RE: AMG: Draft Alternate Section 3.

Hi John,

> -----Original Message-----
> From: john_ibbotson@uk.ibm.com [mailto:john_ibbotson@uk.ibm.com]
> Sent: 27 March 2001 11:34
> To: Williams, Stuart
> Cc: 'xml-dist-app@w3.org';
> Yves Lafon (E-mail)
> Subject: Re: AMG: Draft Alternate Section 3.
> 
> Stuart,
> This is looking good. I think the discussions you started on causality
have
> resulted in a better understanding of the one way issues.

Good.

> 
> In the revised document (section 3.2), [Correlation] appears as an
argument
> to the XMLP_Intermediary.receive primitive but not the
> XMLP_Intermediary.forward one. I believe the intermediary should feed
> forward the Correlation information to the ultimate recipient.

Because the message being forwarded is conceptually *the same* message,
albeit that some processing has been applies (or to use Henrik's analogy,
it's the same shopping basket, but some of the goods may have changed), and
because at least the MessageRef is abstract and local, the Correlation (or
causation) of the message being forwarded is implicitly the same as the one
being received (conceptually they are the same message). If we were to
include Correlation as a parameter in the .forward we might also have to
give some thought to what it might mean for it to be free to be different
from that passed in (if any)  with the .receive.

Another thought that I have had is that we might collapse this all down to
one UnitData operation with .send, .receive, .forward and .status
primitives. That might be something for a 2nd rev. of the document - after
all, if accepted it will be just a WD.

> John
> 
> 
>      XML Technology and Messaging,
>      IBM UK Ltd, Hursley Park,
>      Winchester, SO21 2JN
> 
>      Tel: (work) +44 (0)1962 815188        (home) +44 (0)1722 781271
>      Fax: +44 (0)1962 816898
>      Notes Id: John Ibbotson/UK/IBM
>      email: john_ibbotson@uk.ibm.com

Regards,

Stuart

Received on Tuesday, 27 March 2001 07:30:53 UTC