- From: Joyce Yang <joyce.yang@oracle.com>
- Date: Thu, 31 Oct 2002 16:46:54 -0800
- To: Sanjiva Weerawarana <sanjiva@watson.ibm.com>
- CC: "WS-Desc WG (Public)" <www-ws-desc@w3.org>
n the 80-20 rule, I'm not sure the "80" is on the no-ack case. I think most subscription will need an ack. -Joyce Sanjiva Weerawarana wrote: > On Thu, 2002-10-31 at 13:52, Joyce Yang wrote: > > I'm not a fan of existing outbound operations, but I do see > > them make the pub/sub modeling much easier and maybe > > more close to the original idea of outbound operations. The > > syntax Sanjiva proposed is very attractive (inline below), > > > > <portType name="pt1"> > > <operation name="normal-op1"> ... </operation>* > > <event name="event-1"> > > <subscription message="subscription-message"/> > > <notification message="notification-message"/> > > </event> > > </portType> > > > > however, it's not sufficient, we will have to invent many more > > constructs to catch the concepts on whether the "subscription" > > is with ack, and the "notification" is with ack or not. We > > definitely can add a new "ack" attribute to <subscription> > > and <notification> to serve the needs, but aren't they modeled > > very naturally by two outbound operations as mentioned in my > > proposal? But overall, I'm glad we are conceptually trying > > to solve the same problem. > > I agree there are different variations possible as you noted. > My approach to this kind of stuff is to take the most simple > scenario and address that and see whether its enough. We can of > course add whatever cases we want, but IMHO we should go for the > 80-20 case only and live with that. Other more complicatated > scenarios are left to extensibility. > > So my preference would be to pick the following choices: > - subscriptions are not ack'ed by the event source > - notifications are not ack'ed by the event sync > > Sanjiva.
Received on Thursday, 31 October 2002 19:48:04 UTC