Minutes for CLUE at interim-2013-clue-3
minutes-interim-2013-clue-3-2

IETF CLUE WG Interim meeting, September 17, 2013
Time: 9:30am EDT - 11:30am EDT

Participants:
------------
Paul Kyzivat
Mary Barnes
Roberta Presta
Stephan Wenger
Rob Hansen
Mark Duckworth
Roni Even
Christian Groves
Keith Drage
Stephane Cazeaux
Christer Holmberg

Recording:
------------
https://ietf.webex.com/ietf/ldr.php?AT=pb&SP=MC&rID=16017692&rKey=95edee0ed7b2bdec

Raw notes by Rob Hansen:
------------------------

No comments on WGLC for the telepresence requirements document on the mailing
list - needs to be reviewed and people need to post comments on the mailing list

The current RTCP draft has expired; Roni presented the status about a revised
draft. As yet, Jonathon and Roni have not submitted a revised document because
it requires changes based on the work in other groups. FEC (forward error
correction) in particular was raised an issue - Roni and Jonathon plan to make
a proposal, hopefully keeping it out of the CLUE signalling.

Roni took an action to post a note to the list about the FEC issue.

The work being done on appId and the SDP unified plan for RTCweb both may have
impact on CLUE; appId is a more general form of captureId. The unified approach
is more difficult, as the current signalling in CLUE depends on m-line
attributes such as 'label'.

Keith clarified that the taxonomy draft would be an informative reference.
There was discussion of how much CLUE should be tied to the unified plan. Rob
took an action to look at how the CLUE signalling interacts with the unified
plan.

Mark presented some slides on the framework. One major open issue is switching,
which has individual drafts from both Mark and Christian. More comments are
needed on the mailing list about these drafts. Roles were also raised as an
issue.

There was a suggestion to move the framework to the standards track, as there
are some specifications in the framework that are required for a successful
implementation. Mary clarified that she felt that the framework document was a
normative reference for other documents, but not for an implementer - Keith
wanted to make sure that this was made clear in the abstract. Mary also raised
a ticket that some decisions need to be made before the security section can be
completed.

Mark committed to proposing some changes on the encoding group limitations as
per IETF87, removing codec-specific limitations and pixels-per-second from the
endpoint group constraints. There was some discussion here that clarified that
there was no intention to remove encoding groups as a whole.

Due to time limitation, the 'roles' section of the agenda was skipped, and
Christian presented his proposal on improving how switching would be
implemented in CLUE. The issue is that to solve many problems with CLUE the
original capture information is required, but with switching and composed as
present that is not available. The proposal is to have a new capture type (MCC)
indicating that multiple captures are combined herein. Captures that a part of
a MCC do not have to have an encoding (meaning that they can only be obtained
via an MCC; they can't be requested directly).

There was discussion of how the reciever could determine which capture they
were currently receiving when it was being received - the RTP mapping has been
focused on describing an encoding, so not all the tools will apply there. There
was also discussion of whether there was actually a need for this information
on a packet-by-packet basis - correcting for geometry would be one of the use
cases here.

There was also discussion of the fact that for large conferences this could
result in very large advertisments, and that it would also potentially involve
a lot more advertisments being sent as people joined or left the conference.
Again, for large advertisments this would likely require new advertisments.

There was discussion of how this relates to Mark's switching draft, and how
this can solve the problems raised there. Mark suggested that an example of how
to solve the problems raised in his draft would be usefu; Christian volunteered
to provide such an example.

Paul asked whether people had cycles to work on actually providing the
originating capture information, or whether we were willing to live with the
simple approach currently in the framework. Mark, Christian, Roni and Roberta
all expressed interest in this issue, and again there was a call for more
conversation on the list.

With ten minutes left we skipped the signalling and moved to the wrap-up. Mary
suggested that we kept the protocol and signalling documents seperate for the
time being to simplify the editorial task. There are still many tasks that need
to be solved on the signalling side, with little work being done by anyone but
Paul.