- From: Tammo van Lessen <tammo.van.lessen@iaas.uni-stuttgart.de>
- Date: Mon, 14 Jan 2008 14:39:06 +0100
- To: "public-ws-policy@w3.org" <public-ws-policy@w3.org>
Hi all, I'm reposting this because the original mail did not make it into the public mailinglist archive - so I was not sure whether you've got my mail or not. My apologies if you receive this twice. Best regards, Tammo -------- Original Message -------- Subject: Re: [Bug 5346] Operation parameter in bindingOperation is of type QName Date: Thu, 10 Jan 2008 14:26:19 +0100 From: Tammo van Lessen <tvanlessen@taval.de> To: Paul Cotton <Paul.Cotton@microsoft.com> CC: public-ws-policy@w3.org <public-ws-policy@w3.org>, Christopher B Ferris (chrisfer@us.ibm.com) <chrisfer@us.ibm.com> References: <4D66CCFC0B64BA4BBD79D55F6EBC22574A99DAC7B7@NA-EXMSG-C103.redmond.corp.microsoft> Hi, Paul Cotton wrote: > Note that the comment actually refers to [1] which is an Editor's CVS > version of the spec. But I believe the comment applies to the > published WG Note at [2]. Yes, it also applies to [2]. >> I'm wondering whether the second parameter ('operation') in >> operation binding references should be really a QName. Shouldn't it >> be a NCName? The examples are also referring to NCName, like >> http://...#wsdl11.bindingOperation(TicketAgentSoap/listFlights) > > Here is my initial reaction: .... > In particular the WSDL 2.0 table [4] defines the operation parameter > for the binding references (e.g Binding Fault, Binding Operation, > Binding Message Reference, and Binding Fault Reference) as QName. > Since our table is based on the WSDL 2.0 table, I believe that we > defined the second parameter's type as QNAME to match what was done > in the WSDL 2.0 table [4]. AFAIK, WSDL 1.1 and WSDL 2.0 are somehow different in this case. WSDL 2.0 supports interface inheritance, thus inherited operations still have another namespace (i.e. different to targetNamespace). I believe that's why they are using QNames to identify an operation. This is not the case in WSDL 1.1. The binding is specified for a particular portType (without supporting inheritance), thus nested bindings to operations can obviously only correspond to those defined in this portType. Therefore I think that the tuple (binding, operation) is sufficient to identify the operation uniquely. But of course I might have got it wrong... Best regards, Tammo -- Tammo van Lessen Institute of Architecture of Application Systems |Tel. (+49)711 7816 487 University of Stuttgart |Fax. (+49)711 7816 472 Universitaetsstr. 38, 70569 Stuttgart |Room 1.132
Received on Monday, 14 January 2008 13:42:06 UTC