Thursday, 8 September 2011< ^ >
martin.thomson has set the subject to: RTCWEB WG
Room Configuration

[13:42:43] Fluffy joins the room
[13:54:04] <Fluffy> hiu
[13:54:05] <Fluffy> testt
[13:54:10] Magnus joins the room
[13:54:25] Cary Bran joins the room
[13:54:26] Cary Bran leaves the room
[13:54:36] Dan York joins the room
[13:54:54] <Dan York> ping
[13:54:58] <Fluffy> pong
[13:54:59] burn joins the room
[13:55:21] Olle E. Johansson, Sollentuna, Sweden (GMT +1) joins the room
[13:56:42] Dan York leaves the room
[13:57:04] alan.b.johnston joins the room
[13:57:56] Dan York joins the room
[13:59:42] Neil Stratford joins the room
[14:00:31] Jon Lennox joins the room
[14:01:07] <Dan York> Very pretty chairs slide, Cullen. You score bonus points for not simply having a white background but instead giving us orange circles
[14:01:12] TEd joins the room
[14:01:33] TEd leaves the room
[14:01:47] Ted joins the room
[14:04:04] hta joins the room
[14:04:06] <Magnus> I will take credit. I guess it is a down cut version of last meetings version.
[14:04:07] <Fluffy> I only copied Ted's slides from last meeting
[14:04:07] steely_glint joins the room
[14:04:20] <Fluffy> Oh, Magnus slides from last meeting :-)
[14:04:23] <Ted> <>proceedings/82/slides/
[14:05:23] coopdanger joins the room
[14:05:24] <Ted> the tools is not actually up and running fully for the upcoming meeting, so the names are not associated with the slides, you just have grab them all
[14:05:46] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I hear Swedish discussions in the background
[14:06:00] <Ted> ultimately, we'll get the secretariat to create an 81.5 entry in the proceedings
[14:08:03] Cary Bran joins the room
[14:08:03] <Ted> Note the Note well; for those not on webex, it has been displayed. If you have questions on it, please review before making a contribution.
[14:08:17] <Ted> Now reviewing the Agenda
[14:08:24] coopdanger leaves the room
[14:08:50] <hta> is there a slide with the current agenda? I don't see it on the proceedings site.
[14:09:01] kfleming joins the room
[14:09:02] <Ted> 20 minutes on  Recording (John Elwell)
--40 Review and discussion of other use cases proposed on mailing list and in-draft (Use case draft author team)
Signalling (1 hr)
--15 minutes Issue Overview (Matthew Kaufmann)
--15 minutes Offer/Answer architectural text (Cullen Jennings)
--30 minutes discussion
Security (1 hr)
--Discussion of what assurance we wish to provide to which parties (Eric Rescorla)
Terminology Mapping (30 minutes)
--Mapping WebRTC constructs to RTCWeb terms (Magnus)
Congestion Control (30 minutes)
--One congestion control approach (Harald)
[14:09:10] <Ted> The same as last posted, basically.
[14:09:11] <Jon Lennox> Magnus's audio is somewhat difficult to make anyone else having trouble?
[14:09:18] <kfleming> Jon Lennox: yes
[14:09:18] <Ted> There may be adjustment now
[14:09:52] <Ted> Thank you, John, for the slide numbers on the slides!
[14:10:30] EKR joins the room
[14:10:32] kfleming leaves the room
[14:10:55] Kevin P. Fleming joins the room
[14:14:39] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> There are security issues here as well if SRTP is used.
[14:15:03] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Some sort of consent
[14:15:06] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> is needed
[14:15:43] <Fluffy> on slide 6
[14:15:52] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Just noticed :-)
[14:16:19] <Fluffy> Ole: sorry, I just meant we were now display slide 6 not that , uh something else
[14:16:22] <Kevin P. Fleming> someone who is typing needs to mute :-)
[14:16:24] <Ted> If you are typing near a mic, please mute
[14:16:58] Bernard joins the room
[14:17:18] <Fluffy> Mary - you seeing this ?
[14:19:51] <Kevin P. Fleming> do we need to have people say who they are when speaking, like we do in non-virtual meetings?
[14:20:10] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I agree. Thisis "call-in user 2"
[14:21:10] <Fluffy> uh, yah, I think we do - I will interupt
[14:22:35] <Fluffy> Cary is taking minutes in the following googls doc at
[14:22:49] <Fluffy> If anyone wants to help him edit that in real time, much appreciated
[14:22:52] <Ted> Thanks cary!
[14:23:44] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Bad connection?
[14:27:07] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> We're back to the UI issues. If a session is recorded, all participants needs to know...
[14:29:44] <Dan York> So one action -> John E. to provide text for a recording use case into the use cases document.
[14:30:30] <Cary Bran> @Dan York - got it in the minutes - thanks!
[14:30:40] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Is it *one* use case or multiple? A user recording a session is different from call-center where a manager wants to record session for control…
[14:32:28] <Dan York> I think the key point is to make sure that the rtcweb architecture doesn't preclude session recording in the future.
[14:32:47] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> — New presentation without slide numbers
[14:32:48] <Dan York> I do remember seeing some sample text tossed around... maybe 1,000 emails ago or so
[14:33:04] <Jon Lennox> Is this current slide deck on the materials page?
[14:33:10] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Who's talking?
[14:33:19] <hta> Olle, rtcweb-4.pdf
[14:33:20] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Ahh. Stefan Håkansson
[14:33:43] <Jon Lennox> hta: that was me actually, but thanks. :-)
[14:35:03] <Fluffy> If anyone can tell me which user is adding background noise, I can mute them
[14:36:05] <burn> Cullen, doesn't webex have a way to tell how much audio is coming from each line? (a la Zakim)
[14:36:45] <Fluffy> yah, if you go to the roster display list, you can see it is, but with this many people on the call, it's sort of hard to see who is talking
[14:37:04] <Fluffy> When people talk, they get little animated green lines near their telephone icon
[14:41:52] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Are all TURN servers the same? Do we need a specific IPV6-IPv4 turn server setting that's different from the default TURN server?
[14:43:22] <Jon Lennox> That might be generalizable to the concept of a set of TURN servers...there's no reason in principle you can't create candidate addresses on as many TURN servers as you like, other than setup time and resource consumption issues.
[14:43:35] <Jon Lennox> (Is anyone else seeing the shared slides flashing around like crazy?)
[14:43:51] <Fluffy> uh yes slides flashing for me
[14:44:22] <Kevin P. Fleming> yes, me too
[14:44:32] <Kevin P. Fleming> i assumed it was because i'm running on Linux...
[14:44:41] <Jon Lennox> I'm on Windows 7.
[14:44:43] <Dan York> No, it's flashing on the iPad, too
[14:45:25] <Magnus> I think it flashes when the slides goes out and in of focus at my computer that is sharing the slides.
[14:45:34] <Dan York> Er... yes, it's flashing on the iPad, too. (The No was in response to Kevin)
[14:47:39] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Agree.
[14:49:51] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Emergency service requirements are different everywhere. The Swedish policy is "just get us the damn call" and any metadata you can add is cudos to you.
[14:49:59] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Norway is quite the opposite...
[14:49:59] <Fluffy> Great quote from Fracois :-)
[14:51:42] <Dan York> So what is the action here?
[14:52:27] <Ted> Have Paul and Stephan tWenger alk to ECRIT chairs about someone with expertise to discuss this with.
[14:52:27] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Release control will affect the API, or ?
[14:54:02] <Fluffy> we need to be moving on time wise
[14:56:47] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> It's the same use case - just with T.140 and video added.
[14:58:29] <Ted> I think the key issue on Bernard's issue is different from Stephan's—it's actually a relay requirement, because 911 calls will not have fluent ASL signers. Getting relay services to use RTCWEB is a different issue—but a good alternative to proprietary video calling services.
[14:59:02] <Ted> (Note that sign is not a universal language either, British and US are completely different, so you need to talk to a relay service with the right capability there)
[14:59:41] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> In Europe there's a requirement that all 911 centers should support sign language… It's not there yet, but coming.
[15:00:12] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> But there may be some media routing issues here, where the video and audio needs to be mixed between the emergency handler, the caller and the translator.
[15:00:14] <Ted> For just the local sign dialect, or for multiple dialects?
[15:00:39] <Ted> Ah, I misunderstood. They have to be able to loop in a relay service, then?
[15:00:53] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I think the starting point is local sign dialects, that's a starting point
[15:01:16] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> yes, I worked with some strange relays - we have a governmental translator service that was in the loop.
[15:01:36] <Ted> Someone calling in as a signer to the usual 112 center creates a requirement that the 112 center send media to the translater, which then sends the translated media on?
[15:02:09] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Typically you have video between caller and translator, audio between all of them.
[15:02:51] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> If the emergency handler supports video, they will be in that loop too. Which is where we're heading.
[15:02:52] <Ted> I would have expected video and real-time-text, with the real-time-text shared, but it's the same requirement at the baseline.
[15:03:33] <Ted> Time check: we're now at 1 hour in this meeting.
[15:06:41] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Do we agree on some use cases that are definitely OUT of scope for rtcweb?
[15:09:20] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I think we need to move away from the idea that rtcweb is a "phone".
[15:09:40] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> There are requirements coming from using it in combination with phone software, which is only one app for rtcweb...
[15:10:13] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Yes! Move on.
[15:11:21] giles joins the room
[15:12:38] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Related to SPLICES?
[15:18:24] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Is application specific streams part of rtcweb? It can be handled by SDP...
[15:19:14] <Fluffy> Moving on to Matthew signalling slides
[15:19:33] <Fluffy> on slide labeled "Scope" (2nd slide)
[15:24:17] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Very good presentation!
[15:24:37] <Dan York> +1
[15:27:28] <hta> we don't have mandatory in the IETF. All IETF standards are voluntary-to-use.
[15:27:41] <Kevin P. Fleming> and MUST == SHOULD anyway
[15:28:45] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> XMPP is obvious for a lot of developers too. Especially non-PSTN app developers.
[15:29:58] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I think assuming that we always have web servers on the backend is too limiting. I may be wrong here.
[15:30:37] <Kevin P. Fleming> with what is currently proposed, browsers are going to be served up JavaScript from some place, and be sending XMLHTTPRequest stuff back to that place
[15:31:25] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Ahh. Sandbox issues.
[15:31:40] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I think I was wrong then. Thanks Kevin.
[15:31:44] <Kevin P. Fleming> the same sandbox JS developers already live in, yes
[15:32:16] <steely_glint> websockets may blur that distinction slightly.
[15:32:27] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Aha! An invalid SIP uri without scheme… :-)
[15:32:32] <giles> are these excellent-sounding slides available on the web?
[15:33:06] <Jon Lennox>
[15:33:29] <Kevin P. Fleming> 'equally easy'... another good quote
[15:33:39] <EKR> yeah, "It's just typing"
[15:33:52] <Fluffy> "small matter of programming"
[15:34:21] <Kevin P. Fleming> i am seriously going to use equally easy the next time someone asks me about adding a new major piece of work to a project
[15:34:38] <giles> Jon Lennox: thank you
[15:38:40] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> In SIP we don't select A (one) codec. We agree on a set of codecs and are allowed to change. This is an important feature when using gateways to avoid transcoding if possible.
[15:38:45] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> s/SIP/SDP/
[15:39:11] <steely_glint> @kevin As in 'securing an existing SIP library for use in a web browser will be equally easy'?
[15:39:23] <Kevin P. Fleming> steely_glint: exactly!
[15:39:37] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> ;-)
[15:42:06] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Thank you for a very good presentation and overview.
[15:43:32] <Dan York> Agreed... nicely done by Matthew. Succinct summary of a LOT of discussion.
[15:44:21] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> The only thing missing was page numbers on the slides ;-)
[15:46:15] <hta> the fact that I don't agree with him doesn't subtract from my appreciation of the presentation!
[15:47:03] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Are we going for all of SDP or just a subset?
[15:47:23] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> SDP o/a lite or the full monty?
[15:47:56] <hta> It's impossible to implement all of SDP, since it grows a new extension with every 10 new RFCs published.
[15:48:13] <EKR> And sometimes two new extensions
[15:48:17] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Unfortunately I tend to agree.
[15:48:41] <hta> and the baseline SDP is much too limited (no ICE), so we're talking about a subset.
[15:49:04] <Jon Lennox> hta: do you mean superset?
[15:49:25] <hta> Jon, superset of baseline, subset of the universe of SDP with all extensions.
[15:50:29] <Fluffy> Francois is speaking
[15:50:31] <Kevin P. Fleming> i would also be a big fan of expressing the SDP content as JSON inside the browser, instead of text strings. SDP has gotten to the point where deconstructing the group of lines of text into their conceptual 'structured' format has become very difficult to get right.
[15:50:59] <Kevin P. Fleming> taking into account things like SDPCapNeg, i guess
[15:51:21] <hta> Kevin, I tried that, but nobody seemed to like it..... any time we have a transform, maintaining the transform becomes quite hard.
[15:51:21] <Jon Lennox> To what extent do we care about Jingle, in areas where its semantics differ from SDP?
[15:51:44] <Fluffy> don't mention that elephant - it is under the carpet righ now
[15:51:45] <steely_glint> I liked it :-)
[15:51:50] <hta> Jon, to the extent of filing bug reports on jingle? I'm not sure there are any *intended* differences.
[15:52:15] <Jon Lennox> Jingle has partial updates in places where SDP can only be atomic, and I believe that's intended as a feature
[15:53:23] <Dan York> @Kevin - Yes, I have a related question - do we want web developers to actually USE the RTCWEB APIs? If so, there are X-million web developers who are living in JSON... and how many are out there who actually work with raw SDP?
[15:53:33] <Kevin P. Fleming> right
[15:54:00] <Kevin P. Fleming> teaching a JS developer to walk through an SDP in its text form, applying the global lines to each media session, and all that, seems very wasteful
[15:54:09] <Dan York> agreed
[15:54:17] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Of the smaller amount of developers that actually work with SDP - how many just love it and say "it's darn simple" ?
[15:54:23] <Dan York> :-)
[15:54:28] <Neil Stratford> hah
[15:54:31] <Jon Lennox> It's damn simple until you try to do anything complicated.
[15:54:35] <Dan York> How many fingers do you have, Olle?
[15:54:39] <Kevin P. Fleming> JSON is not XML, but at least it's structured. SDP is essentially unstructured, as SDPCapNeg has shown quite clearly.
[15:54:40] <Jon Lennox> At which point it's a bag of name/value pairs
[15:54:58] <EKR> so when you say JSON you mean: {"sdp":"v=0
o=alice 2890844526 2890844526 IN IP4
c=IN IP4
t=0 0
m=audio 49170 RTP/AVP 0 8 97
a=rtpmap:0 PCMU/8000
a=rtpmap:8 PCMA/8000
a=rtpmap:97 iLBC/8000
m=video 51372 RTP/AVP 31 32
a=rtpmap:31 H261/90000
a=rtpmap:32 MPV/90000
[15:55:10] <Kevin P. Fleming> no
[15:55:14] <Dan York> NOOOOOO
[15:55:19] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> ha ha
[15:55:54] <steely_glint> {"codecs":[{"ptype":0,"rate":8000,"name":"PCMU"},{"ptype":9,"rate":8000,"name":"g722"},{"ptype":101,"rate":8000,"name":"telephone-event"}]}
[15:56:01] giles leaves the room
[15:56:03] <EKR> Mine seemed so much easier :)
[15:56:07] <Kevin P. Fleming> steely_glint: yes
[15:56:15] <Jon Lennox> EKR: Now do a CapNeg one.
[15:56:20] <Kevin P. Fleming> wrapped in a 'media' session
[15:56:31] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> s/telephone-event//g
[15:56:45] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Do we really need DTMF? :-)
[15:56:59] <Kevin P. Fleming> Jon Lennox: CapNeg + ICE... dreamy
[15:57:10] <Jon Lennox> Olle: depends who you're gatewaying too.
[15:57:12] <Jon Lennox> to
[15:57:41] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> You could keep it outside of the media and send it somewhere else though...
[15:58:00] giles joins the room
[15:58:05] <hta> the line saying ...... and a bunch more along these lines ..... is the scary one.
[15:58:44] <Kevin P. Fleming> but in general, SDP is not the *cause* of these issues, they exist regardless of the offer/answer mechanism you might want to use
[16:00:41] <steely_glint> btw the json I showed came from the <> android client. -a web rtc client in use today
[16:01:55] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Aftet two hours exactly my SNOM stopped all audio. Interesting timer somewhere
[16:02:53] <EKR> wow, that was terrible
[16:05:01] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Who is call-in user_20? Harald?
[16:05:09] <burn> that was harald, yes
[16:05:32] <hta> I was kicked off the call (google voice times out after 2 hours) and had to dial back in.
[16:07:13] <Kevin P. Fleming> we've got typing on the call again...
[16:07:23] <Jon Lennox> Call-in User_21, whoever that is
[16:07:38] <burn> will we be having a formal break, or just keep going?
[16:07:51] <burn> (not that we are running ahead of schedule)
[16:08:02] <hta> I certainly hope for a break(age). 2 hours is my sit-still limit.
[16:08:05] <Magnus> I think we need a 5 minute bio break.
[16:08:06] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Fluffy - please mute call-in user_21
[16:08:58] <Jon Lennox> He's probably not watching the jabber room while presenting
[16:09:44] <giles> I just reconnected. I don't know if I'm 20 or 21
[16:09:57] <giles> but my phone does say it's muted, so I claim no to be responsible for the typing :)
[16:13:47] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Yes!
[16:13:52] <Kevin P. Fleming> 5 minute bio break, for those not on the call
[16:14:00] <EKR> I think we can also cut security a hair short :)
[16:14:01] <Ted>
[16:14:15] <burn> resume officially at 20 minutes after the hour
[16:14:40] tterribe joins the room
[16:18:40] joins the room
[16:23:06] spencerdawkins joins the room
[16:23:18] spencerdawkins leaves the room
[16:23:22] spencerdawkins joins the room
[16:26:14] <EKR> does this room still exist?
[16:26:21] <steely_glint> yep
[16:26:22] <Ted> Yes
[16:28:18] <steely_glint> So you could offer the javascript the chance to deal with it, and if it doesn't then the browser does something brutal.
[16:28:56] <Jon Lennox> steely: or even, does something that's sensible for the simple cases.
[16:29:46] <Jon Lennox> This may indeed be brutal for the complex cases
[16:31:24] <steely_glint> There are 2 layers to <> - a simple js API and an SPI underneath it.
[16:31:35] <EKR> thanks!
[16:31:53] <Kevin P. Fleming> it's really out of scope for this meeting (and WG), but i have serious concerns with the amount of stuff that is proposed to be embedded in browsers.
[16:32:22] <EKR> The browser is the new operating system, haven't you heard?
[16:32:38] <Dan York> Agreed, Kevin
[16:32:38] <Kevin P. Fleming> under the model that is being talked about, building an alternate, novel implementation of any part of RTCWeb/WebRTC is certainly possible, but deploying it to users requires convincing them to adopt your modified browser.
[16:32:52] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> "I don't care, I want to make progress" - Fluffy
[16:33:12] <Ted> Can we get someone to repeat what was just said?
[16:33:21] <EKR> the sound quality from whoever is calling now, sounds terrible. Is he using RTCWeb?
[16:33:53] <steely_glint> "Why can't SIP 2.0 be implemented in a browser - just as a framework"
[16:34:30] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> EKR: Comment from the future :-)
[16:35:01] <EKR> Well, I can hear myself just fine
[16:35:03] <EKR> :)
[16:35:43] <Kevin P. Fleming> EKR: but we all need time stretching to understand you :-)
[16:36:30] <hta> so is there a difference between -6 and -7?
[16:37:03] <Ted> He did update the slides quite recently, so grab again
[16:37:21] <Cary Bran> @Ted are you going to schedule the consensus call?
[16:37:32] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Or enjoy the blinking stuff in webex
[16:39:02] <Ted> @Cary The consensus calls will likely start next Wednesday. Some of them may be earlier, but the ones with more drafting of language will likely be then.
[16:39:27] <hta> -6 is missing the last slide (opinions), so use -7.
[16:39:28] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> What's the baudrate here?
[16:39:42] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> THanks!
[16:39:57] <Cary Bran> @Ted thanks
[16:40:38] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I like where this presentation is going.
[16:40:59] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> It will be hard to separate it into w3c and ietf issues though...
[16:44:23] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> …and you trust SocialWeb
[16:44:36] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> in a different way that you would trust beer-drinkers-web (maybe)
[16:45:17] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> This is an ugly example. Very good.
[16:45:42] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> But click-to-call is one of the key apps here.
[16:46:07] <giles> CmdrTaco doesn't work there any more!
[16:47:50] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> …and can we bootstrap p2p confidential calls with peers with long-term identities based on the trust of the identity we already have….
[16:59:40] <Ted> Wow, the webex blink rate is getting pretty high. I'm starting to wonder if it is a decodable side channel.
[17:02:25] <Jon Lennox> I think the thing you can decode is what Magnus is doing on his PC.
[17:03:38] <Magnus> sorry, I will stop doing other stuff
[17:04:26] <Kevin P. Fleming> wow... big difference
[17:05:00] <burn> still blinking for me, though
[17:05:24] <burn> so apparently Magnus' boredom with the call is not the only driver of the blink rate
[17:06:27] <Magnus> It might be each screen update that cause it. Thus posting to the chat room or webexs own participant interface may cause updates.
[17:13:37] <Kevin P. Fleming> based on the recent PKI debacles, probably not well at all
[17:14:02] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> We need to find out what parts of this presentation affect the media API and session management
[17:14:21] <EKR> Yeah, that's honestly what I'm thrashing around a bit about too :)
[17:14:25] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> The rest is propably for W3C to figure out and parts should be BCP for app developers
[17:14:32] <Kevin P. Fleming> yes
[17:15:30] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> WIll javascript need to have access to any SRTP keys?
[17:15:51] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Will the media layer need access to HTTPS certs and trust assurances?
[17:16:25] <Kevin P. Fleming> i believe the current proposal is for media transport over DTLS-SRTP, so to some extent keys will be negotiated in the media plane, and the JS won't need to care about them
[17:16:36] <Kevin P. Fleming> but there would still be certs and trust involved
[17:17:04] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Define certs and trust
[17:17:06] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> ;-)
[17:17:14] <Kevin P. Fleming> well, yeah
[17:17:22] <Kevin P. Fleming> but DTLS uses what we have
[17:18:13] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> SocialWeb (in EKRs presentation) may want to provide keys for you to use in a pkcs package. Is that acceptable?
[17:19:26] alan.b.johnston leaves the room
[17:19:36] <EKR> In general, it's better if clients generate them themselves, for the usual reasons that it's bad to share your private keying material
[17:19:52] <EKR> why would you want to do that?
[17:20:07] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> I want to prevent it :-)
[17:20:27] <tterribe> To verify an identity that's shared across
multiple machines.
[17:20:27] <EKR> Agree. I think that the keys should be self-generated and that you should use a DH exchange for each call for PFS
[17:20:30] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> But it shows how hard it is to say "we just need a pki and some certs"
[17:20:41] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> That sounds reasonable
[17:21:03] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Fluffy: My mike was muted all the time.
[17:21:48] <Jon Lennox> Olle: no, there's someone who identified themselves as "Mike"
[17:21:54] rillian joins the room
[17:22:09] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> THat wasn't my mike… ;-)
[17:22:44] csp joins the room
[17:24:32] <Jon Lennox> Magnus -- your audio is hard to make out. Would you be able to switch your mic (go off speakerphone or whatever)?
[17:24:47] <EKR> Indeed.
[17:26:23] <Olle E. Johansson, Sollentuna, Sweden (GMT +1)> Need to exit. Thanks for the discussin and the good presentations!
[17:28:13] Olle E. Johansson, Sollentuna, Sweden (GMT +1) leaves the room
[17:40:42] <Kevin P. Fleming> this could even be the same video stream, but being sent using two different encoding profiles
[17:43:40] <EKR> I must be missing something Is the concern that these things aren't demuxed at all or just about the mapping into the JS API?
[17:46:15] <steely_glint> I love that congestion control is being pushed out by too much other stuff infront of it. Ironic no ?
[17:46:32] <giles> ha
[17:46:36] <Ted> yep
[17:46:44] <Dan York> :-)
[17:46:52] <Ted> But we'll get the flow started, so the retransmits can begin.
[17:47:04] <Fluffy> This presentation is get dropped by the chair router when the congestion Q gets to a depth of 11:50
[17:47:17] <Kevin P. Fleming> Ted implements ECN, we can already see that
[17:47:25] <Dan York> IETF meetings in general could use "congestion control" :-)
[17:47:42] <EKR> I thought they implmented Random Late Drop
[17:47:44] <Dan York> Fluffy: good one!
[17:48:16] <Fluffy> as usual our buffers are too deep
[17:48:29] <Ted> Alert Gettys!
[17:51:14] <Kevin P. Fleming> how is this problem unique to rtcweb?
[17:51:54] <csp> you might want to do something different for video congestion control than for bulk transfer, to avoid destroying media usability
[17:52:21] <Fluffy> I think most the audio / video congestion control applies to any RTP based audio / video
[17:52:30] <Kevin P. Fleming> Fluffy: right
[17:52:59] <csp> sure, but the answer for RTP audio/video may be different than the answer for other media
[17:53:05] <Jon Lennox> RTCWeb is the point at which you have to stop handwaving the problem away or hoping administrators/applications configure things properly, due to potentially hostile javascript.
[17:53:36] <Kevin P. Fleming> true enough
[17:53:46] <Kevin P. Fleming> since you can't push JavaScript into my phone :-)
[17:55:38] <Magnus> Got dropped out
[17:57:42] <Kevin P. Fleming> MSRP?
[17:58:10] <EKR> you can push JS to my phone
[17:58:19] <Fluffy> That's like the thing that came before XMPP that no one uses , rigth ? :-)
[17:58:29] <Kevin P. Fleming> Fluffy: yes, except for a few people
[17:58:31] <Dan York> Am I hearing a cat on someone's connection?
[17:58:36] <Kevin P. Fleming> Dan York: yes
[17:58:57] Dan York wonders if that is the timekeeper signaling the end
[17:58:59] <EKR> Wow, the name IFRAME is really unfortunate here....
[17:59:01] <Jon Lennox> Looks like allyn?
[17:59:11] <Kevin P. Fleming> EKR: lol
[17:59:19] <Kevin P. Fleming> Fluffy: can you include the cat in the list of attendees?
[17:59:23] <EKR> I tuned out for a second and then I was like "What's wrong with iframes"
[17:59:28] <Fluffy> allyn is muted
[18:00:02] <tterribe> EKR: I always thought I-frame was unfortunate nomenclature, even just in video. "Does it mean intra or inter?"
[18:00:41] <giles> tterribe: we should start calling them 'a' and 'e' frames
[18:01:14] csp leaves the room
[18:01:15] <hta> xxx-frames?
[18:01:37] <Fluffy> XXX-frame, R-Frames, and PG-Frames
[18:02:16] <Kevin P. Fleming> good meeting, all
[18:02:26] Ted leaves the room
[18:03:02] burn leaves the room
[18:03:31] steely_glint leaves the room
[18:03:39] Jon Lennox leaves the room
[18:05:03] Bernard leaves the room
[18:06:07] hta leaves the room
[18:06:11] Fluffy leaves the room
[18:09:56] Kevin P. Fleming leaves the room
[18:10:03] giles leaves the room
[18:39:10] EKR leaves the room
[18:41:17] Olle E. Johansson joins the room
[18:48:19] Olle E. Johansson leaves the room
[19:03:33] rillian leaves the room
[19:07:01] Olle E. Johansson joins the room
[19:15:34] Cary Bran leaves the room
[19:35:47] Olle E. Johansson leaves the room
[20:29:50] EKR joins the room
[21:17:07] leaves the room: Replaced by new connection.
[22:06:23] tterribe leaves the room
[22:28:15] tterribe joins the room
[22:28:23] tterribe leaves the room
[22:29:03] tterribe joins the room
[22:29:23] tterribe leaves the room
[22:30:24] tterribe joins the room
[22:30:53] tterribe leaves the room
[22:32:06] tterribe joins the room
[23:33:12] EKR leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!