Re: Gateway use case

David Hull writes:

> I like Jabber.  My friend uses BorgIM.  I can still send my friend a
> message by sending a jabber message to a specially-constructed jabber
> id.  This is transparent to me using my favorite jabber client.  I
> believe it's also transparent at the API level.
> 
> Now suppose I send a SOAP envelope to my friend's jabber id.  My friend
> receives it and executes the SOAP processing model on it.  The
> ImmediateDestination will be my friend's BorgIM id, rendered as a jabber 
id.
> 
>     * How many instances of the one-way MEP are there?  Who are the
>       senders and receivers?
>     * What bindings are used, and how?

Depends on whether that friend acts in the role of the ultimate receiver. 
If yes, then there is one instance of the one-way MEP taking the message 
to that ultimate receiver.  Anything that happens after that is beyond the 
scope of the one-way MEP as written.  Maybe it writes out another SOAP 
message to do another hop again, maybe it doesn't.  That would be like 
asking: what happens if I send a request/response, and after receiving the 
response send a 2nd request. 

If the friend does not act in the role of the ultimate receiver, then the 
question should be out of scope for now.  As we said on the phone, our 
current story about intermediaries and MEPs in general, is incomplete.  If 
the friend is an intermediary (I.e. not ultimate receiver), then we know 
we don't tell the story in detail.  Let's just acknowledge that and move 
on.  If someone wants to propose to recharter the group with the goal of 
clarifying our stories regarding MEPs and intermediaries that would be a 
possibility, though I'd be currently disinclined to support that effort, 
until I find more users implementing intermediaries, or making the case 
that if only we would give them a better spec they would do so.


--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Friday, 1 December 2006 00:52:15 UTC