- From: David Hull <dmh@tibco.com>
- Date: Thu, 10 Mar 2005 21:25:38 -0500
- To: Jonathan Marsh <jmarsh@microsoft.com>
- Cc: public-ws-addressing@w3.org
- Message-id: <423101A2.8000308@tibco.com>
Jonathan Marsh wrote: > Doesn't the binding of SOAP to an underlying transport such as HTTP > answer that question? > I don't think so, at least not yet. The SOAP/HTTP binding describes POST/response (and plain response with GET or other request mechanism). The cases described don't fit into this. They have a one-way request and a one-way fault (or response) to an unrelated endpoint. They would, however, be covered by Dave Orchard's SOAP one-way proposal, which (I believe) calls for a 2xx response to complete a one-way. It's an interesting case. If there is a fault, the behavior is basically two correlated one-ways. If not, it should act like a regular SOAP/HTTP request/reply (which is still two correlated one-ways, but HTTP is doing the correlation). > > > ------------------------------------------------------------------------ > > *From:* public-ws-addressing-request@w3.org > [mailto:public-ws-addressing-request@w3.org] *On Behalf Of *David Hull > *Sent:* Thursday, March 10, 2005 10:05 AM > *To:* public-ws-addressing@w3.org > *Subject:* What, if anything, comes back on the HTTP reply if fault is > non-default? > > > > Eric Johnson of TIBCO caught this on review. He asks > > For example, if I'm using HTTP as the transport, a "fault" occurs, and > there is a "fault" header [and the reply is anonymous], what gets sent > back on the implied reply channel? A "200 OK"? > > There would be a similar question if the fault is anonymous and the > reply-to is present. >
Received on Friday, 11 March 2005 02:26:13 UTC