- From: David Hull <dmh@tibco.com>
- Date: Wed, 13 Jul 2005 18:13:22 -0400
- To: public-ws-async-tf@w3.org
Per DaveO's request on the concall today, the previous message I had in mind is here: http://lists.w3.org/Archives/Public/public-ws-async-tf/2005May/0037.html Note that this is a separate issue from reliability acks. The reliability-layer ack says "Yep, I got the request." The ack I have in mind says "I've processed the request and I won't be sending anything else back your way." E.g., [fault endpoint] is the backchannel, [reply endpoint] is elsewhere, and the receiver now knows that there will be no fault. However, these two acks could be piggybacked on the same return message in cases where the timing is right. This would (I think) look something like this: <soap:Envelope> <soap:Header> <wsa:Action>http://www.w3.org/@@@@/@@/addressing/ack</wsa:action> <!-- You're done with the MEP --> <wsa:MessageID>requestMessageId1357902468</wsa:MessageId><!-- This was the request ID --> <wsSomeReliability:ack sequence="someSequence">1-7,9</wsSomeReliability:ack><!-- Also acknowledging receipt of some messages --> </soap:Header> <soap:Body/> </soap:Envelope> Come to think of it, the reliability layer might be acknowledging completely different messages. Put another way, if the "MEP complete" ack is a SOAP message, the reliability layer can also use it to send back acknowledgments without further ado.
Received on Wednesday, 13 July 2005 22:13:27 UTC