At 05:13 PM 3/11/2010, Brian E Carpenter wrote:
Document: draft-ietf-ippm-twamp-session-cntrl-04.txt
Reviewer: Brian Carpenter
Review Date: 2010-03-12
IETF LC End Date: 2010-03-15
IESG Telechat date:

Summary: Ready, except for normative reference issue.
--------

Major issue:
------------

   Implementers of this feature MAY also wish to implement the "Reflect
   Octets" feature, described in [I-D.ietf-ippm-twamp-reflect-octets],
   once it has been published as an RFC.

I think this is a normative reference, even though it's optional, since
it uses a normative keyword. That could be fixed by s/MAY/may/, but
the later reference to the same draft is clearly normative:

   If the Control-Client has selected the Reflect Octets feature
   [I-D.ietf-ippm-twamp-reflect-octets] in combination with the
   Individual Session Control feature (after the Server identified its
   capability), AND utilizes the feature to insert a locally-specified
   request number in the Request-TW-Session command, THEN the Control
   Client MAY send more than one Request-TW-Session command to a given
   Server without waiting for the corresponding Accept-Session message.
   In such a case the Access-Session response reflects the locally-
   specified request number.  Note that when the Reflect Octets feature
   is being used all Request-TW-Session command and Accept-Session
   responses MUST include the locally-specified request number.

Brian,

Thanks for your comments.

I think we could move the paragraph above
(the last paragraph in section 3.1) to the
ietf-ippm-twamp-reflect-octets draft, provide a
reference to the session-control draft (as an RFC when approved)
and resolve that particular normative reference issue.

Since TWAMP-Control session requests and ACKs always precede the
messages to start and stop sessions (the focus of this draft),
they are separable topics, to me at least. Perhaps the fact that
we could move this paragraph so easily means that the text should
not use normative "MAY" terms - it's just an observation on how
two features might be used together.

The scenario these features have in common is
"create and control many simultaneous test sessions".
But any system using "reflect-octets" in the way described
can launch simultaneous Request-TW-Session commands.

So, for the purposes of resolving your comments, we would
delete the last paragraph of section 3.1, and s /MAY/may/
as you suggested in the introduction.

Al



_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art

Reply via email to