- From: Doug Davis <dug@us.ibm.com>
- Date: Mon, 14 Feb 2005 15:49:05 -0500
- To: tom@coastin.com
- Cc: Hugo Haas <hugo@w3.org>, public-ws-addressing@w3.org
Received on Monday, 14 February 2005 20:49:40 UTC
Tom wrote on 02/14/2005 02:22:55 PM: > Does "a reply" mean an "application level response to the message sent > using ws-addressing"? > > This is important, since some other soap header based > protocols can have their own orthogonal soap header reply mechanisms > (e.g., for the > ws-Reliability callback reply pattern, a protocol reply soap header is > sent to a callback address contained in a ws-Reliability specified soap > header element sent with the request message ). > > I would like to be sure that such a case is not considered a "reply > using ws-addressing". I think the message you're referring to would be a new message and not considered a reply in the WSA sense. But I suspect you'd still want it to include the WS-A headers (msgId, To....) even if it doesn't include the wsa:RelatesTo, right? Likewise, I think clients who leave off a wsa:FaultTo would still want the Fault to include the WSA headers. Based on the current wording (and a bit of guessing) I'm assuming the Fault might or might not have them. Thus, as I've said before - IMO, having the implicit effect of making wsa:FaultTo a required header. -Dug
Received on Monday, 14 February 2005 20:49:40 UTC