Re: status of ROR (part deux)

The reason I'm not entirely supportive of Chris' proposal is that it's not 
entirely clear to me that in all cases where a 202 is received we want the 
receiving bit of software to act as a SOAP node.  You could imagine 202 
being used as a sort of tunnel, in which the transport is being used to 
move around SOAP envelopes, and only some of the places those envelopes 
land need be SOAP nodes.  Furthermore, if an MEP or binding calls for SOAP 
processing, then I think it should answer some other questions, such as 
where to deliver faults that may result from such processing.  I'm not 
strongly against telling that story about envelopes received with 202's, 
if that's what everyone wants, but I am against telling the story vaguely 
or incompletely.  So, my current inclination is either to stick with what 
I've got (mild preference for that), or to write a somewhat more complete 
story, starting with what Chris drafted, but adding some words about how 
no normative rules are provided for further transmissions of faults 
generated by such processing, maybe say that this binding makes no 
statement as to whether such a node is to act as the ultimate receiver (we 
don't say that about other messages either, though perhaps we should, but 
in this case it seems to me that the situation is so bizarre from an MEP 
point of view that it's of some value to be explicit.)  Not sure how much 
time it's worth tuning up things like this.

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Monday, 7 August 2006 14:41:43 UTC