- From: <paul.downey@bt.com>
- Date: Wed, 29 Oct 2003 08:48:52 -0000
- To: <sanjiva@watson.ibm.com>, <jeffsch@windows.microsoft.com>, <anne@manes.net>, <alewis@tibco.com>
- Cc: <umit.yalcinalp@oracle.com>, <www-ws-desc@w3.org>
+1, but I'd add that the client is only ok if the transport correlates messages for the client (as with HTTP). For asynchronous models some addressing/policy magic beyond the GED is definitely required. Paul -----Original Message----- From: Sanjiva Weerawarana [mailto:sanjiva@watson.ibm.com] Sent: 29 October 2003 02:00 To: Jeffrey Schlimmer; Anne Thomas Manes; Amelia A. Lewis Cc: umit.yalcinalp@oracle.com; www-ws-desc@w3.org Subject: Re: Proposal: Uniqueness on the Wire Requirement for WSDL 2.0 +1 .. I have to say that WSDL deciding whether the message QName is the service selector or the URL or something else is not right. I remember asking about this when we implemented IBM SOAP, before SOAP 1.1 was released. I too was looking for a specific thing to be tagged as the "key" to routing messages (URL, SOAPAction or first body element NS URI) - eventually it turns out that different implementations can choose to route off different of these and that's completely ok. What's necesssary is that clients be able to form a proper envelope and send it to the right address. Whether the server calls a magician to decide which service to invoke or whether it requires unique element QNames so that it can route or whether it uses a SOAP header that its clients are expected to put on (and it presumably conveyed to the world via policy or however) is up to that. Sanjiva. ----- Original Message ----- From: "Jeffrey Schlimmer" <jeffsch@windows.microsoft.com> To: "Anne Thomas Manes" <anne@manes.net>; "Amelia A. Lewis" <alewis@tibco.com> Cc: <umit.yalcinalp@oracle.com>; <www-ws-desc@w3.org> Sent: Wednesday, October 29, 2003 7:50 AM Subject: RE: Proposal: Uniqueness on the Wire Requirement for WSDL 2.0 > > > From: Anne Thomas Manes [mailto:anne@manes.net] > > > > While I agree with you, I'm certain that WS-I will define a constraint > > that > > wire signatures must be unique. > > The future may not resemble the past.
Received on Wednesday, 29 October 2003 03:48:55 UTC