RE: One-way messaging in SOAP 1.2

Hi Noah,

I think I see more clearly where you are coming from. For me you alternate
wording :

> * In the binding framework, state that:  "Binding specifications that
> support more than one MEP MUST specify the means by which the send and
> receiver of a message can agree on the MEP being used.

has some ambiguity in that it is open to an interpretation that bindings may
agree use a different MEP from that requested/required by the SOAP node that
initiates the message exchange. Given what you've said I don't think that
this is how you intended to be interpreted. The way I see it is that the
initiator of a message exchange request the use of an MEP that a binding has
already claimed that it supports.

I agree with the sentiment below:

> Since bindings are allowed to engage in that 
> bi-directional "chatter" for a variety of purposes, why not to establish 
> the MEP?  

The crutial point is that all participant in the exchange become aware of
what MEP is in use. I think it is the 'agree' word that triggered my
response. How about:

* In the binding framework, state that:  "Binding specifications that
 support more than one MEP MUST specify the means by which all binding 
 instances participating in a message exchange become aware of the 
 MEP being used.

Regards

Stuart

> -----Original Message-----
> From: Noah Mendelsohn [mailto:noah_mendelsohn@us.ibm.com]
> Sent: 01 February 2002 03:37
> To: Williams, Stuart
> Cc: xml-dist-app@w3.org
> Subject: RE: One-way messaging in SOAP 1.2
> 
> 
> Stuart Williams says: 
> 
> >> What I dislike about the suggested revision is that it hints at the
choice
> >> of MEP being the subject of a run-time negotiation amongst the
participants
> >> in a message exchange.
> 
> I'm not encouraging such runtime negotiation of MEP's, and I would expect 
> it to be rare, I don't see why it shouldn't be allowed.  I think our 
> design already provides for it, and disallowing it would be 
> artificial. 
> 
> I think it's absolutely crucial to realize that, while the envelopes will 
> flow "downstream" from hop to hop, that the bindings will be sending 
> traffic bi-directionally, possibly in fragmented forms etc.  The most 
> obvious examples of such traffic are low level acknowledgements, flow 
> control window updates, etc.  Since bindings are allowed to engage in that

> bi-directional "chatter" for a variety of purposes, why not to establish 
> the MEP?   Unusual, but perfectly reasonable if that's how the binding 
> happens to be spec'd.
> 
> In short, I think that putting an assymetric responsibility on the 
> receiver is artificial.  The design is already set to give the appropriate

> flexibility in establishing MEPs, and if a binding specification wants to 
> call for low level "negotiation" at run time, I see no reason to prohibit 
> that.
> 
> ------------------------------------------------------------------
> Noah Mendelsohn                              Voice: 1-617-693-4036
> IBM Corporation                                Fax: 1-617-693-8676
> One Rogers Street
> Cambridge, MA 02142
> ------------------------------------------------------------------
> 
> 
>

Received on Friday, 1 February 2002 11:29:11 UTC