RE: T2 - Assertions and Questions

[David Fischer <david@drummondgroup.com>]

David,

I assumed for my example that you had another legal proxy (the To Party MSH,
maybe your own mailroom MSH) other than C1.  This is not unreasonable and may
occasionally occur (in any discussion about IMs we are talking about a minority
use case).  I have to disagree with your assertion that there are no unknown
IMs.  Once you pass to your local IM (in this case Ariba, which has no legal
standing other than to pass the message through) you don't necessarily know
where this message might go and a receipt/ack from your local IM does NOT
guarantee that the message arrived at the final destination nor does a receipt
from that IM, signed or otherwise, constitute any legal obligation on the To
Party to perform, nor will it have any legal standing in court.  Even if you
were to obtain legally binding, signed acknowledgements from each hop, only the
final acknowledgement would make any difference at all.  Even that is not a
contract between the To Party and the From Party, only a DR between the ends
would imply such a contract.

BTW, if the IMs do garbage collection routinely, how would you obtain that chain
of signed receipts at a later date?  Most of the corroborating data would be
gone.  Only the ends keep this information for any duration.  Even if you tried,
the spec has no way of requesting a "signed chain" of Acks.  The value of
AckRequested="signed" may be changed by any hop along the way.

Let's take another case.  If I do not have a persistent connection, I might have
an IM hold messages for me until I retrieve them.  In this case, they are not
the end and they cannot send a legally signed DR for me (why do we talk about
legal when we are not lawyers?).  The signed DR must wait for me (the MSH) to
pick up the message and then send the DR.  In Dan's example, the IM would then
be analogous to a PO Box.  Just because the message gets put in my PO Box does
not prove that I got it.  This is where the analogy breaks down because the
postman must have me sign on delivery where electronically, I sign myself and
send back without a postman.  Electronically, only the possessor of the
private-key may sign.

The only way I know to eliminate the idea of multi-hop is to encapsulate (wrap)
the From-To message as a payload of another ebXML message which would go to the
IM who would then unwrap and send, etc.  This idea looked fine until we added
TraceRouteList.  Now we must have multi-hop.  Much as I would like to support
Assertion #3C (I even made such an assertion myself once), we have been through
all this and it simply doesn't work.

Regards,

David Fischer
Drummond Group.

-----Original Message-----
From: David Smiley [mailto:dsmiley@mercator.com]
Sent: Friday, August 17, 2001 10:32 AM
To: 'David Fischer'; Dan Weinreb; mwsachs@us.ibm.com
Cc: ebxml-msg@lists.oasis-open.org
Subject: RE: T2 - Assertions and Questions


David F.,

I would like you to clarify your example (copied below):

<DavidF>
In an ebXML network example, I (the From Party) might send through Ariba and
then through C1 to you (the To Party).  You might actually have someone
hosting
your MSH which has legal authority to respond with a DR.  If I get a receipt
from Ariba or C1, this means nothing (except that the message is proceeding
on
the path, which is good to know but not critical).  I might not even know C1
is
in the path!  I need a DR (signed?) from you or your proxy (I am assuming
here
that C1 is not your legal proxy).
</DavidF>

<DavidS>
I think your example does not describe the relationships between the various
parties sufficiently. What is the business relationship between From Party
and Ariba? Where did the To Party specify for incoming message delivery? Why
are you assuming that C1 is not the legal proxy?

If the To Party tells the From Party to deliver messages to C1, I contend
that successful delivery to C1 is end of the line as far as the From Party
is concerned. In that case, a DR from C1 is meaningful. If C1 is not the
legal proxy of the To Party, it would not be used by the To Party as its
designated destination for message delivery.

I come back to my basic assertions on this whole topic (omitting detail
which can be found in any of the 18 other messages with this subject):

Assertion #1:
Application-level processing of any kind is out of scope of the Message
Service Specification.

Assertion #2:
The To Party defines the location where messages intended for it are to be
sent.

Assertion #3A:
From a business viewpoint, there is no such thing as an unknown intermediate
party.

Assertion #3B:
This still works in an ecommerce marketplace scenario.

Assertion #3C:
For our purposes, there is no such thing as a multi-hop message.

</DavidS>

David Smiley
Director of Standards
Mercator

<trim/>


----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>