RE: Proposal for new last call issue: Some unprocessed headers should stay from Henrik Frystyk Nielsen on 2002-10-16 (xml-dist-app@w3.org from October 2002)

From: Henrik Frystyk Nielsen <henrikn@microsoft.com>
Date: Wed, 16 Oct 2002 10:05:11 -0700
To: "Jean-Jacques Moreau" <moreau@crf.canon.fr>, "Jacek Kopecky" <jacek@systinet.com>, "Yves Lafon" <ylafon@w3.org>
Cc: "Noah Mendelsohn" <noah_mendelsohn@us.ibm.com>, "XMLP Dist App" <xml-dist-app@w3.org>
Message-ID: <79107D208BA38C45A4E45F62673A434D093740CB@red-msg-07.redmond.corp.microsoft.com>

>Are you suggesting we should use "relayIfNotProcessed" instead? 
>Is that a go-back-to-WD solution? (Yves?)

The concern I have with this model is that I think it tends to
introduces more edge cases than the relay role with respect to
interactions with existing semantics in the SOAP. The examples that I
have run into are:

        <soap:Header>
                <hfn:myHeader role="..any role you like..." 
                                mustUnderstand="true"
                                relayIfNotProcessed="true">
                        ...
                </hfn:myHeader>
        </soap:Header>

and

        <soap:Header>
                <hfn:myHeader role="none" 
                                relayIfNotProcessed="false">
                        ...
                </hfn:myHeader>
        </soap:Header>

and

        <soap:Header>
                <hfn:myHeader role="ultimate receiver" 
                                relayIfNotProcessed="true">
                        ...
                </hfn:myHeader>
        </soap:Header>

I haven't found similar complications with the relay role.

Henrik

Received on Wednesday, 16 October 2002 13:05:49 UTC