W3C home > Mailing lists > Public > xml-dist-app@w3.org > March 2001

[AMG] intermediaries

From: Yves Lafon <ylafon@w3.org>
Date: Mon, 19 Mar 2001 17:56:08 +0100 (MET)
To: <xml-dist-app@w3.org>
Message-ID: <Pine.GSO.4.33.0103191738120.25548-100000@tarantula.inria.fr>
"intermediaries" seems to imply that they are between the sending XMLP
application and the receiving XMLP application, however it may not be the
case:
You are subscribed on a service giving advice on surfing condition,
Q query a server S asking for information about a specific place. The
model is "with weather and surfer profile, I'll say if it is a good time
to surf". The request is then "here is my profile, get the weather and if
is is good to surf now".

Theoriginal request doesn't contain any information about the
subscription. The server S gets the request, get some information from a
Weather service and reply with that information, as the "Surfing Condition
Database" replied with a fault (requiring authentication)

     |-> SCD (Fault)
Q -> S -> W
  <-   <-

The it fill the information for SCD and reissue the query (with the
weather information gathered )
Q -> S -> SCD
  <-   <-

But note that S did the "real" generation of the reply.
Of course S has some knowledge, and we can argue that it will perform
another query rather than just transporting the received one, but still,
the intermediary SCD is after S (and so is W). So it really say "reply to
this" to what we call an intermediary rather than "go and get a reply to
this" (which is the current semantic of an intermediary). This, of course
leads to a more hop-by-hop model.

-- 
Yves Lafon - W3C
"Baroula que barouleras, au tiéu toujou t'entourneras."
Received on Monday, 19 March 2001 11:56:10 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:59 GMT