[07:33:13] cm-msk@jabber.org joins the room
[07:33:21] cm-msk@jabber.org leaves the room
[07:36:57] cm-msk@jabber.org leaves the room
[08:37:50] Meetecho joins the room
[08:50:03] Paolo Saviano joins the room
[08:50:03] Martin Thomson joins the room
[08:50:03] Murray Kucherawy joins the room
[08:50:03] Alessandro Amirante joins the room
[08:50:03] Youenn Fablet joins the room
[08:50:03] Bernard Aboba joins the room
[08:50:03] Juliana Guerra joins the room
[08:50:03] Robert Stepanek joins the room
[08:50:03] Emad Omara joins the room
[08:50:12] Ted Hardie joins the room
[08:50:39] <Ted Hardie> We can hear you.
[08:50:56] <Martin Thomson> Interesting.
[08:51:05] <Martin Thomson> Outbound OK, inbound ... ?
[08:51:48] <Martin Thomson> write-only video conferencing seems like an interesting innovation
[08:52:01] <Ted Hardie> "Broadcast", they could call it.
[08:52:12] <Meetecho> For active audio connections, the browser initiates the SDP offer, while for recvonly ones, they're initiated by the server: this may cause some differences in ICE setup (e.g., in terms of which connectivity checks are done first)
[08:52:15] Kiran Makhijani joins the room
[08:52:36] <Meetecho> Not sure if this is related to your issue, but we'll try to investigate
[08:53:28] Mike Coleman joins the room
[08:53:49] frodek joins the room
[08:54:10] Frode Kileng joins the room
[08:54:48] Youenn Fablet leaves the room
[08:54:52] Youenn Fablet joins the room
[08:55:22] Youenn Fablet leaves the room
[08:55:27] Youenn Fablet joins the room
[08:56:01] Simon Romano joins the room
[08:56:05] Sean Turner joins the room
[08:56:11] Alissa Cooper joins the room
[08:56:31] Richard Barnes joins the room
[08:56:45] Eric Rescorla joins the room
[08:57:01] Sergio Garcia Murillo joins the room
[08:57:01] Colin Perkins joins the room
[08:57:47] Joseph Kalfa joins the room
[08:57:48] Jonathan Lennox joins the room
[08:58:20] Sergio Garcia Murillo leaves the room
[08:58:25] Sergio Garcia Murillo joins the room
[08:58:34] Sergio Garcia Murillo leaves the room
[08:58:39] Sergio Garcia Murillo joins the room
[08:58:45] Lucas Pardue joins the room
[08:59:23] Chi-Jiun Su joins the room
[08:59:24] Patrick Tarpey joins the room
[08:59:41] Shuai Zhao joins the room
[08:59:58] Kazuaki Ueda joins the room
[09:00:18] Saúl Corretgé joins the room
[09:00:34] Timothy Panton joins the room
[09:00:36] Bobo Bose-Kolanu joins the room
[09:00:37] <Murray Kucherawy> Trying to contact Bobo.
[09:00:43] Yaron Sheffer joins the room
[09:00:47] <Murray Kucherawy> Oh there he is.
[09:01:02] Stephan Wenger joins the room
[09:01:03] Xavier de Foy joins the room
[09:01:07] Renan Krishna joins the room
[09:01:11] <Richard Barnes> whoa, it takes like a full 750ms to unmute
[09:01:13] Stephan Emile joins the room
[09:01:29] cm-msk@jabber.org joins the room
[09:01:44] <Emad Omara> is there anyone talking? I can't hear/see anyone
[09:01:46] Dragana Damjanovic joins the room
[09:01:49] <Meetecho> Richard Barnes: I made a full presentation in MOPS this morning explaining how audio and video work
[09:01:54] Subir Das joins the room
[09:02:03] Harald Alvestrand joins the room
[09:02:04] Alexandre Gouaillard joins the room
[09:02:06] <Jonathan Lennox> Emad: yes, we have started
[09:02:13] <Meetecho> You may want to watch it later for more insight, but long story short, unmuting opens a new PeerConnection, so that may take a bit
[09:02:28] Alexandre Gouaillard leaves the room
[09:02:44] <Emad Omara> hmm I'm not getting any media streams
[09:02:52] <Meetecho> Emad Omara: have you tried reloading?
[09:03:01] James Gruessing joins the room
[09:03:01] Mike English joins the room
[09:03:02] James Gruessing leaves the room
[09:03:05] James Gruessing joins the room
[09:03:18] Ahmed Bashandy joins the room
[09:03:21] Alexandre Gouaillard joins the room
[09:03:34] Mo Zanaty joins the room
[09:03:46] Emad Omara leaves the room
[09:03:55] Emad Omara joins the room
[09:04:10] <Emad Omara> reloading worked
[09:04:11] Mohit Sethi joins the room
[09:04:13] <Emad Omara> all good now
[09:04:17] <Richard Barnes> I know you think there are justifications, but clearly opening a new PC on unmute is not necessary and has bad UX consequences.
[09:04:19] <Emad Omara> video conferencing is hard!!!
[09:04:21] Mihail Yanev joins the room
[09:05:29] <Simon Romano> ...and scalable video conferencing is even harder ;-)
[09:05:41] <Juliana Guerra> I anyone talking right now?
[09:05:48] <Juliana Guerra> IS*
[09:05:51] <Meetecho> Richard Barnes: we realize, I was explaining how it currently works. We are already working on fixes for that (persistent audio PC)
[09:05:51] <Jonathan Lennox> Juliana: yes
[09:05:53] <Emad Omara> Yes
[09:06:20] Justin Uberti joins the room
[09:06:30] Robert Stepanek leaves the room
[09:07:19] Juliana Guerra leaves the room
[09:07:42] <Bernard Aboba> Where did everyone go?
[09:07:55] Robert Stepanek joins the room
[09:07:55] <Eric Rescorla> @Bernard: we are all still here. Something is busted
[09:08:07] Juliana Guerra joins the room
[09:08:27] Jonathan Lennox (Jabber) joins the room
[09:09:22] <Jonathan Lennox> I don't see Bernard in the participant list at all in Meetecho, though I see him in Jabber
[09:09:51] <Meetecho> Bernard Aboba: can you try reloading?
[09:09:58] Loren McIntyre joins the room
[09:10:35] <Richard Barnes> @Meetecho: FYI, the fact that you release and re-acquire the audio device causes additional latency with some headsets that mode switch depending on whether the mic is on.  So for me it takes a full ~2sec to unmute or mute, during which i lose audio playout as well
[09:11:20] <Meetecho> Richard Barnes: ack. Releasing the mic is done on purpose to ensure maximum privacy though. We don't like the idea of holding on to a resource when we're not using it.
[09:11:51] <Richard Barnes> (might be that i just have a janky headset though :) )
[09:11:56] Bernard Aboba leaves the room
[09:11:57] <Meetecho> I wasn't aware of this hardware issues, though
[09:12:05] <Eric Rescorla> @Meetecho: I hear what you're saying, but I think that's actually backwards.
[09:12:06] <Richard Barnes> I hear you, but privacy vs. UX is always a balance.
[09:12:41] Jonathan Lennox (Jabber) leaves the room
[09:12:43] <Meetecho> Fair points, and thanks for the feedback!
[09:13:02] <Eric Rescorla> Because in most of these cases, you are holding a permission so there's no meaningful difference between "mic is acquired but not plumbed to the network" and "mic is not acquired but can be acquired with no user input"
[09:13:04] Bernard Aboba joins the room
[09:13:35] <Richard Barnes> :point_up:
[09:13:49] <Harald Alvestrand> @ekr that runs into the "recording indication" question from webrtc....
[09:13:56] <Eric Rescorla> @Harald: sure.
[09:14:32] Magnus Westerlund joins the room
[09:14:32] <Eric Rescorla> We show the indicator when the mic is acquired
[09:14:42] James Gruessing leaves the room
[09:15:21] <Eric Rescorla> FWIW, Meet seems to hold the mic even when muted
[09:15:27] <Eric Rescorla> but not the camera when off
[09:15:29] <Colin Perkins> There's a reason RTP didn't do a generic packetisation, of course
[09:16:48] <Martin Thomson> I'm using my browser mute button and staying active
[09:17:05] <Harald Alvestrand> @ekr that's true. they do that by release rather than mute because chrome doesn't turn off the light on mute.
[09:17:30] <Harald Alvestrand> @barnes you should send me (privately) the make of your mike - I'm sure the team knows it's a bad one, but never hurts to be fully sure.
[09:17:33] <Eric Rescorla> As far as I know, the light is controlled by the OS
[09:17:45] <Eric Rescorla> Or the hardware
[09:18:28] <Eric Rescorla> @Meetecho: what would be really helpful here would be if people's mics were automatically activated when they came to the front of the queue
[09:18:40] <Bobo Bose-Kolanu> did anyone else just lose all audio?
[09:19:14] <Eric Rescorla> It's still live for me
[09:19:19] <Meetecho> Eric Rescorla: actually this changed from 108, the queue is not tied to the media itself anymore. When you click the mic icon, you automatically unmute: the hand icon is only a visual helper for chairs
[09:19:20] Bobo Bose-Kolanu leaves the room
[09:19:22] <Colin Perkins> We don't wan to limit the possibility of partial decoding, just because current implementations don't support it
[09:19:25] Bobo Bose-Kolanu joins the room
[09:19:33] Jingrong Xie joins the room
[09:19:34] <Meetecho> That's what we were asked to do
[09:19:48] <Bobo Bose-Kolanu> reloading restored audio
[09:19:53] <Eric Rescorla> @Meetecho: yes, I'm saying that this is not ideal, because what happens in practice is that people get to the front of the queue and *then* need to unmute
[09:20:06] <Jonathan Lennox> Colin: the problem is that partial decodability precludes the frame-wide auth overhead amortization.  It's a tradeoff.
[09:20:17] <Eric Rescorla> I'd really like to see some data on the overhead
[09:20:30] <Jonathan Lennox> Basically the auth tags
[09:20:30] <Colin Perkins> Sure
[09:21:06] <Eric Rescorla> Jonathan: sorry, what I mean here is how many packets on average a frame covers
[09:21:24] <Eric Rescorla> because that tells you how much it's improving
[09:21:25] <Jonathan Lennox> Ah, I see.  (It depends a whole lot on your bitrate, obviously.)
[09:21:31] <Eric Rescorla> yep
[09:23:26] <Colin Perkins> @magnus: That's a separate issue then I was mentioning, but also a concern
[09:23:56] <Richard Barnes> Youenn is correct that how we packetize is *the* question for this WG.  All the crypto stuff is pretty trivial.
[09:24:42] <Bobo Bose-Kolanu> 👍🏾
[09:24:58] Zaheduzzaman Sarker joins the room
[09:28:03] Simon Romano leaves the room
[09:29:28] Zaheduzzaman Sarker leaves the room
[09:30:28] <Eric Rescorla> I am also happy to bring it back in once someone understands what we are trying to do and has good crypto
[09:32:08] Alessandro Amirante leaves the room
[09:32:11] Alessandro Amirante joins the room
[09:32:22] Hiroyuki Goto joins the room
[09:32:47] <Harald Alvestrand> if there is no signature, the only trust about who sent it is "decrypting with the key I thought was the right one didn't produce garbage".
[09:33:24] Bobo Bose-Kolanu leaves the room
[09:33:39] Bobo Bose-Kolanu joins the room
[09:34:10] <Martin Thomson> I want a tool that can cut the queue.  You people just don't stop.
[09:34:14] Alessandro Amirante leaves the room
[09:34:20] Alessandro Amirante joins the room
[09:34:23] <Jonathan Lennox> It's called saying "I'm cutting the queue".
[09:34:58] <Harald Alvestrand> almost everything in this preso is, IMO, out of scope for the WG.
[09:35:14] <Bobo Bose-Kolanu> @Meetecho - stream dropped to 0kbps again, had to reload to access fyi, not sure why this is happening
[09:36:09] Moshe Pupkin joins the room
[09:36:20] <Mike English> Does this imply a need for some out of band identity provider? Seems like we're circling around this scenario where the SFU and website may be untrusted, but how else do participants establish trust with each other?
[09:36:22] <Eric Rescorla> @Bobo: what browser are you in?
[09:36:32] <Bobo Bose-Kolanu> @ekr - chrome
[09:36:39] <Bobo Bose-Kolanu> osx
[09:37:10] <Martin Thomson> does this assume some sort of acknowledgment/feedback channel?
[09:37:11] <Eric Rescorla> @Bobo: OK that's informative. MT was having a lot of problems of this nature but using Firefox on Windows. This suggests strongly that the problem is on the Meetecho side
[09:37:27] <Richard Barnes> I probably disagree that KID can be per-user.  Assuming that the app provides user distinction seems like a big ask.
[09:37:29] <Colin Perkins> To clarify: the wire format goes into RTP packets?
[09:37:52] <Eric Rescorla> @Richard: well, you need a key per sender, right?
[09:37:52] <Martin Thomson> Richard: do you want to open an issue?
[09:37:55] <Eric Rescorla> Or a very large nonce
[09:38:08] <Richard Barnes> @Eric - My mental model was that the KID would indicate the sender key
[09:38:18] <Eric Rescorla> @Richard: sure, that works
[09:38:21] <Bobo Bose-Kolanu> @ekr - 👍🏾
[09:38:54] <Meetecho> Bobo Bose-Kolanu: looking at the stats the new streams seem fine, so it may be a temporary network hiccup
[09:39:04] Patrick Tarpey leaves the room
[09:39:24] <Bobo Bose-Kolanu> @meetecho - maybe, twice in one session seems odd to me...will let you know if it happens a third time
[09:40:04] <Harald Alvestrand> Signatures over N frames only work if all N frames are delivered. With winnowing SFUs, they are frequently not.
[09:40:07] Mohit Sethi leaves the room
[09:40:10] Mohit Sethi joins the room
[09:40:32] <Meetecho> Bobo Bose-Kolanu: the only thing I notice is a slightly higher RTT but very few video nacks for instance
[09:40:32] <Richard Barnes> point of order: we haven't actually adopted a document :)
[09:40:33] James Gruessing joins the room
[09:40:52] <Eric Rescorla> @harald: right, you need to include redundant packets
[09:40:57] <Eric Rescorla> Sorry, redundant hashes
[09:42:40] <Eric Rescorla> @Bernard: are you back?
[09:44:54] <Richard Barnes> Point of order: This was supposed to be a use case discussion
[09:46:04] Moshe Pupkin leaves the room
[09:46:16] <Colin Perkins> This discussion would be a lot easier to follow if we'd had the use case explained first
[09:47:24] <Richard Barnes> If your codec can't survive fuzzing, you're gonna have a bad time
[09:47:32] Chi-Jiun Su leaves the room
[09:47:39] Chi-Jiun Su joins the room
[09:47:57] <Eric Rescorla> I don't understand why it allows fuzzing. You're already part of the group and so you can just sign with *your* key
[09:48:30] <Richard Barnes> i guess if you have per-user authn, then you get authn of who's fuzzing you
[09:48:37] <Eric Rescorla> Authenticated fuzzing
[09:48:42] <Richard Barnes> precisely
[09:50:59] <Richard Barnes> Good point, Emad -- don't need the aggregation info tied to the crypto, so we could punt to the packetization layer
[09:51:22] Robert Stepanek leaves the room
[09:51:24] <Magnus Westerlund> People need to be clearer on terminology due to the overlap between SFRAME frames and video frames.
[09:52:08] <Richard Barnes> i have been using "chunks" or "encrypted units" in my notes on this topic
[09:52:20] <Justin Uberti> we need some sort of 'independent decodable unit' word that is easy to say
[09:52:26] <Richard Barnes> IDU!
[09:52:27] <Bobo Bose-Kolanu> "IDU"
[09:52:29] <Bobo Bose-Kolanu> lol
[09:52:51] <Eric Rescorla> just say IFRAME. What could go wrong
[09:52:55] <Justin Uberti> lol
[09:52:58] <Bobo Bose-Kolanu> hahaha
[09:53:38] Stephan Wenger leaves the room
[09:53:43] Stephan Wenger joins the room
[09:55:22] Mihail Yanev leaves the room
[09:55:46] <Justin Uberti> multiple dimensions here - each video frame can have multiple encodings, and each encoding can have multiple IDUs
[09:55:50] <Harald Alvestrand> anytime I see "FU payload", I can't unsee the inapproriate substitution.
[09:57:23] Frode Kileng leaves the room
[09:57:43] <Richard Barnes> what is FU supposed to be here?
[09:57:49] <Jonathan Lennox> Fragmentation Unit
[09:57:53] <Richard Barnes> (hta notwithstanding)
[09:57:54] <Richard Barnes> thanks
[09:57:55] <Magnus Westerlund> The video IDU that is protected by a single SFRAME will be larger than a single IP/UDP/RTP can carry, and thus the RTP payload level need fragmentation support for individual SFRAMES?
[09:59:23] <Martin Thomson> Justin can you type that idea up?
[10:00:29] <Mo Zanaty> subframe
[10:00:56] <Richard Barnes> i think the goal is to *minimize* SFU impacts
[10:01:01] <Bobo Bose-Kolanu> @justin in the etherpad would be great if possible 🙏🏾
[10:01:14] <Richard Barnes> https://codimd.ietf.org/notes-ietf-109-sframe?edit
[10:02:34] <Bobo Bose-Kolanu> @justin there's a '...' where your writeup could go
[10:02:50] <Justin Uberti> willdo
[10:02:59] <Martin Thomson> thanks
[10:03:00] <Harald Alvestrand> we need some place to address the question of how to signal in SDP that sframe has been applied. It's been ruled out of scope for this WG.
[10:03:23] <Martin Thomson> if it is masking out the bits we like from the bits we don't, that's gross.
[10:03:23] <Justin Uberti> @magnus westerlund yes, fragmentation support will be needed for when IDUs exceed MTU
[10:03:39] <Jonathan Lennox> Harald: my instinct is that it would be defined as part of the RTP payload format, which is where SDP descriptions usually go.
[10:04:39] <Justin Uberti> ultimately we will have some transforms for existing codecs, hopefully fairly simple, and with new payload formats they will be designed to separate out their metadata from payload so no transform is needed
[10:05:46] <Colin Perkins> Does this mean we need, e.g., h264-frame payload format?
[10:05:56] <Colin Perkins> h264-sframe
[10:06:04] Hiroyuki Goto leaves the room
[10:06:06] <Justin Uberti> I don't think so, but we will need a h264 transform to sframe
[10:06:49] <Colin Perkins> hmmm... I wonder if the RTP signalling may be easier if it's done as a payload format?
[10:06:51] <Justin Uberti> the key question is whether the transform just decides what bits are encrypted or whether it has to map the metadata to a generic common metadata
[10:07:29] <Jonathan Lennox> I think the implementation complexity may be comparable either way.
[10:07:38] <Magnus Westerlund> @Mo, so is this slide wrong that DD format that AV1 developed is actually not generic?
[10:08:24] <Colin Perkins> Question for the crypto people: does the crypto work at a byte level, or does it work on blocks?
[10:08:48] <Jonathan Lennox> The DD format can probably be applied to any older codec, as I understand it.  No guarantee it can be applied to not-yet-defined ones, of course.
[10:09:28] <Colin Perkins> i.e., can we encrypt an arbitrary length, or is padding needed
[10:09:30] <Shuai Zhao> is the DD comply with RFC8285?
[10:09:35] <Richard Barnes> @Colin - arbitrary length
[10:09:36] <Mo Zanaty> DD wants to be generic. So did frame-marking. So will the next metadata format. None will capture the interesting semantics of new formats, which may innovate in ways we don't foresee.
[10:09:50] <Martin Thomson> arbitrary length up to an insane limit
[10:10:13] <Stephan Wenger> +1 to Mo.
[10:10:25] <Jonathan Lennox> Yes, DD uses 8285, but it needs the long format (0x100x) which isn't commonly implemented compared to the short format (0xBEDE).
[10:10:26] <Colin Perkins> that wasn't true of some of the older AES modes, right? Or am I confused
[10:10:45] <Eric Rescorla> That's right
[10:10:53] <Richard Barnes> @Colin - that's correct.  but everything modern operates on unpadded inputs
[10:10:56] <Eric Rescorla> But RTP has always used modes that didn't require padding
[10:11:13] <Eric Rescorla> Because it used to be CTR and is sometimes now GCM
[10:11:23] <Jonathan Lennox> (For values of "always" that don't include the original broken crypto in RFC 1889, which may be what Colin is remembering.)
[10:11:33] <Colin Perkins> that's maybe it - thanks
[10:11:47] <Magnus Westerlund> Colin, are you intending to cut a subset of the encrypted bytes, then decrypting that doesn't work at arbitrary boundaries, then you need a stream cipher.
[10:11:49] <Eric Rescorla> Sorry, I should have said SRTP always used
[10:12:49] Mike Coleman leaves the room
[10:13:02] <Martin Thomson> muscle memory
[10:13:39] <Sergio Garcia Murillo> @Mo have you checked how the dependency descriptor works?
[10:15:37] <Justin Uberti> added some thoughts on STAP-A to the etherpad.
[10:17:15] <Bobo Bose-Kolanu> thanks justin
[10:18:30] Lucas Pardue leaves the room
[10:18:37] Lucas Pardue joins the room
[10:20:20] Shuai Zhao leaves the room
[10:20:35] Shuai Zhao joins the room
[10:20:43] Lorenzo Miniero joins the room
[10:21:05] <Martin Thomson> I don't think that E needs to be flexible
[10:21:48] <Jonathan Lennox> Not for interactive use cases, I think I agree.  We might want to make sure we don't have any stored-media uses cases for SFrame, though.
[10:23:39] Oshani Dayaratna joins the room
[10:24:03] Deepti Rathi joins the room
[10:24:14] Oshani Dayaratna leaves the room
[10:24:40] Lucas Pardue leaves the room
[10:25:56] Deepti Rathi leaves the room
[10:26:25] <Yaron Sheffer> There's some confusion between kid and truncated-kid.
[10:27:18] slm joins the room
[10:27:19] <Martin Thomson> I think that the disconnect is whether you couple SSRC + KID or just use KID.
[10:27:45] <Mo Zanaty> @Sergio, DD was designed to support a specific subset of AV1 scalability structures and reference structures. It does not support every possible structure that AV1 allows, much less what AV2 will allow.
[10:29:27] <Bobo Bose-Kolanu> @ekr - for note-taking; re: cipher suite issue, was your point of view that MLS extension should be used to negotiate this?
or something else?
[10:29:52] Stephan Emile leaves the room
[10:29:55] Stephan Emile joins the room
[10:31:02] <Sergio Garcia Murillo> @mo the DD tracks frame dependencies (past and future) of each frame. AV1 structures are not coded "into" the DD.
[10:32:03] <Martin Thomson> sergio: there is also the risk that sframe is too specific to RTP
[10:32:32] andrew_liu joins the room
[10:33:31] <Sergio Garcia Murillo> I agree
[10:34:12] <Sergio Garcia Murillo> we should try to avoid being too RTP-specific
[10:35:15] Xing Jiang joins the room
[10:35:58] <Richard Barnes> Sergio: Note separate SFrame and MLSContext structs https://github.com/cisco/sframe/blob/master/include/sframe/sframe.h#L102
[10:36:03] <Colin Perkins> Can you say the result for the audio recording?
[10:38:15] <Richard Barnes> i'm impressed by those who are in a superposition of "hand raised" and "hand not raised"
[10:38:26] <Jonathan Lennox> nullable boolean
[10:38:27] <Richard Barnes> Schrödinger's poll
[10:38:40] Xing Jiang leaves the room
[10:41:29] Lucas Pardue joins the room
[10:41:55] Stephan Wenger leaves the room
[10:42:42] <Emad Omara> Is it IETF 110 already ?
[10:42:43] <Mohit Sethi> Slide says ietf 110
[10:43:03] <Jonathan Lennox> Forward-thinking
[10:43:15] <Bobo Bose-Kolanu> rule 110 can be any numbered IETF it wants
[10:46:53] frodek leaves the room
[10:47:07] Kazuaki Ueda leaves the room
[10:47:10] Kazuaki Ueda joins the room
[10:49:12] <Justin Uberti> This seems like a generalized form of the video conferencing scenario.
[10:49:42] Bron Gondwana joins the room
[10:50:18] <Richard Barnes> maybe with less frequent key rotation
[10:50:20] <Bobo Bose-Kolanu> @justin - is this because of interactivity amongst the N?
[10:50:35] <Ted Hardie> What seems different is how you get the user's key to the sender; what acts as the introducer?
[10:50:47] <Justin Uberti> Agreed, the KMS is different here
[10:52:18] <Justin Uberti> @bobo in many ways it's similar to a single conference participant in a very large conference (i.e. they are just focused on distributing their media to N recipients)
[10:53:35] <Harald Alvestrand> the content distributor wants to be able to exclude members from the group too.
[10:53:45] mcint joins the room
[10:53:47] <Bobo Bose-Kolanu> does this hold for very large N? trying to understand if there is only difference in degree or if there is difference in kind at some point in scale
[10:53:51] <Harald Alvestrand> (cable viewer who hasn't paid his bill)
[10:54:25] <Harald Alvestrand> could run mls with hundreds or thousands of leaves/joins per epoch
[10:54:58] <Richard Barnes> my guess is that MLS doesn't add that much value in this case
[10:55:31] <Eric Rescorla> So usually you just encrypt this all with one key and then encrypt that key to the authorized viewers
[10:55:57] Ahmed Bashandy leaves the room
[10:55:58] Kiran Makhijani leaves the room
[10:55:59] Ted Hardie leaves the room
[10:56:00] <Eric Rescorla> The media is enormous so the cost of the key transport is minimal
[10:56:01] Chi-Jiun Su leaves the room
[10:56:02] Emad Omara leaves the room
[10:56:03] Murray Kucherawy leaves the room
[10:56:03] Eric Rescorla leaves the room
[10:56:04] Magnus Westerlund leaves the room
[10:56:04] <mcint> project doing p2p video conferencing, in scope to consider? https://github.com/meething/meething
[10:56:05] Yaron Sheffer leaves the room
[10:56:06] Mohit Sethi leaves the room
[10:56:07] Richard Barnes leaves the room
[10:56:07] Timothy Panton leaves the room
[10:56:07] Renan Krishna leaves the room
[10:56:07] Alissa Cooper leaves the room
[10:56:08] Loren McIntyre leaves the room
[10:56:08] Harald Alvestrand leaves the room
[10:56:10] Bron Gondwana leaves the room
[10:56:12] Shuai Zhao leaves the room
[10:56:14] <Justin Uberti> But you'd still want the distribution infra to be able to do rate matching and all that stuff
[10:56:15] Dragana Damjanovic leaves the room
[10:56:15] Alexandre Gouaillard leaves the room
[10:56:17] andrew_liu leaves the room
[10:56:22] Bernard Aboba leaves the room
[10:56:23] Subir Das leaves the room
[10:56:24] Youenn Fablet leaves the room
[10:56:25] Juliana Guerra leaves the room
[10:56:26] Colin Perkins leaves the room
[10:56:28] Saúl Corretgé leaves the room
[10:56:33] Jonathan Lennox leaves the room
[10:56:34] Lorenzo Miniero leaves the room
[10:56:34] Kazuaki Ueda leaves the room
[10:56:38] Mo Zanaty leaves the room
[10:56:39] Sean Turner leaves the room
[10:56:45] <Justin Uberti> anyway, seems like something worth looking into further
[10:56:55] Justin Uberti leaves the room
[10:56:56] Xavier de Foy leaves the room
[10:56:59] cm-msk@jabber.org leaves the room
[10:57:08] Meetecho leaves the room
[10:57:20] James Gruessing leaves the room
[10:57:23] Lucas Pardue leaves the room
[10:57:44] Mike English leaves the room
[10:58:09] Martin Thomson leaves the room
[10:58:50] Bobo Bose-Kolanu leaves the room
[10:58:58] Alessandro Amirante leaves the room
[10:58:58] Jingrong Xie leaves the room
[10:58:58] Paolo Saviano leaves the room
[10:58:58] Stephan Emile leaves the room
[10:58:58] Joseph Kalfa leaves the room
[10:58:58] Sergio Garcia Murillo leaves the room
[12:14:19] slm leaves the room
[12:22:56] slm joins the room
[12:41:16] slm leaves the room
[18:02:21] halfshot joins the room
[18:02:26] halfshot leaves the room
[18:13:53] undefined joins the room