--- Log opened Wed Apr 07 19:59:32 2004 19:59 -!- xiphmeet_log [~giles@motherfish-II.xiph.org] has joined #xiphmeet 19:59 -!- Topic for #xiphmeet: Xiph.org April MonthlyMeeting. Agenda at http://wiki.xiph.org/MonthlyMeeting200404. Live log at http://xiph.org/~giles/xiphmeet.log for those of us starting late. 19:59 -!- Topic set by rillian [] [Wed Apr 7 19:55:53 2004] 19:59 [Users #xiphmeet] 19:59 [ adiabatic] [ dev0 ] [ Misirlou ] [ purple_haese] [ xiphmeet_log] 19:59 [ DanielWH ] [ j^ ] [ Odin- ] [ rillian ] [ xiphmont ] 19:59 [ derf_ ] [ kjoonlee] [ OggZealot] [ volsung_ ] 19:59 -!- Irssi: #xiphmeet: Total of 14 nicks [0 ops, 0 halfops, 0 voices, 14 normal] 19:59 -!- Channel #xiphmeet created Wed Mar 24 00:25:38 2004 19:59 -!- Irssi: Join to #xiphmeet was synced in 1 secs 20:00 -!- xiph [~xiphmont@HOTASS-7.MIT.EDU] has joined #xiphmeet 20:00 -!- xiphmont [~xiphmont@HOTASS-7.MIT.EDU] has quit [Remote closed the connection] 20:00 < xiph> I *hate* xchat. 20:00 < xiph> ntpd updates my clock and xchat says "OMG! No pings for whole minutes!" 20:00 -!- xiph is now known as xiphmont 20:01 < rillian> you're not supposed to use ntpdate without the slew option. lots of things have that problem 20:01 < rillian> Shall we start? 20:01 < xiphmont> that would work if the sleep clock on the G4 wasn't broken :-) 20:01 < rillian> First item is SXSW report 20:01 < rillian> how'd that go? 20:02 < xiphmont> Is vanguardist expected? 20:02 < volsung_> He's been idle for 5 hours. 20:02 < volsung_> I'll sub for him. 20:02 * volsung_ assumes xiph will go first. :) 20:03 < rillian> volsung_, maybe you should go first 20:03 < volsung_> Okey. SXSW was very successful, I think. 20:03 -!- xiph [~xiphmont@HOTASS-8.MIT.EDU] has joined #xiphmeet 20:03 < volsung_> Monty went out and worked the other booths much of the time. 20:03 -!- J_Bullet [~jbullet@213.106.98.148] has joined #xiphmeet 20:03 < volsung_> I'll let him explain who he talked to. 20:04 < xiph> All the iMacs set up as listening stations were supposedly Ogg enabled through iTunes and whatever plugin; I never once got any one of them to work (iTunes plugin needs work) 20:04 < xiph> Dammit 20:04 < xiph> OK, what did folks see and what did folks not see? IRC is sucking donkeys through a straw tonight. 20:04 < volsung_> But for the most part, there were a lot of interested indie labels. 20:04 < rillian> nothing you posted showed up 20:04 < xiph> Dammit 20:04 < xiph> [20:02:19] <xiphmont> Is vanguardist expected? 20:04 < xiph> [20:02:33] <xiphmont> [he and Stan were the most fired up by it it seems :-] 20:04 < xiph> [20:03:05] <xiphmont> SXSW went more or less as expected. I talked to most of the booths on the floor. Doppler and his crew pulled off the streaming. 20:04 < xiph> [20:03:24] <xiphmont> Two things stuck in my mind though: 20:04 < xiph> [20:04:02] <xiph> All the iMacs set up as listening stations were supposedly Ogg enabled through iTunes and whatever plugin; I never once got any one of them to work (iTunes plugin needs work) 20:05 < volsung_> Oh yeah, Creative Commons and Magnatune were very awesome. 20:05 < rillian> volsung_, so our target audience actually showed interest? that's great. 20:05 < xiph> Folks were generally very happy to see us there. 20:05 < volsung_> We need to make sure to keep in contact with those two orgs especially. 20:05 < rillian> we've been wondering where the labels were 20:06 < xiph> Magnatune, CC and w3k were very happy to meet us and talk :-) 20:06 < volsung_> Most of the labels we saw had never heard of us, so I think we're at the beginning of infecting that mindshare. 20:06 < purple_haese> w3k? 20:06 < xiph> Well, there were just as many indies there trying desperately to figure out who they had to sleep with to get the bigger labels to notice them. 20:07 -!- J_Bullet [~jbullet@213.106.98.148] has left #xiphmeet [] 20:07 < xiph> Yes, awareness of who we are was very low there. 20:07 < volsung_> Oh yeah, and big props go out to doppler for scoring us the booth and passes. 20:07 -!- EvilOverlord [~jbullet@213.106.98.148] has joined #xiphmeet 20:07 < volsung_> Anyway, the link on the agenda points to the stuff Manuel and I thought of after the meeting was over. 20:08 < xiph> BTW, stan, stickers and stuff made it back to me OK 20:08 < rillian> xiph: what was the second thing that struck you? 20:08 < volsung_> Mostly the strong need for Xiph.org to have an obvious "orientation" area for interested people. 20:09 < xiph> rillian: that the WiMP plugin was similarly crippled when I tried it :-) 20:09 < volsung_> Manuel and I have been sketching a mockup of such pages that we'll check-in once we clean it up. 20:10 < rillian> I like the things you came up with as well 20:10 < rillian> volsung_, you said you're also working on a demo cd? 20:10 < volsung_> yeah, Manuel and I figured we'd want something like that again in the future. 20:10 < volsung_> So we hooked up with CC and Magnatune to come up with some interesting content. 20:11 < volsung_> John Buckmann (magnatune) was all about us putting ogg versions of the demo CDs he was handing out onto a Xiph.org CD. 20:11 < rillian> found any video we could include? 20:11 < volsung_> Actually yes. 20:11 < volsung_> While I was at a film panel, I saw a guy was releasing his 90 min film under a CC license. 20:12 < volsung_> See http://fourthwall.creativecommons.org 20:12 < volsung_> It's not publically up yet since he hasn't figured out how to do that. 20:12 < volsung_> But we emailed him and he uploaded the 19 GB DV file onto my computer. 20:12 < volsung_> So I now have it, and I'm trying to figure out what to do with it. :) 20:12 < rillian> oh, you have it already! excellent 20:12 < volsung_> Yeah, he was very responsive. 20:13 < rillian> I'd appreciate it if you could push a copy to motherfish so I can play with it 20:13 < volsung_> 'k. 20:13 < volsung_> Do we have the disk space? 20:13 < rillian> yes 20:13 < volsung_> Alright, my asymmetric cable modem cap means it will finish sometime by the next meeting. :) 20:14 < rillian> you could also dump it to a couple of tapes and mail me a copy :) 20:14 < volsung_> heh. Upload will be easier. :) 20:14 < volsung_> I have no DV hardware. 20:14 < rillian> Anyway, sounds good 20:14 < EvilOverlord> if I may interject, when you say he hasn't figured it out, do you mean what format to put the video in? 20:14 < purple_haese> Snail-Mail a harddisk with the file ;) 20:14 < volsung_> EvilOverlord: Yeah, I got the impression he was still evaluating how to put this much data online. 20:15 < EvilOverlord> volsung_: isn't Xvid+ogg the obvious choice? 20:15 < volsung_> Especially since one of his goals is to put the raw footage up to so people can make derived works. 20:15 < volsung_> EvilOverlord: Not necessarily. 20:15 -!- MikeS [~msmith@CPE-144-137-40-253.vic.bigpond.net.au] has joined #xiphmeet 20:15 < volsung_> EvilOverlord: Certainly wasn't the first I thought of. :) 20:15 < EvilOverlord> volsung_: oh okay 20:15 < rillian> Next item is the website revision 20:15 < MikeS> Sorry I'm late; forgot about the effects of the end of daylight saving. 20:16 < rillian> we've already touched on the stan and manuel's ideas 20:16 < rillian> Nathan has also requested a separate meeting about this since he couldn't make this one 20:16 < rillian> xiph: you were also working on a rewrite. anything new there? 20:16 < purple_haese> Speaking of Demo content: Mike Oldfield is using Ogg for his Music Virtual Reality "games." Maybe he'd be willing to release a bit of material from those games for a Xiph demo. 20:17 < xiph> rillian: no, nothing new. Sorry. 20:17 < purple_haese> I hereby volunteer trying to ask him about it, although it's probably going to be like calling Elvis :) 20:17 < xiph> Rather... 20:17 < xiph> several people said "But we have new stuff to put up / have ideas, don't stomp on them" and I became unfocused. 20:18 < volsung_> rillian: Are we officially going to be putting website development into svn? 20:18 < rillian> yes 20:18 < rillian> nathan's point about passwords in php source might need to be addressed for some sites 20:19 < rillian> but I think that's what we should do for xiph.org 20:19 -!- xiphmont [~xiphmont@HOTASS-7.MIT.EDU] has quit [Read error: 113 (No route to host)] 20:19 < volsung_> Manuel and I are focusing on content for the website, so if someone comes up with a non-ugly template (with all the CSS evil), we'll be more than happy to plug into it, with tweaks. 20:19 -!- xiph is now known as xiphmont 20:19 < rillian> that sounds like a good division 20:20 < rillian> though of course monty should work with you on text if he wants :) 20:20 < volsung_> I know Nathan is the resident CSS Nazi. :) 20:20 < rillian> I look forward to seeing test.xiph.org 20:20 < volsung_> Ah, also on website, we need to replace all CVS docs with SVN docs. 20:20 < xiphmont> Well, I want to actually do as little as possible personally on that front ;-) 20:20 < xiphmont> yes 20:21 < volsung_> There are a few people still checking cvs thinking they are getting latest code. 20:21 < rillian> it's not on the agenda, but now might be a good time to talk about repository layout 20:21 < rillian> since that's sort of what we were waiting for with the website updates 20:21 < purple_haese> Sorry to rewind the discussion a bit: Are there any comments about my Mike Oldfield suggestion, like Go For It or Don't Bother? 20:21 < rillian> I originally thought per-project directories would be simplest 20:22 < volsung_> purple_haese: I'm not familiar with what he has. If you do the footwork, we'll make the space. 20:22 < rillian> but now I think that only makes sense for icecast 20:22 < volsung_> rillian: per project means? 20:22 < rillian> purple_haese, certainly it would be great to have stuff from him. 20:22 < purple_haese> volsung_: He's probably most famous for Tubular Bells. I'll see how approachable he is. 20:22 < MikeS> Yes, SVN docs are important. Even our developers don't know how to set things up properly. 20:22 < rillian> /theora/trunk instead of /trunk/theora 20:23 < volsung_> Which is per project? :) 20:23 < rillian> the first, like theora is now 20:23 < volsung_> Ah. 'k. 20:23 < rillian> If no one has an opinion, I'll move theora back under trunk with all the other codecs 20:23 < rillian> and we'll leave it at that 20:23 < volsung_> Yeah. 20:24 < volsung_> I think that is the easiest to deal with. 20:24 < rillian> ok, good. 20:24 < volsung_> Especially given the huge number of projects we have. 20:24 < rillian> Next item is release schedule 20:24 < rillian> theora a3 went out 20:24 < rillian> as did a new libao 20:25 < volsung_> And positron. 20:25 < rillian> *nod* 20:25 < rillian> we have libogg 1.1.1 gated by a bug monty needs to fix 20:25 < volsung_> (With another on the way by the end of the month. It seems I disagree with DI's database usage again.) 20:25 < rillian> and libvorbis 1.0.2 ready to go 20:25 < rillian> vorbis-tools 1.0.2 needs some build system work, iirc 20:26 < xiphmont> "The neverending build system development" :-) 20:26 < MikeS> Yes (1.0.1 vorbis-tools was a significant backwards step from 1.0 in terms of buildability on most systems) 20:26 < rillian> I think I might be the only "release team" member here, so I'm happy to leave it at that 20:26 < rillian> (or we can discuss build issues) 20:26 < volsung_> libogg and libvorbis both have the line-ending fixes for the Win32 project files now, right? 20:26 < rillian> yes 20:27 < rillian> we should probably apply the 'native' line ending flag to the source files, although that's not critical like it is for the MSVC project files 20:28 < rillian> Ok. next item 20:28 < xiphmont> well, we did in CVS... 20:28 < volsung_> Oh yeah, who's going to write the SVN page? 20:28 < rillian> the first person to get to it? 20:29 * rillian will if there are no volunteers 20:29 < xiphmont> I'm still being lame about mailsystem.... I could add that to lthe list. 20:29 < xiphmont> Despite not having used SVN yet :-) 20:29 < xiphmont> "learning experience" 20:29 < volsung_> Oh yeah, are we all moved to mailman now? 20:29 < rillian> xiphmont: I'd rather you fix the libogg bug and write OggFile 20:30 < rillian> volsung_, right. 20:30 < rillian> we're moving to mailman 20:30 < rillian> as the better known of the evils 20:30 < volsung_> How much does that impact the mailing list pages? 20:30 < rillian> announce and commits lists have already moved 20:30 < rillian> the rest will move when I get a change to integration the archive pages and spam filter 20:30 < rillian> s/change/chance/ 20:31 < volsung_> Are you still hand-approving commits posts? 20:31 < rillian> yes 20:31 < volsung_> Any solution in sight? Or just not worth it now? 20:31 < rillian> it's easy enough to fix 20:32 < rillian> though getting upstream to unbreak it might take longer 20:32 < rillian> It's just not been worth it so far 20:32 < rillian> The next item really needs some discussion, 20:33 < rillian> which is what to do about granulepos 20:33 < xiphmont> Ah, right, I did not mail out final proposal. 20:33 < xiphmont> There is a final proposal: 20:33 < xiphmont> a) yes, we're (generally) moving to start time because it results in more efficient buffering. 20:33 < xiphmont> b) we're using a reserved flag (currently mandated to be zero) to flag a page sorted by start time. 20:34 < rillian> (for those who haven't been following this, the ogg spec says page timestamps are the last time decodable with a page. for various reasons, it's been proposed that we change that to first time) 20:34 < xiphmont> c) Codecs that make use of a start-time-sorted page and also use end-of-stream short packets must be able to determine when packets on an ending page are short. 20:35 < DanielWH> every page gets flagged? 20:35 < xiphmont> As for start/end? Yes, it's the most generic, strightforward way to declare it. 20:35 < xiphmont> *because* 20:36 < rillian> xiphmont, do you mean flag a page that records start time, instead of sorts by? 20:36 < xiphmont> rillian: they;'re the same thing 20:36 < xiphmont> and yes, that needs to be stated clearly. 20:36 < rillian> absolutely not 20:37 < rillian> the page spec shouldn't assume the multiplexed stream mapping 20:37 < xiphmont> there are two cases: records/sorts by start time and records/sorts by end time. 20:37 < xiphmont> what do you mean? 20:38 < rillian> I mean it's possible to screw up the sort order without screwing up the granulepos fields 20:38 < rillian> those are different layers of the spec 20:38 < xiphmont> no, they're both ogg-level. 20:38 < xiphmont> Ogg itself is thin but monolithic, layer-wise 20:38 < DanielWH> 1x3x9? 20:39 < DanielWH> sorry. 20:39 < rillian> I really disagree there 20:39 < rillian> interleave is a mapping issue, not an ogg issue 20:40 < DanielWH> what's the demultiplexer supposed to do? when it gets a stream with rillianora in it will it have to demultiplex differently 20:40 < rillian> but then, you've still not convinced me that forcing a sort order is necessary 20:40 < xiphmont> we either have implicit or explicit sort order, yes? 20:41 < xiphmont> With implicit, it;s locked and you can;t mix. acolwell had some interesting possible uses for mixing sort order within a single stream... so long as its declared, it doesn;t bother seeking any. 20:41 < rillian> I'm just saying the flag should have nothing to do with the sort order 20:41 < rillian> through of course the sort order can depend on the flag 20:42 < xiphmont> If you don;t know if a granpos is beginning or ending, you can't seek accurately. 20:42 < rillian> xiphmont, I don't follow your implicit vs. explicit terms 20:43 < xiphmont> You either have to explicitly flag how pages are sorted, or you have to know some other way through inferred knowledge not in the stream itself (implicit) 20:44 -!- j^ [~j@pD9E36759.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)] 20:44 < rillian> or, the sorting can be implicit while the starttime/endtime status of the granulepos is explicit 20:45 < rillian> anyway, the proposal for the flag sounds like a good idea. we can argue about the definition later 20:45 < xiphmont> we've already established that if granpos == beginning of page, you have to sort it by that. 20:45 < xiphmont> if granpos is not monotonically increasing, then the search alg no longer works. 20:45 < rillian> no one's been able to convince me of that 20:46 < rillian> so we'll use start time for all our new codecs? 20:46 < xiphmont> How do you bisection search a partially unsorted list? 20:46 < rillian> what about vorbis in theora streams? 20:46 < xiphmont> rillian: we'll encourage it. 20:46 < MikeS> "Everything works either way, but please do this, it's more efficient". 20:46 < xiphmont> vorbis I has to be by end time becasue Vorbis I defined it that way. 20:47 < rillian> Vorbis I is defined to only work in degenerate streams 20:47 < xiphmont> Current thinking is that in Theora it still has to be by end time because demuxing will otherwise produce invalid Vorbis I streams 20:47 < rillian> if we have a flag we could change that 20:47 < rillian> ok, that I'll buy 20:47 < xiphmont> My worry is that the default state of the current tools would be broken... but I'm not actually feeling adamant about it like I was in earlier discussions. 20:48 < xiphmont> pages are ordered by granpos regardless of whether it's start or end time. 20:48 -!- Parn [~lodoss@148.Red-217-126-33.pooles.rima-tde.net] has joined #xiphmeet 20:49 < xiphmont> mixing start/end sorted pages in a stream, at worst, just results in buffering that requires more space thna is theoretically optimal. 20:49 < xiphmont> nothing breaks. 20:50 < rillian> my point has always been that if the ordering isn't perfect, you lose some seeking resolution, and again buffering is less optimal, but there's still no reason it should break 20:50 < rillian> thus it's also in the category mike mentioned 20:51 < xiphmont> No, you really do fuck the bisection search. 20:51 < rillian> no 20:52 < xiphmont> The simple three-line version no longer works. 20:52 < DanielWH> they're always sorted by time when you're multiplexing, right? 20:52 < xiphmont> You need something that tolerates the inherent fencepost errors, etc... 20:53 < xiphmont> I do not see any value to suddenly unsorting the multiplexing, or using some implicit, fuzzy sorting information. 20:53 < rillian> you still need to find the start of all the streams for the bit you want to play 20:53 < xiphmont> rillian: no, you don;t./ 20:53 -!- volsung_ [~volsung@sirius.hep.utexas.edu] has quit ["Time to leave."] 20:53 < xiphmont> you only find a requested point in time. 20:54 < xiphmont> (simplest case) 20:54 < xiphmont> ...and the more complex cases are built out of 'seek to point in time'. 20:54 < rillian> so for theora, 20:55 < rillian> theora pages have start times 20:55 < rillian> vorbis pages have end times 20:55 < xiphmont> When my point in time is scattered throughout a small section of file.... 20:55 < xiphmont> yes. 20:55 < adiabatic> Incidentally, I'm here. 20:55 < rillian> you sort by time equivalent of the granulepos, regardless of whether it's start or end? 20:55 < rillian> hi adi 20:55 < xiphmont> yes. 20:56 < rillian> so your 3 line bisection search find the page with the greatest granulepos before the time you request 20:57 < xiphmont> Most efficient case is if everything is start time. Second most efficient is everything end time. Mixing is least efficient but works fine. 20:57 < rillian> you scan forward to find a video page 20:57 < rillian> you see calculate where the last keyframe was 20:57 < rillian> you seek *there* 20:57 < xiphmont> if we bother with keyframe seekage, yes :-) 20:57 < xiphmont> (*we* will; most other app developers probably will not) 20:58 < rillian> and you start decoding, until you have what you need to start playback at the requested seek point 20:58 < xiphmont> yes 20:59 < xiphmont> btw, as a clarification of: "find the page with the greatest granulepos before the time you request" 20:59 < rillian> anyway, in general you need to back up anyway 21:00 < xiphmont> if it's start-sorted, you begin with that page. If it's end-sorted, you begin right after it. And that's the only modification to the seeking we have now. 21:00 < rillian> if you're doing it wrong, yeah 21:00 < xiphmont> Backing up is not what I'm arguing against. I'm arguiong that "seek to point in time" must be well defined and uniquely defined. 21:01 < DanielWH> so I'm seeking to frame 500. associated with frame 500 are captions, beginning on frame 450, sound, beginning between 459 and 500, and simulator yaw beginning at frame 486. 21:01 < rillian> what I don't like is that if you do it right, you have to handle the broken streams anyway. you're specing something that's only helpful if you don't do things right 21:01 < xiphmont> If the stream is broken it works and you get the wrong answer. 21:01 < xiphmont> If the stream is not broken, it works and you get the right answer. 21:01 < xiphmont> rather, 21:01 < xiphmont> If the stream is broken it works and you get the wrong answer or an error. 21:02 < DanielWH> so I suppose I want to seek to the earliest of all the codecs in the stream before video frame 500. 21:02 < xiphmont> DanielWH: there are no bare frames. Frames are in pages, and that's a different layer that we're discussing now. 21:03 < xiphmont> Seeking is an operation ont he stream. If you want an operation of bits within the stream, that's harder and it should be. 21:04 < DanielWH> okay so you just want to make sure I can find theora's page with granulepos = x among a theora+vorbis stream, and if I can (in 3 lines) then all is well. 21:04 < xiphmont> DanielWH: granpos x == time y, seek to time y, done. 21:04 < rillian> xiphmont, somehow, when you move the line instead of the thing it's on the wrong side of it, I can agree with you :) 21:04 < xiphmont> did I braino an earlier statement? 21:05 < xiphmont> [looks back] 21:05 < DanielWH> so then after ogg's taken me to theora page 500, my application has to do something at a higher level to find the captions that are still being displayed on 500 in continuous playback (but are in the stream between theora pages 490 and 491). 21:06 < rillian> DanielWH, yup. 21:06 < DanielWH> no worries then. 21:06 < xiphmont> DanielWH: yes, captions are a discontinuous type. 21:06 < DanielWH> but you want to make sure ogg's seeking algorithm can use vorbis granulepos and theora granulepos to help it seek to theora 500 21:06 < DanielWH> hence the flags. 21:07 < rillian> xiphmont, I feel like we're arguing about an ogg api rather than an format specification 21:07 < xiphmont> rillian: hm, we might be. I'mwondering where confusion set in, because we did agree at the point I set the 'current plan'. 21:08 < DanielWH> don't mind me. 21:08 < xiphmont> Anyway, this has been beyond scope of meeting for a while now. 21:08 < xiphmont> I have to go forage food soon; we should move on. 21:08 < rillian> ok. thanks for stating the proposal 21:09 < rillian> that's the main thing I wanted out 21:09 < rillian> Project status reports 21:09 < xiphmont> Yeah, that it's not in a formal doc already is unforgivable. I really can only think about one thing at a time in depth, and that wasn't it recently. 21:09 < rillian> as discussed last time, we have funding from fluendo.com to work on the theora spec 21:09 < xiphmont> "Yay!" 21:09 < rillian> derf's been writing 21:09 < xiphmont> rather, "Yeeeehaw!" 21:09 < rillian> and we have a good start 21:09 < adiabatic> "Huzzah!" 21:10 < rillian> alpha 3 was releasedd 21:10 < rillian> I've been working on OggMNG again 21:11 < rillian> acolwell build plugins for realplayer 10 that have gotten a few hundred downloads 21:11 < rillian> and we made ./ with the release (thanks Arc) 21:11 < rillian> so things are going pretty well 21:11 < rillian> MikeS, how's ice* coming along? 21:12 < xiphmont> We may have put a few folks into a stupor with the granpos thing. 21:12 < rillian> quite 21:13 < xiphmont> [Is mike hooked up to an SNMP-aware zapper device?] 21:13 < rillian> we've already covered some of the other projects 21:13 < xiphmont> Yes, I already have Vorbis 1.0.3 mapped out, we might call it 1.1. 21:13 < rillian> jm isn't here, but I've not seem much new with speex; mostly adopter questions 21:14 < DanielWH> last I checked MikeS was doing fine getting ice* and shout* to do theora + timing via theora instead of just vorbis. 21:14 < rillian> xiphmont, what's the plan for 1.0.3? 21:14 < DanielWH> sorry I mean KarlH I think 21:14 < rillian> DanielWH, great! I'm glad he's involved with that 21:14 < xiphmont> fixing noise normalization (well, making it more sophisticated) and generalizing stereo coupling. 21:15 < rillian> DanielWH: oh, yes, that I knew about 21:15 < rillian> xiphmont, so this would support channel coupling in 5.1 audio? 21:15 < xiphmont> yes. 21:16 < xiphmont> I've only played a bit; I'm still putting it behing the mux spec and OggFiel API 21:16 < purple_haese> xiphmont: Did you get a chance to read the paper about the cause of the HF boost? 21:16 < xiphmont> purple_haese: yes, had a long discussion with him. 21:16 < purple_haese> Good. I 21:16 < rillian> what about doing a lowpass .1 only mode in the meantime? 21:16 < purple_haese> ugh 21:16 < purple_haese> I'm glad he's being listened to. 21:16 < xiphmont> rillian: hee hee hee 21:17 < purple_haese> I was worried cause there were no follow-ups on the list. 21:17 < purple_haese> Can the long discussion be summarized in a nutshell? 21:18 < xiphmont> No. But I have logs. 21:18 < xiphmont> Bug me for them offline. 21:18 < rillian> Shall we move on to the last item? 21:18 < xiphmont> "We have a call tomorrow. That is all." 21:19 < xiphmont> I need that doc from Peter. 21:19 < rillian> heh 21:19 < xiphmont> S'about it. 21:19 < MikeS> Sorry, was off doing something else. 21:19 < rillian> We're still pursuing the creation of a 'Xiph Forum' standards body with ieee-isto.org 21:19 < rillian> your jobs, as evangenlists 21:19 < MikeS> ice* are currently fairly stagnant, but working well. 21:20 < rillian> is to remind/convince the desktop linux people 21:20 < rillian> that they should be pushing our codecs as hard as they can 21:20 < rillian> particularly, gnome, kde, and freedesktop.org 21:20 < rillian> because this will encourage the people with money to support our work 21:21 < adiabatic> Would here and now be a good place to brainstorm about such things? In particular, defaulting to Ogg Vorbis or flac when encoding. 21:21 < xiphmont> I think I must be off... 21:21 < xiphmont> Anything left for me? 21:21 < rillian> nope. I think we can be done. 21:21 < MikeS> I think one crucial thing that we really need to push harder on (which Monty mentioned earlier) is player support for the "big 3" - real, qt, wmp. I think the real part is looking good. The other two - much less so. 21:22 < purple_haese> By the way... are there any plans to "encourage" game publishers to show Xiph their "appreciation"? 21:22 -!- xiphmont [~xiphmont@HOTASS-8.MIT.EDU] has quit ["He travels fastest who travels alone"] 21:22 < MikeS> There's a guy working on a new set of dshow filters - perhaps we should work with him, get his stuff into SVN if he's interested? Or perhaps one of the other existing projects is better. 21:22 < rillian> purple_haese, not that I'm away 21:22 < rillian> aware 21:23 < rillian> MikeS, yes. if we can them them to maintain their stuff in our repository that would be wonderful 21:23 < rillian> Ok, I move we adjourn to informal discussion 21:23 < MikeS> ok. 21:24 < purple_haese> Seconded. 21:24 < rillian> Ok. Thanks everyone for coming 21:24 < rillian> and for your contributions 21:24 < rillian> logs will be posted --- Log closed Wed Apr 07 21:24:33 2004