[Ffmpeg-devel-irc] ffmpeg-devel.log.20140609

burek burek021 at gmail.com
Tue Jun 10 02:05:02 CEST 2014


[00:12] <J_Darnley> Does anyone know of a website that will match text input with known software licenses?
[00:18] <J_Darnley> Or a website that lists licenses?
[00:19] <J_Darnley> I wonder if FSF or OSI do.
[00:20] <J_Darnley> Looks like they both do.
[00:58] Action: Daemon404 finds ubitux's python script disturbing
[00:58] <Daemon404> well not the script... the fact that it had to exist at all
[00:58] <ubitux> well, i tried several weeks to do it from the head
[00:59] <ubitux> and it wasn't a good idea.
[00:59] <Daemon404> did a human even write the filter
[00:59] <ubitux> actually, factor.py is freaking slow to run on all the mode
[00:59] <ubitux> dunno
[00:59] <ubitux> (but not optimized at all)
[01:10] <cone-834> ffmpeg.git 03James Almer 07master:6d408495b595: x86/dct32: don't build ff_dct32_float_sse on x86_64
[01:58] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:3b977a6ded41: avcodec/alsdec: Clear MPEG4AudioConfig so that no use of uninitialized memory is possible
[01:58] <cone-834> ffmpeg.git 03Dale Curtis 07release/0.10:2facb10f705a: matroska: Fix use after free
[01:58] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:f24246a8891e: avcodec/h264_mp4toannexb_bsf: prepend global headers before any in stream parameter sets
[01:58] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:6cf254ea6814: avcodec/wma: use av_freep(), do not leave stale pointers in memory
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:ec29aec61854: swscale/x86/swscale_template: loose hardcoded dstw_offset
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:71a3ad42b8c9: avformat/mpegts: Remove redundant check
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:90b69d1d020b: avcodec/diracdec: fix undefined behavior with shifts
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:363cf196c904: avfilter/vf_deshake: fix loss of precission with odd resolutions
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:dd8464bc9951: avformat/h263dec: Fix h263 probe
[01:59] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:6a679279f7f2: avcodec/mjpegdec: Fix undefined shift
[02:34] <cone-834> ffmpeg.git 03Michael Niedermayer 07release/0.10:da97174dcb14: update for 0.10.13
[03:16] <cone-834> ffmpeg.git 03James Almer 07fatal: ambiguous argument 'refs/tags/n0.10.13': unknown revision or path not in the working tree.
[03:16] <cone-834> Use '--' to separate paths from revisions
[03:16] <cone-834> refs/tags/n0.10.13:HEAD: x86/dct32: don't build ff_dct32_float_sse on x86_64
[03:56] <Youka> I'd like to decode video files with unicode file names on windows. avformat_open_input seems to use fopen, so no unicode. Where do i find the implementation or a way around?
[03:58] <BBB> Youka: Im pretty sure it uses avpriv_open
[03:58] <BBB> which uses a win32-compatible version
[03:58] <BBB> Youka: see libavutil/file_open.c
[03:59] <Youka> Is that new since v1.1?
[04:00] <Youka> this win32 file open with utf8 conversion?
[04:00] <wm4> yes it was added at some later point
[04:01] <Youka> very cool (and needed), thx BBB & wm4
[04:01] <wm4> I guess it's also possible that it's somehow not enabled on all builds?
[04:02] <Youka> found no option that could exclude it
[04:03] <jamrial> seems to have been added in ffmpeg 2.1
[04:34] <cone-834> ffmpeg.git 03James Almer 07master:345f2234d171: x86/scale: fix xmm register count for hscale*_sse2
[05:52] <cone-834> ffmpeg.git 03Michael Niedermayer 07master:52e563bb2f78: avformat/cavsvideodec: Fix probing when the file extension is avs
[05:52] <cone-834> ffmpeg.git 03Michael Niedermayer 07master:0accf24b15ac: cavsdec: fix qp fixed slice handling
[05:58] <Timothy_Gu> names
[05:59] <Timothy_Gu> whoops sorry PEBCAK attack
[09:45] <UtUser> Is anyone around on here?
[09:48] <UtUser> What's the deal with the "development mailing list" anyways?
[09:48] <UtUser> Somebody submitted a patch for an issue which I re-opened and there hasn't been any word on it since, and it's been over a week!
[09:49] <thardin> try bumping it. sometimes people forget
[09:49] <thardin> or better: link it here first
[09:49] <UtUser> http://trac.ffmpeg.org/ticket/2661#comment:9
[09:49] <UtUser> http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/178313
[09:49] <UtUser> I think it would be better to email everybody
[09:50] <UtUser> nobody seems to read the issue tracker
[09:51] <thardin> patch looks simple enough
[09:51] <UtUser> ok then
[09:51] <thardin> ffmpeg has an ut encoder nowadays?
[09:51] <UtUser> yeah
[09:51] <UtUser> native and built-in
[09:52] <UtUser> thanks to I think Daemon404
[09:52] <UtUser> can you help?
[09:52] <thardin> nope, sry
[09:52] <UtUser> you were just giving me advice just now
[09:52] <thardin> speaking of bugs, I think I have an mxf bug to fix
[09:52] <thardin> yeah, but I don't know anything about ut
[09:52] <UtUser> ok
[09:53] <UtUser> I'm not asking you to code anything
[09:53] <UtUser> I want the management to take another look at the patch
[09:53] <UtUser> and merge it if possible
[09:53] <UtUser> how can they be gotten to do that?
[09:54] <thardin> well either pinging them in here or bumping the thread on the list
[09:54] <thardin> probably in that order
[09:54] <UtUser> ok
[09:54] <thardin> regressions are usually considered important
[09:54] <UtUser> the @ people or the + people?
[09:54] <UtUser> ^ I'm glad they have that much sense at least
[09:55] <thardin> carl is usually in here, but not now it seems
[09:55] <thardin> cehoyos that is
[09:55] <TimNich> thardin: whats the mxf issue?
[09:55] <thardin> oh, uh.. EssenceKind stuff
[09:56] <thardin> basically I want to just do whatever the most developed version of libmxf does
[09:56] <TimNich> the usual quagmire then...
[09:56] <TimNich> so the bmxlib github then.
[09:57] <thardin> I want it so it's relatively easy to update when it changes
[09:57] <thardin> like a subtree (which ffmpeg abhors)
[09:57] <thardin> hopefully the relevant file can be used as-is, so updating is just a copy-paste job
[09:58] <thardin> or some small script could do it
[09:58] <TimNich> some of the libmxf stuff uses c++
[09:58] <thardin> blasphemy
[09:59] <thardin> hopefully the API is still C
[09:59] <thardin> oh yeah, it had a C part and a C++ part
[09:59] <TimNich> I think its basically just a c++wrapper round c to gwet some of the nice c++ libs like timecode.
[09:59] <thardin> no worries then
[10:00] <thardin> I still think mxf is too much hassle to pretend to have a native decoder for
[10:00] <thardin> no one can really handle all of mxf
[10:00] <TimNich> :)
[10:00] <thardin> you just asymptotically approach decent support for some set of features
[10:01] <TimNich> and then rgress...
[10:01] <TimNich> *regress
[10:02] <thardin> yeah, or someone uploads a file from yet another broken encoder
[10:03] <UtUser> Anssi BBB BuxiNess drv j-b llogan merbanan michaelni ods15 pengvado reynaldo ruggles siretart superdump ubitux wbs kierank nevcairiel uau: speaking of regressions, can you look at http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/178313  &  http://trac.ffmpeg.org/ticket/2661
[10:04] <nevcairiel> It's obvious why it didn't continue. No one answered the question asked on the ML
[10:05] <ubitux> this is just for Daemon404 (Derek) and cehoyos (Carl Eugen) afaict, no need to hl everyone
[10:05] <ubitux> nevcairiel: the question was answered on the trac by UtUser afaict
[10:06] <nevcairiel> Should answer on the ML or its hard to find
[10:06] <nevcairiel> Or the patch submitted should hint at the answer on the ML
[10:06] <UtUser> ubitux: I am fluffrabbit on the trac
[10:07] <UtUser> or just fluff I guess
[10:08] <nevcairiel> In any case, should ping it on the ML then
[10:09] <UtUser> I'm replying to the ML right now
[10:09] <UtUser> not sure what to say though
[10:14] <thardin> quote/link trac?
[10:20] <UtUser> thardin: Done.
[10:20] <UtUser> gave them a sample too
[10:21] <UtUser> and urged somebody to examine the patch
[10:23] <UtUser> test
[10:25] <thardin> sounds good
[10:27] <UtUser> do you ever find yourself struggling to find a balance between ambition and satisfaction, always being depressed or disappointed at the fruits of your labor, leading you to question how much work you really put in and the tangibility of life and reality?
[10:28] <UtUser> addiction studies have shown that an immediate feeling of pleasure is not necessary to fall back on the same behavior patterns or work towards an arbitrary goal.
[10:29] <UtUser> it saddens me what a "contributing member of society" I am
[10:29] <thardin> so do something else then
[10:30] <UtUser> I want to, I really do
[10:30] <UtUser> but there is a long story that goes along with why I am doing what I am doing, and I can't change
[10:30] <UtUser> I only want to be happier about it most of the time
[10:31] <UtUser> I wish computer work didn't feel like computer work
[10:31] <thardin> considering seeking help then
[10:31] <UtUser> I am considering it
[10:32] <UtUser> but psychological therapy won't help me
[10:32] <thardin> I think most computer folk are kind of mad
[10:32] <UtUser> I agree
[10:32] <UtUser> the worst part is, I'm only doing this to support my main goal: video and film
[10:32] <UtUser> that is my connection to ffmpeg
[10:33] <UtUser> you codec developers must work long and miserable hours
[10:33] <thardin> many have a passion for it
[10:33] <UtUser> a passion does not necessarily equal joy or ease in doing it
[10:33] <UtUser> there are always bugs
[10:34] <UtUser> and I react violently to bugs
[10:34] <UtUser> such is the way of computers
[10:35] <UtUser> they cheapen the whole process (a good thing) but also make it more stressful and unhappy
[10:35] <UtUser> I feel like I am stretched too thin
[10:35] <thardin> you can't take time off?
[10:36] <UtUser> I can't right now
[10:36] <thardin> :/
[10:36] <UtUser> technically, I'll have a couple of weeks off
[10:36] <UtUser> but unfortunately I'll be spending them with family
[10:36] <UtUser> I'm a different person around them
[10:37] <UtUser> I don't want to have to worry about my income, my education, and my projects
[10:37] <UtUser> I want to direct films
[10:38] <UtUser> that would be a great career
[10:38] <UtUser> it's also unrealistic right now
[10:38] <UtUser> I take such pleasure in telling people how to make art
[10:39] <thardin> being creative is fun
[10:39] <UtUser> absolutely
[10:39] <UtUser> and there can be no creativity on computers
[10:39] <UtUser> absolutely none
[10:39] <UtUser> they suck the creativity right out of everything, at least for me
[10:39] <UtUser> I don't know how those VFX guys keep their sanity
[10:40] <UtUser> for instance, I've got a blender render going right now
[10:40] <UtUser> I've also got a dozen tabs open related to FFMPEG
[10:41] <UtUser> it's also 1:30 in the morning
[10:41] <UtUser> I hate doing this shit
[10:41] <UtUser> my laptop is going to make noise all night
[10:41] <UtUser> it's a small apartment, and I don't feel like sleeping in the kitchen
[10:42] <thardin> take a break
[10:42] <UtUser> and do what?
[10:42] <UtUser> sit?
[10:42] <thardin> dunno
[10:42] <UtUser> I'll storyboard
[10:43] <thardin> :/
[10:43] <UtUser> it's fun
[10:43] <UtUser> I get to draw pictures without a computer
[10:43] <thardin> ok
[10:43] <UtUser> I also need to take a piss
[10:43] <thardin> I'm going back to working on things
[10:43] <UtUser> ok then bye
[10:53] <j-b> 'morning
[10:54] <UtUser-afk> good morning
[10:54] <UtUser-afk> pretend I'm not here
[14:10] <BBB> michaelni: I sent a different way of handling that 64bit overflow in swr, totally untested but seems like its a more pleasant way to deal with it
[14:10] <BBB> (imo)
[14:41] <ubitux> FYI
[14:41] <ubitux> libav is going to drop avserver
[14:41] <ubitux> while i'm not motivated to maintain ffserver, i know that it's used in production by several companies
[14:42] <ubitux> and so it would be a mistake to drop it
[14:43] <ubitux> s/going to/willing to/
[14:43] <cone-870> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/heads/zopan': unknown revision or path not in the working tree.
[14:43] <cone-870> Use '--' to separate paths from revisions
[14:43] <cone-870> refs/heads/zopan:HEAD: cavsdec: fix qp fixed slice handling
[14:44] <cone-870> ffmpeg.git 03Paul B Mahol 07master:4b6cb1cc6901: Zoom & Pan filter
[15:01] <ubitux> mmh it's strange
[15:02] <ubitux> i'm trying to v4l2 ’ webm
[15:02] <ubitux> and it fails pretty quickly&
[15:02] <ubitux> like, it seems to be recording but in fact no
[15:02] <ubitux> mmh probably because it's too slow to encode
[15:17] <michaelni> ubitux, reynaldo, i think we (will) need a regression test for ffserver
[15:18] <michaelni> or fix the disabled one we have
[15:19] <ubitux> yeah definitely
[15:21] <cone-870> ffmpeg.git 03Carl Eugen Hoyos 07master:673716c54b39: avcodec/utvideodec: Increase vlc len
[15:21] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:334aafe56500: avcodec/huffman/ff_huff_gen_len_table: support skiping stat=0 entries
[15:21] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:29a1164e6029: avcodec/utvideodec: fix use of get_vlc2()
[15:21] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:637093a2db53: avcodec/utvideoenc: skip unused huffman entries
[15:25] <reynaldo> michaelni, ubitux: agree
[15:32] <cone-870> ffmpeg.git 03Ronald S. Bultje 07master:f341340552e2: swr: handle 64bit overflow check in multiple_resample().
[15:49] <Daemon404> orite i cant use PKG_CONFIG_LIBDIR to help ffmpeg see libx264
[15:49] Action: Daemon404 stabs C-E
[17:13] <cone-870> ffmpeg.git 03Janne Grunau 07master:7b06ddb8352f: configure: use .altmacro for gnu as check on arm
[17:13] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:383d8ddf2951: Merge commit '7b06ddb8352fe7f434414d7911ced94956bb25b2'
[17:27] <cone-870> ffmpeg.git 03Janne Grunau 07master:1619274fb393: av_dict_set: fix potential memory leak with AV_DICT_DONT_OVERWRITE
[17:27] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:375fa061255a: Merge commit '1619274fb393f55a365cc10f88faa173c9a8e772'
[18:51] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:2ffe55a64f78: avfilter/vf_zoompan: fix mixing of declarations and statements
[18:51] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:2d3842f541bd: avformat: add mjpeg_probe()
[19:14] <jamrial> http://fate.ffmpeg.org/log.cgi?time=20140609151549&log=compile&slot=x86_64-netbsd-clang-noopt is the error at the end ccache's fault?
[19:14] <jamrial> checks like "if (EXTERNAL_SSE(cpu_flags) && ARCH_X86_32)" are used in many other places. I don't see why it would fail here
[19:31] <michaelni> jamrial, fails here on ubunu with "make distclean ; ./configure --cc=clang --disable-optimizations && make -j12" too
[19:32] <michaelni> jamrial, solution found, will push in a moment
[19:37] <cone-870> ffmpeg.git 03Michael Niedermayer 07master:06f576c4ab3b: avcodec/x86/dct_init: fix build failure with clang && disable-optimizations
[20:24] <llogan> from twitter: "time.h is causing problems in Xcode, it's importing that instead of the system version..."
[20:24] <llogan> what i say !!!
[20:47] <BtbN> Is there any documentation on how to use vdpau with libavcodec? The doxygen documentation is not realy usefull on how to actualy use it. Ultimately i want to bind the video to some OpenGL texture.
[21:04] <Daemon404> llogan, that doesnt even make sense
[21:04] <Daemon404> its in a sub-dir
[21:40] <Plorkyeran> xcode has an off-by-default option to search dirs recursively
[21:40] <Plorkyeran> so he probably has that on and unsurprisingly it's breaking shit
[21:42] <Daemon404> that sounds like a terrible idea for an option
[21:43] <Plorkyeran> it's intended to be purely for backwards compatibility with retarded old shit
[21:43] <Daemon404> since when has apple cared about that
[21:44] <Plorkyeran> they old don't care about bc with their own stuff
[21:44] <Daemon404> o
[21:45] <Plorkyeran> and it's more of forward compatibility that they don't care about
[21:45] <Plorkyeran> they make it excessively hard to support anything but the most recent version or two of ios and os x
[21:45] <Plorkyeran> but old software still runs
[21:56] <llogan> Plorkyeran: shall i quote you as the reply to his tweet?
[21:57] <jamrial> why is there a "zopan" branch in the repo?
[21:57] <llogan> Plorkyeran: or i can leave out your name if you prefer
[22:13] <michaelni> j-b, can you delete to zopan branch from ffmpeg repo, iam not sure who pushed that but it shouldnt be there
[22:15] <j-b> I can, but tomorrow. remind me to do it
[22:21] <jamrial> michaelni: thanks for fixing clang. i'm still not sure how checking ARCH_X86_32 first makes a difference, though
[22:25] <michaelni> if(0 && func()) -> func() call can be optimized out if(func() && 0) -> call cannot be optimized out as it may have side effects
[22:25] <michaelni> and clang with disabled optimizations then probably fails to remove the code in the second case i would guess
[22:26] <michaelni> s/the code/the if() body/
[22:33] <iive> hum... afair the C standard demands that the second expression should not be calculated if the first is false
[22:33] <iive> so if this is true, then it is a bug and should be reported to clang.
[22:34] <Daemon404> ?
[22:34] <Daemon404> look closer.
[22:35] <Daemon404> that is exactly what michaelni described.
[22:37] <iive> i thought the problem is in the func() not the body.
[22:37] <iive> gcc also have issue with removing code from the body with -O0
[22:37] <Daemon404> the problem is the body not being optimized out
[22:37] <Daemon404> whhch can only be safely done if the 0 comes first
[22:37] <Daemon404> as per the C spec
[22:39] <wm4> why not replace the if with #if
[22:39] <Daemon404> because ffmpeg relies on DCE
[22:39] <Daemon404> everywhere
[22:39] <wm4> that's stupid
[22:40] <Daemon404> less bad than #if soup IMO, but YMMV
[22:40] <Daemon404> and considering ffmpeg compiles with every compiler under the sun
[22:40] <Daemon404> im not too worried.
[22:47] <iive> Daemon404: could you show me where the C standard says the body of if(0) should be discarded ?
[22:48] <wm4> nobody claimed that
[22:49] <iive> if(0 && func()) and if(func()&&0) are not equivalent. in the seconds func() would always be called.
[22:58] <iive> wm4: then the other statement is not true. you can always safely remove the body of the function in case the if() always evaluates to 0. However the code inside the if() may not be removed (if it changes variables and calls functions).
[22:59] <wm4> what of "ffmpeg relies on DCE" do you not understand
[22:59] <wm4> DCE = dead code elimination
[23:02] <iive> dce is optimization.
[23:03] <wm4> and ffmpeg relies on it
[23:03] <iive> sure.
[23:03] <cehoyos> Hi!, is there a reason why "zopan" cannot be deleted?
[23:04] <wm4> <michaelni> j-b, can you delete to zopan branch from ffmpeg repo, iam not sure who pushed that but it shouldnt be there
[23:04] <wm4> <j-b> I can, but tomorrow. remind me to do it
[23:04] <cehoyos> Thank you!
[23:04] <cehoyos> (I am sure who)
[23:05] <cehoyos> And should we remove "We need a volunteer Debian Package Maintainer." from the topic? It is not very nice to our Debian Package Maintainers...
[23:06] <wm4> are there debian packages now?
[23:06] <cehoyos> http://anonscm.debian.org/gitweb/?p=collab-maint/ffmpeg.git
[23:07] <cehoyos> Done, flame me if I did something wrong...
[23:08] <ubitux> wm4: ffmpeg is in queue since a month
[23:08] <wm4> I hope they post packages soon; there's none of this in unstable or experimental yet
[23:08] <wm4> ok
[23:08] <ubitux> wm4: https://ftp-master.debian.org/new.html
[23:08] <iive> i think they've been posted and await ftp masters approval
[23:08] <cehoyos> They are checking copyrights now.
[23:09] <cehoyos> I wonder who checks copyrights on existing packages...
[23:09] <cehoyos> Anyway, good night!
[23:23] <cone-168> ffmpeg.git 03Billy Shambrook 07master:3c2d9f86c718: ffprobe: add rc_max_rate to stream output
[23:23] <cone-168> ffmpeg.git 03Carl Eugen Hoyos 07master:958168d50569: lavf/mpegts: Reduce number of "Predefined SLConfigDescriptor" warnings shown.
[00:00] --- Tue Jun 10 2014


More information about the Ffmpeg-devel-irc mailing list