RE: Proposed resolution for Issue 50 (Misallignment of faut to and reply to )

While I can understand the pragmatism of using the presence of a ReplyTo as an indicator that a reply is expected, I dislike it. I'd rather that the absence of a ReplyTo be regarded as an indicator that any reply would return on the default channel (often the same one the request came in on). If anything, I'd think that a ReplyTo might be seen as an indicator that the reply was expected to be going down a DIFFERENT channel, and therefore this might be seen as a flag that the current connection would not be used (with the obvious exception of a ReplyTo with the anonymous address - I guess that's what you intended).
 
Still, if we retain the required ReplyTo, let us at least allow it to be empty to specify the use of the default (binding specified) channel.
 
FWIW
 
Tony Rogers

	-----Original Message----- 
	From: public-ws-addressing-request@w3.org on behalf of Jonathan Marsh 
	Sent: Thu 03-Mar-05 7:26 
	To: Mark Peel; public-ws-addressing@w3.org 
	Cc: 
	Subject: RE: Proposed resolution for Issue 50 (Misallignment of faut to and reply to )
	
	


	I thought this was on the list somewhere but I couldn't find it, so I'll
	restate it here.  One benefit we would lose from making wsa:ReplyTo
	optional is an indication in the message whether a reply is expected.
	An intermediary or proxy might use this information to, for example,
	keep a connection open without having a description of the message's MEP
	available.  The information content of a specified wsa:ReplyTo and a
	default value is different.
	
	> -----Original Message-----
	> From: public-ws-addressing-request@w3.org [mailto:public-ws-
	> addressing-request@w3.org] On Behalf Of Mark Peel
	> Sent: Wednesday, March 02, 2005 12:08 PM
	> To: public-ws-addressing@w3.org
	> Subject: Re: Proposed resolution for Issue 50 (Misallignment of faut
	> to and reply to )
	>
	>
	>
	> +1 to Dave's approach.  But as for quoting Ockham's Razor as written,
	> I
	> feel we have enough cryptic language to deal with already...  Latinum
	> est; non potest legi.
	>
	> Cheers,
	>
	> Mark Peel
	> Web Services Infrastructure
	> Novell, Inc.
	>
	>
	> >>> David Hull <dmh@tibco.com> 03/02/05 9:43 AM >>>
	> I think we're on the same page semantically.  I believe the difference
	>
	> is between saying
	>
	>     * missing => anonymous => binding-specified
	>
	> and
	>
	>     * missing => binding specified
	>
	> For me the latter wins. /Entia non sunt multiplicanda praeter
	> necessitatem/.
	>
	>