During last week's teleconference, I was asked to post the following issue, with respect to the current RPC faults proposal. The proposal includes the following sentence: "Procedure level errors, ie those that might normally be expected when executing a procedure, SHOULD be indicated in a successful RPC response containing some application-dependent indication of a failure." My reading of this proposal is that if we go that way, we will assimilate procedure errors, ie service execution errors, to a successful execution of the procedure (at least at first glance), which I think is wrong. There is a strong difference in semantics between a successful execution and a failed execution, and I think we should make that difference explicit and unambiguous. Clearly, this is what the Java (or C#) programmer would do, raising an exception. I propose that we report "procedure level errors" as an explicit "rpc:ProcedureFailed" fault. Jean-Jacques.Received on Monday, 27 August 2001 04:06:50 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:03 GMT