[Ffmpeg-devel-irc] ffmpeg-devel.log.20141125

[burek]

[01:12] <Theywentviral> im having trouble using a new software called stream-m
[01:13] <Theywentviral> has anyone heard of this?
[01:15] <Theywentviral> im using a method called concat and its causing me to use 90% cpu usage when its looping
[01:18] <Theywentviral> ok
[01:18] <Theywentviral> well i can't see anyones replies
[01:25] <TheyWentViral> ok
[01:26] <J_Darnley> Why would you come here for help with other software?
[01:28] <TheyWentViral> _Darnley it says for the development of ffmpeg , and the development of other software that uses ffmpeg clearly in the documentation
[01:29] <TheyWentViral> This software stream-m uses ffmpeg
[01:30] <llogan> This is the channel for discussion of FFmpeg development only. user support is at ffmpeg-user IRC and mailing list, but we only support FFmpeg stuff, and not 3rd party stuff (including stream-m)
[01:30] <TheyWentViral> im having trouble with the concat of a stream i beleive its causing me to use 90% CPU usage i was just wondering if anyone ran into a similar problem when it loops using this method...
[01:31] <J_Darnley> So give us some ffmpeg command lines and terminal output, then maybe we can help you.
[01:31] <llogan> ...so you'll have to get help from stream-m, or if you are using ffmpeg directly then use the mailing list or IRC channel.
[01:31] <TheyWentViral> ok sorry llogan maybe i inturpreted the doucmentation wrong im just looking for a little help i'll search there
[01:31] <llogan> (ffmpeg-user that is, and not here or ffmpeg-devel mailing list)
[01:35] <TheyWentViral> Thanks again
[01:55] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:03a17f2bbf36: avcodec/mjpegdec: Print the number of bits in the unsupported pixel format error
[01:55] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:172d22a071b5: avcodec/mjpegdec: Add YUVA420 formats to *scale asserts
[02:12] <cone-890> ffmpeg.git 03Uwe L. Korn 07master:40665d27e38e: flvdec: Document how the duration is retrieved at the end of the file
[02:12] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:b83beb131f0d: Merge commit '40665d27e38e6a2f65037878202bd1a398c7683e'
[02:48] <cone-890> ffmpeg.git 03Martin Storsjö 07master:b9d08c77a443: lavf: Don't try to update files atomically with renames on windows
[02:48] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:71ecfcf2d31e: Merge commit 'b9d08c77a44390b0848c06f20bc0e9e951ba6a3c'
[04:04] <cone-890> ffmpeg.git 03Martin Storsjö 07master:960aff379da4: lavf: Use wchar functions for filenames on windows for mkdir/rmdir/rename/unlink
[04:05] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:ecfafc5f2b63: Merge commit '960aff379da46dcaff61504a57714d4d4e758e41'
[04:05] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:60420fa3dc88: avdevice/dshow: fix build, ensure that feature enable #defines are set before includes
[04:05] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:c724c82a7828: doc/print_options: Fix build on mingw after 960aff379da46dcaff61504a57714d4d4e758e41
[04:19] <cone-890> ffmpeg.git 03Carl Eugen Hoyos 07master:0e3ea5b28a32: Include stddef from snow_dwt.h.
[04:19] <cone-890> ffmpeg.git 03Carl Eugen Hoyos 07master:20f3cdf0c894: Include config.h from huffyuvdsp.h.
[04:19] <cone-890> ffmpeg.git 03Michael Niedermayer 07master:67dbf8bec977: Merge remote-tracking branch 'cehoyos/master'
[04:38] <Zeranoe> michaelni: Any chance you could commit the attached patch for this bug https://trac.ffmpeg.org/ticket/4130 
[04:56] <cone-890> ffmpeg.git 03Carl Eugen Hoyos 07master:1f3d4788983d: avdevice/decklink_common: Fix "Cross-compiling FFmpeg on Debian for Windows with MinGW-w64"
[04:56] <michaelni> Zeranoe, applied
[05:01] <Zeranoe> michaelni: Thanks
[11:17] <cone-477> ffmpeg.git 03Vittorio Giovara 07master:bc75b64cff37: vc1pred: remove logically dead code
[11:17] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:0aa208837e24: Merge commit 'bc75b64cff37d58f3944e2da3da45c37f35f019a'
[11:55] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:c117da9d3e0d: lpc: remove unneeded {}
[11:55] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:62e52b94e684: vorbis_parser: Move vp check to avoid a null pointer dereference
[11:55] <cone-477> ffmpeg.git 03Luca Barbato 07master:b67138598ce1: vc1: Simplify a little setting the intra variables
[11:55] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:a63ec9d5737a: Merge commit 'c117da9d3e0db7dc311d817054988364b3ef4587'
[11:55] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:bcccb2c2986b: Merge commit '62e52b94e684491dfc5a6b7ca688bb86f7cd0f3f'
[11:55] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:ed25ca166171: Merge commit 'b67138598ce158e3083f6295a27b63e2065d5ecb'
[12:06] <cone-477> ffmpeg.git 03Luca Barbato 07master:d25afb579fac: vc1: Set the is_intra bitfield to all 1 when needed
[12:06] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:79f77ce31c6c: Merge commit 'd25afb579facc83fd3a839f21411124d0b09f0ba'
[12:26] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:51946d2de8bd: vc1: Use the correct shift amount
[12:26] <cone-477> ffmpeg.git 03Luca Barbato 07master:16158da9607f: hnm4: Use av_image_check_size
[12:26] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:0e85a28fa434: Merge commit '51946d2de8bd4a4aada43b6ab41340b0f5eb4ecb'
[12:26] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:ba1a19bc35e4: Merge commit '16158da9607f2f84232d3dd381406b2f2449ec74'
[12:44] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:57ed5a64feec: hnm4: change width/height to int to fix hypothetical integer overflows
[12:44] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:ca5c3ff90972: vf_interlace: x86: improve asm performance
[12:44] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:8c9945285e30: Merge commit '57ed5a64feec4af1f16f9a74c63cfa9aa8147242'
[12:44] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:3fe3c8abb15a: Merge commit 'ca5c3ff90972a5c97aabda2ace57ba72dcd7d83b'
[12:44] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:ca59b5b6eceb: avfilter/x86/vf_interlace: remove redundant instructions
[13:35] <wm4> ubitux: went through utils.c; I wonder why everything checks whether pb is NULL (except some code doesn't)
[13:36] <wm4> ubitux: I don't think the API allows the demuxer to close the pb at all, if that's what you planned
[13:37] <wm4> even setting to NULL might not be fine (due to the API)
[13:43] <__gb__> hi kierank, you can use vaapi without rendering, yes -- even without a display server actually (using the VA/DRM backend)
[13:45] <__gb__> and, if you have a kernel version recent enough (>= 3.15 to have that feature enabled by default), then you can use render nodes too for using vaapi remotely through ssh, even if another user has locally started a display server
[13:46] <__gb__> if you don't want to go over elaborated libudev stuff, you could just open /dev/dri/renderD128 -- that's the render nodes companion to /dev/dri/card0
[13:47] <__gb__> and provide the resulting fd to vaGetDisplayDRM() and use the resulting VADisplay wherever needed
[14:10] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:0eecf40935b2: avcodec/mjpegdec: Fix context fields becoming inconsistent
[14:56] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:3d5d95db3f5d: avcodec/utils: Check that the data is complete in avpriv_bprint_to_extradata()
[15:28] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:8cd80b5fcbfa: avformat/jacosubdec: Cleanup when avpriv_bprint_to_extradata() fails
[15:48] <wm4> "Getting board?"
[15:48] Action: wm4 facepalms
[16:11] <Zeranoe> michaelni: I think ecfafc5f2b632780dd310def1d6b3a2c10565465 or 960aff379da46dcaff61504a57714d4d4e758e41 introduced a new compiling error http://paste.ubuntu.com/9233871/
[16:18] <Daemon404> Zeranoe, that looks like a C++ error... weird
[16:22] <Zeranoe> Daemon404: More context: http://paste.ubuntu.com/9234011/ looks like decklink_enc.cpp is being included
[16:25] <michaelni> Zeranoe, does a cast to wchar_t* fix that ?
[16:26] <Daemon404> michaelni, a cast is the wrong fix
[16:26] <Daemon404> this is not a valid error in C
[16:27] <wm4> well it's compiled as C++ apparently?
[16:27] <Daemon404> that seems majorly wrong
[16:27] <wm4> these are just stubs for win32
[16:27] <wm4> so nobody cares whether they're inline
[16:27] <wm4> same with win32 pthread wrappers
[16:28] <Daemon404> doesnt matter, it can still be done right in C++
[16:28] <wm4> with the new operator?
[16:28] <Daemon404> let me sort it
[16:28] <Daemon404> also
[16:28] <Daemon404> #include "libavformat/internal.h"
[16:28] <Daemon404> in libavdevice
[16:28] <Daemon404> yum yum
[16:29] <wm4> apparently this is "ok"
[16:29] <wm4> because it's a compile time dep only
[16:30] <Daemon404> i think vfwcap.c is misleading
[16:30] <Daemon404> Zeranoe, were you building multithreaded?
[16:30] <Daemon404> i think ti *actually* failed in the .cpp
[16:32] <Daemon404> ok so...
[16:32] <Daemon404> https://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavdevice/decklink_enc.cpp;h=6c5450f4ec5f36d1b397737b5dde9063d5ec6acc;hb=HEAD#l29
[16:32] <Daemon404> what the fuck g++
[16:32] <Daemon404> you should not be doing that when the func is under extern "C"
[16:32] <wm4> uh
[16:32] <wm4> extern C doesn't mean it's C code
[16:32] <nevcairiel> extern C only controls linkage
[16:32] <nevcairiel> nothing else
[16:32] <wm4> just that C mangling is used
[16:33] <nevcairiel> common misconception
[16:33] <Daemon404> i see
[16:33] <Daemon404> that's kinda lulz
[16:33] <wm4> also you could have tortured the contributor into writing that file in C
[16:33] <Daemon404> i figured it would be smarted than to try and inline C into C++
[16:34] <Daemon404> because inline is a hint to the compiler
[16:34] <Daemon404> not a guarantee
[16:34] <nevcairiel> the standard can probably not assume that every C++ compiler cna also speak C
[16:34] <nevcairiel> while turning off name mangling is easy
[16:34] <Zeranoe> Daemon404: I am, pthread
[16:34] <Daemon404> Zeranoe, not what i meant, but its not relevant now
[16:35] <Daemon404> i guess a cast it is
[16:35] <Daemon404> c++/c interop sure is fugly
[16:35] <Zeranoe> Daemon404: Yes to what you meant, make -j4
[16:37] <BtbN> https://github.com/BtbN/FFmpeg/commit/13b5cceea6300e36c2d8d7d6b8b2cfacbcdc5dc1 Is this the correct way to add a dependency on an external header file?
[16:37] <BtbN> This configure script is extremely confusing and way too much stuff happens automagicaly based on naming
[16:38] <BtbN> I'm still not sure if it would propperly disable my code if nvenc is disabled
[16:41] <BtbN> Seems like it does, but i have no idea why
[16:41] <BtbN> It also doesn't seem to matter that in allcodecs there is no compile-time removal of anything
[16:42] <Zeranoe> Daemon404, michaelni: Do you need anything else from me on this?
[16:42] <Daemon404> Zeranoe, no, the fix is clear
[16:43] <BtbN> Hm, what qualifies a library to be flagged as non-free? As far as i know, linking ffmpeg with nvenc doesn't imply any redistribution restrictions. You just can't include its header in ffmpeg.
[17:01] <michaelni> Zeranoe, a tested patch would be best but i can try to blindly fix it and maybe after a few tries we will have found all cases th
[17:01] <michaelni> that need changing
[17:04] <Zeranoe> michaelni: I would be happy to test any patchs
[17:17] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:d0879a93eac6: avformat/os_support: try to fix build when included from a c++ file like libavdevice/decklink*cpp
[17:18] <michaelni> Zeranoe, pushed a fix for that one line, does it build now or are there more issues ?
[17:21] <Zeranoe> michaelni: running with a single make job now
[17:43] <Zeranoe> michaelni: Everything seems fine now, thank you for the prompt fix
[17:44] <michaelni> np
[18:02] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:2f1de5ca139f: avcodec/huffyuvdec: apply vertical filter in steps of 1 line for interlaced BGRA
[18:27] <cone-477> ffmpeg.git 03Yu Xiaolei 07master:bc3d02fa88c4: lavf/avio: clarify the buffer parameter of avio_alloc_context
[18:30] <cone-477> ffmpeg.git 03Benoit Fouet 07master:155f4dd668b8: doc: add entry for APNG demuxer where needed.
[18:48] <Daemon404> so uh....
[18:48] <Daemon404> would a patch be accepted to disable xmp printing in ffprobe?
[18:48] <Daemon404> because some files print shit for 5+ minutes
[18:48] <Daemon404> its ridiculous
[19:22] <ubitux> wm4: alright, so what do you propose (re:closing pb)
[19:22] <wm4> let the API user handle it
[19:23] <wm4> just export the necessary information
[19:26] <ubitux> wm4: but, i don't see what the user is supposed to do here
[19:27] <BtbN> ffmpeg has pull requests now?
[19:27] <wm4> ubitux: the user can do things with the AVIOContext, or even unset it (if lavf allows that)
[19:28] <ubitux> so the patch as is looks ok to you?
[19:28] <wm4> yes
[19:29] <wm4> although granted I'm not perfectly sure how I'll use it
[19:29] <ubitux> i'll push it if you're able to make use of it then
[19:29] <wm4> fair enough
[19:29] <ubitux> maybe we forgot something so i don't want to change the api yet
[19:54] <BtbN> Are github PullRequests accepted nowdays?
[19:54] <kierank> no
[19:54] <BtbN> That's highly annoying...
[19:55] <Daemon404> BtbN, some of us would say github PRs are highly annoying
[19:55] <Daemon404> they have a number of disadvantages.
[19:55] <BtbN> Setting up mails, managing changes, and getting tons of ML mail is annoying
[19:56] <Daemon404> im saying its personal preference, and the overwelming majority of people here prefer the ML.
[19:56] <BtbN> I have that stuff flying around since ages now, and didn't send it in just because of having to do that git mail stuff
[19:56] <Daemon404> ok?
[19:57] <Daemon404> and you can attach patches too.
[19:57] <BtbN> Makes it kinda hard to discuss some of the changes
[19:57] <BtbN> I could send a mail pointing to my github branch, as request for comments
[19:59] <Daemon404> someone might review it, but i cant guaranatee it
[20:00] <BtbN> I'm still not sure if it qualifies as nonfree or not
[20:00] <llogan> BtbN: some github pull requests have been accepted, IIRC. we prefer patches sent to the mailing list, but you can try a PR. not nearly as many people see them.
[20:00] <BtbN> It's a new h264 encoder, needs some feedback
[20:00] <BtbN> nvidia nvenc stuff
[20:01] <Daemon404> non-free with dlopen/loadlibrary is a massive grey area
[20:01] <Daemon404> i dont think youll ever know conclusively
[20:01] <BtbN> There is no non-dlopen way though. That's the official way nvidia wants it to be used
[20:01] <wm4> ML + github PRs are the worst combination
[20:01] <Daemon404> BtbN, i know 
[20:02] <wm4> pick either one, and ffmpeg picked ML-based workflow
[22:15] <cone-477> ffmpeg.git 03Martin Storsjö 07master:f20141d73f08: vorbis_parser: Include stdint.h in the header, to make it work standalone
[22:15] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:23e91a1bfd7f: Merge commit 'f20141d73f08ed0c8e875bd993a7143e19b266e3'
[22:18] <cone-477> ffmpeg.git 03Lukasz Marek 07master:e29153f414f5: lavc/avuienc: fix mem leak in case of init failure
[22:18] <cone-477> ffmpeg.git 03Lukasz Marek 07master:2a89afb376ae: lavc/libxvid: return meaningful error codes
[22:42] <kierank> kurosu_ / jamrial : any more asm comments?
[22:44] <kurosu_> kierank, nope
[22:45] <kurosu_> was the unrolling actually helpful? you said it was but it looked more than I would have assumed
[22:45] <kurosu_> and I'm not sure if it breaks the C part
[22:46] <kurosu_> I mean, handling batches of 12 instead of 6 as I thought it assumed
[22:47] <kurosu_> anyway, I don't really care: if it breaks, I bet you'll be eager to fix it
[22:48] <kierank> for 8-bit yuv422p -> v210 you have to unroll
[22:48] <kierank> whereas in 10-bit mode you only have 8 pixels in your register
[23:10] <cone-477> ffmpeg.git 03Lukasz Marek 07master:ea0d8938173e: lavu/opt: handle NULL obj in av_opt_next
[23:10] <cone-477> ffmpeg.git 03Lukasz Marek 07master:1907ff0a67ed: lavc/utils: free private options on avcodec_open2 fail
[23:19] <jamrial> kierank: nothing aside from what i mentioned for the 10bits patch
[23:19] <kierank> ok
[00:00] --- Wed Nov 26 2014