- From: Amelia A. Lewis <alewis@tibco.com>
- Date: Wed, 01 Oct 2003 14:41:39 -0400
- To: Sanjiva Weerawarana <sanjiva@watson.ibm.com>
- Cc: www-ws-desc@w3.org
Hmm, On Thu, 02 Oct 2003 00:36:21 +0600 Sanjiva Weerawarana <sanjiva@watson.ibm.com> wrote: > > Should we consider the case in which a fault may associate with > > several messages? There is no such case in the current patterns > > set, because all use fault-replaces-message and have zero or one > > replaceable messages. In message-triggers-fault, two messages in a > > pattern means two possible references. Hypothetical patterns with > > message-replaces-fault and number of messages > 2 would have the > > same issue. Allow a list of ncname in @messageReference or just ask > > users to specify multiply? I think it is probably more > > straightforward to have a single ncname. > > Yep- this is the same issue Roberto brought up .. I think the > simplicity of single NCName outweighs the flexibility of the > other option. Slightly different, actually. Roberto suggests that messageReference:detail is 1:N, I am suggesting that messageReference:detail is N:1; what we have is 1:1 and the reality may be N:N. But I agree that we have sufficient capability in the XML syntax to simply list them, and that it is clearer to do so using the 1:1 syntax than to introduce either form of one to many, much less many to many. Amy! -- Amelia A. Lewis Architect, TIBCO/Extensibility, Inc. alewis@tibco.com
Received on Wednesday, 1 October 2003 14:41:35 UTC