[FFmpeg-devel-irc] IRC log for 2010-03-23

irc at mansr.com irc at mansr.com
Wed Mar 24 01:00:21 CET 2010


[00:33:23] <Dark_Shikari> bah, how do I make a script that iterates over files with spaces in their names?
[00:33:43] <mru> depends
[00:33:53] <Dark_Shikari> i.e. for file in ls .... ; type
[00:33:56] <Dark_Shikari> of thing
[00:34:11] <mru> for file in *; do ...; done not working?
[00:34:29] <Dark_Shikari> ah, that works
[00:34:31] <mru> iirc wildcard expansion does the right thing
[00:34:35] <Dark_Shikari> yup, it does.
[00:34:59] <mru> $(command) expansion happens before word splitting though
[00:35:13] <Dark_Shikari> yup.
[00:37:22] <mru> if a simple wildcard didn't match the right files it would be a bit more complicated of course
[02:17:55] <CIA-24> ffmpeg: michael * r22640 /trunk/libavcodec/ (7 files):
[02:17:55] <CIA-24> ffmpeg: Change ref_index structure so it matches how its organized in h264.
[02:17:55] <CIA-24> ffmpeg: Also revert the related error concealment hotfix.
[03:35:52] <CIA-24> ffmpeg: mru * r22641 /trunk/libavcodec/ (fft.h arm/Makefile arm/rdft_neon.S rdft.c arm/fft_init_arm.c): ARM: NEON optimised RDFT
[04:38:20] <peloverde> mru, I noticed the NEON RDFT... any word on how HE-AAC performs on ARM (vs libfaad2)?
[06:28:52] <av500> maybe ffmpeg should at least move to http://svnhub.com/
[06:29:22] <kshishkov> prepare to be moved by KotH there
[06:30:26] <pJok> god morgon kshishkov :)
[06:30:44] <kshishkov> goda morgnar
[06:31:26] * pJok hands kshishkov the view of skåne going by
[06:31:42] * kshishkov goes to webkameror.se
[06:33:36] <thresh> moroning
[06:33:54] <pJok> nothing like the smell of öresundståg brakes in the morning
[06:34:58] <kshishkov> try smell from IC3 diesel engine
[06:35:45] <pJok> they aren't that bad
[06:35:56] <pJok> try the old dsb wagons
[06:36:10] <pJok> the sound of the brakes makes you think that you are still in 1960
[06:36:26] <kshishkov> our railroads still have to reach 1960s
[06:36:52] <pJok> hehe
[06:36:57] <kshishkov> I think heaters there still use wood
[06:37:14] <kshishkov> (better ones use coal though)
[06:37:39] <thresh> aaaaand you cant ask to make the heat lower
[06:38:14] <kshishkov> thresh: wrong country. You can't ask for _anything_
[06:38:23] <av500> thresh: that is a universal fact! :)
[06:38:33] <av500> existing heaters are always too hot
[06:39:07] * av500 loves to watch people fight about opening windows on trains and busses
[06:39:25] <pJok> av500, its easy here... you can't open the windows
[06:39:31] <pJok> at least not on the trains
[06:39:42] <kshishkov> no need to do that either
[06:39:46] <av500> http://german-way.com/blog/2009/02/13/german-phobia-killer-draft/
[06:39:55] <thresh> kshishkov: you can ask for a tea
[06:40:17] <kshishkov> thresh: and hot water is free when available
[06:40:55] <pJok> kshishkov, depends on if you are in the airconned section of the train or not... it made a big difference this summer in the öresundstrains if you were in the non-aircon section  or not
[06:40:57] * kshishkov tries to forget his travels on local railroads and not having nightmares about it
[06:42:00] <thresh> well, moscow-kiev 'branded' train is alrite
[06:42:13] <kshishkov> pJok: I travelled by IC3 and electric version of IC3 last August
[06:42:17] <thresh> moscow-sevastopol otoh is not
[06:42:36] <pJok> IR4
[06:42:47] <pJok> they've been rebuilt, most of them for regional traffic
[06:42:52] <pJok> ie not as comfy anymore
[06:43:15] <kshishkov> you have not seen out trains
[06:43:27] <pJok> no, i haven't
[06:43:32] <kshishkov> lucky you
[06:43:54] <kshishkov> kiev-kharkiv "express" 1st class still gives me nightmares
[06:44:13] <kshishkov> while 2nd class there was merely sucky
[06:45:13] <pJok> there is a 1sst class?
[06:45:26] <kshishkov> it is called so
[06:45:58] <kshishkov> but actually it's worse than 2nd
[06:46:20] <pJok> http://www.kharkovapartmentsforrent.com/images/tourinfo/kharkov_train.jpg
[06:46:55] <kshishkov> yes, that's it
[06:48:12] <pJok> looks about as modern as the old dsb trains
[06:48:34] <thresh> but they're modern
[06:48:44] <thresh> at least in this part of the world, they are still produced
[06:49:01] <kshishkov> cars are produced in Ukraine
[06:49:35] <kshishkov> and you have power outlets only in 1st class cars (and they don't work there either)
[06:50:22] <pJok> http://www.jernbanen.dk/Fotos/Motor/DSB_ME1507_2006.jpg
[06:51:16] <kshishkov> here they haven't grasped an idea of rolling stock not looking like brick
[06:52:23] <pJok> annoyingly i can't seem to find the wagons that those old trains drag along
[06:52:29] <pJok> http://lh3.ggpht.com/_SxGMP8rXRho/RW43J_SNABI/AAAAAAAAArY/-L6Ksd0Ky1k/061004_131751_5220.jpg
[06:58:18] <pJok> http://www.l-eriksen.dk/toge/materiel/andre/blandet/1/3200.jpg
[06:58:25] <pJok> they are blue now, but still as horriblle
[07:01:18] <kshishkov> ours are more horrible
[07:03:05] <kshishkov> http://rupoezd.ru/vagon/platskart.php
[07:05:08] * av500 likes the seat numbering
[07:05:43] <thresh> yeah, is it some kind of quest to buy seats in the middle of a wagon
[07:05:56] <thresh> so you're not buggered by a nearby toilet
[07:06:21] <kshishkov> боковая полка возле туалета - это классика!
[07:06:24] <av500> pick a middle number, like 36 :)
[07:06:36] <thresh> 38 is waay worse :)
[07:06:39] <kshishkov> yes, and you'll be near the toilet for sure
[07:08:12] * kshishkov still gets warm feeling when he remembers SJ Liggvagn
[07:08:23] <pJok> kshishkov, those are quite comfy
[07:08:34] <thresh> that's why I miss a lot of good bike races in Crimea: there is no suitable flight there, and you spend ~24 hours in a train like that. Sum that with my height (i don't quite fit on a sleeping berth') and you will know my feelings.
[07:08:37] * pJok took one from Stockholm to Malmö
[07:09:10] <pJok> seat numbering is a silly international standard
[07:09:11] <kshishkov> pJok: I took it other way around (and Stockholm-Oslo-Stockholmk)
[07:09:32] <pJok> didn't get as much sleep
[07:09:36] <pJok> but thats not unusual
[07:09:42] * thresh has plans on testing french trains this summer
[07:09:42] <pJok> since i sleep badly in general
[07:09:47] <kshishkov> thresh: my feet usually reach sideway bench
[07:10:01] <thresh> we somehow need to get from Andorra to Geneve
[07:10:06] <pJok> the best thing about sleeping while traveling is that your effectiive traveltime is reduced to 0
[07:10:22] <kshishkov> thresh: and while Swedesh are not that high I was able to sleep extremely well in their wagons
[07:10:50] <thresh> kshishkov: well, if you're travelling in a compartment, it is just about ok for me :)
[07:11:00] <kshishkov> thresh: http://upload.wikimedia.org/wikipedia/commons/thumb/c/c4/Liggvagnskupe.jpg/250px-Liggvagnskupe.jpg
[07:11:11] <thresh> ha
[07:12:20] <kshishkov> and I've tried travelling in a coupe carriage here - still got nightmares. They should have _ventilation_ at least
[07:12:58] <pJok> ventilation is overrated
[07:13:04] <thresh> i've only been to Moscow-Kiev train in coupes
[07:13:08] <kshishkov> and still not enough place for feet not mentioning strange metal thingies on the walls that make you fell uncomfortable
[07:13:09] <thresh> s/to/on a/
[07:15:06] <kshishkov> pJok: if you'll ever be my enemy then I'll invite you to try Ukrainian railroads
[07:16:48] <pJok> hehe
[07:18:29] <pJok> i will try not to reach that point
[07:19:23] <kshishkov> another thing is that railroad car toilets here still made from metal
[07:21:59] <pJok> i think they are too in those old dsb wagons
[07:22:45] <votz> thresh: how tall are you?
[07:23:16] <thresh> ~ 190 cm
[07:23:25] <votz> ah nice, getting up there :)
[07:23:40] <kshishkov> thresh: same height here
[07:24:27] <votz> ~200cm
[07:24:58] <kshishkov> pJok: our railroads are on the same quality as automobile roads, so riding our intercity "expresses" (with max speed less than Öresundståg) is quite uncomfortable too
[07:26:38] <pJok> hehe
[07:27:02] <pJok> even though the öresundståg have a top speed of 180kmh, they still have an average speed of around 50kmh
[07:27:35] <kshishkov> express kiev-kharkiv has an average speed ~90 kmh
[07:27:50] <kshishkov> but it makes only two stops on whole 500km route
[07:28:11] <votz> thresh: kshishkov: do you know what percentile 190cm puts you where you guys live?
[07:28:25] <kshishkov> no idea
[07:28:58] <kshishkov> oh, here's another tall guy
[07:29:13] <votz> superdump? :)
[07:29:22] <kshishkov> yep
[07:29:29] <kshishkov> as tall as me IIRC
[07:29:46] * av500 looks down on kshishkov
[07:29:49] <superdump> mmm, about the same i guess, yes
[07:29:53] <votz> av500: how bout you?
[07:29:59] <av500> 195
[07:30:03] <votz> when not on stilts :P
[07:30:45] <av500> there's always hiking boots and wild hair...
[07:30:56] <votz> lots of tall people :D
[07:32:04] <kshishkov> who cares?
[07:33:26] <votz> why a fellow tall person of course
[07:34:34] <kshishkov> in my case it's rather fellow thick person
[07:48:09] <KotH> hoi zämä
[07:48:48] <av500> salut
[07:48:55] <kshishkov> shalom
[07:56:06] <CIA-24> ffmpeg: mstorsjo * r22642 /trunk/libavformat/rtsp.c:
[07:56:06] <CIA-24> ffmpeg: Reconstruct the RTSP URL, in order to remove the auth part from the URL sent to the server
[07:56:06] <CIA-24> ffmpeg: Don't modify the user-specified s->filename at all, keep all modifications
[07:56:06] <CIA-24> ffmpeg: locally and in rt->control_uri.
[08:00:12] <CIA-24> ffmpeg: mstorsjo * r22643 /trunk/libavformat/rtsp.c:
[08:00:12] <CIA-24> ffmpeg: Parse options in the RTSP URL only from the last question mark onwards
[08:00:12] <CIA-24> ffmpeg: This helps if the URL (erroneously?) contains question marks within the path.
[08:01:25] <CIA-24> ffmpeg: mstorsjo * r22644 /trunk/libavformat/rtsp.c: Handle multiple RTSP transport options properly by adding all of them into the mask
[08:30:43] <bcoudurier> hi guys
[08:30:59] <kshishkov> hi script
[08:31:15] <bcoudurier> I'm behind currently :)
[08:31:25] <andoma> :)
[08:31:33] <kshishkov> something went wrong
[08:33:07] <bcoudurier> Dark_Shikari, I retested --threads 0 in lossless mode and in default settings is still slower than ffv1 on my 720p24 source
[08:34:23] <Dark_Shikari> x264's default settings aren't really useful for lossless
[08:34:26] <Dark_Shikari> speed-wise
[08:34:37] <Dark_Shikari> anything slower than --preset faster is basically wasteful
[08:34:46] <Dark_Shikari> gives like 1% total compression at most
[08:34:59] <Dark_Shikari> also, as I said, the compression difference between ffv1 and x264 depends heavily on how noisy the source is
[08:35:06] <Dark_Shikari> if spatial prediction is more useful than temporal, ffv1 will win
[08:37:16] <twnqx> if you want a fast lossless h.264 encoder... there was a 30line C source recently :P
[08:37:46] <kshishkov> you can have 30K line crappy H.264 encoder too
[08:38:05] <twnqx> but that one is FAST
[08:38:21] <Dark_Shikari> what one
[08:38:53] <twnqx> the 30line one :P
[08:39:00] <twnqx> no compression at all etc
[08:39:37] <bcoudurier> humm ffv1 dates 2004
[08:40:07] <Dark_Shikari> twnqx: heh, cpm-olnly?
[08:40:10] <Dark_Shikari> *pcm
[08:40:29] <twnqx> Dark_Shikari: the discussin was in #x264 Oo
[08:40:38] <kierank> he wrote a blog post about it
[08:40:47] <twnqx> two days ago or so
[08:40:55] <Dark_Shikari> I don't read every line of chat >_>
[08:43:23] <twnqx> Dark_Shikari: http://www.cardinalpeak.com/blog/?p=488
[08:45:27] <av500> ...A great exercise to get to know h.264 and encoders in general....
[08:45:30] <av500> ???
[08:45:43] <kshishkov> yes, not a little exercise
[08:48:28] <Dark_Shikari> that's cheating
[08:48:34] <Dark_Shikari> he hardcoded the SPS/PPS ;)
[08:49:14] <bcoudurier> rofl
[08:49:23] <Dark_Shikari> and mb headers
[08:49:26] <Dark_Shikari> and slice headers
[08:49:28] <Dark_Shikari> yeah, that's cheating =p
[08:50:18] <twnqx> ;)
[08:50:22] <twnqx> and the image size!
[08:50:34] <Dark_Shikari> that's implicit from hardcoding the sps
[08:50:59] <twnqx> i was a bit surprised he didn't use the defines in the constants though
[08:52:24] <twnqx> but i like the comment "There you have it—a complete h.264 encoder that uses minimal CPU cycles, with output larger than its input!"
[08:53:08] <kshishkov> and fixed input size
[08:53:35] <twnqx> argh, i lost my flac to alac command :(
[13:45:39] <BBB___> merbzt1: done
[13:45:44] <BBB___> hm, my nick is weird
[13:46:07] <merbzt1> tnx
[13:46:08] <jai> Big Buck Bunny with a tail
[13:49:06] * Terminating due to: TERM
[14:12:29] * /join #ffmpeg-devel ...
[14:12:31] *** 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 0.5.1 has been released! | this chat is now publicly logged.
[14:12:31] *** TOPICINFO: superdump!~rob at unaffiliated/superdump, 1267604523
[14:46:51] * Terminating due to: TERM
[14:48:48] * /join #ffmpeg-devel ...
[14:48:50] *** 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 0.5.1 has been released! | this chat is now publicly logged.
[14:48:50] *** TOPICINFO: superdump!~rob at unaffiliated/superdump, 1267604523
[15:35:17] <BBB> spyfeng: did you send the patch yet?
[15:41:48] <twnqx> broken mov: http://www.biol.uzh.ch/filme/molekulare_zellbiologie/animations/enzyme_catalysis.mov
[15:42:00] <twnqx> [mov,mp4,m4a,3gp,3g2,mj2 @ 0x13cc3c0]Could not find codec parameters (Data: 0x0000, 159 kb/s)
[15:42:26] <twnqx> well, at least not decodable by ffmpeg :P
[15:54:23] <merbzt1> twnqx: what says mplayer ?
[15:54:45] <twnqx> same thing, since he tries LAVF
[15:54:57] <merbzt1> vlc ?
[15:55:07] <twnqx> "command not found"
[15:55:09] <twnqx> :P
[15:55:17] <merbzt1> \o/
[15:55:27] <merbzt1> almost quote worthy
[15:57:49] <twnqx> nah, forget it i'll blame broken file
[15:57:57] <twnqx> just tried quicktime in winblows
[16:10:09] <jishnu7> hello all... i hav a problem after compiling...
[16:11:45] <jishnu7> any thr one to help me..?
[16:11:52] <jishnu7> any one thr to help me..?
[16:14:12] <av500> -> #ffmpeg
[17:23:18] <CIA-24> ffmpeg: cehoyos * r22645 /trunk/libavformat/dv.c:
[17:23:18] <CIA-24> ffmpeg: Fix seeking in DV when filesize is unknown.
[17:23:18] <CIA-24> ffmpeg: Patch by Tomas H?rdin, tomas D hardin A codemill D se
[17:40:41] <CIA-24> ffmpeg: cehoyos * r22646 /trunk/ (doc/ffplay-doc.texi ffplay.c):
[17:40:41] <CIA-24> ffmpeg: Add a loop option to ffplay.
[17:40:42] <CIA-24> ffmpeg: Patch by Robert Kr?ger, krueger signal7 de
[17:55:59] <CIA-24> ffmpeg: diego * r22647 /trunk/libavcodec/h264.c:
[17:55:59] <CIA-24> ffmpeg: Remove unused variable, fixes the warning:
[17:55:59] <CIA-24> ffmpeg: libavcodec/h264.c:1562: warning: unused variable `s'
[17:59:31] <CIA-24> ffmpeg: cehoyos * r22648 /trunk/libavcodec/mpegvideo_enc.c:
[17:59:31] <CIA-24> ffmpeg: Always check if ff_alloc_picture() succeeds.
[17:59:31] <CIA-24> ffmpeg: Patch by Pavel Pavlov, pavel summit-tech ca
[18:20:22] <DonDiego> ramiro: you around?
[18:20:24] <DonDiego> hi guys
[18:20:45] <mru> hi DonDiego
[18:21:53] <DonDiego> siretart: you around?
[18:24:51] <kierank> any idea what a "bitstream key" could refer to?
[18:25:01] <mru> context?
[18:25:05] <kshishkov> encryption or ID
[18:25:10] <mru> some kind of decryption key?
[18:26:13] <kierank> only context I have is a string saying "16-bit Dolby E streams must use bitstream keys. Enabling bitstream keys and continuing."
[18:44:26] <jermy_> are the git and svn repositories kept closely in sync?
[18:45:40] <mru> yes
[18:45:58] <mru> the svn post-commit hook updates the git repos
[18:46:39] <jermy> fair enough. Just wondering if the first thing to do about a dependency issue would be to switch back to svn
[18:46:58] <mru> what's the problem?
[18:46:59] <jermy> (and yes, I realise that asking any more would be more suitable over in #ffmpeg)
[18:47:06] <jermy> "No rule to make target `libavformat/rtp_asf.h', needed by `libavformat/rtpdec.o'.  Stop.
[18:47:23] <mru> seems like a header file was deleted
[18:47:35] <mru> make clean or rm libavformat/rtpdec.d
[18:48:32] <DonDiego> -MP could solve that
[18:48:49] <mru> and break god knows what else
[18:49:01] <mru> these things are fragile
[18:49:17] <jermy> thanks, mru. I'm an idiot for not cleaning first
[18:49:19] <DonDiego> removed headers are a known problem
[18:49:35] <jermy> I assumed a new configure would have fixed the dependencies
[18:49:42] <mru> yes, but the side-effects of the workaround are not known
[18:49:45] <jermy> (to require a new compile)
[18:50:20] <DonDiego> just make a run with -MP and one without
[18:50:27] <DonDiego> compare the .d files
[18:50:27] <mru> not good enough
[18:50:32] <mru> I know what it does
[18:50:40] <mru> I don't know for sure what effect that has on make
[18:50:51] <mru> playing with empty rules is a dangerous game
[18:51:04] <mru> especially when we have headers that actually are generated
[18:51:16] <DonDiego> but these have hardcoded rules
[18:51:25] <DonDiego> as they have to have
[18:51:25] <mru> gcc doesn't know that
[18:52:48] <mru> I've broken things enough times to know I need to be careful
[18:52:58] <mru> and don't see this fixing any big problem
[18:53:09] <mru> headers are rarely deleted
[18:59:08] <BBB> Vitor1001: thanks for the patch, I'll ry to get rid of the +1 offset at the top of the code, that way the yasm code should work well also
[18:59:26] <BBB> might need a few nights of coffee-drinking+screen-staring
[19:00:54] * kshishkov knows some devs who manage to live without coffee
[19:01:21] * mru thought code quality was directly tied to strenght of coffee
[19:02:13] <kshishkov> Benjamin said he does not like coffee
[19:05:03] <kshishkov> so I'd like to see what he can writeafter a mug of strong coffee
[19:05:11] <BBB> I was about to suggest the same thing
[19:05:24] * BBB kicks kshishkov back into wvp2 REing
[19:05:32] * kshishkov resists
[19:05:35] <BBB> hmpf
[19:05:45] * mru throws a pack of coffee beans at kshishkov
[19:05:46] <BBB> can you copy rtmpe into ffmpeg's rtmp?
[19:06:28] <kshishkov> mru: what brand is that?
[19:06:51] <mru> the best
[19:06:55] <kshishkov> rtmpe requires some encrypting
[19:07:38] <BBB> so A) go write encrypting for lavu, or B) link to libssl
[19:07:47] <kshishkov> here coffee is roasted from the best acorns
[19:08:00] <kshishkov> BBB: there's no option B in FFmpeg ;)
[19:08:10] <mru> over here acorns are used to make risc machines
[19:08:21] <BBB> what's so bad about libssl?
[19:08:27] <BBB> everyone uses it already anyway
[19:08:29] <Dark_Shikari> over here squirrels eat them
[19:08:37] <jai> the openssl code is a bit ugly iirc
[19:08:45] <mru> understatement of the day
[19:08:57] <mru> and some distros are afraid of its licence
[19:08:59] <BBB> whereas ffmpeg's code is ultra-clean
[19:09:06] <elenril> isn't libssl incompatible with gpl?
[19:09:08] <jai> comparatively, yes
[19:09:11] <mru> elenril: no
[19:09:12] <BBB> not to mention that all distro's love ffmpeg's licensing "problems"
[19:09:17] <jai> :/
[19:09:32] <BBB> as long as you don't copy ssl code into a gpl project, you're fine
[19:09:34] <BBB> linking is ok
[19:09:42] <kshishkov> and I told we need libavcrypt long time ago
[19:09:47] <BBB> kshishkov: go go go!
[19:09:55] * BBB sends a sixpack coke cola to kshishkov 
[19:10:02] <mru> we already have des and aes
[19:10:12] <kshishkov> BBB: nah, I accept only Trocadero
[19:10:30] <BBB> is that vodka?
[19:10:33] <jai> and mru's patch for a zlib replacement
[19:10:34] <elenril> http://en.wikipedia.org/wiki/OpenSSL#Licensing ?
[19:10:35] <jai> :)
[19:10:54] <mru> as a general rule, wikipedia is wrong
[19:11:00] <mru> especially on legal matters
[19:11:10] <mru> it's written by freetards
[19:11:29] <kshishkov> BBB: no, http://en.wikipedia.org/wiki/Trocadero_(drink)
[19:11:56] <BBB> swedish
[19:11:57] <BBB> should've known
[19:11:58] <kshishkov> probably the only caffeinated stuff I drank last year (not counting tea of course)
[19:13:49] <saintd3v> weeeeeee ffh264 :P
[19:14:12] <kshishkov> and rejected again!
[19:15:06] * saintd3v pokes peloverde
[19:15:15] <peloverde> pong
[19:16:48] <saintd3v> a little birdie said you had some local changes to the aac encoder?
[19:17:24] <kshishkov> to aac decoder too, so what?
[19:18:41] <ramiro> DonDiego: pong
[19:18:41] <peloverde> I do have some local changes, but they would make more sense after some changes I have yet to make but those won't be ready until I finish PS and a paid side project
[19:19:12] <saintd3v> peloverde: exactly what I was wondering, thanks :D
[19:19:44] <DonDiego> ramiro: i have a fix for the altivec vs. runtime-cpudetect failure, can you review?
[19:20:20] <ramiro> I thought you had fixed that already...
[19:20:28] <ramiro> I can take a look
[19:23:58] <DonDiego> ramiro: http://www1.mplayerhq.hu/~diego/altivec_runtime.diff
[19:40:30] <BBB> bcoudurier: that was a joke...
[19:41:00] <BBB> DonDiego: gives me an empty document
[19:43:28] <kshishkov> BBB: you may be gravely wrong about it beong a joke
[19:43:34] <kshishkov> *being
[19:43:39] <BBB> potentially
[19:44:44] <DonDiego> BBB: the patch i posted?
[19:44:53] <DonDiego> it's definitely not empty...
[19:45:04] <Vitor1001> BBB: sorry was away from the keyboard
[19:45:20] <BBB> oh wait, firefox downloads it instead of displaying it
[19:45:28] <BBB> isn't .diff served as text/plain by mphq?
[19:45:57] <DonDiego> no, as text/x-diff
[19:46:05] <DonDiego> firefox is stupid for not displaying it inline
[19:46:13] <kshishkov> maybe as original mimetype?
[19:46:13] <BBB> Vitor1001: is ok, I'll try to 16-align all calls to rdft() over the next few days, thanks for the patch
[19:46:17] <BBB> Vitor1001: should be possible
[19:46:18] <Vitor1001> BBB: If you have any problem with that DCT stuff, feel free to ask me
[19:46:20] <mru> text/* means it's, guess what, text...
[19:46:22] <BBB> sure
[19:46:53] <mru> there are unaligned rdft calls?
[19:46:54] <DonDiego> safari does it
[19:47:09] <DonDiego> is there a way to override such stupidity in firefox?
[19:47:44] <mru> yes, use a different browser
[19:47:46] <kshishkov> patch the source
[19:47:51] <DonDiego> ramiro: reping
[19:47:58] <mru> the android browser displays that url btw
[19:48:02] <DonDiego> mru: well, what do you use?
[19:48:39] <mru> still firefox
[19:48:48] <mru> chrome insists on having dbus installed
[19:48:51] <mru> and I'm not that stupid
[19:49:05] <CIA-24> ffmpeg: vitor * r22649 /trunk/ (5 files in 2 dirs): Implement the discrete sine/cosine transforms DCT-I and DST-I
[19:49:25] <kshishkov> ELinks displays attachments without any fuss
[19:49:36] <kshishkov> and loads faster that Firefox on my hw too
[19:50:03] <elenril> kshishkov: http://tvtropes.org/pmwiki/pmwiki.php/Main/CaptainObvious
[19:50:37] <kshishkov> elenril: that link reflects its contents
[19:55:01] <ramiro> DonDiego: what was the configure command that failed?
[19:57:18] <DonDiego> --enable-runtime-cpudetect --disable-altivec
[20:04:11] <CIA-24> ffmpeg: vitor * r22650 /trunk/ (libavcodec/avcodec.h doc/APIchanges): Bump minor version, I forgot in the last commit
[20:06:48] <ramiro> DonDiego: your patch seems correct to me.
[20:07:35] <ramiro> actually I don't even think it needs ARCH_PPC anymore, since HAVE_ALTIVEC implies that.
[20:09:24] <DonDiego> right..
[20:25:05] <CIA-24> libswscale: diego * r30952 /trunk/libswscale/utils.c:
[20:25:05] <CIA-24> libswscale: Do not compile AltiVec code when AltiVec is not available; not even in
[20:25:05] <CIA-24> libswscale: runtime cpudetection mode.
[20:25:05] <CIA-24> libswscale: Fixes compilation with '--enable-runtime-cpudetect --disable-altivec'.
[20:27:13] <Vitor1001> DonDiego: Does it compiles now on PPC with "--disable-optimizations"?
[20:36:21] <DonDiego> dunno
[20:36:24] <DonDiego> i can try..
[20:37:08] * DonDiego starts a compile
[21:21:38] * BBB needs a good gpg howto
[21:21:43] <BBB> or pgp
[21:21:47] <BBB> what's that stuff called again?
[21:23:59] <mru> man gpg
[21:31:09] <j-b> don't read it, you will never understand it and you'll save some time
[21:33:08] <BBB> I see firepgp supports gmail
[21:33:10] <BBB> hmmm
[21:33:17] <mru> I don't trust crypto/security systems I can't understand
[21:33:41] <mru> selinux being a good example
[21:33:50] <mru> that's totally fucking incomprehensible
[21:34:18] * elenril heard a few people saying it's NotThatBad
[21:34:34] <elenril> tomoyo is quite nice though
[21:35:47] <BBB> people around me keep telling me the dutch tax system is NotThatBad
[21:38:17] * elenril kicks spherical coordinates
[21:41:42] <mru> BBB: is it like the swedish system?
[21:41:49] <mru> it works like this:
[21:41:54] <BBB> that's the one that is worse
[21:41:57] <mru> 1. How much money did you make last year?
[21:42:04] <mru> 2. How much have you got left?
[21:42:08] <mru> 3. Send it in.
[21:42:11] <Dark_Shikari> lol
[21:42:14] <BBB> lol :)
[21:42:19] <BBB> dutch system is like this:
[21:42:27] <BBB> how good is your accountant?
[21:42:38] <mru> that's universal
[21:42:41] <BBB> take that number as a percentage of your total pre-tax income and that's your return
[21:42:42] <Dark_Shikari> american system is like this:
[21:42:50] <Dark_Shikari> 1) send us all the money in your right pocket.
[21:42:58] <Dark_Shikari> 2)  are you self-employeed?  take everything from your left pocket too, we want that.
[21:43:17] <Dark_Shikari> 3) make a list of your business expenses and deduct them.  But 90% of the things on your list don't actually count.
[21:43:36] <Dark_Shikari> 4) You didn't send us enough.
[21:43:39] <mru> here's mine:
[21:43:47] <mru> 1. Get paid into an offshore account.
[21:43:52] <mru> 2. Don't tell anyone.
[21:43:55] <Dark_Shikari> 2. get audited by the IRS
[21:44:11] <mru> so tell them about enough to keep them happy
[21:45:29] <thresh> in Russia it's 1. Nobody cares 2. If they do, it doesn't matter if you paid it right.
[21:48:22] <BBB> uhm...
[21:48:37] * BBB thinks this thing is confusing
[21:49:11] <mru> BBB: if you find it confusing you're not making enough money
[21:49:33] <BBB> no, pgp is confusing
[21:49:37] <BBB> my taxes are just annoying
[21:49:42] <kierank> all countries have an extra step. If you earn above XXX taxes are optional
[21:49:50] <BBB> it says "an unknown error occurred, this cannot be fixed right now"
[21:49:52] <BBB> o so
[21:50:12] <BBB> as in, thanks, that was incredibly useful and a true testimony to the quality of free software in general and firepgp specifically
[21:50:21] <mru> BBB: if you earn enough, that's what the taxman says too
[21:50:28] <BBB> :)
[22:00:06] * BBB frustrated
[22:00:09] * BBB disables firepgp again
[22:00:21] <BBB> this is yet another piece of shit software like only freetards can make
[22:00:24] <BBB> I cannot believe this
[22:00:44] <BBB> don't people go to design school before they try to expose something to end users?
[22:00:51] <mru> no
[22:03:28] * BBB feels like shooting the firepgp developer
[22:10:25] <jermy> So... what's the usual process for sponsoring some ffmpeg development? Contact a likely developer directly, or announce centrally?
[22:11:07] <peloverde> jermy, usually sending an inquiry to the list unless you have a specific developer in mind
[22:11:40] <peloverde> A few developers are listed at http://ffmpeg.org/consulting.html
[22:13:04] <jermy> That'd suggest Baptiste a good person to talk to, then
[22:13:24] <jermy> bcoudurier: Likely to be at NAB?
[22:17:03] <Dark_Shikari> depends on what you want done basically
[22:17:51] <jermy> Well, it's AVC-Intra - and I'm sure it's been mentioned a few times
[22:18:05] <mru> what's that?
[22:18:14] <mru> another bastardised avc variant?
[22:18:47] <jermy> Pretty much. Intra-only, and 10bit at either 50 or 100Mbit
[22:19:04] <mru> what? fixed frame size?
[22:19:23] <jermy> Yeah
[22:19:27] * mru looks at dnxhd
[22:19:53] <mru> and you want to decode that stuff?
[22:19:55] <mru> or encode it?
[22:19:58] <jermy> There's a veritable clusterfsck of these things. Also check out Apple's Pro-Res
[22:20:19] <mru> fixed frame size does make some things simpler
[22:20:21] <jermy> decode only, so we don't mind throwing away two bits if ffmpeg will only otherwise support 8-bit
[22:20:25] <mru> at the expense of quality
[22:20:43] <BBB> I think ffmpeg already does 10bit formats, no?
[22:20:45] <jermy> Very easy to write to tape :)
[22:20:53] <mru> are the frames normal avc otherwise?
[22:21:00] <jermy> Oh, sorry then, my misunderstanding
[22:21:31] <mru> does ffh264 do 10-bit?
[22:22:02] <jermy> I _think_ it's just pushing quite a few parameters to the limits of what would be considered usual, and using a few extra features not normally seen
[22:24:06] <mru> if it's all in the spec we should support it
[22:24:13] <mru> except probably the 10-bit part
[22:24:25] <jermy> I'd probably have to try to work out what's currently missing
[22:24:40] <mru> what happens if you try to decode such a file?
[22:24:41] <jermy> There's 3D transforms in the spec - is that supported?
[22:24:56] <jermy> gimme one second - I'm just building current head
[22:25:02] <mru> in the h264 spec?
[22:25:06] <mru> I don't remember seeing those
[22:25:12] <mru> mpeg4-2 sure
[22:25:20] <mru> it has everything and then some
[22:26:08] <kierank> is there an avc intra spec?
[22:26:24] <BBB> didn't x264 already support that?
[22:26:31] <BBB> if so, there must be
[22:26:31] <BBB> no
[22:26:32] <BBB> ?
[22:27:00] <mru> x264 supports fixed frame size encoding
[22:27:03] <mru> iirc
[22:27:19] <mru> I don't recall anything about that being per some specific spec
[22:27:50] * BBB goes home
[22:30:05] <Dark_Shikari> we have soc projects for both 10-bit and 4:2:2...
[22:30:09] <Dark_Shikari> necessary for avc intra
[22:30:29] <jermy> I'm currently getting http://jeremy.publication.org.uk/ffmpeg_avcidecode
[22:30:58] <Dark_Shikari> or do you mean decoding as opposed to encoding
[22:31:09] <jermy> just decoding for us
[22:31:20] <Dark_Shikari> that would actually be really good
[22:31:26] <Dark_Shikari> it would reduce the amount of work we would have to do
[22:32:10] <mru> looks like the mxf demuxer isn't recognising the stream properly
[22:32:54] <peloverde> mxf was recently failing regressions on some configs, did that get fixed?
[22:32:58] <jermy> Hrm. I probably have an old version that has some MXF fixes on it
[22:34:59] <peloverde> MXF fails on x86_32 / Linux / gcc svn 156187
[22:38:10] <jermy> More fun - this is attempting to decode an extracted stream from the file directly: http://jeremy.publication.org.uk/ffmpeg_avcidecode_stream
[22:38:34] <kierank> jermy: does jm decode it?
[22:39:27] <jermy> Yes, I think it does
[22:39:39] <jermy> but I'll build and test it quickly
[22:42:24] <kurosu> Could someone tell me if Ramiro Polla ever joins this IRC channel, and if yes, under what nickname?
[22:43:30] <DonDiego> kurosu: ramiro has the nickname ramiro ...
[22:44:08] <kurosu> thanks
[22:49:14] <jermy> Yup, JM looks happy with it
[22:49:37] <jermy> slow, but happy (that might be to do with writing out the raw YUV to a NFS share, though)
[22:52:43] <mru> no, that's just jm being jm
[23:06:03] <jermy> Although clearly I'm not having a very good attempt at converting from JM's output - http://jeremy.publication.org.uk/00014F.avi (450K, resized to 320x180)
[23:36:27] <CIA-24> ffmpeg: ramiro * r22651 /trunk/libavdevice/x11grab.c:
[23:36:27] <CIA-24> ffmpeg: x11grab: AVFormatParameters are always passed from av_open_input_stream(),
[23:36:27] <CIA-24> ffmpeg: there is no need to check for it.
[23:47:14] <j-b> peloverde: did you find the fix?
[23:48:30] <j-b> jermy: you have AVC-Intra files? From Panasonic cameras?
[23:59:44] <CIA-24> ffmpeg: jbr * r22652 /trunk/libavformat/ (Makefile flacenc.h flacenc_header.c):
[23:59:44] <CIA-24> ffmpeg: Put ff_flac_write_header() in a separate C file to allow it to be shared without
[23:59:44] <CIA-24> ffmpeg: duplicating code or adding a dependency on vorbiscomment.o.


More information about the FFmpeg-devel-irc mailing list