- From: Umit Yalcinalp <umit.yalcinalp@oracle.com>
- Date: Thu, 22 Jan 2004 15:22:31 -0800
- To: ygoland@bea.com
- Cc: www-ws-desc@w3.org
Yaron Goland wrote: >One of the ugly problems we ran into in BPEL is that in WSDL 1.1 it is >theoretically possible to define a portType which does not have bindings for >all of its operations. It would be a good thing were WSDL to explicitly >require that any binding for an interface MUST bind all operations in that >interface. After all, the interface is the true definition, the binding is >simply the dirty details. It would seem reasonable that the dirty details >match the true definition. > > I believe this is the current intent although I agree that this is not explicitly stated. You don't have to have binding operation components for each operation, having one binding would be sufficient as it is assumed that the binding would apply to all the operations in the interface. Could I ask the editors to take this as a clarification item on their list? >This requirement would in no way mandate that every binding must explicitly >reference all or any operations in an interface. Rather, it requires that >either explicitly or implicitly the binding MUST provide for the expression >of all operations in an interface. > >For example, my understanding of our current SOAP binding is that if one >binds to SOAP there is no need to explicitly list each operation in order to >make that operation available via SOAP. Explicitly references to operations >are only needed if one wants to explicitly define SOAP header related data a >specific operation. Therefore the existing SOAP binding would be compliant >with the previously stated requirement. > >Issue 16 also brought up this issue but the issue is marked as closed and I >couldn't find an explanation for how it was resolved. > > Thanks, > > Yaron > > > > -- Umit Yalcinalp Consulting Member of Technical Staff ORACLE Phone: +1 650 607 6154 Email: umit.yalcinalp@oracle.com
Received on Thursday, 22 January 2004 18:30:02 UTC