- From: Jean-Jacques Moreau <moreau@crf.canon.fr>
- Date: Tue, 20 Mar 2001 13:16:01 +0100
- To: frystyk@microsoft.com
- CC: "'Mark Nottingham'" <mnot@akamai.com>, Williams Stuart <skw@hplb.hpl.hp.com>, xml-dist-app@w3.org
Henrik Frystyk Nielsen wrote: > >This brings another question to mind - will there ever be a > >case where a block is targeted at a node in the request, the > >node processes the block during the request, and the response > >also needs processing by the node, without targeting? > > > >I can imagine that your challenge/credentials module would be > >targeted in both directions. Would there ever be a case where > >the response would be implicitly targeted, based on its > >correlation with the request? > > I think so. The reason is that there is only one "from" (i.e. a party > responsible for a message) and that party is located at the initial > sender. The initial sender may outsource processing to other nodes in > the message path but that doesn't change the party responsible for the > message in the first place. > > In other words, each block doesn't have an individual "from" associated > with it and when a receiver receives a message it always looks like a > single message with a single "from". Any party in the path can of course > send additional messages, which allows these parties to be associated > with the "from" for the messages they generate. Allowing multiple parties (handlers, intermediaries) to add blocks to a request is like allowing multiple conversations to be carried out on the same channel. In such a multi-speaker context, doesn't it matter to be able to identify individual speakers (this is probably implied by Fig. 2.1@AM)? If so, shouldn't we explicitely tag individual blocks with a "from" attribute? Jean-Jacques.
Received on Tuesday, 20 March 2001 07:18:38 UTC