W3C home > Mailing lists > Public > public-ws-addressing@w3.org > March 2005

Re: New issue: Notification relationship IRI

From: Anish Karmarkar <Anish.Karmarkar@oracle.com>
Date: Tue, 08 Mar 2005 16:57:24 -0800
Message-ID: <422E49F4.2070102@oracle.com>
To: "Conor P. Cahill" <concahill@aol.com>
CC: WS-Addressing <public-ws-addressing@w3.org>


The WS-Addressing WG in its 7th march 2005 concall [1] discussed the 
notification relationship IRI issue that you raised. The WG decided not 
to accept this issue and I took an action to inform you of the same.

The relationship types in the [relationship] message addressing 
properties are extensible. Therefore, users of WS-Addressing are allowed 
to define IRIs (such as a notification relationship IRI) that specify a 
particular relationship between messages. Therefore the WG did not see 
the need to accept this ('notification relationship IRI') as an issue.


[1] http://www.w3.org/2002/ws/addr/5/03/07-ws-addr-minutes.html

Conor P. Cahill wrote:
> AOL propses that the working group add an additional IRI to the list of 
> predefined IRIs for the relationship values for the case when the 
> message is a notification associated with another message.
> In our messaging model, we have many situations where data will be 
> returned on a response that has a documented lifetime (TTL) which may 
> later be impacted by asynchronous changes made on the responding site 
> (such as a change in the authorization data used to generate an 
> authorization response) and we plan to send a subsequent message related 
> to the incoming request that isn't a reply but is in fact a notification 
> related to the request -- in our case it would be an invalidation 
> notification.
> One could argue as to whether such a notification should be related to 
> the request or response, and while I think that may be out of scope for 
> the WG, I think there are good reasons to make it related to the request 
> as the sender will already have the means of associating an incoming 
> mesage with the request message ID, so they don't have to add additional 
> functionality to support associating an incomming message with a 
> previous response messageID.
> The proposed IRI would be:
>     http://www.w3.org/@@@@/@@@/addressing/notification
> with the Description:
>     Indicates that this is a notification related to the message 
> identified by the message IRI
> Slight side note:  I think the description for the reply IRI should be 
> clarified to indicate that the IRI in the description refers to the 
> message IRI ("the IRI" doesn't make it totally clear which IRI we are 
> talking about since they are both IRIs).
> Conor
Received on Wednesday, 9 March 2005 01:01:52 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:28:24 UTC