xcon bof

[Patrik Fältström <paf@cisco.com>]

Participants control centralized conferences.
  conference policy control
    general policy and membership
    media policy
  floor control

Sip conferencing initiated this...

Protocols developed will not be SIP or SIP extensions

Conference policy control

  draft-koskelainen-xcon-cpcp-reqs-00.txt
  draft-koskelainen-xcon-xcap-cpcp-usage-00.txt
  draft-levin-xcon-cpcp-00.txt

Media policy control

  draft-even-xcon-media-policy-requirements-00.txt
  draft-mahy-xcon-media-policy-control-00.txt

Floor control

  draft-koskelainen-xcon-floor-control-req-00.txt

Example Use Scenario

  draft-even-xcon-conference-scenarios-00.txt


Discussion...

The discussions were mixed between "scope in the charter" and
problem solution. Especially around use of multicast or not.
Floor control in a multicast environment would be extremely
complicated -- how do you prohibit someone from participating?

For me, there are a number of tricky details that needs
to be resolved, and I hope they will resolv them sooner
rather than later. There is just too much ability to
have features being added to the protocol as hacks, and
not as integrated parts in the protocol.

Example of things I didn't see discussed (enough) today:

 - When talking about roles, how is the abstraction
   mapping between roles and physical participants
   made? One would like at least some ideas early
   on regarding:
     +physical-user <-> alias <-> role
      (where physical user is for example SIP URI
      or email address or whatever)
     +is ability to admin locked to the role or
      user or both?
     +if this is extended to handle cascading,
      will we get an UID-collision problem
      and will there be conflicts between admin
      on the two systems?
 - Too many things are basically defined as being
   "blobs of XML". How authentication and authorisation
   and access control or the rest of the "base protocol"
   is to work is not known. Key management anyone?
   I saw some words saying "let's use SOAP". The
   design of this protocol which carries the
   XML-commands they are interested in is pretty
   important. I actually thing it is _MORE_ important
   (given the simple conference model they will use)
   to get the carrying protocol right than the
   XML blobs.

Humming regarding "interest", which indicated clear interest in this.