[Ffmpeg-devel-irc] ffmpeg-devel.log.20160828

[burek]

[03:37:31 CEST] <kierank> durandal_1707: going to vdd?
[03:56:53 CEST] <kode54> is it to be expected that a developer needs to implement their own pre-roll when seeking?
[03:57:11 CEST] <kode54> when calling avformats seeking functions, that is
[03:57:49 CEST] <kode54> I found that when using FFmpeg in VGMStream, I needed to implement a maximum of 2 seconds of pre-roll (or up to the beginning of the file, if its less than that)
[03:58:07 CEST] <kode54> or else the WMA Pro decoder would ramp up from silence
[03:59:03 CEST] <kode54> so I manually pre-roll and discard samples at the decode stage
[04:20:03 CEST] <Compn> kierank : no, durandal_1707 said he isnt. i hope he shows though
[04:24:24 CEST] <Compn> kode54 : sorry, not sure, try libav-user mailing list, for people using ffmpeg in different projects. http://ffmpeg.org/pipermail/libav-user/
[07:24:01 CEST] <kode54> I also had another question, but Ive since posted it to the mailing list
[07:24:19 CEST] <kode54> specific to the use of the AAC decoder, versus parsers for ADTS or ADIF
[11:10:15 CEST] <cone-906> ffmpeg 03Paul B Mahol 07master:88bcdf109a44: avfilter: hflip,swapuv,vflip: add timeline support
[11:22:14 CEST] <cone-906> ffmpeg 03Jai Luthra 07master:0c023d181e58: lavc/lpc: Add min_shift parameter in LPC
[12:13:20 CEST] <BtbN> michaelni, if "make fate-source" doesn't throw an error anymore, I should be fine?
[12:13:46 CEST] <BtbN> Or is there something else that needs changing except for the error it shows
[12:15:29 CEST] <nevcairiel> you just need to add the exception line since its not a ffmpeg header, but third-party
[12:16:12 CEST] <BtbN> Yeah, that's what I did.
[13:40:18 CEST] <durandal_1707> can I get reviews? Do I need to bribe someone?
[13:52:16 CEST] <cone-906> ffmpeg 03Davinder Singh 07master:fecf5ae9aa74: MAINTAINER: add myself for Motion Estimation and Interpolation filters
[14:49:17 CEST] <Compn> durandal_1707 : kierank wants you to come to vdd
[14:50:30 CEST] <durandal_1707> I want vdd/kierank comes to me
[16:04:44 CEST] <cone-906> ffmpeg 03Marton Balint 07master:13b90ff2c127: avformat: fix decoded creation_time timestamps
[17:11:30 CEST] <BtbN> I really wish Nvidia would release some GT Multimedia Updated cards
[17:11:40 CEST] <BtbN> Kepler is still the most recent you can get there
[17:12:09 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:325e56479ff6: avcodec/nvenc: include nvEncodeAPI v7 SDK header
[17:12:10 CEST] <cone-906> ffmpeg 03Oliver Collyer 07master:d1bf8a3aa878: avcodec/nvenc: added support for 10 bit HEVC encoding
[17:12:11 CEST] <cone-906> ffmpeg 03Oliver Collyer 07master:a81b398e869e: avcodec/nvenc: added support for rate control lookahead
[17:17:37 CEST] <durandal_1707> michaelni: how to properly take frames once they are returned by separate thread?
[17:18:03 CEST] <durandal_1707> in lavfi, attempting to write mt frame support
[18:11:34 CEST] <durandal_1707> michaelni: every thread instance needs own private context.
[18:12:50 CEST] <durandal_1707> because each filter need allocate own temp memory 
[18:13:26 CEST] <durandal_1707> and I don't see nice way to do it
[18:31:37 CEST] <cone-906> ffmpeg 03Marton Balint 07master:2ee8a4f8873b: ffmpeg: fix -stream_loop with -re
[18:36:22 CEST] <BtbN> MSVC finds an amazing amount of warnings in ffmpeg.
[18:49:01 CEST] <JEEB> BtbN: clang probably as well
[18:49:15 CEST] <BtbN> most of them are bullshit though
[18:49:29 CEST] <JEEB> yes, although clang's % of non-bullshit is probably higher
[18:49:33 CEST] <JEEB> MSVC also has static analysis
[18:49:38 CEST] <JEEB> (ok, clang has it as well)
[18:49:45 CEST] <BtbN> There are a few valid ones, but most of them are pointless
[18:50:32 CEST] <nevcairiel> it doesnt help that ffmpeg builds with W4, which really includes all sorts of warnings that most people would find "normal code"
[18:50:39 CEST] <JEEB> yup
[18:50:48 CEST] <JEEB> -Wall -Wextra level
[18:51:15 CEST] <BtbN> not even that
[18:51:24 CEST] <BtbN> it's _way_ above -Wall -Wextra -pedantic
[18:51:25 CEST] <nevcairiel> more like -Wpedantic
[18:51:35 CEST] <nevcairiel> no W in that? oh well
[18:52:06 CEST] <BtbN> for extra fun, add -pedantic-errors
[18:52:08 CEST] <JEEB> I wonder how many times someone has tried adding -Werror to FFmpeg compilation
[18:52:22 CEST] <JEEB> and then facedesked hard
[18:52:33 CEST] <BtbN> Werror is just plain wrong to add to productive cflags.
[18:52:51 CEST] <BtbN> You don't want a compiler update with new warnings break your build for no good reason.
[18:53:06 CEST] <JEEB> I know :P
[18:53:27 CEST] <JEEB> I just wondered out loud if anyone had even tried that without looking at what FFmpeg produces during compilation
[18:53:31 CEST] <BtbN> It's also lovely that it's impossible to change the lanauge of MSVC warnings
[18:53:38 CEST] <BtbN> They are all german, with no way to change that.
[18:53:39 CEST] <JEEB> as in, a random person maintaining a random package
[18:54:18 CEST] <nevcairiel> dont install german msvc if you dont want german warnings =p
[18:54:26 CEST] <BtbN> It is english msvc
[18:54:29 CEST] <BtbN> but on a german windows
[18:54:37 CEST] <BtbN> and that's what determines the compiler output apparently
[18:54:54 CEST] <nevcairiel> well dont use gmerna windows then!
[18:55:19 CEST] <JEEB> I think it might be using the language for non-unicode applications thing
[18:55:22 CEST] <JEEB> not sure, though
[18:55:22 CEST] <iive> can't you switch windows language?
[18:55:26 CEST] <JEEB> you can
[18:55:30 CEST] <BtbN> I don't want to switch Windows language.
[18:55:44 CEST] <JEEB> anyways, try switching the code page for non-unicode apps
[18:55:50 CEST] <JEEB> not sure if that'll do the trick
[18:55:52 CEST] <JEEB> but worth a try
[18:56:04 CEST] <iive> nah... 
[18:56:18 CEST] <iive> whey are not guessing the language from the codepage
[18:56:28 CEST] <JEEB> you'd be surprised how many do that :P
[18:56:39 CEST] <JEEB> it's called "locale" in some things even
[18:56:50 CEST] Action: iive facedesk
[18:56:57 CEST] Action: iive facedesks
[18:57:35 CEST] <JEEB> oh, and it's called "change system locale" in the control panel thing as well
[18:58:22 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:19e75fd8838d: avcodec/nvenc: fix library names on cygwin
[18:58:23 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:a19989cae581: avcodec/nvenc: fix potantially uninitialized free
[18:58:46 CEST] <nevcairiel> presumably you can control this in visual studio options -> environment -> international settings
[18:58:54 CEST] <nevcairiel> not sure if this only applies to the gui or what
[18:58:57 CEST] <BtbN> That changes the IDE language
[18:58:57 CEST] <nevcairiel> since mine is all english
[18:59:01 CEST] <BtbN> not the compiler output
[18:59:22 CEST] <BtbN> I have spent quite some time trying to figure this out, and it is hard-tied to the system language.
[18:59:36 CEST] <nevcairiel> well then its your fault for installing german windows
[18:59:36 CEST] <JEEB> yeah, I just tried and it's not the system locale
[18:59:50 CEST] <JEEB> because my locale is .jp and I'm getting en_us messages
[19:00:06 CEST] <BtbN> I don't want an English Windows. But English compiler messages. Which is plain impossible.
[19:00:59 CEST] <nevcairiel> supposedly you can delete/rename the locale dir and it falls back to english
[19:06:50 CEST] <BtbN> https://github.com/FFmpeg/FFmpeg/blob/master/compat/nvenc/nvEncodeAPI.h#L3078 https://github.com/FFmpeg/FFmpeg/blob/master/compat/nvenc/nvEncodeAPI.h#L117 do i interpret this right, that these don't match up? oO
[19:30:08 CEST] <durandal_1707> michaelni: I got idea how to start, put thread_crc to AVframeInternal
[19:30:23 CEST] <durandal_1707> *thread_ctx
[19:58:31 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:df615efcf275: avcodec/nvenc: check maximum driver API version
[19:58:32 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:26a5cbd7811e: avcodec/nvenc: use proper soname for cuda/nvenc libraries
[19:58:33 CEST] <cone-906> ffmpeg 03Timo Rothenpieler 07master:cac2df230e73: avcodec/nvenc: update license header
[20:03:38 CEST] <michaelni> durandal_1707, possible yes
[21:18:40 CEST] <cone-906> ffmpeg 03Thomas Hebb 07master:a37e6dd2ba85: avformat/mov: aax: pass proper AVClass to av_log()
[23:25:08 CEST] <durandal_1707> michaelni: so what would ff_filter_frame do if run inside worker thread? I need to cache input frames and output frames?
[23:42:57 CEST] <michaelni> durandal_1707, i didnt think deeply about it but if you have 2 filters each filtering multiply frames with a thread per frame. A worker passing a frame on would likely put it in a fifo (possibly of size 1 frame) where the next filter would take it from when the next thread is available or so
[23:49:31 CEST] <durandal_1707> michaelni: and how handle ordering? using pts is fragile
[23:50:31 CEST] <nevcairiel> like frame threading in avcodec, make the threads themself fifo ordered
[23:50:52 CEST] <michaelni> something would need to keep track of order
[23:52:53 CEST] <durandal_1707> eh, I think I got it
[00:00:00 CEST] --- Mon Aug 29 2016