[Ffmpeg-devel-irc] ffmpeg-devel.log.20150926

burek burek021 at gmail.com
Sun Sep 27 02:05:04 CEST 2015


[00:07:17 CEST] <baptiste> is the experimental flag for the aac encoder removed yet ? :)
[00:12:22 CEST] <atomnuker> not yet
[00:20:16 CEST] <nevcairiel> claudio still busy, I assume
[00:20:58 CEST] <atomnuker> yes, he's busy working on the encoder
[00:21:06 CEST] <Compn> ooo baptiste , long time no see :)
[00:22:17 CEST] <iive> are there still critical bugs remaining?
[00:24:22 CEST] <atomnuker> problem is the amount of testing Claudio does on every single patch is an overkill
[00:24:54 CEST] <baptiste> the work is awesome though 
[00:25:01 CEST] <baptiste> hey Compn :)
[00:25:28 CEST] <Compn> atomnuker : yes, we are all amazed someone fixed up aac encoder. thats why i was so interested in your working on it
[00:41:50 CEST] <DHE> I think I'm ready to submit my patch to the mailing list
[00:57:55 CEST] <jamrial> just found a case of xmm register clobbering on libswresample that the fate suit didn't catch with --enable-xmm-clobber-test
[00:58:12 CEST] <jamrial> because we apparently have little to no coverage of multichannel tests
[01:01:05 CEST] <BBB> checkasm!!!!!
[01:01:09 CEST] <BBB> dude, seriously
[01:01:12 CEST] <BBB> checkasm all these functions
[01:01:39 CEST] <jamrial> heh
[01:07:52 CEST] <BBB> I see you dont like the idea :-p
[01:19:05 CEST] <jamrial> i might give it a try. this is the audioconvert module which is a lot like the flac dsp i added to checkasm the other day
[01:19:43 CEST] <jamrial> although it deals with float audio, so a checkasm test may not be feasible
[01:20:37 CEST] <Gramner> it should handle floats fine now
[01:26:34 CEST] <cone-387> ffmpeg 03Christophe Gisquet 07master:e19ea0218aee: dnxhddec: indicate colorspace
[01:50:02 CEST] <Compn> carl : did you mirror these new samples yet? https://trac.ffmpeg.org/ticket/2690#comment:26
[02:18:23 CEST] <cone-387> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:5c1acf0a09b0: ffserver: rm whitespace c&p leftovers from macro
[02:18:24 CEST] <cone-387> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:83411d73313e: ffserver: drop error counting when parsing ACL row
[02:18:25 CEST] <cone-387> ffmpeg 03Reynaldo H. Verdejo Pinochet 07master:086718e8f783: ffserver: drop unneeded branching
[03:39:30 CEST] <cone-387> ffmpeg 03Rodger Combs 07master:f559812a8437: tests/checkasm: make randomize_buffers a function for easier debugging
[09:51:56 CEST] <cone-626> ffmpeg 03Claudio Freire 07master:0f98fd30e2d3: AAC encoder: fix OOB access in search_for_pns
[10:02:37 CEST] <TheRyuu> v2'ed the HEASLR patch and sent the seperate dynamicbase/nxcompat combined patch to the ML
[11:08:51 CEST] <cone-626> ffmpeg 03wm4 07master:f290e48d86e1: mmal: drop the h264 BSF
[11:08:52 CEST] <cone-626> ffmpeg 03wm4 07master:49623f531972: mmal: Remove setting extradata on input format
[11:08:53 CEST] <cone-626> ffmpeg 03wm4 07master:a9b8c638cfe2: mmal: Fix AVBufferRef usage
[11:08:54 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:c5a312dd6afb: Merge commit 'f290e48d86e10f34b5ddc519127636bcebec7c43'
[11:08:55 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:f3310df8538d: Merge commit '49623f531972be5dc2dd8c1b4b8748cad7c424ff'
[11:08:56 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:2b61cc44433e: Merge commit 'a9b8c638cfe2f82191db65e3e3a39f3b35df81f5'
[11:15:55 CEST] <cone-626> ffmpeg 03Vittorio Giovara 07master:588a5619da0d: dxv: Parse ancillary encoder information
[11:15:56 CEST] <cone-626> ffmpeg 03Vittorio Giovara 07master:fb2889691cb7: dxv: Support the original first version
[11:15:57 CEST] <cone-626> ffmpeg 03Vittorio Giovara 07master:bbf71d46db34: dxv: Print texture information after header parsing
[11:15:58 CEST] <cone-626> ffmpeg 03Vittorio Giovara 07master:1bcd4a476ba4: dxv: Support RAW intermediate compression
[11:15:59 CEST] <cone-626> ffmpeg 03Vittorio Giovara 07master:b2417ee6d1ee: dxv: Improve error message
[11:16:00 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:d0aec0aae872: Merge commit '588a5619da0d041e55b365f63d0fa9c72bdbd4d3'
[11:16:01 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:234c7378ca19: Merge commit 'fb2889691cb7720d2680e188eb6036a35afa2392'
[11:16:02 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:f6048e4920de: Merge commit 'bbf71d46db3417b43bcbd745cbf235e8e2ff69ae'
[11:16:03 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:b4ea8a730521: Merge commit '1bcd4a476ba45a7fdf59d1701b8f0e274418cc32'
[11:16:04 CEST] <cone-626> ffmpeg 03Hendrik Leppkes 07master:9c3c8d2c5602: Merge commit 'b2417ee6d1ee0c5e9b170a642d73bdf68908966f'
[11:27:54 CEST] <anshul> is there any encryption or encoding done when mails are sent to mailing list, because mails from ffmpeg-devel mailing list when saved to .eml file looks like encrypted
[11:35:41 CEST] <TheRyuu> isn't there some base64 stuff going on
[12:21:08 CEST] <Compn> anshul : probably just your mail client compressing emails
[12:32:32 CEST] <DHE> anshul: could be user error. (ie mine)
[12:32:50 CEST] <nevcairiel> the patch on the ML is fine
[12:32:59 CEST] <nevcairiel> its user error alright, but not yours =p
[12:34:12 CEST] <nevcairiel> well, mostly fine. copy-pasting a patch into a mail will often break,  so either use git send-email to send it directly as a mail, or just attach it
[12:38:38 CEST] <DHE> yeah I'll do that next time...
[12:39:11 CEST] <DHE> anyway, that closed caption patch has been tested (by me) on a set top box vendor and some mobile video players.
[12:49:54 CEST] <cone-626> ffmpeg 03Ronald S. Bultje 07master:7a4b97e946f5: checkasm: clip vp9 loopfilter test pixels inside allowed bitdepth range.
[12:49:58 CEST] <anshul> DHE: testing that new attached patch
[12:50:35 CEST] <anshul> and with which set top box you tested it, just for info
[12:50:48 CEST] <DHE> anshul: Entone Kamai 500 and Amulet 450 series
[12:51:08 CEST] <DHE> also is there a command that checks the whitespace thing?
[12:52:24 CEST] <anshul> I get those warning with git apply, and if you reset and do git diff then it show in red marks
[12:52:57 CEST] <JEEB> anshul: just fyi the ARIB (ISDB) captions are in a separate PID so they're not together with the things that use stuff within the video stream
[12:53:15 CEST] <JEEB> unless the countries that adopted ISDB that are not japan went for something else :D
[12:54:28 CEST] <DHE> why have I never seen that before?
[12:54:49 CEST] <DHE> I'm gonna blame github for making me complacent
[12:56:23 CEST] <anshul> JEEB: yes I know, but there ts is totally different, i ran into lot of problem when I was implemting closed caption and using ffmpeg mpegts muxer
[12:57:15 CEST] <anshul> so I was asking if we have some flags in ts that could suggest whether they are dvb or isdb
[12:58:47 CEST] <anshul> Compn: I see that message source same in web gmail ui and thunderbird, and if it is compression then why its done only to mails from mailing list
[13:00:46 CEST] <anshul> DHE: it was not your error on that mail since i see source of simple mails with question in ffmpeg-user, i see them encrypted if i save it as .eml or in gmail use show source
[13:05:41 CEST] Action: DHE is testing out how git-send-email works now...
[13:16:21 CEST] <anshul_> DHE: I am not able to use the command here is pastebin http://pastebin.com/gqjJkrzC
[13:17:40 CEST] <anshul_> am i using that option incorrect
[13:19:44 CEST] <DHE> use "-a53cc 1"
[13:20:02 CEST] <DHE> -x264opts are given the x264 library directly and it doesn't have such a command
[13:23:08 CEST] <anshul_> ok got it thanks
[13:24:19 CEST] <DHE> does this suggest the doc part needs fixing?
[13:24:26 CEST] <anshul_> DHE: if you find it ok then add the command with -a53cc in encoder.texi
[13:24:44 CEST] <anshul_> yes there may be more lame user like me
[13:26:52 CEST] <anshul_> I tested it your patch works fine with ccextrator too. :)
[13:27:44 CEST] <DHE> now it shows "a53cc _boolean_". the description still says 0 means off so I'm hoping that's enough.
[13:30:00 CEST] <DHE> still messing with send-email before I actually use it to post to the list
[13:31:17 CEST] <anshul_> DHE: generally I dont use git send-email, i prefer git format-patch -1. but majority of people here prefer git-send-email
[13:51:36 CEST] <cone-626> ffmpeg 03Michael Niedermayer 07master:10bbf6cf622f: avcodec/ffv1dec: Explicitly check read_quant_table() return value
[13:51:37 CEST] <cone-626> ffmpeg 03Michael Niedermayer 07master:2d221d9e069e: avcodec/ffv1dec: Fix off by 1 error in quant_table_count check
[13:53:03 CEST] <BtbN> I guess the OpenCL maintainer(s) isn't/aren't the most active
[14:14:18 CEST] <mahesh> Hi hello i am mahesh . i would like to start contributing to ffmpeg
[14:14:26 CEST] <mahesh> how should i start with ?
[14:14:35 CEST] <mahesh> got my git cloned . THen?
[14:14:41 CEST] <Compn> mahesh : try building ffmpeg
[14:14:45 CEST] <JEEB> start by compiling it, and then looking at what you wanted to work on
[14:14:47 CEST] <Compn> make sure you can build and test ffmpeg.
[14:15:05 CEST] <JEEB> for the tests you do need the FATE suite though, so that can take quite a bit of time :P
[14:15:13 CEST] <JEEB> or at least a part of the suite
[14:15:23 CEST] <mahesh> any page how to compile , build 
[14:15:25 CEST] <mahesh> ?
[14:15:28 CEST] <mahesh> Please
[14:15:39 CEST] <Compn> mahesh : yes on ffmpeg.org , one sec
[14:15:54 CEST] <Compn> mahesh : https://trac.ffmpeg.org/wiki/CompilationGuide
[14:16:06 CEST] <mahesh> thank you comp
[14:16:15 CEST] <JEEB> those compilation guides tend to have the kitchen sink, it's really not that hard :)
[14:16:20 CEST] <JEEB> what OS are you on?
[14:16:27 CEST] <mahesh> i am on ubuntu
[14:16:40 CEST] <JEEB> then you just need build-essential(s) and yasm
[14:17:12 CEST] <JEEB> then make an ffmpeg_build directory somewhere and run the configure script from the cloned directory
[14:17:41 CEST] <mahesh> yes 
[14:17:43 CEST] <mahesh> thank you 
[14:17:46 CEST] <JEEB> so for example if I have directories ffmpeg and ffmpeg_build next to each other, I go into ffmpeg_build and do `../ffmpeg/configure`
[14:17:50 CEST] <JEEB> that should be enough
[14:18:01 CEST] <JEEB> it should output the configuration
[14:18:12 CEST] <JEEB> and after that by using make you should be able to get it building
[14:19:01 CEST] <mahesh> referring to the document or page  i can do it right ? without hiccups
[14:19:46 CEST] <JEEB> yes, the document should be alright, but it has all kinds of libraries etc. there :P
[14:19:56 CEST] <mahesh> yes thank you 
[14:20:09 CEST] <mahesh> do i need to take all the libraries
[14:20:12 CEST] <JEEB> no
[14:20:21 CEST] <JEEB> as I said you can just have build-essential and yasm installed
[14:20:26 CEST] <JEEB> and then you can run the configure script
[14:20:42 CEST] <JEEB> that will give you a basic configuration ffmpeg
[14:21:23 CEST] <JEEB> you might need to add more libraries depending on if you're working on a wrapper or whatever, but the very basic thing is what you should start with
[14:22:14 CEST] <mahesh> superb ... thats a lot help jeeb
[14:22:19 CEST] <mahesh> thank you so much
[14:22:30 CEST] <JEEB> lorf
[14:22:50 CEST] <Compn> and thus another developer was born
[14:22:56 CEST] <Compn> good job JEEBster
[14:23:53 CEST] <JEEB> I can never remember if the package is build-essential or build-essentials, and I always end up ssh'ing into a ubuntu box and searching
[14:24:08 CEST] <BtbN> build-esse<TAB>
[14:26:45 CEST] <JEEB> BtbN: that only works if you're already in a terminal so the ssh part doesn't change :P
[14:26:58 CEST] <JEEB> also it only works if you're not already sudo -s -H'd as root
[14:27:15 CEST] <BtbN> why wouldn't bash-completion work as root?
[14:27:31 CEST] <JEEB> many things don't enable it for the root users
[14:27:50 CEST] <JEEB> *user
[14:27:57 CEST] <JEEB> if you do just sudo -s it works in ubuntu
[14:28:02 CEST] <JEEB> but -H strips that
[14:28:26 CEST] <JEEB> I think debian's sudo default might be different, I think there even without -H it doesn't work
[14:28:47 CEST] <BtbN> Use -i then
[14:29:23 CEST] <JEEB> but yeah, I agree there's enough of ways :)
[14:37:12 CEST] <mahesh> i am compiling ffmpeg . can i directly install those packages without needing to build , but directly from sudo apt-get install ? for eg:yasm,libx264-dev
[14:37:52 CEST] <durandal_1707> BtbN: you have questions about keyspill removal for colorkey filter
[14:38:51 CEST] <JEEB> mahesh: if you haven't even built basic ffmpeg yet, just install yasm from apt-get for now (it's a tool btw, not a library used in ffmpeg) together with build-essential
[14:38:56 CEST] <BtbN> I haven't bothered with that at all yet.
[14:39:11 CEST] <BtbN> The results are quite ok with out it already though
[14:39:17 CEST] <mahesh> what about other libs ?
[14:39:26 CEST] <mahesh> @jeeb
[14:39:50 CEST] <mahesh> libx264-dev
[14:40:00 CEST] <BtbN> I think i'd have an idea how to do it with the colorkey filter, as it's RGB based. But not too sure about (Y)UV chromakey.
[14:40:58 CEST] <BtbN> And the colorkey filter is the one where it basicaly doesn't matter, as it's intended for removing pink backgrounds from static images and stuff like that.
[14:41:14 CEST] <iive> does somebody know why firefox includes libvpx encoder and what does it use it for?
[14:41:46 CEST] <BtbN> Because they are years behind on their audio/video agenda.
[14:42:41 CEST] <iive> my question is more like... why they need encoder... 
[14:43:00 CEST] <JEEB> mahesh: your first thing is to get ffmpeg itself building. don't care about any other libs for now
[14:43:08 CEST] <JEEB> I mean, you get most decoders within ffmpeg itself
[14:43:13 CEST] <JEEB> and a lot of encoders
[14:43:26 CEST] <JEEB> also please don't send private messages to me
[14:43:29 CEST] <JEEB> it's rude
[14:43:52 CEST] <iive> BtbN:  and to be clear... perf top indicates that vp9_ssim_parms_8x8_sse2 is the main function slowing down my startup/shutdown .
[14:44:25 CEST] <mahesh> Jeeb : sorry 
[14:44:55 CEST] <mahesh> Jeeb : so just with the above yasm and the build essentials are enough . no need to go below it for installation
[14:44:58 CEST] <mahesh> ?
[14:45:29 CEST] <JEEB> yes, after that you should have the configure script work without any parameters
[14:45:47 CEST] <JEEB> and a nice output of all decoders, encoders, demuxers etc. that are enabled with that configuration
[14:48:50 CEST] <BtbN> iive, uhm, what?
[14:49:07 CEST] <iive> BtbN: yeh... my question exactly.
[14:50:59 CEST] <iive> BtbN: something in firefox uses the encoder, and that function works a lot.... for not so short periods of time, hanging the browser...
[14:51:42 CEST] <mahesh> Jeeb : wget http://ffmpeg.org/releases/ffmpeg-snapshot.tar.bz2 . Is this necessary ?
[14:52:14 CEST] <JEEB> didn't you say that you already cloned it?
[14:52:19 CEST] <JEEB> I mean, FFmpeg
[14:52:59 CEST] <JEEB> if you are going to try and develop FFmpeg then you will most probably want to use git to be able to work with your changes and FFmpeg updates easily :P
[14:53:22 CEST] <JEEB> for just compilation of a random ffmpeg version that of course doesn't matter
[14:53:22 CEST] <mahesh> thats the place where i had confusion . because i had git cloned 
[14:53:38 CEST] <JEEB> then you just use the git clone you have
[14:53:58 CEST] <mahesh> thank you jeeb ... 
[14:54:00 CEST] <JEEB> it's not rocket science: have build-essential and yasm, then run the configure script within the FFmpeg code :P
[14:54:21 CEST] <JEEB> then you get the output of the configure script on the terminal, most probably without errors and showing you a very long list of stuff that was enabled
[15:06:14 CEST] <durandal_1707> iive: its encoding your camera video and seending away
[15:06:42 CEST] <iive> durandal_1707: good thing i don't have camera plugged in :P
[15:20:06 CEST] <cone-626> ffmpeg 03Henrik Gramner 07master:2ab65b652dc5: checkasm: Use a self-balancing tree
[15:34:26 CEST] <cone-626> ffmpeg 03Paul B Mahol 07master:4cf60b4fa12f: avfilter/vf_blend: add addition128 mode
[16:44:03 CEST] <cone-626> ffmpeg 03Christophe Gisquet 07master:e4b7c3a9f21b: dnxhddec: proper rule for interlaced mb flag
[17:44:09 CEST] <cone-626> ffmpeg 03Christophe Gisquet 07master:dc218bd08305: dnxhddec: parse and print adaptive color transform
[18:44:42 CEST] <cone-626> ffmpeg 03Ganesh Ajjanagadde 07master:db9de9b954a7: avformat/hls: remove unused function
[18:53:55 CEST] <Gramner> is anything other than two's complement supported in ffmpeg? e.g. can I use -1 to set all bits in an integer?
[18:54:20 CEST] <wm4> (unsigned type)-1 should always be valid to set all bits
[18:54:32 CEST] <wm4> (even in 100% language-lawyer conformant C)
[18:55:48 CEST] <BtbN> Are there seriously platforms that don't use it?
[18:55:54 CEST] <wm4> no
[18:57:44 CEST] <Gramner> I belive two's complement, one's complement, and sign-magnitude are allowed according to the C standard
[19:00:26 CEST] <nevcairiel> that concept is used in other places as well
[19:00:30 CEST] <nevcairiel> so just do
[19:00:40 CEST] <Gramner> ok, thanks
[20:34:53 CEST] <cone-626> ffmpeg 03Timo Rothenpieler 07master:31ee86cd9869: avutil/opencl: Fix volatile pointer
[20:43:25 CEST] <iive> Gramner: two's complement is mandatory for ffmpeg.
[20:45:18 CEST] <Gramner> i see, good to know
[20:47:28 CEST] <wm4> iive: where does it say that?
[21:11:01 CEST] <kurosu> IEEE754 32bits floats is also a requirement in ffmpeg - a lot of code does bitwise operations according to this
[21:11:27 CEST] <kurosu> (audio codecs obviously do it a lot)
[21:36:11 CEST] <cone-626> ffmpeg 03James Almer 07master:af990d72b7e3: checkasm/Makefile: add missing testclean target
[21:36:12 CEST] <cone-626> ffmpeg 03James Almer 07master:4e03f0ab08e2: checkasm/vp9dsp: add const to suppress "discards const qualifier" warnings
[21:51:00 CEST] <J_Darnley> iive Gramner wm4: I'm sure I've read that two's complemet is required, but I can't find it in any of the files in git.
[22:06:02 CEST] <Gramner> maybe it should be documented somewhere if that's the case. I know that x264 does, along with signed right-shift and >=32 bit int
[22:15:49 CEST] <J_Darnley> Ah, maybe it was in x264 some where that I read it.
[22:16:17 CEST] <J_Darnley> oh yes: doc/standards.txt
[22:16:24 CEST] <cone-626> ffmpeg 03Henrik Gramner 07master:ad9a543e9351: avutil/avstring: Inline some tiny functions
[22:18:15 CEST] <kurosu> ">=32 bit int" is also a requirement in ffmpeg
[22:19:09 CEST] <kurosu> some code used (int scalar product dsp) or is using int instead of int32_t
[00:00:00 CEST] --- Sun Sep 27 2015


More information about the Ffmpeg-devel-irc mailing list