith. Jiri - looks like this will need to be raised as an enhancement and then we can all discuss options for next version.
Changed to enhancement :-) IIRC it would be nice to at least handle situation when the content is descendant of java.util.Throwable