so,
there is no major diagreement.
the architecture talks about out-of band.
abbie
-----Original Message-----
From: Graham Klyne [mailto:GK-lists(_at_)ninebynine(_dot_)org]
Sent: Tuesday, May 21, 2002 3:13 PM
To: Markus Hofmann
Cc: OPES Group
Subject: Re: WG Last Call: draft-ietf-opes-architecture-00
At 10:55 AM 5/21/02 -0400, Markus Hofmann wrote:
I think the requirement here should be that tracing is
feasible *without*
using extensions on the application protocol. (Not saying that
extensions MAY NOT be used, just not required to get at
tracing information.)
If there's no extensions/addition at all, how would we get the trace
pointers to the enduser, so that the enduser can retrieve the more
detailed information fro mthe OPES boxes?
The short answer is: out of band. In-band would be nice if
it could be
achieved without protocol extensions.
I presume that the OPES intermediary would be communicating with the
endpoint in whose admin domain it is operating, so some local
arrangement
is not unreasonable. Returning to my earlier example (ISP
web cache), the
link for tracing information could be provided through a web
page that is
accessed through a login mechanism similar to that used for
retrieving mail
via a web browser.
Note, I'm not saying extensions cannot be used, just that
obtaining the
diagnostics should not depend on the endpoint having implemented such
extensions.
#g
-------------------
Graham Klyne
<GK(_at_)NineByNine(_dot_)org>