- From: Jeffrey Schlimmer <jeffsch@windows.microsoft.com>
- Date: Thu, 2 Oct 2003 08:16:39 -0700
- To: "Sanjiva Weerawarana" <sanjiva@watson.ibm.com>, <www-ws-desc@w3.org>
> Sanjiva Weerawarana [mailto:sanjiva@watson.ibm.com] writes: > "Jeffrey Schlimmer" <jeffsch@windows.microsoft.com> > > > > Perhaps I have misunderstood your point here, but this is not the > > current design of the patterns, so this proposal has the difficult task > > of proposing a simultaneous change there. The current abstraction of the > > pattern URI is quite consistent: it does not define specific message > > types nor specific fault types. Requiring it to do the latter when it > > does not do the former seems completely unreasonable. Would the > > request-response pattern URI definition define application-level faults > > to the request? > > (I sent a direct response to Jeffrey by mistake .. this that + more > thoughts.) > > Oops, you're right .. but faults are of course different from other > messages in that other messages in a pattern have a 1-1 relationship > with an actual message. Hmm, I guess in general even that may not be > the case. > > I see we need more thought/work in this area. Can we live with a 1-1 > relationship for message references to real messages? Yes > Otherwise > even that gets messy. Agreed > For faults, I guess I see that a single fault ref can be implemented > by different app level fault types and there's no way the pattern > def can take that into account. Given that, I'd be ok with allowing > one to repeat <fault> elements with the same messageReference value > to have the semantic that Roberto indicated. Roberto's syntax pref > is ok too, OK > but I'm a bit concerned with syntactic non-similarity with > input/output. Agreed > Sanjiva.
Received on Thursday, 2 October 2003 11:17:06 UTC