Re: [ebxml-bp] RE: [uddi-spec] Publishing UN/CEFACT UMM models to aUDDI registry

["Monica J. Martin" <Monica.Martin@Sun.COM>]

Jean-Jacques Dubray wrote:

>Dubray: Steve:
>Frankly, I don't see why we allow a retry at the business level
>different from the RM level. For me once a message has been reliably
>delivered, the sender is done. All should be handled at the responding
>side. I am assuming that this is a legacy from the time where RM was not
>widely available. 
>
>Dale, would you consider removing that feature from the specification?
>  
>
mm1: Let's discuss on ebBP list before further redress on UDDI list. 
Thanks. The legacy was based on the UMM. See from Chapter 8, R10 
(Section 8.3.1) particularly the last part of the sentence. In addition 
this is further

"Both partners agree to the number of times to retry a transaction when 
a time-out-exception condition is signaled. This parameter only applies 
to time-out signals
and not business process controls or document content exceptions."

mm1: This is further addressed in the next section 8.3.2, the business 
transaction is to be reinitiated (but based on the timeout of lack of 
response receipt). What this (and other discussions I am involved in 
such as UBAC) point out is that some premises in the UMM may need 
redress. Retry count is only one instance.

>Incidentally, I don't think a retry would allow for any content change.
>This is counter intuitive to a retry.
>
mm1: See comment above.

>With respect to WSDL, we are not working on the same level of
>granularity that you are proposing, but our proposal will include your
>mapping as well so that's good. The idea is that if you have a
>correlation method (e.g. WS-Addressing, as one amongst others), you can
>establish a complex mapping between all message/signal interchanges of a
>BTA and several operations either from the same port type or more than
>one (could have signals handled by a separate port type). We would not
>constrain to use the ebXML signals in the WSDL mapping case. Any
>operation message could count for a signal if mapped as so.
>
>Let me know if you have more questions.
>
>Jean-Jacques
>  
>