RE: Announce: WS-Callback, WS-MessageData, and WS-Acknowledgement specifications

I stand corrected.  Sorry.

Another possible insight here is that the TCP/IP <-> RM situation is
analogous to the HTTPS <-> WS-Security situation.  In both cases the
first offers a protocol-based solution, the latter a message-based.  In
both cases, the protocol-based solutions may work fine, but unless you
work very hard they tend to "disappear" once the message is delivered.
They do not leave very much to put into your backend database as an
audit trail or basis for nonrepudiation, for example (again, unless you
work rather hard at it).

-----Original Message-----
From: Jon Dart [mailto:jdart@tibco.com] 
Sent: Friday, May 16, 2003 12:20 PM
To: www-ws-arch@w3.org
Subject: Re: Announce: WS-Callback, WS-MessageData, and
WS-Acknowledgement specifications



WS-Acknowledgement and WS-Callback provide a framework for an 
acknowledgement-based reliability protocol that is an alternative to 
WS-ReliableMessaging (there is overlap in the authorship of these - 
there is less connection IMO to WS-Reliability). It is possible there 
will be some convergence of these specs.

One of the architectural concerns I have had about WS-Acknowledgement 
and WS-Callback is that they are basically using SOAP headers to extend 
the WSDL 1.1 MEPs. This is mixing up two architectural layers that 
really should be kept separate, IMO. WSDL 1.2 will provide a better way 
to do this. IMO it would be better if this extension could be done in a 
way that provides a better migration path to WSDL 1.2.

--Jon

Cutler, Roger (RogerCutler) wrote:
> Following are my personal impressions of the situation:
> 
> I believe that WS-Acknowledgement is one of the specification 
> candidates related to Reliable Messaging that have come out over a 
> period of time. Although I would not be qualified to say that it is 
> dated or has been superceded, I believe that at least some of the 
> authors of this spec later went on to participate in the creation of 
> one (or possibly)of what most people consider the two primary RM 
> candidate specs at present, WS-Reliability and WS-ReliableMessaging.

Received on Saturday, 17 May 2003 04:47:12 UTC