- From: Prasad Yendluri <pyendluri@webmethods.com>
- Date: Fri, 24 May 2002 15:31:22 -0700
- To: www-ws-desc@w3.org
- CC: Jonathan Marsh <jmarsh@microsoft.com>, Jeffrey Schlimmer <jeffsch@windows.microsoft.com>
Folks, In the Description Group call yesterday Jonathan Marsh recommended that we recast the "Can One-Way operations return faults?" issue under a new (and appropriate) name to capture the real essence of the issue. What this issue is really calling for is the need to support 'communication of Failures only' scenario or as Jeff Schlimmer noted the Negative Acks (NACKs) scenario. It is a realistic and practical case that in certain scenarios the invoker of a service does not need a response or want to know if the operation succeeded but only if the invocation *failed*. That is, the invoker wants to learn of the failures (so that corrective measures can be taken for the failed cases) even for operations that do not return a response. This is analogous to the invoking a method that returns a void with a catch block. As an example, if party-A invokes the same service offered by 100 different parties (e.g. a broadcast type situation), party-A would not want to filter through all the positive empty responses to find the failed ones; where as party-A would want to learn of the failures so that corrective action (e.g. retry) can be taken. This is the situation with an operation (WSDL) not returning a positive or "empty" response but only a Fault, conditionally if the fault-condition occurs. This could be as simple as adjusting the grammar for request/response in the current spec to make wsdl:output optional. <wsdl:definitions .... > <wsdl:portType .... > * <wsdl:operation name="nmtoken" parameterOrder="nmtokens"> <wsdl:input name="nmtoken"? message="qname"/> <wsdl:output name="nmtoken"? message="qname"/> <wsdl:fault name="nmtoken" message="qname"/>* </wsdl:operation> </wsdl:portType > </wsdl:definitions> Regards, Prasad P.S. Ref: ACTION: Prasad, JeffS close the current issue, and have Prasad open a new issue re-titled "Negative Acknowledgement" from 5/23/02 minutes...
Received on Friday, 24 May 2002 18:40:01 UTC