Date: Fri, 21 Sep 2001 10:11:09 -0700
From: Dale Moberg <>
For the Ferris solution, the IMSHes ack
when they hand off to the RoutingApp.
So when there is a downstream
failure in the remaining MSH RM agents,
our IMSH is in the embarassing situation
of having to somehow withdraw its ack.
Indeed, having to withdraw an ack is so embarassing that we clearly
don't want it to be part of the protocol. But I think all this can be
cleaned up if we just think of ebXML MS as really having two layers,
an "ebXML RM" lower layer and an "ebXML MS" higher layer. The ack
from the IMSH is at the lower layer and refers only to the hop-to-hop
communication. If an IMSH determines that things are hopeless and the
message cannot get through, it sends back a DFN, which is an entity of
the higher layer. The lower layer treats the DFN like any other
higher-layer message and starts bucket-brigading it back to the From
Party. All this is invisible to the application.
This means that while successful RM
occurs at the MSH level, failure
handling is in part pushed (conceptually)
up to the application layer.
No, not up to the application layer, just up to the "ebXML MS" layer,
still entirely within the MSH.
--
Dan
