W3C home > Mailing lists > Public > public-ws-chor@w3.org > November 2006

Re: Exchange type issue

From: Gary Brown <gary@pi4tech.com>
Date: Mon, 06 Nov 2006 14:36:36 +0000
Message-ID: <454F4874.8060605@pi4tech.com>
To: Martin Chapman <martin.chapman@oracle.com>
CC: charlton_b@mac.com, "'Monica J. Martin'" <Monica.Martin@Sun.COM>, 'Steve Ross-Talbot' <steve@pi4tech.com>, 'WS-Choreography List' <public-ws-chor@w3.org>

Hi Martin

I think this is dealing with a very specific situation - i.e. a 
notification being sent without a previous request. This may well be 
modelled using a channel from B to A, and send a request.

However the situations I am primarily concerned with are the situations 
where a dialogue is already under way between A and B, A being the 
client and B being the server. B then wishes to notify A of some change 
in situation. For this to be modelled using a request from B to A, it 
would require a second channel to be established in the CDL, and for the 
endpoint reference for A to be passed to B as part of the preceding 
dialogue. This all complicates the choreography unnecessarily, and 
creates a bi-directional dependency between the client and server that 
may not be desirable.


Martin Chapman wrote:
> Can someone please tell me the real difference between a notify and a in-only? If I have two participants A and B, when and why
> would I use notify instead of in-only if B needs to interact with A without a preceeding "request"?
> Martin.
>> -----Original Message-----
>> From: Charlton Barreto [mailto:charlton_b@mac.com] 
>> Sent: Sunday, November 05, 2006 10:57 PM
>> To: Monica J. Martin
>> Cc: Steve Ross-Talbot; Martin Chapman; 'Gary Brown'; 
>> 'WS-Choreography List'
>> Subject: Re: Exchange type issue
>> Monica J. Martin wrote:
>>>> Steve Ross-Talbot wrote: Monica,
>>>> I take your point about religiosity. As regards clarity around the
>>>> new exchange type and semantics I do not think it changes the  
>>>> semantics of anything in WS-CDL at all. Rather it makes explicit  
>>>> something that is today implicit. So in a sense it tidies things up.
>>>> Cheers
>>>> Steve T
>>> We have yet to consider that the only difference is the 'respond' is
>>> not tied to a 'request.' Therefore, this could be accommodated by 
>>> allowing a respond that may or may not be tied to a request. As Gary 
>>> said there is no other difference.  Thanks.
>> True, there is no other difference. However, having the new exchange 
>> type makes explicit the exchange pattern represented by the choreo. As 
>> there is no semantic difference, I see no logical reason not 
>> to have the 
>> new exchange type.
>> -Charlton.
Received on Monday, 6 November 2006 15:19:34 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:30:37 UTC