marshal and all,
please use the thread
[ OPES architecture] Final Points of Discussion: Tracing
that i have just started.
abbie
-----Original Message-----
From: Marshall Rose [mailto:mrose(_at_)dbc(_dot_)mtview(_dot_)ca(_dot_)us]
Sent: Wednesday, May 22, 2002 12:18 PM
To: Mark Baker
Cc: ietf-openproxy(_at_)imc(_dot_)org
Subject: Re: Callout (was Re: WG Last Call:
draft-ietf-opes-architecture-00
I just have one major comment ...
As I've stated here before, I can see no reason why
"callout" receives
any mention in an architectural document. Callout is an
implementation
detail, where an OPES entity determines that it needs to go
elsewhere to
accomplish its task. Furthermore, not only do I not see a need to
*identify* a callout protocol, I don't even see a need to
standardize on
one. The important interface is the one between entities,
such as that
provided by HTTP, or other transfer protocols. If you need
to "callout"
to a service, give it an HTTP interface and call it an
entity. This is
how things work today with HTTP intermediaries (firewalls, caches,
etc..).
mark - the reason that callout is in the architecture is
because we standardize a callout protocol. if we didn't care
to have the protocol standardized, then you and i are in
agreement; however, everyone i talk to in this space tells me
that a standardized callout protocol is a real-world
requirement. so, there you have it.
/mtr