[Ffmpeg-devel-irc] ffmpeg-devel.log.20141009

[burek]

[00:16] <rcombs> ubitux: so, there's still no way to get MKV-style data from avcodec_decode_subtitle2, then?
[01:03] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:20a5956b8dae: dump: split audio and video probing on multiple lines
[01:03] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:f22f873ef013: Merge commit '20a5956b8daeee4cb59d6fa00ec809b02c04d7f8'
[01:03] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:e051d09c58c6: avcodec: Allow choosing the dump format field separator.
[01:03] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:f0a6874de85f: avformat: Allow choosing the dump format field separator.
[01:45] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:41e86146e724: dump: print detailed color space information
[01:45] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:af2896041244: Merge commit '41e86146e7240f9d42e8cdb86a94bb3eb2bbe15f'
[01:56] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:9f3a70c44224: dump: print the original coded dimensions when available
[01:56] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:5f43a7957b70: Merge commit '9f3a70c44224ed92d91df65bb9470a2649e7d019'
[02:13] <cone-470> ffmpeg.git 03Sylvain Rabot 07master:167e6f148936: dump: Update streams and chapters printout format
[02:13] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:04d8f616fddb: Merge commit '167e6f1489368a4693145da61cb022c113a285bc'
[02:22] <cone-470> ffmpeg.git 03Marc-Antoine Arnaud 07master:9d5fed6dc4cc: avcodec: fix descriptions on color transfer options
[02:22] <cone-470> ffmpeg.git 03Marc-Antoine Arnaud 07master:91eef402037a: avcodec: add enum values on chroma sample location
[02:22] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:7652b176fcb3: Merge commit '9d5fed6dc4cccb0689f694112f5026988bebc472'
[02:22] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:923a242282c1: Merge commit '91eef402037ae5a51d89ade6cb71cdd70b291c8b'
[02:36] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:586ea0379c20: avcodec: fix colorspace option description
[02:36] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:0d8a3656ba4c: mov: support the Color Parameter Atoms 'colr'
[02:36] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:6414c3ecd065: Merge commit '586ea0379c207a942c1f2d61e0b3a63c458cc598'
[02:36] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:349a32c436d9: Merge commit '0d8a3656ba4c0ae8e4e0c91ff6b07d72a317f9f6'
[02:45] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:65875a8b3b07: vp8: initialize color space and range properties
[02:45] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:3bc86462a2ae: Merge commit '65875a8b3b079752da25a61ec188d2e3d90a569f'
[02:54] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:eddf95283fbb: vp9: initialize color space and range properties
[02:54] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:5b134a0544eb: Merge commit 'eddf95283fbba1b469b76e7518bc3941f671b193'
[03:11] <cone-470> ffmpeg.git 03Vittorio Giovara 07master:1a02e78653a5: vc1: initialize color properties
[03:11] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:f6777ce6fe7c: Merge commit '1a02e78653a5c7674e6c43b6e6f2748dbd2091d7'
[04:26] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:cf7b90dea219: fix seperator typo
[05:42] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:f9a13174794a: avcodec/utils: Add missing AV_PIX_FMT_YUVJ411P to color_range override code.
[05:42] <cone-470> ffmpeg.git 03Michael Niedermayer 07master:057862303904: avformat/dump: use format dump_separator if the codec one is not set
[07:19] <ubitux> rcombs: i'm still working on it
[07:19] <ubitux> rcombs: problem is that i have to update our weird parser so i'm not breaking the subtitles encoders
[07:20] <rcombs> alright, gotcha
[07:33] <jamrial> so many OPW applicants but only one slot...
[08:02] <hima> ubitux: while making changes in my code for katedec.c how can i see where the error exactly is. using this ffmpeg -i in.kate out.srt is as follows
[08:02] <hima> http://fpaste.org/140536/28345111/
[08:16] <jamrial> that error happens when ffmpeg couldn't detect the format of the file. the probe function may be wrong in some way (assuming the demuxer is properly registered in allformats.c and such)
[08:17] <jamrial> use av_log() in different parts of your code to see where it failed
[08:18] <hima> ohh thanks alot :) i will check right now
[08:18] <jamrial> do you get the same error if you force the kate demuxer?
[08:18] <hima> force the kate demuxer i did not get it?
[08:18] <jamrial> something like "./ffmpeg -f kate -i sample_kate.kate out.srt"
[08:19] <hima> yes
[08:19] <hima> says unknown input for kate 
[08:20] <jamrial> unknown input format: 'kate'? if so, the demuxer isn't properly registered
[08:21] <jamrial> you need to add the usual stuff to allformats.c, Makefile and such
[08:21] <hima> yes i have done that 
[08:21] <hima> after configuring it added a file katedec.c
[08:22] <hima> though i see katedec.o missing 
[08:22] <hima> like all other demuxers
[08:23] <hima> *katedec.d
[08:23] <hima> i meant
[08:24] <hima> in all formats added    REGISTER_DEMUXER (KATE,             kate);
[08:25] <hima> in Makefile   OBJS-$(CONFIG_KATE_DEMUXER)              += katedec.o subtitles.o
[08:25] <hima> is it right?
[08:26] <jamrial> yes, it should work with that
[08:27] <hima> my katedec.o file is missing actually. how is that created ?
[08:27] <jamrial> that line in libavformat/Makefile should take care of that
[08:28] <hima> umm then what is going wrong according to you?
[08:29] <jamrial> check config.mak in your build folder, search for CONFIG_KATE_DEMUXER
[08:31] <hima> build folder under ffmpeg folder?
[08:31] <jamrial> the folder where you ran configure. it can be the actual ffmpeg source folder, or some other outside of it
[08:36] <hima> yes its there      CONFIG_KATE_DEMUXER=yes
[08:37] <hima> can that error arise if there is error in my katedec.c? cuz i am unsure of my code there are errord
[08:38] <hima> *errors
[08:38] <jamrial> yes, but make would have stopped there
[08:38] <jamrial> what do you get if you run "make libavformat/katedec.o"?
[08:38] <jamrial> ok, hold on, that paste above says "built on Oct  4 2014"
[08:39] <jamrial> you're running a very old ffmpeg build
[08:40] <jamrial> i'm guessing make is failing at katedec.o and you still have this old ffmpeg binary you successfully built five days ago
[08:42] <hima> make libavformat/katedec.o shows my code errors though  i need to check them and i did not understand the second part. old ffmpeg build?
[08:45] <jamrial> make is failing at katedec.o, so the ffmpeg binary is not being built (that only happens when every object is sucessfully compiled)
[08:45] <jamrial> the ffmpeg binary you ran in the paste above is old. one you successfully compiled about five days ago before you added the kate demuxer code, most assuredly
[08:46] <hima> yeah okay that might be i will check 
[08:46] <hima> thanks 
[08:47] <jamrial> no problem
[09:02] <prachi> @ubitux sir you asked me to implement a demuxer on a file.Actually i am not able to open that file.Can you please help me?
[09:05] <ubitux> prachi: open that file? what do you mean?
[09:10] <prachi> Whoops.
[09:10] <prachi> you seem to have missed the gist of that gist you were looking for
[09:10] <prachi> sir,this i9s the message i am getting
[09:10] <prachi> I am new.I don't know what this exactly means
[09:10] <prachi> is*
[09:16] <prachi> https://gist.githubusercontent.com/11rcombs/ece7488b06e9a73a1057/raw/00b8c37773164861f47e403a42ee160
[09:17] <prachi> this is the link i am talking about
[09:45] <ubitux> prachi: your link is incomplete
[09:45] <ubitux> https://gist.githubusercontent.com/11rcombs/ece7488b06e9a73a1057/raw/00b8c37773164861f47e403a42ee16081cbb9908/gistfile1.txt
[09:45] <ubitux> you truncated it somehow
[09:46] <prachi> okay sir, thanks
[09:47] <prachi> So,i am supposed to build a demuxer for this file?
[09:48] <ubitux> you can use libavformat/vplayerdec.c as a template
[09:49] <ubitux> except you will output AV_CODEC_ID_TEXT instead of AV_CODEC_ID_VPLAYER
[09:49] <ubitux> you need to add an entry in libavformat/Makefile
[09:49] <ubitux> and libavformat/allformats.c
[09:49] <ubitux> then run the configure again, and make
[09:50] <ubitux> the ff_smil_* functions declared in libavformat/subtitles.h will help you to parse that format
[09:51] <prachi> okay sir,Thanks
[10:30] <benoit-> hi, is there any issue with the git server right now? I fail to fetch, and want to be sure it's on our side before pinging the IT
[10:31] <ubitux> works for me
[10:31] <ubitux> i just pulled stuff
[10:31] <ubitux> > NTSC widescreen 240p 427x240
[10:31] <ubitux> how do you even encode in such width?
[10:36] <nevcairiel> carefully?
[12:14] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:8cfb9ccfef5f: avcodec/utils: Fix "warning: pointer type mismatch in conditional expression"
[12:14] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:5d36e8527f08: avformat/mov: Fix 'warning: format specifies type unsigned short but the argument has type int'
[12:33] <cone-314> ffmpeg.git 03Mika Raento 07master:2bc220958c75: ismindex: handle discontinuous streams better
[12:33] <cone-314> ffmpeg.git 03Mika Raento 07master:d3834f290138: ismindex: handle discontinuous streams better
[12:33] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:7b973e734111: Merge commit '2bc220958c7570020d87ee353db65460b518998a'
[13:09] <Jul13t> t> Is Yasm really necessary in the compilation of ffmpeg? would ffmpeg work appropriately with
[13:10] <Jul13t> --disable-asm?
[13:10] <Daemon404> it would work, but it would be very slow.
[13:10] <Daemon404> it is not recommended.
[13:13] <Jul13t> ok, x264 was not finding the --bindir=/ for Yasm
[13:13] <Jul13t> so was facing some compilation probs
[13:15] <Jul13t> so installed it directly with the yum mgr
[13:19] <Jul13t> Daemon404, I also wish to know If i'm to complete my qualification task before the last date of OPW applications, since I was planning on finishing my proposal and give mentors for early review before final submission and then continue working on my qualification task thereafter
[13:19] <J_Darnley> I am always surprised at the people who think the message printed by configure is somehow overblown.
[13:27] <Daemon404> --bindir does not tell a configure script where to look for binaries
[13:27] <Daemon404> it tells it where to install them
[13:30] <Jul13t> Daemon404, So how do I make ./configure check for dependencies in the $home/bin dir? using prefix?
[13:32] <cone-314> ffmpeg.git 03Yu Xiaolei 07master:d5a6f1127263: lavf: fix 2GB file seek limit on Android
[13:35] <pa> Daemon404, did you work on the mkv demuxer?
[13:37] <Daemon404> Jul13t, set your PATH
[13:37] <Daemon404> pa, no, why
[13:37] <Jul13t> Daemon404, did that already ;) works
[13:37] <pa> Daemon404, i remembered (wrong) that someone said you did :)
[13:38] <Daemon404> if someone said that about me, is should sue them for slander
[13:38] <Daemon404> s/is/i/
[13:39] <pa> dont worry, iam sure wrong
[13:48] <hima> ubitux: hi 
[13:54] <wm4> android sure is terrible
[13:55] <wm4> not only because they didn't support LFS properly, but also because they implemented LFS in the first place (it's an ancient compatibility hack)
[13:57] <J_Darnley> linux from scratch?
[13:58] <nevcairiel> linux filesystem standard?
[13:58] <wm4> large file support
[13:58] <wm4> I was commenting on "lavf: fix 2GB file seek limit on Android"
[13:58] <J_Darnley> oh
[13:59] <ubitux> hima: hi, i'm here now
[14:00] <j-b> of course, this patch is wrong
[14:00] <j-b> or at least, very limited
[14:01] <j-b> changing lseek without modifying off_t...
[14:01] <nevcairiel> i dont think lavf uses these types
[14:01] <j-b> sure...
[14:01] <j-b>        off_t lseek(int fildes, off_t offset, int whence);
[14:02] <j-b> from the man page
[14:02] <hima> umm i did successfully did make libavformat/vplayerdec.o. and then i get this 
[14:02] <hima> ubitux  make libavformat/vplayerdec.o
[14:02] <j-b> so it is wrong...
[14:02] <hima> http://fpaste.org/140593/85608314/
[14:02] <hima> segmentation fault :(
[14:02] <wm4> j-b: I assume libav* just uses int64_t everywhere
[14:03] <nevcairiel> if lavf doesnt use off_t for some variable or something, the define will automatically work
[14:03] <ubitux> hima: rebuilding your .o is not enough, just use make ffmpeg
[14:03] <nevcairiel> since lseek64 uses off64_t 
[14:03] <ubitux> hima: i can't guess where you segfault is; use gdb or valgrind
[14:03] <wm4> libavutil/file.c:56:    off_t off_size;
[14:03] <wm4> oops.
[14:04] <nevcairiel> av_file_map is evil anyway!
[14:04] <hima> make ffmpeg works fine too
[14:04] <hima> http://fpaste.org/140594/56268141/
[14:04] <Daemon404> [11:01] [psy]  -!- Mode change [-r] for user D404|Sleep
[14:04] <nevcairiel> and like only used in lavfi
[14:04] <Daemon404> ffff
[14:04] <Daemon404> woops
[14:05] <j-b> wm4: no idea
[14:05] <hima> i did not understand gdb and valgrind?
[14:05] <ubitux> hima: valgrind ./ffmpeg_g -f kate -i sample_kate.kate out.srt
[14:05] <ubitux> hima: or gdb --args ./ffmpeg_g -f kate -i sample_kate.kate out.srt
[14:05] <ubitux> and then "r" to run
[14:06] <j-b> anyway, the fix is simple
[14:06] <j-b> if you actually read the manpage
[14:08] <hima> ubitux http://fpaste.org/140595/28565111/
[14:08] <ubitux> i told you ./ffmpeg_g
[14:09] <ubitux> ffmpeg_g is the version with the symbols
[14:09] <ubitux> it will give you more insight
[14:09] <hima> oh sorry i will do it
[14:13] <hima> ubitux http://fpaste.org/140598/41285669/ seems lots of errors :|
[14:13] <ubitux> well then you know where the problem is then
[14:14] <hima> yeah. okay i will make the required changes 
[14:14] <hima> thanks 
[14:18] <hima1> Ubitux: Is there a last date for the submission of the qualification task?
[14:18] <ubitux> isn't it in less than 2 weeks?
[14:19] <ubitux> i'm going to read that opw doc once and for all in a few hours i guess
[14:19] <ubitux> need to find some time..
[14:19] <hima> Um i did not read it anywhere. So asked
[14:19] <hima> Okay 
[14:20] <ubitux> or maybe i was mistaken with the application deadline
[14:20] <ubitux> check the doc, i probably know less about opw than you
[14:20] <hima> Will ask my doubts tonight then . there is an application deadline
[14:20] <Jul13t> ubitux, that's what I was asking
[14:20] <hima> That is 22nd oct
[14:21] <hima> Ok
[14:21] <Daemon404> 1/g 8
[14:59] <J_Darnley> Yay!  I recompile with libfontconfig thinking I wouldn't have to specify font options but that still doesn't let drawtext work.
[15:00] <ubitux> really?
[15:00] <ubitux> it should
[15:01] <J_Darnley> I probably haven't configed fontconfig.
[15:01] <J_Darnley> or it is just cygwin's fault
[15:05] <J_Darnley> Yay!  text=metadata=title doesn't work anyway.
[15:45] <cone-314> ffmpeg.git 03Benoit Fouet 07master:7cda31cb924c: avformat/mov: E-AC-3 streams need parsing too.
[17:09] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:f3b54ee98993: avutil/Makefile: add softfloat to TESTPROGS
[17:09] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:44198a728d15: avutil/softfloat: Fix dependence on signed overflow in av_normalize1_sf()
[17:09] <cone-314> ffmpeg.git 03Michael Niedermayer 07master:99afec08f8a1: avutil/softfloat: add some asserts
[18:26] <cone-314> ffmpeg.git 03James Almer 07master:0c0694931442: compat/w32pthreads: use the CONDITION_VARIABLE typedef if available
[18:35] <anshul_mahe> I am writing closed caption decoder, will it be in ASS category
[18:35] <wm4> anshul_mahe: currently all sub "decoders" output ASS, so maybe yes
[18:39] <anshul_mahe> ok
[18:55] <ubitux> anshul_mahe: if it's text, yes, use ass for the markup
[20:04] <anshul_mahe> ok
[20:11] <J_Darnley> Damn!  drawtext didn't like my em-dashes
[20:12] <anshul_mahe> I have added my decoder file to ffmpeg, but looks like make does not like it
[20:13] <anshul_mahe> is there any way to increase verbosity level in ffmpeg make
[20:13] <J_Darnley> Oh, that's because it was masqerading as ansi
[20:13] <J_Darnley> anshul_mahe: gcc should be printing errors
[20:13] <J_Darnley> but you can use V=1 to see the commands make is running
[20:14] <anshul_mahe> it looks like i have broken makefile http://pastebin.com/fXJ6iTNT
[20:16] <anshul_mahe> I does not show anything, same behaviour
[20:16] <J_Darnley> post the changes you have made to the makefile
[20:16] <anshul_mahe> ohh I got it, I added .c file in makefile instead of .o
[20:21] <J_Darnley> To people familiar with ffmpeg's expression evaluator, how can I do this: or(lt(t,5),gt(t,214))
[20:21] <J_Darnley> There is no or function.
[20:21] <J_Darnley> wait, I should ask that on #ffmpeg
[20:56] <ubitux> J_Darnley: with +
[20:56] <ubitux> but you can use between
[20:56] <ubitux> between(t,5,214)
[20:57] <ubitux> not(between(t,5,214))
[20:57] <ubitux> actually
[20:57] <ubitux> or lt(t,5)+gt(t,214) if you prefer
[21:33] <J_Darnley> ubitux, thanks, I got the + suggestion on #ffmpeg
[21:34] <ubitux> not(between()) is more readable
[21:34] <ubitux> :P
[21:34] <J_Darnley> Ture
[21:34] <J_Darnley> True
[21:34] <wm4> when will this eval stuff support strings?
[21:34] <wm4> I could need it
[21:35] <J_Darnley> I still had flashbacks of my time learning Winamp's ATF expressions
[21:35] <J_Darnley> wm4: when we replace it with a proper scripting language?
[21:49] <J_Darnley> BTW, would anyone like to see the first nicely-rendered output of the vis I've been working on?
[21:52] <ubitux> of course
[21:53] <J_Darnley> Are you okay with a magnet link, or would you prefer a plain http download?
[21:53] <wm4> I want to use lavfi visualization stuff in mpv eventually, but for now it's too much of a PITA
[21:53] <ubitux> magnet is a pain, i need a few indirection; my client is elsewhere
[21:56] Action: J_Darnley curses this 50M limit
[22:08] <cone-314> ffmpeg.git 03Frank Galligan 07master:241b306b1eaa: Fix writing first audio Cues in dash mode.
[22:10] <J_Darnley> finally: http://users.telenet.be/darnley/pachelbel.mp4
[22:11] <J_Darnley> (34 MiB)
[22:11] <J_Darnley> ubitux ^^
[22:12] <ubitux> canon~
[22:14] <J_Darnley> No, it isn't that
[22:15] <ubitux> oh? ok
[22:15] <J_Darnley> The author just thought that he managed to recreate that pachelbel canon riff/melody/sound/(insert correct term)
[22:16] <ubitux> heh mpv doesn't play it
[22:16] <ubitux> but ffplay does
[22:16] <ubitux> oh well, probably my video driver shit
[22:16] <ubitux> whatever
[22:16] <ubitux> oh wow
[22:17] <ubitux> that's cool!
[22:17] <J_Darnley> maybe I shouldn't have used fullrange yuv444
[22:18] <ubitux> so is this written in lua?
[22:18] <ubitux> or the scripting is with eval?
[22:18] <J_Darnley> Just the scripting
[22:18] <ubitux> yeah sure the scripting
[22:18] <ubitux> in lua then?
[22:19] <J_Darnley> Lua is replacing the original expression code
[22:19] <ubitux> how much code does all of this represents?
[22:20] <J_Darnley> This uses only a few of the total number of components, but it does use the most popular one (Superscope)
[22:21] <J_Darnley> It is that whch draws the shape
[22:21] <ubitux> it's really nice
[22:21] <ubitux> how long before you submit?
[22:21] <J_Darnley> other than that there is a crappy blur, and a fade out
[22:22] <J_Darnley> I could submit my patch to ffmpeg at almost any time
[22:22] <ubitux> (mirrored @ http://lucy.pkh.me/pachelbel.mp4)
[22:22] <J_Darnley> I haven't made any significant changes to the filter in a while.
[22:24] <ubitux> so will it depend on a lib or something?
[22:26] <J_Darnley> Yes, I didn't think I should try to add 14k (at the moment) lines to ffmpeg
[22:26] <ubitux> :)
[22:27] <ubitux> as long as it's easily packageable&
[22:27] <ubitux> x-deps to ffmpeg or not at all?
[22:27] <ubitux> (like for fft or stuff like that)
[22:28] <J_Darnley> I do have one at the moment (av_get_cpu_flags) but I'm really thinking about dropping it.
[22:28] <ubitux> ok
[22:28] <J_Darnley> fft is done in the lavfi filter
[22:28] <ubitux> oh, ok
[22:53] <cone-314> ffmpeg.git 03James Almer 07master:8c6992bf0cc2: compat/w32pthreads: use the condition variable API directly when targeting newer versions of Windows
[00:00] --- Fri Oct 10 2014