[FFmpeg-devel-irc] IRC log for 2010-01-30

[irc at mansr.com]

[00:36:55] <Compn> [17:39] <peloverde> Why is DivX EOLing DivX 7 without DivX 8 being out?
[00:37:11] <Compn> probably they found some nasty bugs, easier to upgrayydd than to make 7.1 and look like an idiot
[00:37:31] <peloverde> "End of support for DivX 7.x for Windows" http://support.divx.com/faq/view/supportFAQen118/End-of-support-for-DivX-7x
[00:38:59] <mru> divx look like idiots whatever they do
[00:39:05] <Dark_Shikari> wait WHAT already?
[00:39:46] <peloverde> Yeah, they haven't even announced an upgrade path as far as I can tell
[00:40:59] <ShadowJK> do they have any developers
[00:41:58] <Dark_Shikari> of course they do
[00:42:19] <peloverde> I think they are too busy whining that the iPad doesn't support their formats
[00:42:54] <ShadowJK> N900 does! (I think...)
[00:43:40] <peloverde> "The iPad does not have .divx support. They want you to convert your stuff to Apple formats rather than play what you have." As if H.264+AAC in MP4 weren't a widely implemented international standard.
[00:44:14] <ShadowJK> heh
[00:44:27] <ShadowJK> what about mpeg4ASP+AAC in mp4? :)
[00:44:47] <Dark_Shikari> peloverde: the funny thing being they've deprecated .divx
[00:44:52] <Dark_Shikari> they now support h264 + aac/ac3 + srt in mkv
[00:45:18] <mru> and what is .divx? avi?
[00:45:21] <peloverde> yes
[00:45:26] <peloverde> .divx = .avi
[00:45:41] <peloverde> They have some nutty private streams for chapters and whatnot
[00:46:10] <Dark_Shikari> and menus etc
[00:46:17] <Compn> divx subtitles
[00:46:29] <Dark_Shikari> it was actually a good idea IMO
[00:46:30] <mru> who the heck wants menus?
[00:46:33] <Dark_Shikari> rather than make their own shitty container format
[00:46:37] <Dark_Shikari> they stuck private streams onto a real one
[00:46:40] <mru> avi is shitty
[00:46:41] <Dark_Shikari> so that it's backwards-compatible
[00:46:44] <Dark_Shikari> well, this was made a long time ago
[00:46:47] <Dark_Shikari> .divx is old
[00:46:57] <mru> there have always been better formats than avi
[00:47:01] <Dark_Shikari> back when it was originally conceived I don't think mp4 or mkv existed
[00:47:12] <mru> mov is old as the hills
[00:47:32] <Dark_Shikari> sure but it was also proprietary too
[00:47:40] <Dark_Shikari> and it was An Apple Format
[00:47:44] <Dark_Shikari> and that was back before apple was cool
[00:47:51] <peloverde> It has been open spec for forever
[00:47:57] <mru> yeah
[00:48:01] <Dark_Shikari> well back then apple was still using SVQ1
[00:48:01] <ShadowJK> i suspect backwards compatibility was their main goal
[00:48:04] <peloverde> even when apple super sucked in the 90s people still thought mov was good
[00:48:09] <Dark_Shikari> with that fucking bizarre audio format
[00:48:15] <Dark_Shikari> I forgot, which one did they use with SVQ1
[00:48:16] <peloverde> QDM
[00:48:22] <peloverde> QDMC and QDM2
[00:48:28] <ShadowJK> although, if backwards compatibility was their main goal, why did they make it .divx
[00:48:30] * Compn still lol's at quicktime free wont do fullscreen , only quicktime pro
[00:48:40] <peloverde> Compn, they changed that
[00:48:43] <Compn> oh hah
[00:49:00] <peloverde> It did take them forever though
[00:49:07] <mru> ShadowJK: because they're one of those annoying companies that takes standard stuff and sells as if it were something fantastic they've made
[00:49:32] <peloverde> I still cringe when people call AAC an apple format
[00:49:32] <Dark_Shikari> mru: like apple
[00:49:37] <Dark_Shikari> or Nero
[00:49:43] <Dark_Shikari> DivX, Apple, Nero, hmm, who else
[00:50:04] <ShadowJK> real
[00:50:06] <mru> nero are best known for their h264 codecs
[00:50:09] <Compn> on2
[00:50:15] <mru> on2 made their own stuff
[00:50:16] <Yuvi> sorenson
[00:50:17] <peloverde> I would rather them pretend they invented it than use proprietary shit that blows
[00:50:33] <Compn> i guess so
[00:50:36] <peloverde> I think Real is the worst
[00:50:37] <Dark_Shikari> mru: but they sold it as "Nero Digital"
[00:50:46] <mru> hmm, never noticed
[00:50:51] <Dark_Shikari> they have a logo just like divx
[00:50:55] <Dark_Shikari> Nero Digital-compatible devices
[00:51:00] <Dark_Shikari> Nero's ripping software says H.264 nowhere
[00:51:05] <Dark_Shikari> and just says "Nero Digital" format
[00:51:08] <peloverde> Real takes standard stuff, modify it enough to be incomparable, then brand it as their own
[00:51:17] <Dark_Shikari> Yeah, Nero doesn't even bother with the modify step =p
[00:51:25] <mru> nor does divx
[00:51:35] <peloverde> Nero admits their Audio is AAC
[00:51:44] <mru> divx is just all-out annoying
[00:55:21] <peloverde> If I'm doing an FFT with a pretwiddle can I skip ff_fft_permute if I bitreverse while applying the pretwiddle?
[00:57:09] <Dark_Shikari> divx is really annoying yes
[01:13:22] <hyc> annoying == adobe ...
[01:13:58] <peloverde> Which reminds me, Mike's latest blog entries have been entertaining
[05:41:23] <peloverde> How frowned upon is it to bust open an FFTContext and replace the twiddle tables?
[05:42:55] <peloverde> err... it's late here, i'll worry about that tomorrow
[09:53:25] * Terminating due to: __DIE__
[11:33:15] * /join #ffmpeg-devel ...
[11:33:17] *** TOPIC: Welcome to the FFmpeg development channel. That is, development of FFmpeg, not using FFmpeg, nor libav*. | Users should redirect their questions to #ffmpeg | FFmpeg 0.5 has been released! | this chat is now publicly logged.
[11:33:17] *** TOPICINFO: Compn, 1264791848
[13:11:13] * Compn drops wheel troll and runs
[13:13:33] <kshishkov> award silver ffmpeg2theora to Compn
[13:13:46] <Dark_Shikari> Compn: "ffmpeg is a wheel factory"
[13:13:47] <Dark_Shikari> awesome
[13:16:29] <kshishkov> ... err, providing means to reinvent the bicycle?
[13:16:54] * mru puts pedals on ffmpeg
[13:17:49] <thresh> so, its 'bike' already, now to go with 'shed'
[13:18:06] <mru> I want it blue
[13:18:23] <kshishkov> Faluröda
[13:42:24] <Kovensky> ramiro: < LigH> syntax error before "mode_t" ... why there? Is the type "mode_t" undefined? Or is the cause really a line above?
[13:42:40] <Kovensky> 10:25.51 < golgol7777> LigH: Hi, I confirmed the build failure of pthread-w32
[13:42:43] <Kovensky> 10:28.22 < LigH> Thanks golgol7777 ... now let's try to guess a reason; "make" defines the existance of mode_t.
[13:42:46] <Kovensky> 10:28.53 < LigH> gcc -c -o attr.o -D__CLEANUP_C -O3 -finline-functions  -I. -DHAVE_CONFIG_H -Wall  attr.c
[13:42:50] <Kovensky> 10:29.12 < golgol7777> That patch breakes build because its removes sys/types.h inclusion.
[13:42:53] <Kovensky> 10:29.48 < golgol7777> I recommend this patch for 2.8.0.0
[13:42:55] <Kovensky> 10:30.00 < golgol7777> http://up-cat.net/wiki.cgi?action=ATTACH&page=Pthreads%2Dw32+2%2E8%2E0%A4%CE%A5%D0%A5%B0&file=pthreads%2Dw32%2D2%2D8%2D0%2Dfixposix%2Ediff
[13:42:59] <Kovensky> 10:30.46 < golgol7777> It's another patch for the same 'pid_t not defined' issue.
[13:43:09] <Kovensky> he used the patch you link on the arrozcru wiki
[16:02:38] <DonDiego> moint
[16:02:40] <DonDiego> moin
[16:02:49] <DonDiego> my powerbook died :-(
[16:02:53] * kshishkov looks for pross-au
[16:03:01] <kshishkov> it's unlikely to be "moin"
[16:03:13] <kshishkov> okay, when is funeral?
[16:03:27] <DonDiego> does anybody know the libschroedinger CVS repository location?
[16:03:30] * kshishkov has DVD reading function died in his MacMini
[16:03:36] <kshishkov> sf?
[16:03:49] <DonDiego> kshishkov: moin works at every time of the day
[16:04:19] <DonDiego> http://schrodinger.cvs.sourceforge.net/viewvc/schrodinger/
[16:04:29] <DonDiego> that only appears to contain the website..
[16:07:02] <kshishkov> probably they just take frech Dirac release, slap some patch and ship it
[16:08:19] <thresh> DonDiego: git://diracvideo.org/git/schroedinger.git
[16:08:43] <thresh> or at least it was there last time i pulled it
[16:09:32] <DonDiego> did libdirac switch to git as well?
[16:09:59] <thresh> no idea
[16:28:06] <Compn> DonDiego : btw if p-p-p-p-powerbook died, do you still want replacement battery? if not, update donation page :P
[16:29:12] <kshishkov> depends on the way it died
[16:29:31] <kshishkov> in theory notebooks should be able to work even without a battery. Desktops do
[16:30:17] <Compn> yes, i have two laptops without batteries that must be plugged in at all times
[16:30:29] <kshishkov> only when they work :P
[16:30:52] <Compn> i'm just saying, if its dead, he probably doesnt need the battery for it anymore
[16:31:08] <mru> I have a macbook w/o battery...
[16:31:25] <mru> not mine, strictly speaking
[16:32:11] <DonDiego> macbook != powerbook
[16:33:26] <Compn> so howd it die ?
[16:33:28] <kshishkov> still, what can die in it? NVRAM battery? Display? MacOS?
[16:33:43] <Compn> laptops #1 death = mobo overheats
[16:33:47] <DonDiego> the display flickered
[16:33:49] <DonDiego> then went out
[16:33:52] <DonDiego> for a few seconds
[16:33:55] <DonDiego> then revived
[16:34:06] <DonDiego> this went back and forth a few times
[16:34:16] <DonDiego> then the display stopped working entirely
[16:34:22] <DonDiego> but the backlight still works..
[16:34:27] <kshishkov> ah, tough luck then
[16:34:39] <kshishkov> could be that video signal strip breaking
[16:34:48] <mru> sounds about like KotH's thinkpad a few years ago...
[16:34:58] <kshishkov> or mine
[16:36:38] <DonDiego> what happened to those machines?
[16:36:58] <mru> they died
[16:37:28] <kshishkov> mine got display matrix replaced
[16:39:32] <DonDiego> what is the display matrix?
[16:39:49] <kshishkov> that thingy with liquid crystals
[16:39:58] <Compn> the lcd ?
[16:40:11] <kshishkov> yep
[16:40:18] <Compn> dats a bummer
[16:40:29] * Compn forgets price for replacement powerbook screen
[16:40:37] <kshishkov> anyway, any mru should say "if display does not work, ssh to it"
[16:40:42] <Compn> DonDiego : external monitor works ?
[16:41:02] <mru> could be burned out gpu
[16:42:15] <Compn> external monitor would tell if gpu was dead
[16:42:17] <DonDiego> no, external monitor fails as well
[16:42:17] <Compn> ouch
[16:42:27] <Compn> $100-$400 for replacement lcd
[16:42:45] <Compn> $100 for older, $200 for newer powerbook
[16:42:51] <DonDiego> more likely logic board or cables i think
[16:44:53] <Compn> ssh work ?
[16:45:08] <DonDiego> haven't tested
[16:45:20] <DonDiego> i gave it away to the apple shop
[16:45:28] <DonDiego> and iirc ssh was not enabled
[16:45:36] <kshishkov> nummer
[16:45:41] <Compn> ah
[16:45:54] <DonDiego> nummer?
[16:46:06] <Compn> maybe he meant bummer ?
[16:46:10] <kshishkov> yep
[16:46:16] <mru> or number
[16:46:17] <kshishkov> keys are close
[16:46:23] <mru> hummer
[16:46:36] * Compn wonders if kshishkov has to work by candlelight in 3rd world country
[16:47:11] <kshishkov> Compn: no, I use light from display
[16:48:01] <mru> light? who needs light?
[16:49:57] <kshishkov> Compn: but yes, I have such experience - thanks to occasional blackouts
[16:50:37] <Compn> my dad taught me how to make candles, so i have a few candles in case of blackouts
[16:50:56] <kshishkov>  you Benjamin Franklin
[16:51:33] <Compn> its interesting to recycle old melted candles into new candles
[16:51:48] <Compn> and hard to grasp just how many lightbulbs we trash each year
[16:52:37] <Compn> or that we used to. now using cfl's. anyone use led lights yet ?
[16:53:09] <kshishkov> me
[16:53:25] <kshishkov> from different hardware around here
[16:56:16] <Compn> blinking modem lights dont count :P
[16:56:38] <kshishkov> why?
[16:56:49] * kshishkov has USR Courier V.everything
[16:57:20] <Compn> oh i remember that
[16:57:45] <kshishkov> still, some blinks from network cards, some blinks from BeagleBoard, some blinks from laptop, some constant light from adaptors
[17:00:47] <Compn> poor mike, stephen warren, gwenole, iive and me commented on his swf blog. and then everyone else piled on him calling him 'not a linux developer' :(
[17:01:10] <DonDiego> url?
[17:01:12] <kshishkov> call him also 'not a FFmpeg developer'
[17:02:45] <kshishkov> also Mike is the one to blame for Theora success
[17:03:14] <Compn> http://blogs.adobe.com/penguin.swf/2010/01/solving_different_problems.html#comments
[17:03:24] <iive> kshishkov: he is?
[17:03:37] <Compn> iive : didnt he write the spec ?
[17:03:53] <kshishkov> iive: well, even Theora spec still features remains from his description of VP3
[17:04:11] <iive> i thought on2 opensourced vp32 or something.
[17:05:00] * Compn checks wikipedia
[17:05:14] <kshishkov> so what? it takes too much time to figure out of their code
[17:05:19] <Compn> There is no formal specification for the VP3 bitstream format beyond the VP3 source code published by On2 Technologies. In 2003, Mike Melanson created an uncomplete description of the VP3 bitstream format and decoding process at a higher level than source code, with some help from On2 and Xiph.Org Foundation. The Theora specification adopted some portions of this VP3 description.
[17:05:30] <Compn> haha
[17:05:31] <Compn> http://en.wikipedia.org/wiki/Theora
[17:05:42] <iive> We all know that Mike likes to reverse engineer obscure crappy codecs. It's not his fault that somebody thought it could become descent codec:)
[17:05:46] * Compn laughs at 'too much time to figure out the code'
[17:06:13] * Compn laughs harder at theora people hating on mike for adobe now
[17:06:27] <Compn> hes god and the devil, your brain must now explode
[17:06:33] <kshishkov> Compn: actually my blog started from my attempts to understand their code for TrueMotion 2
[17:08:49] <kshishkov> Mike - the only man who has solid ground to hate Theora
[17:13:34] <Compn> kshishkov : you wish there was code for tm2
[17:13:47] <Compn> understand their binary :P
[17:14:25] <Compn> or is there code? hmm
[17:15:15] * Compn looks at truemotion2.c
[17:15:19] <Compn> lots of these "+= 4"
[17:15:34] <Compn> could those be factored out ?
[17:16:36] <Compn> probably not worth it
[17:17:20] <kshishkov> Compn: there is their VpVision product sources (still at duck.com) which contains decoders for TM1, TM2 and TrueMotion VP3
[17:17:29] <Compn> ah
[17:19:31] <kshishkov> Compn: specially for you - http://duck.com/vpvision/
[17:19:44] <Compn> haha no i'd rather not read those sources
[17:19:59] <kshishkov> indeed
[17:21:31] <ohsix> people still seem to miss that videos in flash are still clips, they need to act like any other clip; you cant just composite them
[17:24:24] <kshishkov> browsers just need to mimic YouTube player so it can simply use <video type="youtube" ...> tag
[17:27:38] <thresh> j-b asks for +v here
[17:28:37] <DonDiego> j-b: do you have any patches in ffmpeg?
[17:28:49] <j-b> I just want to speak!
[17:28:51] <thresh> he's unable to send anything here
[17:28:52] <thresh> woah
[17:28:54] <j-b> gosh
[17:29:07] <kshishkov> DonDiego: ask Compn instead
[17:29:11] <j-b> no idea what changed. "#ffmpeg-devel Cannot send to channel" I had that
[17:29:12] <Compn> ??
[17:29:23] <j-b> Anyone is interested in Dolby-E ?
[17:29:34] <kshishkov> Compn: do you have any patches in FFmpeg?
[17:29:36] <thresh> is it still +R?
[17:29:53] <kshishkov> no
[17:30:16] <Compn> kshishkov : why ask me? i dotn work on vlc
[17:30:17] <thresh> weird
[17:30:30] <Compn> and yes its weird, i had same problem before i identified, +R is off but cant chat
[17:30:46] <ohsix> ircd is up to shenanigans
[17:30:48] <thresh> we can all blame freenode for being stupid
[17:30:58] <j-b> I was identified, weirdly
[17:31:03] <j-b> sorry for the mess.
[17:31:16] <j-b> who is going to be in charge for you SoC 2k10 ?
[17:31:57] <kshishkov> j-b: whom you were asking?
[17:32:30] <j-b> anyone that can answer :)
[17:32:54] <j-b> about Dolby-E, the question is more because I believe I can provide additional funding
[17:33:06] <kshishkov> probably it will be a bunch of mentors, as usual
[17:33:27] <j-b> about SoC, to discuss about whether we can do the same as last year about qualification task
[17:34:11] <Compn> j-b : best to ask on list for our non-irc members :)
[17:34:11] <j-b> meaning, allowing to people having succeeded in FFmpeg qual task to skip VideoLAN tasks
[17:34:36] <j-b> Compn: everyone is on IRC except Michael, no? :D
[17:34:46] <kshishkov> multimedia SoC cabal, eh?
[17:35:31] <Compn> j-b : good question :P
[17:35:40] <Compn> reimar isnt on irc :P
[17:37:27] <Compn> have to ask DonDiego how that went last year (vlc using ffmpeg small/qual tasks as vlc qual tasks)
[17:37:42] <Compn> or whoever oversaw soc
[17:38:08] <Compn> i thought it was a good idea anyhow :)
[17:39:29] * Compn looks for more small tasks to add
[17:41:19] <j-b> is MSS2 in the list for SoC ?
[17:41:30] <kshishkov> probably not
[17:41:54] <j-b> "AACS implementation" why not use libbluray ?
[17:42:21] <Compn> i think aacs implementation was written before libbluray appeared
[17:42:37] <Compn> i mean, the wiki page was
[17:42:40] <Compn> feel free to change it
[17:43:03] <DonDiego> libavcodec/h261dec.c:292: warning: implicit declaration of function ‘ff_set_qscale’
[17:43:07] <DonDiego> how lame is that?
[17:43:22] <ohsix> 7
[17:43:45] * Compn checks 2010 soc page
[17:43:54] <Compn> wasnt there a patch for flashsv2 ?
[17:44:19] <kshishkov> Compn: drv is working on that. Or at least supposed to
[17:44:59] <Compn> yeah, i think ronald copied too many old things
[17:46:10] <Compn> i think its wise to drop vp6 encoder after google bought them
[17:46:19] <Compn> or is that not finalized either hmm
[17:46:45] <kshishkov> thank pengvado for that!
[17:47:22] <Compn> ?
[17:47:47] <kshishkov> it's solely to his effor we don't have opensource VP6 encoder
[17:47:51] <kshishkov> *effort
[17:47:56] <peloverde> BSAC added: http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010#BSAC_AAC_Decoder
[17:49:02] <Compn> j-b : nevermind, i added note about libbluray, thanks
[17:50:06] <j-b> Compn: ok, I have to try it a bit more thoug
[17:52:29] <kshishkov> jai: looks like MN is not against discontinuing animating dead SoC JPEG2K code
[17:53:07] <Compn> arg double negative
[17:53:20] <kshishkov> where?
[17:53:32] <Compn> NOT against DIS
[17:53:39] <mru> nothing wrong with that
[17:53:47] <mru> how would you put it?
[17:53:57] <Compn> mn is willing to discontinue...
[17:54:05] <mru> antidisestablishmentarianism...
[17:54:07] <kshishkov> that's wrong emotion
[17:54:13] <Compn> or mn is ok with killing jpeg2k
[17:54:30] <Compn> errr
[17:54:31] <Compn> hehe
[17:54:44] <Compn> whats that mean tho? you guys hate jp2k code ?
[17:55:01] * Compn still confused about the double negatives
[17:55:06] <mru> not the code so much as the codec...
[17:55:28] * kshishkov does not have any feelings about it. Maybe except for wavelets...
[17:56:12] <mru> wrong double negative would be something like "I ain't got nothing against ..."
[17:56:21] <mru> that's wrong in other ways too of course
[17:56:30] <peloverde> Does anyone have a tool that can remux from TS+LATM to MP4?
[17:57:10] <mru> I gave up on latm, flexmux, SL, and all that crap long ago
[17:58:13] <peloverde> I'm trying to get the audio out of some dmb files
[17:58:57] <peloverde> We could just pretend that AAC doesn't exist in the broad cast world, but that's kind of silly
[17:59:42] <kshishkov> why not pretend that whole broadcast world does not exist? Saves a bit on Dolby E too
[18:00:14] <j-b> :)
[18:04:06] <peloverde> Weren't people just complaining that the DVB muxer is being done elsewhere
[18:11:26] <kshishkov> Hasn't The Great Chairman said "Let all flowers bloom"?
[18:12:03] <peloverde> MPlayer seems to support LATM?
[18:19:12] <elenril> Playlist/Concatenation Support << isn't this already done/almost done?
[18:19:50] <kshishkov> it was, but IIRC not in main SVN
[18:20:02] <elenril> i know, it wasn't committed
[18:20:43] <elenril> but if all it needs is a little cleanup, then it probably doesn't deserve a soc project
[18:22:53] <Compn> elenril : i agree, i think ronald copied too much
[18:22:57] <Compn> post on the wiki ;p
[18:23:18] <elenril> :effort:
[18:25:53] <peloverde> If I have an FFT where the second half of the input is blank, am I better off doing one DIF manually?
[18:30:03] <kshishkov> yes
[18:30:18] <kshishkov> one lesser FFT + posttwiddling
[18:42:30] <Compn> oh gogo the bee joined in that swf blog post of mikes haha
[18:42:34] <Compn> havent seen him around
[18:50:50] <peloverde> Chrome uses FFmpeg; flash doesn't. I think the choice is pretty clear.
[18:51:12] <kshishkov> yep - make FFmpeg use flash
[18:51:54] <peloverde> FFmpeg doesn't support binary codec loaders
[18:52:02] <peloverde> But I'm sure mplayer would welcome a patch
[18:52:11] <kshishkov> bah, binary codec loaders
[18:52:23] <kshishkov> we are talking about decoders in pure ActionScript!
[18:53:07] <peloverde> i think someone tried vorbis in actionscript once
[18:54:09] <kshishkov> was it fast in IE?
[18:54:34] <peloverde> Is flash faster in IE?
[18:55:17] <kshishkov> who knows?
[18:55:40] <kshishkov> could have an additional penalty on something
[18:57:40] <Compn> peloverde : once i heard someone say they would accept gstreamer input into ffmpeg
[18:58:55] <kshishkov> yuck
[18:58:56] <mru> that would be ridiculous
[18:59:01] <mru> gstream *uses* ffmpeg
[18:59:46] <kshishkov> would be nice to make FFmpeg reading input from gst-ffmpeg reading input from gst-ffmpeg reading input from ...
[19:00:37] <ohsix> with gst in the middle they can join in on the covered ears crowd :> (even if the only possible use-case is what they feel is infringing, they; and their lawyers can say it wasn't the idea)
[19:01:46] <ohsix> its nebulous possible litigation/claims vs. big ball of reasonable doubt; they like at least having that =)
[19:02:45] <Compn> there are a lot of 'concern trolls' running around
[19:02:55] <Compn> 'concerned about patents' 'concerned about lawsuits' etc
[19:03:14] <kshishkov> such talks put me in berserk mode
[19:03:26] * kshishkov has a really good reason to hate IP laws
[19:03:37] <ohsix> everyone has a reason to hate IP laws
[19:03:43] <peloverde> I always like when people throw around the vague notion that our code is stolen from somewhere
[19:03:58] <ohsix> but discounting their actual motivation for doing what they do and coming up with your own alternate narrative isn't useful
[19:04:08] <Compn> kshishkov : aside from patenting medicine ?
[19:04:20] <ohsix> peloverde: you dont mean me, right? that wasn't my intent
[19:04:46] <Compn> peloverde : thats a good troll too. but it only works on reporters (who then repeat it without question)
[19:04:54] <KotH> DonDiego: on my laptop, the graphic chip got fried
[19:05:10] <KotH> DonDiego: luckily i got a replacement on warranty :)
[19:05:41] <ohsix> Compn: heh even if they do repeat it, the audience they're reporting to has no real idea or concern :> when they do they verify stuff for themselves; not listen to pulp news "reports"
[19:06:19] <Compn> peloverde : its double fun when the person who uses that troll, has already been accused of violating gpl
[19:06:30] <Compn> a troll defense, if you will
[19:06:44] <Compn> it is not us who stole, but you stole from us first!
[19:06:52] <ohsix> the problem as I see it is that everyone wants to throw their two cents in without really putting in the effort to assess what they're actually talking about
[19:07:09] <peloverde> That's what I hate about reddit
[19:07:16] <Compn> thats common problem everywhere
[19:07:30] <peloverde> It's worse on reddit
[19:07:54] <peloverde> astrange constantly gets downmodded on technical issues as did I before I left
[19:08:15] <ohsix> yea, but you can stay above the noise :> theres been a lot of hullaballoo and overlap, that i were in a position to say, i would be telling people to stfu
[19:08:18] <peloverde> once I was super down modded for saying DivX 4-6 - MPEG-4 ASP in AVI
[19:08:31] <ohsix> peloverde: haha
[19:08:50] <kshishkov> yes, it's no good telling the truth
[19:08:51] <ohsix> was that back when divx was just a modified bit of someone elses code?
[19:08:59] <peloverde> The upmodded response was If that were the case how could they sell expensive device certifications
[19:09:02] <kshishkov> it won't make you a good politician either
[19:09:44] <DonDiego> Compn: what about patenting medicine?
[19:09:51] <ohsix> telling the truth doesn't get you much; and if you repeat yourself it weakens your position, you need to know wether the people you are addressing can even reasonably use the information they might glean from you
[19:10:41] <kshishkov> DonDiego: guess how prices change when certain pharmacy product is patented?
[19:11:59] <mru> medical patents are as absurd as any other
[19:12:11] <mru> patents on genes, wtf?
[19:12:23] <mru> if I have that gene, do I need to pay royalties?
[19:12:33] <kshishkov> shush, walking patent law breakage
[19:12:34] <Compn> patents on crops :(
[19:13:18] <ohsix> Compn: for real (crops), have you seen food, inc.? monsanto utterly ruins people if they find their genes in peoples fields, and you know how genes work
[19:13:27] * elenril should patent method and apparatus of bikeshedding
[19:13:43] <Compn> i didnt see food inc , but i did read about monsato ruining people :(
[19:14:12] <ohsix> when you can sue people for someething that people couldn't EVER possibly control, like the _wind_ or the very good chance that someone tracks something in on their fucking boot; you are an awful entity
[19:19:57] <kshishkov> isn't whole USA built on principle "if you neighbor sneezes, sue him"?
[19:20:23] <mru> s/if.*,//
[19:20:40] <kshishkov> ah yes
[19:54:07] <KotH> .o0(where is bbb when one needs him?)
[21:16:53] <BBB> hi
[21:17:04] <BBB> Vitor1001, do you want to mentor amr-wb decoder for next soc?
[22:07:50] <BBB> mru, can you ok my msg to ffmpeg-devel@ again?
[22:08:55] <mru> done
[22:08:58] <BBB> ty
[22:09:07] <BBB> I wanted to do something else also...
[22:09:11] <BBB> Vitor1001, ping?
[22:09:16] <BBB> KotH: ping?
[22:12:21] <Vitor1001> BBB: Sorry, was away from keyboard
[22:12:52] <Vitor1001> I'm not mentoring anything this year. I have a PhD thesis to write and a work to find :(
[22:14:22] <BBB> ok, got it
[22:14:33] <BBB> I also submitted a new wmavoice patch, please look :)
[22:18:20] <jez9999> BBB: hi
[22:19:08] <Vitor1001> BBB: ok, will review it today :)
[22:19:09] <BBB> hi jeremy
[22:19:25] <BBB> Vitor1001, oh, and I have a local modification already to add a flush function just like michael did for wma1/2
[22:19:30] <BBB> we should do that for wmapro also
[22:19:44] <Vitor1001> BBB: Does it work?
[22:19:49] <BBB> I suppose
[22:19:51] <Vitor1001> in the sense that seeking work?
[22:19:55] <BBB> seeking works
[22:19:56] <BBB> yes
[22:20:08] <BBB> why wouldn't it?
[22:20:32] <Vitor1001> That was what was broken in wmapro
[22:20:39] <BBB> ?
[22:20:41] <Vitor1001> BTW, what do you have against enums?
[22:21:05] <Vitor1001> What the flush function was needed was to seek
[22:21:30] <BBB> oh, enums
[22:21:38] <BBB> hm, I think I did't address that
[22:21:43] <BBB> if you want enums, I'll add enums
[22:22:19] <DonDiego> amr-wb decoder as soc task sounds good
[22:22:26] <BBB> hi dondiego
[22:22:30] <Dark_Shikari> amr-nb encoder? ;)
[22:22:32] <Vitor1001> Or I want you to tell me why you decided not to use it
[22:22:54] <BBB> it'd be nice if we could come up with suggested qualification tasks per suggested soc job
[22:22:58] <BBB> like x264
[22:23:14] <BBB> Vitor1001, uhm, I think I read over it and forgot, so I didn't really go either way yet in my thinking
[22:23:16] <BBB> sorry
[22:23:27] <BBB> I know you told me twice already
[22:23:36] <Vitor1001> No problem, just to be sure there wasn't a reason it hasn't been done
[22:23:42] <jez9999> BBB: say you were going to implement a callback in ffmpeg allowing calling code to reject individual packets based on source address (hypothetically).  how would you go about it?
[22:23:55] <Vitor1001> Not all my suggestions result in better code, it is better to take them with a grain of salt
[22:23:58] <Dark_Shikari> I'd be happy to try to come up with x264-style qual tasks if anyone wants
[22:24:01] <Dark_Shikari> for a given project
[22:24:06] <BBB> Dark_Shikari, please do
[22:24:19] <BBB> Dark_Shikari, I think it's poentially much better than what we did last years
[22:24:23] <BBB> and we have time to prepare now
[22:24:36] <Dark_Shikari> where's a list to look at
[22:24:40] <BBB> uh...
[22:24:41] <Dark_Shikari> I'm lazy
[22:24:46] <BBB> http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010
[22:25:00] <BBB> Vitor1001, I'll need to think of names for them, so give me a sec
[22:25:13] <KotH> BBB: pong
[22:25:19] <Vitor1001> BBB: No need to hurry...
[22:25:26] <BBB> jez9999, I don't think I would do it :-)
[22:25:31] <Vitor1001> I'll probably review it after dinner anyway
[22:25:45] <Vitor1001> Just another question, in the nit:
[22:25:49] <Vitor1001>  > Is this bit array there just to optimize a range-checking?
[22:25:49] <Vitor1001> No, decrease stack size usage.
[22:25:50] <Dark_Shikari> BBB: I'll do it in a bit, doing a school lab tm
[22:25:50] <Dark_Shikari> *atm
[22:25:51] <jez9999> BBB: i knew you'd say that, which is why i said hypothetically
[22:25:55] <BBB> Dark_Shikari, sure
[22:26:00] <jez9999> BBB: if someone held a gun to your head
[22:26:03] <BBB> jez9999, alright, you got me
[22:26:13] <BBB> jez9999, give me a sec to handle everyone else and I'll think of a way
[22:26:49] <Vitor1001> BBB: I understand that you use a bit array instead of a int array to save stack size, but why do you need an array at all?
[22:26:59] <Vitor1001> Why is this value not calculated where it is used?
[22:27:44] <BBB> my impression is that it's a list of "have I handled a pulse starting at this offset already", or "do I want this offet to be excluded as a pulse starting point"
[22:27:56] <BBB> so I'm not sure if you can calculate it on-site
[22:28:05] <BBB> it's a list, sort of
[22:28:11] <Vitor1001> ok, I'll have to re-read the code to understand it
[22:28:18] <BBB> I removed the other array
[22:28:28] <BBB> I did as you just suggested, i.e. just calculate the one I use where I use it
[22:28:38] <BBB> (I only used one item from the array)
[22:29:45] <Vitor1001> Ok, it is just I don't get much what the function do (i.e., what is the formula for calculating the pulses)
[22:32:09] <Vitor1001> But I'll give a decent review of it soon.
[22:32:24] <BBB> the code is about as much of what I get of it myself
[22:32:37] <BBB> which is admittedly not much
[22:32:51] <BBB> there's some papers on this type of pulses that I'd like to read
[22:32:53] <BBB> but they cost money
[22:32:56] <BBB> which is a bit wasteful
[22:33:26] <Vitor1001> I understand. I'll try to get something by reading the code.
[22:33:40] <Vitor1001> Another thing,
[22:33:56] <Vitor1001> About the division that could be done with integer math
[22:34:17] <Vitor1001> division is slow, but this is wmavoice_decode_init()
[22:34:30] <BBB> true
[22:34:39] <BBB> so .0025 -> / 400
[22:35:00] <Vitor1001> ROUNDED_DIV(ctx->sample_rate << 8, 400)
[22:35:04] <BBB> and .0185 -> *37/2000?
[22:35:04] <Vitor1001> or something like that
[22:35:23] <BBB> the rounding is different :)
[22:35:27] <BBB> it's not 0x80, it's 50
[22:35:31] <BBB> another one of those things :(
[22:35:44] <Vitor1001> :p
[22:35:52] <Vitor1001> The hallmark of misdesigned codecs
[22:36:08] <Vitor1001> I remember a way to do av_log2(x)
[22:36:08] <BBB> :)
[22:36:16] <BBB> ?
[22:36:20] <Vitor1001> int a = log2(x) + .1;
[22:36:41] <Vitor1001> People like crazy roundings, nonsense floating-point use, etc
[22:37:03] <Vitor1001> Part of the beauty of RE is getting rid of that kind of junk :D
[22:37:43] <BBB> I added the float, the original code was junkie-style bit-truncation 64 bit multiplications
[22:37:47] <BBB> while that's not necessary
[22:37:58] <BBB> I mean, samplerate=8000
[22:38:06] <BBB> x2000/37 would easily fit in 32 bits
[22:38:09] <BBB> yet they do it in 64
[22:38:21] <BBB> with a value that looks like they actually wrote it as float
[22:38:26] <BBB> and suggesting that their dll uses floats here
[22:38:30] <BBB> (hence I did it in float)
[22:38:41] <Vitor1001> Probably because original code was float and was converted to fixed-point by someone at ms
[22:38:52] <BBB> yeah
[22:39:54] <Vitor1001> About using get_vlc(), kshishkov is someone who could give us a tip
[22:40:04] <Vitor1001> he has written quite a few decoders that use it
[22:40:49] <BBB> ok
[22:40:54] <BBB> it has a max_depth==3
[22:40:58] <BBB> which is not good
[22:41:01] <BBB> I need a max_depth==6
[22:41:24] <Vitor1001> Ok, I think I understood it the other way around
[22:41:36] <Vitor1001> vlc_get has a smaller max_depth.
[22:42:03] <Vitor1001> Of course there is no documentation for vlc functions in lavc :p
[22:42:26] <mru> BBB: max max_depth is 3
[22:42:52] <Vitor1001> mru: How can someone do something useful with max_depth = 3?
[22:43:03] <mru> uh?
[22:43:10] <BBB> mru: eh, yes that is what I mean
[22:43:11] <Vitor1001> It looks pretty limited as vlc coding
[22:43:14] <mru> we're talking about vlc tables, right?
[22:43:15] <BBB> sorry
[22:43:22] <Vitor1001> yes
[22:43:50] <Vitor1001> Unless I think depth is another thing
[22:43:55] <mru> say you have 9 bits per level
[22:44:05] <BBB> dondiego: could I convince you somehow to write the website for the foundation?
[22:44:09] <mru> depth 3 then gives you a max code length of 27 bits
[22:44:12] <BBB> dondiego: e.g. beer, hookers, both?
[22:44:13] <mru> do you need more?
[22:44:21] <BBB> mru: I need 6
[22:44:25] <BBB> and they're only 2 bits each
[22:44:35] <mru> then you're doing it wrong
[22:44:40] <BBB> ?
[22:44:45] <BBB> MS designed this, not me :)
[22:44:45] <Vitor1001> ok, I misunderstood what the depth is.
[22:44:54] <mru> BBB: show us the codes
[22:44:59] <BBB> read 2 bits, if 0/1/2, return that
[22:45:03] <BBB> else read another 2 bits
[22:45:09] <BBB> if 0/1/2, return that+3
[22:45:11] <BBB> else, read another 2 bits
[22:45:12] <BBB> etc.
[22:45:15] <DonDiego> BBB: why me?
[22:45:16] <BBB> until you did that 6 times
[22:45:30] <BBB> then return 5x3+4 (?)
[22:45:34] <BBB> er, +3
[22:46:03] <BBB> dondiego: you made ffmpeg.org
[22:46:11] <BBB> excellent victim
[22:46:28] <mru> BBB: superdump make ffmpeg.org
[22:46:28] <DonDiego> umm, not really
[22:46:32] <BBB> oh
[22:46:37] <BBB> sorry
[22:46:40] <BBB> misunderstood then
[22:46:48] <DonDiego> i maintained it before superdump took over
[22:47:00] <BBB> I suppose he's not online now...
[22:48:18] <mru> BBB: so your codes look like this? http://pastebin.ca/1772488
[22:49:32] <BBB> mru: yes
[22:50:09] <BBB> mru: and that is probably suboptimal, but I didn't design it so I plead non-guilty anyway
[22:50:19] <mru> and just to clarify, what are the values returned for those codes?
[22:50:33] <jez9999> BBB: im going to bed now, but i'll read your suggestions in the morning ;-)
[22:50:40] <BBB> jez9999, ok
[22:51:22] <BBB> mru: they're an index in a table, just like get_vlc(), so they return the index in the table, and that is then any value from 0-25
[22:51:24] <BBB> or something
[22:51:26] <BBB> probably 0.21
[22:51:30] <BBB> maybe the table is too big
[22:51:51] <mru> ok, so the index can be whatever it wants
[22:51:59] <mru> you just adapt the other table
[22:52:16] <BBB> yes
[22:52:23] <BBB> it returns frame type
[22:52:37] <BBB> for low bitrate, that's generally 0 or 2 (pitch-adaptive windowing)
[22:52:50] <BBB> for higher bitrate, that's 0,3-6 or 0,9-11 or so
[22:53:12] <BBB> so it's just a way of decreasing bits spent on coding frametype
[22:53:19] <BBB> (well duh, that is what vlc is)
[23:04:53] <mru> BBB: http://pastebin.ca/1772506
[23:05:17] <mru> you can do that with on level of 12 bits or two levels of 6 bits
[23:05:29] <mru> what's the distrubution like?
[23:05:38] <BBB> almost always 2 bits only
[23:05:59] <BBB> actually I was wrong, max runs is 7 not 6
[23:06:03] <BBB> so it goes even further
[23:06:04] <BBB> sorry :)
[23:06:29] <mru> add another level then
[23:06:35] <mru> I think you'll spot the pattern
[23:06:40] <BBB> yes
[23:06:40] <mru> and make it two levels of 7 bits
[23:06:47] <BBB> ok
[23:06:59] <BBB> and what do those bits[] and codes[] resemble?
[23:07:16] <mru> bits is the length in bits of each codes
[23:07:27] <mru> codes is the codes
[23:07:33] <BBB> yes
[23:07:36] <BBB> but how do I use this? :-p
[23:07:48] <BBB> can I do something with them?
[23:08:22] <mru> you feed them to init_vlc()
[23:08:36] <BBB> ah, ok
[23:09:44] * BBB will look into this
[23:32:59] <DonDiego> gnite