Martin W Sachs wrote:
>
> I believe that the point is not that the originating MSH doesn't receive
> the ACK but that the originating MSH has exhausted the specified number of
> retries without receiving an ACK. Manual inquiry may be the right answer
> but this has to be clearly stated in the specification in the hopes that it
> will be propagated up to where the people who have to make the inquiries
> will see it.
>
> Someone else suggested making the RM protocol transactional, presumably
> with the idea that the received message would not be given to the
> application unless the RM transaction completed successfully. I don't know
> what that protocol would be but I would guess that it would have extra
> messages back and forth, so it might slow things down for the normal cases.
>
> Regards,
> Marty
>
<snip/>
Making the RM protocol transactional would complicate the protocol
considerably. This would be for the realm of ExactlyOnce which the
TR&P team felt was outside our scope.
Cheers,
Chris
