IETF
xrblock@jabber.ietf.org
Thursday, November 13, 2014< ^ >
Room Configuration
Room Occupants

GMT+0
[01:05:36] Meetecho joins the room
[01:12:10] Meetecho leaves the room
[01:14:36] Alan Clark joins the room
[01:16:17] csp joins the room
[01:20:59] rachel huang joins the room
[01:26:51] chair joins the room
[01:27:15] <chair> Ok.. I am relaying the jabber room.. We have 4 people in the room including AD :-)
[01:27:37] <chair> Can you guys hear the room??
[01:27:39] <csp> Do we have more on jabber than in person? :-)
[01:27:42] <csp> can hear fine
[01:28:06] <chair> Yes, we have more on jabber than in person :-)
[01:28:20] <csp> A virtual working group?
[01:29:16] <Alan Clark> IMHO XRBLOCK should not create a separate registry but should discuss with IPPM
[01:30:10] <csp> I suspect we need a registry of what’s legal for WebRTC, but it can refer to the IPPM registry for the metric definitions.
[01:31:14] <Alan Clark> PMDIR has also discussed metrics registries in general - so this should also be discussed with them (would probably also recommend IPPM)
[01:31:25] <csp> Is the point of the registry to define the metrics, or to define the metrics that are legal for WebRTC? If the former, use the IPPM metric; if the latter, we need a new registry
[01:32:22] <csp> csp = Colin Perkins
[01:32:36] <Alan Clark> Purpose of registry is to ensure that metrics are defined once (somewhere)
[01:32:39] <chair> Did you hear that Colin?
[01:32:44] <chair> Latter.
[01:33:12] <csp> Agree with Varun: we need a webrtc xrbloc registry
[01:34:34] <chair> Alan, as far as I know, which ever path we take, it will be defined once (somewhere).
[01:35:11] <chair> For developers it will probably easier to parse the registry if it's specific for webrtc.
[01:35:13] <chair> Thoughts?
[01:35:44] <csp> spec req makes sense to me
[01:36:11] <Alan Clark> My concern is that if there are multiple registries then we will have equivalent metrics with incompatible definitions
[01:36:40] sunseawq joins the room
[01:36:47] <chair> Alan, did you hear that?
[01:36:50] <Alan Clark> ok
[01:36:55] sunseawq is now known as Qin Wu
[01:37:22] Alissa Cooper joins the room
[01:39:08] <chair> Roni presenting RTCP XR Block for Loss Concealment Metrics Reporting on Video Applications <http://tools.ietf.org/agenda/91/slides/slides-91-xrblock-0.pdf>
[01:41:26] <csp> Hi Roni… yeah, 1:40am here :-(
[01:41:50] <chair> Thanks Colin! We appreciate your participation!!
[01:42:37] <csp> The alternative was far too many hours in an aeroplane…
[01:42:59] <Alan Clark> Generally would only know that loss concealment has been applied, not whether it was effective
[01:43:26] <Alan Clark> Should be .. Call duration = Loss free video image duration + loss impaired video image duration
[01:43:43] <csp> Alan: agree
[01:46:23] <rachel huang> Okay. if that's the case, we only need one metric, that is loss free video image duration. Because we can know call duration from Measurement Identity report block.
[01:47:36] <Alan Clark> There is also the question of how we report segments of video where the loss concealment method was frame freeze - there is a major difference between 2 seconds of video with loss concealment using interpolation vs a 2 second frame freeze
[01:54:12] <Alan Clark> Kind of agree .. replace Loss Impaired Video Image Duration with metrics that report frozen vs interpolated duration?
[01:56:04] <csp> Freeze-frame vs other concealment could make sense. Not sure we want to go into too much detail about the different types of interpolated concealment?
[01:56:16] <Alan Clark> Interpolation in the general sense - some form of pixel replacement
[01:56:30] <csp> sure, but not broken down further
[01:57:01] <Alan Clark> Agreed - frozen vs some form of image recontruction
[01:59:12] <Alan Clark> Ok
[01:59:14] <rachel huang> ok
[01:59:37] <chair> Presenting the next step.
[02:00:41] <chair> Who have read the draft?
[02:01:10] <chair> ??
[02:03:59] Meetecho joins the room
[02:07:14] <Alan Clark> Maybe this could be the first fully virtual WG ?
[02:07:38] <chair> Yes indeed :-)
[02:07:59] <csp> Working groups don’t have to meet, but we need a home for new XR blocks (like we need PAYLOAD to remain)
[02:09:33] <chair> Thanks for participating remotely.. We couldn't have done it without you all :-)
[02:09:46] <csp> Thanks, and goodnight
[02:10:25] <rachel huang> Thanks. Bye.
[02:10:30] <Alan Clark> Bye !!
[02:10:31] <chair> Especially Colin for staying up late.
[02:10:43] Meetecho leaves the room
[02:10:43] Alan Clark leaves the room
[02:10:51] chair leaves the room
[02:11:15] rachel huang leaves the room
[02:11:25] csp leaves the room
[02:11:56] Qin Wu leaves the room
[02:13:50] Alissa Cooper joins the room
[02:14:07] Alissa Cooper leaves the room
[02:15:26] Alissa Cooper joins the room
[02:15:37] Alissa Cooper leaves the room
[02:18:28] Alissa Cooper leaves the room