Issue LC50 - MEPs

Jeffrey/Gudge and Amy,

At the WSDL 2.0 F2F meeting last week, we discussed the MEP issue that I
had raised:
http://www.w3.org/2002/ws/desc/4/lc-issues/issues.html#LC50
I had proposed adding an MEP (similar to p2c in 
http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl12/meps-vs-iops/meps-vs-iops_clean.htm#p2c
or http://tinyurl.com/4pjo4 ) that would not require the response
message to go back to the original requester.  (At present, our in-out
pattern is essentially p2e:
http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl12/meps-vs-iops/meps-vs-iops_clean.htm#p2e ,
which is a specialization of p2c.  I.e., p2c is more general or
less constrained than p2e.)

Sanjiva argued that the response is still going back to the requester
even if it is diverted to a different physical location.  In other 
words, it is still the same requester node even if the message is
being sent to a different physical address.  

I think this is a reasonable viewpoint.  I am slightly worried that
there could be other undiscovered issues if an underlying addressing
mechanism has the ability to send the message to an arbitrary recipient
that could in fact be a *different* node, but I haven't yet thought
of any.

GlenD also pointed out that from the client's point of view, if the
response is *not* going back to the same (logical) requester, then
from the client's point of view, p2c is equivalent to using two
one-way patterns (one going in, the other going out).  

Since a WSDL document 
should only contain information that is needed by *both*
the service and the client, I think Glen's point is compelling.
However, I seem to remember that both of you were strong proponents 
of the WG adopting adopting the more general (less constrained)
pattern (such as p2c) instead of p2e, so I would like to hear your
view of this issue before I rescind my proposal to add p2c.

-- 

David Booth
W3C Fellow / Hewlett-Packard

Received on Thursday, 18 November 2004 17:18:36 UTC