RE: A minor question

My 2c worth, all IMHO:
 
getLastFault should wrap the Fault so it can be distinguished from a "real" fault. 
 
getLastFault should return a Fault if it fails (eg: there was no fault to return), or a Fault in some kind of wrapper (perhaps xxx:LastFault) if it succeeds. So the successful return would be something like:
soap:Envelope/soap:Body/xxx:LastFault/soap:Fault
 
The idea that a Fault could be a regular return is mildy offensive :-)  I believe JAX-RPC will choke on such a design, too.

	-----Original Message----- 
	From: public-ws-addressing-request@w3.org on behalf of Mark Baker 
	Sent: Fri 18-Mar-05 16:23 
	To: Martin Gudgin 
	Cc: public-ws-addressing@w3.org 
	Subject: Re: A minor question
	
	


	On Thu, Mar 17, 2005 at 03:07:03PM -0800, Martin Gudgin wrote:
	> A fault is any message for which the following XPath expression
	> evaluates to true;
	>
	> /soap:Envelope/soap:Body/soap:Fault
	>
	> See[1], specifically;
	>
	> "To be recognized as carrying SOAP error information, a SOAP message
	> MUST contain a single SOAP Fault element information item as the only
	> child element information item of the SOAP Body"
	
	Right, but that doesn't mean that all messages carrying SOAP error
	information are faults.  Using the example from the posting I
	referenced, would you consider the response to a "getLastFault"
	request to be "carrying SOAP error information"?  If so, how would
	you distinguish that message from the message you'd get back as a
	result of "getLastFault" faulting?
	
	Consider also that in the HTTP binding, the only way that the
	receiving node is prescribed[1] to enter the fail state is by receiving
	a 4xx or 5xx HTTP response code ... which is of course, completely
	independent of whether there's a fault in the SOAP envelope or not.
	
	If you want to continue this discussion, I suggest we take it to
	www-archive.
	
	 [1] http://www.w3.org/TR/2003/REC-soap12-part2-20030624/#tabreqstatereqtrans
	
	Mark.