--- Log opened Wed Dec 01 15:08:34 2004
15:54 -!- xiphlog [~giles@140.211.166.32] has joined #xiphmeet
15:54 -!- Topic for #xiphmeet: Welcome to the Xiph.org meeting and discussion channel | Next montly meeting at 23:59 GMT December 1 : http://wiki.xiph.org/MonthlyMeeting200412 | log of the last meeting at http://westfish.xiph.org/~giles/200411_meeting.txt
15:54 -!- Topic set by ChanServ [] [Wed Dec  1 06:43:50 2004]
15:54 [Users #xiphmeet]
15:54 [ Arc    ] [ kfish ] [ nessy  ] [ xiphlog] 
15:54 [ HackRip] [ murdos] [ rillian] 
15:54 -!- Irssi: #xiphmeet: Total of 7 nicks [0 ops, 0 halfops, 0 voices, 7 normal]
15:54 -!- Channel #xiphmeet created Wed Dec  1 06:43:50 2004
15:54 -!- Irssi: Join to #xiphmeet was synced in 0 secs
15:54 -!- [freenode-info] why register and identify?  your IRC nick is how people know you.  http://freenode.net/faq.shtml#nicksetup
15:57 -!- rillian changed the topic of #xiphmeet to: Welcome to the Xiph.org meeting and discussion channel | Next montly meeting at 23:59 GMT December 1 : http://wiki.xiph.org/MonthlyMeeting200412 | Live log at http://westfish.xiph.org/~giles/200412_meeting.txt
15:57 < rillian> hi nessy!
15:59 < nessy> hi rillian
15:59 < Arc> hello.
15:59 < Arc> do we have an agenda?
16:00 < rillian> hi arc
16:00 < rillian> I put something in yesterday
16:00 < rillian> monty will be chairing
16:00 < nessy> hi arc
16:00 < rillian> I'm only half here
16:01 < kfish> morning all
16:01 -!- myidiym [~myidiym@24-50-57-104.sbtnvt.adelphia.net] has joined #xiphmeet
16:04 < Arc> monty is coming?
16:04 < Arc> awesome
16:05 < Arc> have had Qs for him
16:07  * HackRip wonders if Emmet will come to annoy Rillian ;) Emmet: Answer my question! Rillian: Mute ... was not productive ... but funny to read last month ;)
16:11 -!- xiphmont [~xiphmont@BLOOPFISH-SIPB.MIT.EDU] has joined #xiphmeet
16:11 < kfish> morning monty
16:11 < xiphmont> Sorry, haven't had a moment to think all day.  I literally have people in tears screaming at me here about shit.  Gahhhhh.
16:11 < nessy> hey
16:11 < xiphmont> Actors.  Musicians.  Man.
16:12 < xiphmont> OK, so, here now...
16:12 < xiphmont> rillian: did you start anything, and did an agenda ever happen?
16:13 < kfish> http://wiki.xiph.org/index.php/MonthlyMeeting200412
16:13 < xiphmont> Ah, something did get posted.
16:13 < xiphmont> (Waiting for Galeon to start up still)
16:13 < kfish> we haven't started yet though, rillian said you're chairing
16:14 < xiphmont> Yes.  He just called me to see where I was :-)
16:14 < kfish> :)
16:14 < xiphmont> I'm nearly caught up, got the meeting page.
16:15 < xiphmont> OK, so default agenda, mostly project reports.
16:15 < xiphmont> Starting with:libogg and OggStream.  This one is Arc
16:15 < xiphmont> Arc: I replied to your mail last night but didn't have any time today to look for a response... I assume you sent one.
16:16 < xiphmont> Argh, people are yelling at me again here.  Perhaps you should start talking for a bit.
16:17 < Arc> I just got your reply.
16:17 < xiphmont> Should we come back to you?
16:18 -!- MikeS [~msmith@CPE-144-137-46-109.vic.bigpond.net.au] has joined #xiphmeet
16:18 < xiphmont> we'll come back to you.,
16:18 < xiphmont> Ah, MikeS arrives just in time.
16:18 < xiphmont> icecast project report?
16:20 < xiphmont> apparently not.
16:20 < MikeS> Sorry.
16:20 < MikeS> I didn't realise the meeting was on.
16:21 < xiphmont> Ah, you logged in and wandered off :-)
16:21 < MikeS> (daylight savings, etc.)
16:21 < MikeS> I thought it was starting in ~40 minutes.
16:21 < MikeS> Icecast: well, we've done the 2.1 release, it's been well received by users. 
16:21 < xiphmont> Anyway go for it.
16:21 < MikeS> Oddsock is working on improved shoutcast compatibility.
16:22 < MikeS> (argh. I can't get the meeting log, because westfish won't resolve correctly for me)
16:22 < MikeS> The reasoning behind this is that all (?) the shoutcast developers have quit or been fired, so we expect that shoutcast will cease to be a viable option for people in the not-too-distant future, so having an easier switchover path is desirable,.
16:23 < MikeS> Unfortunately, the shoutcast protocol is incredibly braindead.
16:23 < MikeS> I think that's about all. 
16:24 < xiphmont> Interesting, I didn't know about Shoutcast.
16:24 < xiphmont> We should indeed capitalize on that.  I'll make me a note to call around.
16:24 < xiphmont> Arc: you ready to talk?
16:24 < MikeS> (can anyone put the meeting log somewhere that isn't westfish for me? Sorry for the trouble...)
16:25 < kfish> MikeS, we've hardly started, you were the first actual meeting item :)
16:25 < kfish> http://wiki.xiph.org/index.php/MonthlyMeeting200412
16:26 < MikeS> shoutcast is still available, and the shoutcast directory still running - so for the moment, it's not so compelling. However, there are suggestions that it might not remain so forever...
16:26 < MikeS> oh, ok
16:26 < kfish> http://www.annodex.net/~conrad/tmp/200412_meeting.txt
16:26 < kfish> (for a snapshot of everyone's polite good-mornings :)
16:26 < xiphmont> MikeS: still, the threat is looming.  We can capture the userbase, then fix the protocol.
16:27 < xiphmont> But it should be in that order :-)
16:27 < kfish> MikeS, is the bizarreness the crap where they put the filename after the semicolon, etc?
16:28 < MikeS> xiphmont: right, that's the intent. There's not much we can do about the protocol: icecast's main protocol is vaguely (if not entirely) sane, and we encourage source clients to support that. The backwards compatibiltiy issues mean we can't do much about the shoutcast protocol.
16:28 < MikeS> kfish: no (that's almost certainly just a "trick buggy user-agents into working" thing, if it's what I think it is).
16:29 < MikeS> kfish: the bizarreness is the source client protocol, which looks almost, but not quite, entirely unlike http once you actually investigate it in detail, even though it looks almost, but not quite, exactly like http at first glance.
16:29 < kfish> yuk :)
16:30 < myidiym> No doubt this is a stupid question... but, why couldn't icecast simply supplant shoutcast?  i.e., why bother fixing shoutcast?  Is the only issue the install base?
16:30 < MikeS> Their interleaved-metadata-in-mp3 protocol is pretty ugly too. But since it's the only thing supported by _listening_ clients, we're stuck with that.
16:30 < MikeS> myidiym: yes. 
16:31 < Arc> MikeS: thats why step 3) migrate them to using Ogg Vorbis
16:31 < Arc> :-)
16:31 < MikeS> myidiym: there are many, many people running internet radio on shoutcast, and though they might not care much about shoutcast itself, they're entirely unwilling to change the way they actually create the stream (the 'source client', to use icecast terminology)
16:32 < MikeS> Arc: we try :-)
16:32 < kfish> any chance nullsoft will open-source shoutcast?
16:33 < MikeS> kfish: a) I think it's unlikely, b) it'd be of no particular interest to us.
16:34 < MikeS> shoutcast itself isn't particularly great at anything it does, as far as I can tell. It does the job... but fairly unimpressively. icecast is fast getting to the point where it has a strict superset of shoutcast's functionality, as well as better portability and scalability.
16:34 < kfish> cool :) let's move on i guess ...
16:35 < MikeS> Anyway, _just_ getting more icecast users isn't really that interesting to us (well, at least to me): as Arc suggests, moving them away from mp3 is where it becomes worthwhile :-)
16:35 < rillian> amen
16:37 < MikeS> I'm not sure how to get that message out to our users better - any suggestions would be welcome (on the mailing list? Let's move on here...)
16:37 < xiphmont> yeah
16:37 < Arc> want a Prog. Rep. on libogg/oggstream?
16:37 < Arc> someone put it on the agenda
16:38 < xiphmont> yes
16:38 < Arc> ok. first of all, documentation
16:38 < xiphmont> I asked you earlier :-)
16:38 < xiphmont> go for it
16:39 < Arc> Duelist has volunteered to do much of the formatting/etc.  he's putting together a first draft based on the convo we had and the ogg2/ogg.h include file
16:39 < Arc> first draft being just api documentation
16:39 < Arc> I'll correct his and make sure monty has a good look over it
16:39 < Arc> from there the "how to use this" and "migrating from libogg1" needs to be written.
16:40 -!- Azra`el [~nerim@65.73.38.26] has joined #xiphmeet
16:40 < Arc> kfish apparently had a talk f2f with monty re: libogg2, and asked me to look into adding libogg1 backwards compatability.
16:40 < Arc> it's possible, and really not that difficult I think, so that will probobally just appear on svn in a few weeks
16:40 < xiphmont> right, kfish convinced me.  It will also ease migration
16:41 -!- adiabatic [~adiabatic@dsl-206-55-130-248.tstonramp.com] has joined #xiphmeet
16:41 < Arc> I agree.  I didn't at first, but yea I think it will.
16:41 < Arc> Now the controversial aspect is OggStream.  I've had alot of ideas over the last few months, between working on theora, writ, and then iceshare, of what OggFile should have looked like
16:42 -!- derf_ [~lomeando@planetmath.cc.vt.edu] has joined #xiphmeet
16:42 < Arc> it's a different design, tho, and I didn't want to take Monty's name. plus it's not focused on "file" anymore, more of a generic "stream" processor
16:42 -!- AndrewBachmann [andrew_bac@node-40244c8a.sfo.onnet.us.uu.net] has joined #xiphmeet
16:42 < Arc> I'm considering it experimental and we'll see where it goes.  I'm being semi-regular in svn commits so you can see where I'm at through there.
16:43 < xiphmont> the name 'OggFile' was merely to sound like vorbisfile, which also isn't just for files.
16:43 < xiphmont> I have a request re: OggStream.  Let's work on it first and talk about it publically later.
16:43 < Arc> I know, but it worked on files by default. anything else needed callback hooks, iirc
16:43 < xiphmont> no, it worked on anything that's a file handle by default, stream *or* file.
16:43 < Arc> well the public talk has been requesting "needs", which I've gotten from key people (xmms guy, kfish, etc)
16:44 < xiphmont> And file handles were used becasue they're ANSI.
16:44 < xiphmont> that's fine.
16:44 < xiphmont> that counts as 'internal' for my purposes.
16:44 < Arc> btw, annodex looks fucking awesome.
16:44 < xiphmont> I don't want to see publicity attempts until somethign is running.
16:44 < Arc> may I just say that.
16:45 < Arc> only the wiki. :-)
16:45 < nessy> Arc, thanks :)
16:45 < Arc> this is a solo project, i need not recruit help until atleast the foundation is working.
16:46 < Arc> and just FYI, I dont want to go on in depth about it's structure, but it really is minimalistic.
16:46 < xiphmont> OK, then also running code first, Wiki after.
16:46 < xiphmont> it need not be much running code.
16:46 < Arc> it's just a common API for codec plugins/io plugins and enough core code to get everything working.
16:47 -!- alobodig [~CRM-114@69.242.100.24] has joined #xiphmeet
16:47 < xiphmont> See, I got called off of OggFile after alot of prep work, but even more talk.  I'm not writing it now.  That's a black eye for us.  Don't repeat it.  I don't mind getting distracted and droping projects.  I do mind hyping stuff that doesn't exist (and yes, I'm the guilty one wrt OggFile)
16:47 < Arc> the wiki is for internal collaboration
16:48 < Arc> it was being updated while it was being discussed. im not doing much to update it now, just getting it finished so it can be talked about further
16:48 < xiphmont> I'm not calling you on anything you've done, just giving you a heads up on my thinking.
16:49 < Arc> unless someone has a major need that's not being addresses, the api/structure on the paper next to me is what its going to look like.
16:49 < xiphmont> It's 'be careful, don't do this' niot 'you should not have done something'
16:49 < xiphmont> OK.  I do want to see that.
16:49 < Arc> well, I have let IceShare pass a few goal markers myself...
16:49 < Arc> you want me to email you the details?
16:50 < xiphmont> yes, definitely.  I'll review libogg2 stuff first.
16:50 < Arc> as i said in email, i'd like to get whatever libogg2-core stuff with my branch resolved before piling even more stuff.. im working on a branch of a branch right now and that makes me uncomfortable :-)
16:51 < xiphmont> agreed.
16:51 < Arc> I'm going to do a quick rename (ogg- to ogg2-) to resolve the namespace issues in libogg2, pyogg2, and libtheora
16:52 < xiphmont> yes, kfish and I talked abotu that too.
16:52 < kfish> let's discuss the mechanics of the renaming on libogg-dev, please
16:52 < Arc> that will allow me to easily add libogg1 backwards compatability support.
16:52 < kfish> cool
16:53 < Arc> oh and where the hell are the includes going to be?
16:53 < Arc> we have ogg2/ogg.h right now
16:53 < MikeS> "ogg-dev", that is (for those who don't know, we have a new mailing list)
16:53 < Arc> are we going to have libogg2 contain ogg/ogg.h and ogg2/ogg.h?
16:53 < kfish> Arc: backwards compatability means exactly that: #include <ogg/ogg.h> continues to work
16:53 < Arc> grr, ok. 
16:53 < kfish> and functions prefixed with ogg_ continue to work
16:54 < Arc> ill ask Duelist to copy (w/ his new docs formatting scheme, whatever that is) the libogg1 stuff to the new docs base
16:54 < kfish> (ie. source and binary compatability)
16:54 < xiphmont> no
16:54 < Arc> no?
16:54 < xiphmont> ogg2/ogg.h
16:54 < xiphmont> sorry, I'm lagged attentionwise
16:54 < kfish> xiphlog, that's not API compatible
16:54 < xiphmont> I'm still multitasking.
16:54 < kfish> s/xiphlog/xiphmont/ :)
16:54 < Arc> ogg2/ogg.h for libogg2 native
16:54 < xiphmont> oh!
16:55 < xiphmont> I understand now.
16:55 < Arc> but dont we need ogg/ogg.h included with libogg2 to provide backwards support?
16:55 < xiphmont> Sorry, yes, for libogg1 compat that makes sense.
16:55 < Arc> cool :-)
16:55 < Arc> we ARE going to eventually phase out libogg1 compatability right?
16:55 -!- Azra`el [~nerim@65.73.38.26] has quit ["My damn controlling terminal disappeared!"]
16:55 < Arc> as in, when nobody uses it anymore?
16:55 < xiphmont> if it works and it's tiny, *shrug*
16:55 < xiphmont> it should be gone by ogg3, yes ;-)
16:56 < xiphmont> 1/2 smilie
16:56 < Arc> you do realise that in order to generate each stage of the process that, ie, the ogg2_page objects need to be passed through a bitpacker and copied into a ogg_page struct buffer
16:56 < xiphmont> We can cheat, and I plan to.
16:56 < Arc> it's going to be slower than libogg1. maybe not by much because libogg2 is so efficient, but it is going to be slower.
16:57 < Arc> there's a cheat?
16:57 -!- Azra`el [~nerim@65.73.38.26] has joined #xiphmeet
16:57 < xiphmont> I had a few in mind.
16:57 < xiphmont> Depends on how much we want to trade 'tivial wrapper' toward 'performance'
16:57 < Arc> i think, considering the eccentricity in implementations, that we need a 100% compatable wrapper
16:58 < Arc> lower performance will encourage them to upgrade 
16:58 < xiphmont> I'm not aorried about compatible
16:58 < xiphmont> I'm worried about complexity.
16:58 < Arc> ok :-)
16:59 < Arc> so you're going to share your ideas on how to do this more efficient, or are you actually putting on your hackers-hat again and jumping in?
16:59 < xiphmont> A little of both possibly.
16:59 < Arc> cool
16:59 < xiphmont> But not for discussion right now :-)
17:00 < Arc> yea lets move on
17:00 < Arc> we can finish this over email
17:00 < xiphmont> OK
17:00 < xiphmont> Vorbis
17:00 < rillian> I hear there's an aotuv3 now
17:00 < xiphmont> Well, a few things came out of our merge effort that he wanted to look into.
17:01 < xiphmont> Also, Mercora is looking at commissioning a higher performance OggEnc to use in their P2P product (and yes, we're allowed to talk about it)
17:01 < rillian> cool
17:02 < xiphmont> I'm actually thinking I might want to tackle that
17:02 < MikeS> Never heard of them (www.mercora.com is presumably their website), but that'd be pretty cool.
17:02 < xiphmont> I did mockup work for such a beast twice, but the company lost interest.
17:02 < xiphmont> So, I'm already prepared for it anyway.
17:02 < MikeS> Whilst the performance of the current encoder is fine for many people, we _do_ still get complaints about the speed compared to other codecs, so that'd be really good.
17:03 < xiphmont> Well, people complain compared to WMA
17:03 < xiphmont> WMA was designed to be fast at the expense of quality
17:03 < MikeS> (just yesterday, someone on the icecast lists was complaining because, when he set up ices encoding 12 (!!) streams, it was taking 80% of his cpu :-)
17:03 < xiphmont> It's part of the very design.
17:03 < MikeS> Yeah
17:03 < xiphmont> But we can do a bit better ;-)
17:03 < MikeS> Right, we don't have to be Number One for speed.
17:04 < xiphmont> next
17:04 < xiphmont> Theora
17:05 < rillian> not much new with me
17:05 < rillian> derf_?
17:05 < rillian> fluendo released flumotion
17:05 < derf_> I can't say I've done anything in the past month with Theora.
17:05 < rillian> and wim's 'cortado' java player
17:05 < rillian> which support theora nicely
17:05 < derf_> Andrey Filippov is making good progress with his FPGA implementation.
17:06 < rillian> oh, right.
17:06 < rillian> that's kind of exciting
17:06 < derf_> A couple of corrections were submitted against the spec, which is good.
17:06 < derf_> That means at least two people are reading it.
17:07 < xiphmont> anythign else?
17:07 < derf_> That's all I can think of at the moment.
17:07 < rillian> https://sourceforge.net/projects/elphel is the software link for anyone who's interested. www.elphel.com makes video cameras with fpga back ends
17:07 < rillian> sort of open source hardware
17:07 < rillian> that's all for theora
17:07 < xiphmont> ah, right, good to mention.
17:08 < rillian> (he's implementing a theora encoder so you can stream ogg theora over the network directly from the camera)
17:08 < derf_> I've been kind of busy with a paper deadline, that I just found out less than an hour ago has been pushed back 3 months.
17:10 < xiphmont> eek
17:10 < xiphmont> oops, blackout.  I can't touchtype.
17:10 < MikeS> Looks like you're doing ok ;-)
17:10 < xiphmont> Speecx?
17:11 < kfish> jean-marc's not here
17:11 < xiphmont> FLAC?
17:11 < MikeS> jm doesn't appear to be here, does anyone else know anything about recent speex progress?
17:11 < kfish> not much has happened recently
17:11 < kfish> jm is busy finishing his phd atm
17:11 < kfish> afaik :)
17:12 < rillian> jm has been answering support questions on the lists, but that's about all
17:12 < xiphmont> I don't see Josh wihther.
17:12 < xiphmont> OK, that's the agenda.  Any random things people would like to cover>
17:12 < xiphmont> ?
17:13 < rillian> re libao
17:13 < xiphmont> OK
17:13 < rillian> I committed a polypaudio driver someone contributed
17:13 < rillian> (sorry, don't remember who)
17:13 < rillian> polypaudio is rumoured to be replacing esd as the gnome audio daemon
17:13 < kfish> ah good, someone was asking about that in polypaudio forums lately :)
17:13 < rillian> and the api is a little more sane
17:13 < rillian> I'd like to do a release with that
17:14 < rillian> but it would be nice if someone would fix http://bugs.xiph.org/show_bug.cgi?id=490 first
17:14 < rillian> I have to fix my ppclinux boot before I can look at it
17:14 < rillian> so any help would be appreciated
17:15 < rillian> we could also go ahead on the principle that it would be not worse than 0.8.4
17:16 < xiphmont> OK
17:17 < xiphmont> Looks like we're done.
17:17 < rillian> yep
17:17 < MikeS> Ok. 
17:17 < Arc> ok.
17:17 < MikeS> Thanks to all who came for coming.
17:17 < rillian> everyone ok with the 12 GMT meeting time next month?
17:18 < xiphmont> yes
17:18 < myidiym> fine with me
17:18 < rillian> great
17:18 < rillian> xiphmont: thanks for chairing
17:18 < myidiym> are we done? (i.e. are we still logging?)
--- Log closed Wed Dec 01 17:18:51 2004