[Ffmpeg-devel-irc] ffmpeg-devel.log.20150412

[burek]

[00:22:39 CEST] <jamrial> michaelni: ok, thanks
[00:22:59 CEST] <cone-750> ffmpeg 03James Almer 07release/2.6:c1c245e1a347: doc: add missing x86 cpuflags to fftools documentation
[00:23:00 CEST] <cone-750> ffmpeg 03James Almer 07release/2.6:6f236d37746e: doc: add missing arm cpuflags to fftools documentation
[00:23:01 CEST] <cone-750> ffmpeg 03James Almer 07release/2.6:04fd0250e1fd: doc: add aarch64 cpuflags to fftools documentation
[00:29:16 CEST] <cone-750> ffmpeg 03Timothy Gu 07master:a202fe00ab9d: fate: Add atrac3+ tests
[00:58:29 CEST] <cone-750> ffmpeg 03James Almer 07release/2.2:57f271ce4149: doc: add missing x86 cpuflags to fftools documentation
[00:58:30 CEST] <cone-750> ffmpeg 03James Almer 07release/2.2:733cb1c6c59d: doc: add aarch64 cpuflags to fftools documentation
[00:58:31 CEST] <cone-750> ffmpeg 03James Almer 07release/2.4:5598d62a0703: doc: add missing x86 cpuflags to fftools documentation
[00:58:32 CEST] <cone-750> ffmpeg 03James Almer 07release/2.4:4b817e2fff3a: doc: add missing arm cpuflags to fftools documentation
[00:58:33 CEST] <cone-750> ffmpeg 03James Almer 07release/2.4:3a5b4afd911f: doc: add aarch64 cpuflags to fftools documentation
[00:58:34 CEST] <cone-750> ffmpeg 03James Almer 07release/2.5:7e05c70bb025: doc: add missing x86 cpuflags to fftools documentation
[00:58:35 CEST] <cone-750> ffmpeg 03James Almer 07release/2.5:60f2f332a397: doc: add missing arm cpuflags to fftools documentation
[00:58:36 CEST] <cone-750> ffmpeg 03James Almer 07release/2.5:43fdd89a3f07: doc: add aarch64 cpuflags to fftools documentation
[01:02:09 CEST] <cone-750> ffmpeg 03Timothy Gu 07n2.6.2:HEAD: fate: Add atrac3+ tests
[01:46:25 CEST] <michaelni> Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in  #ffmpeg | This channel is publicly logged | FFmpeg 2.6.2 has been released!
[01:59:06 CEST] <azk>     
[02:11:30 CEST] <BBB> do we have a way to indicate in e.g. the mpeg12/hevc/h264 decoders to output data in coding instead of display order?
[02:16:26 CEST] <Timothy_Gu> yay I setup an FATE instance for Haiku!
[02:16:29 CEST] <iive> i think there was such a way.
[02:16:41 CEST] <iive> if it haven't been removed.
[02:17:31 CEST] <michaelni> theres SLICE_FLAG_CODED_ORDER but its only for ff_draw_horiz_band()
[02:18:06 CEST] <michaelni> AVCodecContext.draw_horiz_band() rather
[02:24:24 CEST] <kierank> hopefully my opw student can write a draw_horiz_band test
[03:17:31 CEST] <cone-750> ffmpeg 03Mark Reid 07master:415f1fab8de7: libavformat/mxfenc: added support mxf opatom audio muxing
[03:39:55 CEST] <Timothy_Gu> michaelni: http://fate.ffmpeg.org:8080/report/x86-haiku-gcc-4.8/20150412012810
[03:41:41 CEST] <michaelni> Timothy_Gu, very nice, thanks alot
[03:45:09 CEST] <Timothy_Gu> michaelni: also are you gonna merge the old fateserver patch
[04:00:41 CEST] <michaelni> Timothy_Gu, applied to the git repos
[04:01:02 CEST] <Timothy_Gu> michaelni: thx
[04:04:39 CEST] <michaelni> Timothy_Gu, should i update the checkout that is used on fate.ffmpeg.org too ?
[04:12:11 CEST] <michaelni> you can update it too, ill leave it to you, dont want to somehow step onto some work there
[04:24:42 CEST] Action: michaelni falls asleep
[05:48:41 CEST] <Timothy_Gu> michaelni: i'll do it
[05:49:04 CEST] Action: Timothy_Gu googles "time in austria"
[05:54:50 CEST] <Timothy_Gu> michaelni: done. No work is really being done *there* haha. My brain would explode from all the perl.
[12:14:30 CEST] <MelchiorGaspar> anybody alive in here O_o loi ;D
[12:16:24 CEST] <phh> ok.
[12:16:55 CEST] <MelchiorGaspar> I found a format that FFmpeg does not support though I thought it did  
[12:17:10 CEST] <MelchiorGaspar> .pss   PLaystation 2 (Suikoden 3 and 4)
[12:18:07 CEST] <MelchiorGaspar> I extracted from my PS2 DvD discs so I convert to watch (oohh the memories) soI can upload a file if a Dev wants to test it
[13:49:20 CEST] <kierank> MelchiorGaspar: open a ticket and upload a sample
[13:50:06 CEST] <Compn> MelchiorGaspar : we have samples here http://samples.ffmpeg.org/playstation/
[13:50:10 CEST] <Compn> and in a few other dirs
[13:50:13 CEST] <Compn> but sure more samples are good
[14:01:20 CEST] <kierank> woah aac encoder patches
[14:17:41 CEST] <JEEBsv> finally patches?: o
[14:18:41 CEST] <JEEBsv> oh, gsoc
[15:31:10 CEST] <MelchiorGaspar> ok thx... I see PLaystation Folder... to reorigize formats I would suggest having a Playstation folder for each version
[15:31:28 CEST] <MelchiorGaspar> - PlayStation1
[15:31:31 CEST] Last message repeated 1 time(s).
[15:31:31 CEST] <MelchiorGaspar> - PlayStation2
[15:31:33 CEST] <MelchiorGaspar> - PlayStation3
[15:31:38 CEST] <MelchiorGaspar> - PlayStation4
[15:32:23 CEST] <MelchiorGaspar> where you BugTracker located?
[15:33:15 CEST] <JEEBsv> http://trac.ffmpeg.org/
[15:33:33 CEST] <MelchiorGaspar> ty.
[15:39:10 CEST] <MelchiorGaspar> k tracker account created
[15:39:14 CEST] <MelchiorGaspar> I'm Melchior
[16:04:21 CEST] <MelchiorGaspar> LOL OMG AUOMATIC POSTING OF BUGTRACKER TICKETS OOH YAH!!!!
[16:08:05 CEST] <MelchiorGaspar> FTP access to the http://samples.ffmpeg.org/playstation/ does not seem to work... ;_; because I cannot upload files createer then 2.5MB to the tracker and most of these vid files are easly 50MB+
[16:09:14 CEST] <MelchiorGaspar> is there anyway to get upload permissions to the samples server?
[16:12:32 CEST] <michaelni> MelchiorGaspar, http://ffmpeg.org/bugreports.html see "Submitting Sample Media"
[16:14:01 CEST] <MelchiorGaspar> ty
[16:14:59 CEST] <BtbN> Is ffmpeg supposed to play those but fails, or is that something entirely new?
[16:18:00 CEST] <MelchiorGaspar> ok got it I attached a few small files to the bug report as can be seen just above, I have reg'd the upload url to Filezilla I will create a sub folder withthe tarcker id#
[16:22:53 CEST] <MelchiorGaspar> the upload server does not like me.... 
[16:25:27 CEST] <MelchiorGaspar> ok it likes me know I had to set it too one connection only
[16:28:35 CEST] <cone-235> ffmpeg 03Paul B Mahol 07master:d2fc702ace34: avfilter: add chorus filter
[16:37:35 CEST] <MelchiorGaspar> one set of uploads are done
[16:37:43 CEST] <MelchiorGaspar> ./incoming/Ticket-4478_Suikoden3
[16:39:28 CEST] <MelchiorGaspar> ./incoming/Ticket-4478_Suikoden4  <-- Is uploading now  , 756MB worth of .pss files
[17:01:32 CEST] <MelchiorGaspar> ok uploads are done
[17:02:19 CEST] <sfan5> MelchiorGaspar: EV50511.PSS plays fine for me in mpv
[17:15:01 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:4270a9f547bd: avformat/sierravmd: use av_malloc_array()
[17:15:02 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:870ec3f69e04: avformat/segafilm: Use av_malloc_array()
[17:56:37 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:445a02b1ec5e: avformat/rtpdec_asf: Fix potential pointer overflow
[17:57:25 CEST] <wm4> Assertion !s->out_ch_layout || out->ch_count == av_get_channel_layout_nb_channels(s->out_ch_layout) failed at libswresample/rematrix.c:446
[17:57:28 CEST] <wm4> what does this mean
[17:57:33 CEST] <wm4> it works with libavresample
[17:58:16 CEST] <wm4> libswresample unhelpfully prints "Input channel layout has a different number of channels than the number of used channels, ignoring layout" before that, but opening the context succeeds
[18:00:18 CEST] <michaelni> how can that be reproduced ?
[18:08:56 CEST] <wm4> michaelni: with mpv git master (from moments ago) with a sample that switches from  48000Hz stereo 2ch floatp to 48000Hz 5.1 6ch floatp and  -v --speed=1.1 --no-audio-pitch-correction --ao=null
[18:09:19 CEST] <wm4> michaelni: it actually triggers reusing the same swresample context for a format change (while normally the context is destroyed and recreated)
[18:10:22 CEST] <MelchiorGaspar> [11:02] <sfan5> MelchiorGaspar: EV50511.PSS plays fine for me in mpv   <-- I use AvantiGUI w/ FFmpeg 
[18:11:11 CEST] <sfan5> i use ffmpeg version N-71399-g816bbc1 Copyright (c) 2000-2015 the FFmpeg developers (in case that helps in some way)
[18:12:14 CEST] <michaelni> wm4, where can i find the/a input sample for this ?
[18:16:58 CEST] <wm4> michaelni: https://0x0.st/Tv.mkv
[18:30:44 CEST] <BtbN> What's the best video/image filter to look at for a simple pixel-modification-filter example?
[18:36:13 CEST] <Timothy_Gu> how can you check the GPT of an object again?
[18:40:18 CEST] <MelchiorGaspar> I get my FFmpeg bin from, http://ffmpeg.zeranoe.com/builds/
[18:40:24 CEST] <MelchiorGaspar> nap time
[18:41:00 CEST] <wm4> BtbN: maybe vf_eq?
[18:44:42 CEST] <Timothy_Gu> for some reason `-pie` is needed for ffmpeg to link on haiku x86-64
[18:44:48 CEST] <Timothy_Gu> is that the right thing to do?
[18:45:58 CEST] <wm4> btw. why care about Haiku at all
[18:46:09 CEST] <Timothy_Gu> fun :p
[18:46:25 CEST] <wm4> it's 100 times more obscure than using BSD on the desktop, and 1000 times more obscure than using Linux on the desktop
[18:46:33 CEST] <sfan5> what about GNU/Hurd
[18:46:44 CEST] <wm4> Hurd is actively developed
[18:47:04 CEST] <wm4> but there are probably more people caring about Windows 95...
[18:47:18 CEST] <Timothy_Gu> pretty sure more people care about haiku than hurd
[18:50:13 CEST] <iive> talking about hurd - http://xkcd.com/1508/
[18:54:39 CEST] <cone-235> ffmpeg 03Timothy Gu 07master:7ad27f122156: fate: Include branch information in the payload header
[19:19:07 CEST] <Timothy_Gu> fate passes so meh
[20:20:30 CEST] <Timothy_Gu> Can someone please explain how `cextern` works in conjunction with shared library? Does it use the GOT? If not how/why?
[20:21:08 CEST] <Timothy_Gu> michaelni, jamrial, ubitux, BBB_ ^^
[20:24:12 CEST] <Timothy_Gu> disassembling objects built with PIC seems to reveal that they do not use the GOT and uses R_X86_64_PC32, but http://eli.thegreenplace.net/2011/11/11/position-independent-code-pic-in-shared-libraries-on-x64 says that R_X86_64_PC32 cannot be used with shared libs, so I'm really confused right now&
[20:35:18 CEST] <BBB_> Timothy_Gu: I think its just an external reference resolved by the linker
[20:35:26 CEST] <BBB_> so no, it doesnt do pic relocation resolution for you
[20:35:31 CEST] <BBB_> thats all up to you
[20:35:48 CEST] <BBB_> but it allows you to use stuff (e.g. tables) stored in anoher file
[20:39:37 CEST] <BBB> Timothy_Gu: youll notice various places in our asm code doing special PIC magic
[20:39:42 CEST] <BBB> under %if PIC
[20:39:48 CEST] <BBB> or %ifdef PIC
[20:39:55 CEST] <BBB> thats how it works
[20:45:23 CEST] <Timothy_Gu> BBB: the "external reference" is using R_X86_64_PC32, which fails on Haiku
[20:45:24 CEST] <michaelni> wm4, you should be able to workaround this by setting the input/output channel counts ("ich"/"och") but ill push a fix soon and will backport it
[20:45:41 CEST] <Timothy_Gu> (and I wonder why it doesn't fail on linux as well)
[20:45:50 CEST] <BBB> theres a lot of magic in x86inc.asm
[20:46:00 CEST] <Timothy_Gu> who wrote x86inc
[20:46:01 CEST] <Timothy_Gu> ?
[20:46:38 CEST] <Daemon404> pengvado, 
[20:46:40 CEST] <Daemon404> did
[20:46:45 CEST] <Timothy_Gu> oh
[20:46:50 CEST] <Daemon404> a lot of bugmaster too iirc
[20:46:58 CEST] <BBB> ok, so
[20:47:01 CEST] <BBB> first of all
[20:47:03 CEST] <BBB> is PIC defined?
[20:47:06 CEST] <Timothy_Gu> yes
[20:47:10 CEST] <BBB> in the assembly header?
[20:47:16 CEST] <BBB> (config.asm)
[20:47:16 CEST] <Timothy_Gu> config.asm? yes
[20:47:24 CEST] <BBB> is this a 64bit build?
[20:47:26 CEST] <Daemon404> BBB, this is haiku so i bet its just toolchain bugs
[20:47:29 CEST] <Daemon404> just sayin.
[20:47:38 CEST] <Timothy_Gu> yes
[20:47:42 CEST] <BBB> Daemon404: I know, but x86inc.asm has some magic that Ive played with, so we can rule out that
[20:47:54 CEST] <Timothy_Gu> or rather PIC is defined in Makefile (-DPIC)
[20:49:28 CEST] <BBB> and that is included in ASMFLAGS (not CFLAGS)?
[20:49:36 CEST] <BBB> ASFLAGS I guess
[20:49:51 CEST] <cone-235> ffmpeg 03James Almer 07master:14edc9f3a392: configure: add missing h263p decoder dependencies
[20:49:52 CEST] <Timothy_Gu> YASMFLAGS
[20:50:30 CEST] <Timothy_Gu> ok so YASMFLAGS and CPPFLAGS have -DPIC, while CFLAGS and ASFLAGS have -fPIC
[20:50:39 CEST] <wm4> michaelni: thanks, then I will wait for the fix
[20:52:19 CEST] <Timothy_Gu> Daemon404: probably, but it's fun to figure out why it's a bug in the toolchain so that the next i see this i can say "hey this is OBVIOUSLY a toolchain bug"
[20:52:21 CEST] <kierank> what's h263i
[20:52:22 CEST] <kierank> ?
[20:52:48 CEST] <Timothy_Gu> kierank: i = intel?
[20:52:57 CEST] <jamrial> intel, yes
[20:53:08 CEST] <kierank> and p is?
[20:53:36 CEST] <BBB> Timothy_Gu: and what is the exact error message?
[20:53:45 CEST] <j-b> kierank: Plus
[20:53:54 CEST] <Timothy_Gu> BBB: http://fate.ffmpeg.org:8080/log/x86_64-haiku-gcc-4.8/20150412162931/compile
[20:54:06 CEST] <Timothy_Gu> ld: libavcodec/libavcodec.a(deinterlace.o): relocation R_X86_64_PC32 against symbol `ff_pw_4' can not be used when making a shared object; recompile with -fPIC
[20:54:35 CEST] <Timothy_Gu> kierank: "H.263+ / H.263-1998 / H.263 version 2"
[21:02:25 CEST] <BBB> Timothy_Gu: is it only deinterlace.asm?
[21:02:47 CEST] <Timothy_Gu> nope, everything using the masks
[21:03:11 CEST] <Timothy_Gu> BBB: Also if I link with -pie the error goes away
[21:04:45 CEST] <BBB> hm ok thats very obscure then
[21:04:47 CEST] <Timothy_Gu> BBB: I tried compiling a very simple C program that references a mask, and objdump shows that it's using R_X86_64_GOTPCREL rather than _PC32
[21:05:58 CEST] <BBB> so, my understanding is that were doing all gots directy in our code
[21:06:15 CEST] <BBB> so the indicator is kind of unnecessary
[21:06:30 CEST] <BBB> but Im not an expert or anything
[21:07:27 CEST] <Timothy_Gu> let me boot the 32-bit vm and see what it does there
[21:09:28 CEST] <BBB> on 32bit we disable pic
[21:09:37 CEST] <Timothy_Gu> oh wait yeah
[21:09:40 CEST] <BBB> since pic would cost you one precious register and we dont want that
[21:09:46 CEST] <BBB> (you only have 7)
[21:09:50 CEST] <Timothy_Gu> x86
[21:10:41 CEST] <Timothy_Gu> i tried using qemu without kvm this morning&
[21:57:33 CEST] <iive> i actually wonder why x86_64 is not pic by default.
[21:57:48 CEST] <iive> i mean, the ffmpeg configure/build
[21:58:18 CEST] <Daemon404> because ricing
[21:58:23 CEST] <Daemon404> non-PIC is faster
[21:58:39 CEST] <iive> is that true on 64bit too?
[21:58:47 CEST] <Daemon404> im not sure actually
[22:17:07 CEST] <rcombs> I was under the impression it wasn't on x64
[22:23:58 CEST] <Daemon404> dont call it x64... thats a MS thing
[22:28:37 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:d7b9cb2f7a51: swresample: Allow reinitialization without ever setting channel counts
[22:28:38 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:80a28c7509a1: swresample: Allow reinitialization without ever setting channel layouts
[22:28:39 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:3c77bb5f23b2: swresample: Check channel layouts and channels against each other and print human readable error messages
[22:28:40 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:ff50b1b13be9: swresample/swresample-test: Randomly wipe out channel counts
[22:49:53 CEST] <Timothy_Gu> iive: well having one more register is always a good thing right?
[22:54:23 CEST] <iive> Timothy_Gu: well, wasn't the point of x86_64 that it doesn't need one extra register? RIP-relative addressing and stuff?
[23:23:11 CEST] <Timothy_Gu> shoot I broke fate
[23:27:19 CEST] <Timothy_Gu> there, fixed
[23:53:41 CEST] <cone-235> ffmpeg 03Himangi Saraogi 07master:f03dfe118b64: avformat/rtsp: Fix dereference after null check
[00:00:00 CEST] --- Mon Apr 13 2015