[Ffmpeg-devel-irc] ffmpeg-devel.log.20150309

burek burek021 at gmail.com
Tue Mar 10 02:05:02 CET 2015


[00:06:37 CET] <cone-370> ffmpeg 03Vittorio Giovara 07master:7b9cb7b36543: avprobe: Export coded_{width,height} in -show_streams
[00:06:38 CET] <cone-370> ffmpeg 03Michael Niedermayer 07master:fa14685a578d: Merge commit '7b9cb7b36543c2a0e46d99b0e41824b9e7dd0c8f'
[00:26:38 CET] <cone-370> ffmpeg 03Luca Barbato 07release/2.4:cf3523c6e7dd: prores: Extend the padding check to 16bit
[00:26:39 CET] <cone-370> ffmpeg 03Reinhard Tartler 07release/2.4:7c1fe3161769: Prepare for 11.3 Release
[00:26:40 CET] <cone-370> ffmpeg 03Reinhard Tartler 07release/2.4:fc3c1156e361: doc: Update changelog for v11.3
[00:26:41 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:d4afc388e38e: Merge commit 'cf3523c6e7dde33a513e003639d5a8c0b7f3a49d' into release/2.4
[00:26:42 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:d8f6c5a79c10: Merge commit '7c1fe31617699ddefe6b0f39f16e7c3d79e998e2' into release/2.4
[00:26:43 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:61b5dee49e74: Merge commit 'fc3c1156e361202ab97ad63ffb4dacc416906d33' into release/2.4
[00:43:24 CET] <cone-370> ffmpeg 03Martin Storsjö 07release/2.4:450b02307cb6: arm: Suppress tags about used cpu arch and extensions
[00:43:25 CET] <cone-370> ffmpeg 03Luca Barbato 07release/2.4:7136a0bf88f3: vorbis: Check the vlc value in setup_classifs
[00:43:26 CET] <cone-370> ffmpeg 03Vittorio Giovara 07release/2.4:f3dafb63d058: aic: Fix decoding files with odd dimensions
[00:43:27 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:0abd926d64a7: Merge commit '450b02307cb631f501793b52b98b610c3a54378b' into release/2.4
[00:43:28 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:d208977cd1c4: Merge commit '7136a0bf88f31bb8d40a3bbd251963706fb14578' into release/2.4
[00:43:29 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:e2c82f7d9362: Merge commit 'f3dafb63d05896aacf84caf0e4c81c216476d60e' into release/2.4
[01:18:23 CET] <cone-370> ffmpeg 03Anton Khirnov 07release/2.4:77eb3d9a60a9: tiff: Check that there is no aliasing in pixel format selection
[01:18:24 CET] <cone-370> ffmpeg 03Vittorio Giovara 07release/2.4:4070e02dfcf8: configure: Properly fail when libcdio/cdparanoia is not found
[01:18:25 CET] <cone-370> ffmpeg 03Federico Tomassetti 07release/2.4:2c63081b48d9: mdec: check for out of bounds read
[01:18:26 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:16c838b69a4e: Merge commit '77eb3d9a60a9c2bb6d87c960ac186af242bbcc9e' into release/2.4
[01:18:27 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:1f44b09d58d9: Merge commit '4070e02dfcf8c7d871b4a41d8b591ec0c130c70a' into release/2.4
[01:18:28 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:bb2a5ea2cf10: Merge commit '2c63081b48d98f3a0d0bed7b0ec3c0347b99144c' into release/2.4
[01:32:41 CET] <cone-370> ffmpeg 03Federico Tomassetti 07release/2.4:e818da772401: eamad: check for out of bounds read
[01:32:42 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:8aee35acb1b4: rv10: check size of s->mb_width * s->mb_height
[01:32:43 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:905172d75c9c: webp: validate the distance prefix code
[01:32:44 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:5bd9c540d6a6: Merge commit 'e818da77240146b36d6669b1c4e0565239dc55d3' into release/2.4
[01:32:45 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:d08db138e210: avcodec/rv10: check size of s->mb_width * s->mb_height
[01:32:46 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:f2a84d04331d: Merge commit '8aee35acb1b40e51a4fc8d7f7c561088e25d6d2e' into release/2.4
[01:32:47 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:897a51f47b38: avcodec/webp: validate the distance prefix code
[01:32:48 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:c99b8d0cd299: Merge commit '905172d75c9cfd93c757b09fa4b8afa0e926a13c' into release/2.4
[01:50:24 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:2ef2f60b4f03: rmenc: limit packet size
[01:50:25 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:a73b2c288e3d: adxdec: set avctx->channels in adx_read_header
[01:50:26 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:73ca672fb6e6: avformat/rm: limit packet size
[01:50:27 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:20655dd2ec1b: Merge commit '2ef2f60b4f0308d1c871091c9c1a9641d14ec585' into release/2.4
[01:50:28 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07release/2.4:72f83ad277ca: avformat/adxdec: set avctx->channels in adx_read_header
[01:50:29 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:cda143aaa46f: Merge commit 'a73b2c288e3dace6e054a5b48640978be1d5df84' into release/2.4
[02:02:05 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:3a417a86b330: utvideodec: Handle slice_height being zero
[02:02:06 CET] <cone-370> ffmpeg 03Michael Niedermayer 07release/2.4:2f323df8339b: Merge commit '3a417a86b330b7c1acf9db4f729be7d619caaded' into release/2.4
[02:16:19 CET] <cone-370> ffmpeg 03Michael Niedermayer 07master:0ce3a0f9d952: utvideodec: Handle slice_height being zero
[02:16:20 CET] <cone-370> ffmpeg 03Michael Niedermayer 07master:b02323017090: Merge commit '0ce3a0f9d9523a9bcad4c6d451ca5bbd7a4f420d'
[02:38:25 CET] <Bigpet> alright, that was more complicated than it should have been, it took me like 6 tries and each time was like a 1 hour + run on my lappy http://fate.ffmpeg.org/report.cgi?time=20150309011512&slot=x86-Win10-msvc-19.00.22609
[02:39:13 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07master:66879ee12594: ffmdec: initialize f_cprv, f_stvi and f_stau
[02:39:32 CET] <jamrial> it's shouldn't take 1 hour with that cpu
[03:23:30 CET] <cone-370> ffmpeg 03Andreas Cadhalpun 07master:6fa98822eba5: ffmdec: fix infinite loop at EOF
[07:13:15 CET] <builder_> Hi
[07:13:45 CET] <builder_> I have question about NVENC and Intel QSV
[07:14:39 CET] <builder_> both are non-free license like like libfdk_aac?
[07:16:10 CET] <builder_> If enable NVENC or intel QSV, can not distribution to public freely ?
[10:44:05 CET] <cone-112> ffmpeg 03Carl Eugen Hoyos 07master:bf4bd427dd31: lavc/ffv1enc: Auto-select -coder 1 for >8bit also for yuv.
[10:48:01 CET] <cone-112> ffmpeg 03Carl Eugen Hoyos 07master:b69238d9badc: Fix the intra_dc_precision API doc.
[10:52:41 CET] <cone-112> ffmpeg 03Carl Eugen Hoyos 07master:4a94f7b22f24: doc: Improve video creation examples (from image2).
[10:58:39 CET] <cone-112> ffmpeg 03Carl Eugen Hoyos 07master:6453fddb58a0: doc: Clarify documentation for fade duration.
[11:22:11 CET] <cone-112> ffmpeg 03Martin Storsjö 07master:00d751d4fc20: movenc: Set tfhd default sample flags based on actual samples, if possible
[11:22:12 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:e8821e74bb95: Merge commit '00d751d4fc20ec88d2cc2c9f39ec8b9e9c8cdeba'
[11:36:10 CET] <cone-112> ffmpeg 03Martin Storsjö 07master:46d4d8575979: movenc: Avoid writing separate flags for the first sample if not necessary
[11:36:11 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:ba0198a76728: Merge commit '46d4d8575979a24a8d026d9805039b724e0e3e5f'
[12:00:30 CET] <cone-112> ffmpeg 03Reinhard Tartler 07release/2.4:00abc0080d2f: doc: More changelog updates for v11.3
[12:00:31 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.4:6bf31ae3f929: Merge commit '00abc0080d2f5f179f18534713659ce79b22e647' into release/2.4
[12:24:11 CET] <cone-112> ffmpeg 03Anton Khirnov 07release/2.2:03fbb6ff3d28: h264: only ref cur_pic in update_thread_context if it is initialized
[12:24:12 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:b837d5bf83f9: Merge commit '03fbb6ff3d28f639ea5a35aba3c6dca09c17225d' into release/2.2
[12:35:45 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:fa4604d80580: h264: Do not share rbsp_buffer across threads
[12:35:46 CET] <cone-112> ffmpeg 03Anton Khirnov 07release/2.2:3670942fae7b: h264: initialize H264Context.avctx in init_thread_copy
[12:35:47 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:91ef250713d0: h264_cabac: Break infinite loops
[12:35:48 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:e3654b25cfba: Merge commit 'fa4604d80580dde45bfce32ebe04a5c13c233895' into release/2.2
[12:35:49 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:019b4b065095: Merge commit '3670942fae7beb2bfde52557ee95eab5f536e624' into release/2.2
[12:35:50 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:45eb2a0d49c1: Merge commit '91ef250713d04d675a16e5b030d7226baafe3f82' into release/2.2
[12:48:32 CET] <cone-112> ffmpeg 03Vittorio Giovara 07release/2.2:c47cdf837c1b: img2dec: correctly use the parsed value from -start_number
[12:48:33 CET] <cone-112> ffmpeg 03Reinhard Tartler 07release/2.2:f74f4a540151: Prepare for 10.6 Release
[12:48:34 CET] <cone-112> ffmpeg 03Reinhard Tartler 07release/2.2:470fd8e64e29: Update Changelog for v10.6
[12:48:35 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:71c63e799f3a: Merge commit 'c47cdf837c1b52681a84f434443e1f993757a5e9' into release/2.2
[12:48:36 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:e48f2847aba7: Merge commit 'f74f4a540151aacb38306f2e41a160c326be3d51' into release/2.2
[12:48:37 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:895ddf1a80f2: Merge commit '470fd8e64e292d2336b2b860437dcbc053ba9eec' into release/2.2
[12:49:10 CET] <BtbN> Anyone with a gtx980 and nvidia binary drivers(Windows or Linux) around? I added experimental(untested, don't have the needed hardware) h265 encoding support to nvenc: https://github.com/BtbN/FFmpeg/tree/nvenc
[12:52:02 CET] <BtbN> gtx970/960 might also work, nvidia only lists the 980 though. But everything 2nd gen Maxwell should, in theory, work.
[12:57:41 CET] <cone-112> ffmpeg 03Martin Storsjö 07release/2.2:7fd11fbeeb41: arm: Suppress tags about used cpu arch and extensions
[12:57:42 CET] <cone-112> ffmpeg 03Vittorio Giovara 07release/2.2:0ad8d751337e: configure: Properly fail when libcdio/cdparanoia is not found
[12:57:43 CET] <cone-112> ffmpeg 03Federico Tomassetti 07release/2.2:b2b359f12465: mdec: check for out of bounds read
[12:57:44 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:bec350611eb1: Merge commit '7fd11fbeeb41990427b475dc0d8800d2cf15a8c4' into release/2.2
[12:57:45 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:4e6f1df65722: Merge commit '0ad8d751337efbbd61c0d78762448b043100b653' into release/2.2
[12:57:46 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:ce167a49ba70: Merge commit 'b2b359f12465c8445484be24278b324da8ebb0e1' into release/2.2
[13:00:34 CET] <mateo`> BtbN: would it work on a 860m ?
[13:26:04 CET] <cone-112> ffmpeg 03Federico Tomassetti 07release/2.2:0eb8786eac1f: eamad: check for out of bounds read
[13:26:05 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:772f50c1f3e4: rv10: check size of s->mb_width * s->mb_height
[13:26:06 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:61c966ef3012: webp: validate the distance prefix code
[13:26:07 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:c0dc649e81bf: webp: ensure that each transform is only used once
[13:26:08 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:fc6a00d20d02: avcodec/rv10: check size of s->mb_width * s->mb_height
[13:26:09 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:b55be8b26676: avcodec/webp: validate the distance prefix code
[13:26:10 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:25b73ae265fd: avformat/rm: limit packet size
[13:26:11 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:51c580102d4d: Merge commit '0eb8786eac1f4a2132ad80dc49f90d5f81665c5c' into release/2.2
[13:26:12 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:1d9a93682da0: Merge commit '772f50c1f3e4a50ce3f35e31a6f0cd64e7cbe818' into release/2.2
[13:26:13 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:4b70e6e6e563: Merge commit '61c966ef30129a0e4dba485242c039a895914d33' into release/2.2
[13:29:24 CET] <himangi> hey! what does the planeheight attribute stand for in filter contexts.. It seems it has something to do with the different planes per image but not sure what it really means?
[13:38:41 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07release/2.2:0051174c7081: rmenc: limit packet size
[13:38:42 CET] <cone-112> ffmpeg 03Anton Khirnov 07release/2.2:eb9041403d82: tiff: Check that there is no aliasing in pixel format selection
[13:38:43 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:e032e647dd79: utvideodec: Handle slice_height being zero
[13:38:44 CET] <cone-112> ffmpeg 03Reinhard Tartler 07release/2.2:8ae0d702a1ba: doc: More changelog updates for v10.6
[13:38:45 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:19640e99dfd6: Merge commit '0051174c70810b66378cf8ea093eab01302f6049' into release/2.2
[13:38:46 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:17b2ad6f43ac: Merge commit 'eb9041403d820634c45ed4ee98570246a252507a' into release/2.2
[13:38:47 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:66342aab7316: Merge commit 'e032e647dd79e7748145792dfee0358eccb1982e' into release/2.2
[13:38:48 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.2:1177e047f166: Merge commit '8ae0d702a1ba1c3c8d88a29c181f8434f25bf53c' into release/2.2
[13:48:49 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07master:4c91d81be23f: ffmdec: make sure the time base is valid
[14:42:32 CET] <cone-112> ffmpeg 03Arwa Arif 07master:b4ec6afd3d37: Add dependencies to configure file for vf_fftfilt
[14:48:02 CET] <ninten> Is floating point encoding for ALS  implemented already ??
[14:54:00 CET] Last message repeated 1 time(s).
[14:58:40 CET] <gb27> what are the compression techniques available in exr codec?
[15:21:36 CET] <wm4> kierank: didn't you have a nice sample for testing lipsync?
[15:21:42 CET] <kierank> yes
[15:21:53 CET] <kierank> not sure if i can share though
[15:22:00 CET] <kierank> probably privately
[15:22:18 CET] <Daemon404> proprietary lipsync tests lol
[15:22:31 CET] <wm4> lol really?
[15:23:47 CET] <aetas> can't just feed it some brittany spears videos?
[15:27:34 CET] <kierank> Daemon404: it's a BBC test
[15:27:45 CET] <ubitux> as a player, what do you guys do when a (decoded) frame has no pts?
[15:27:48 CET] <ubitux> interpolate?
[15:27:54 CET] <kierank> Yes
[15:27:58 CET] <wm4> use dts
[15:29:24 CET] <ubitux> i'm using av_frame_get_best_effort_timestamp() and get a NOPTS
[15:30:06 CET] <TimNich> kierank:  which bbc test is it, if its mine, you can use it ;)
[15:30:08 CET] <Daemon404> ubitux, if you asked nicholas he'd probably say you should assert() or exit the player
[15:30:15 CET] <Daemon404> judging by his subtitle changes he proposed.
[15:30:16 CET] Action: Daemon404 runs
[15:30:23 CET] <ubitux> :)
[15:30:25 CET] <wm4> yeah
[15:30:38 CET] <kierank> TimNich: it's the one on BBC HD, I guess Simon's doing
[15:30:41 CET] <ubitux> i'll do that then
[15:30:43 CET] <ubitux> ;)
[15:30:47 CET] <wm4> yet hacks that make the decoder vomit pixel soup on invalid data are ok
[15:30:56 CET] <wm4> ffmpeg sure has strange policies
[15:31:11 CET] <wm4> TimNich: anything is fine
[15:31:32 CET] <TimNich> I will see if I can find mine
[15:31:49 CET] <kierank> wm4: "policies"
[15:34:24 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07master:6b8263b03ab3: ffmdec: limit the backward seek to the last resync position
[15:35:06 CET] <kierank> ubitux: in TS you're allowed to have nopts
[15:35:30 CET] <Daemon404> not only allowed
[15:35:32 CET] <Daemon404> recommended
[15:35:35 CET] <Daemon404> due to pcr overhead
[15:38:18 CET] <ubitux> how much successive nopts will i have to interpolate?
[15:38:26 CET] <ubitux> i mean, maximum
[15:38:34 CET] <Daemon404> there is no limit
[15:38:48 CET] <ubitux> :(
[15:38:50 CET] <Daemon404> if there is no fps change, you can successfully interpolate
[15:39:01 CET] <Daemon404> im sure there's recommendations or restrictions for broadcast though
[15:39:23 CET] <wm4> does lavf generate timestamps for ts?
[15:39:26 CET] <wm4> it probably does
[15:39:30 CET] <Daemon404> yes.
[15:39:39 CET] <Daemon404> but it also returns NOPTS
[15:39:41 CET] <ubitux> i'm doing my tests with matrixbench_mpeg2.mpg currently, so no ts
[15:39:50 CET] <ubitux> but av_frame_get_best_effort_timestamp() still gives me a nopts
[15:40:08 CET] <ubitux> i just wonder how i'm going to deal with that...
[15:40:50 CET] <wm4> interpolation is dead simple
[15:41:09 CET] <wm4> just add 1/fps each frame
[15:42:41 CET] <ubitux> fps being avg_frame_rate i suppose
[15:42:52 CET] <ubitux> ...and assuming i have previous frame
[15:43:03 CET] <ubitux> i wonder if setpts filter could be used
[15:43:09 CET] <ubitux> to set only missing TS
[15:43:42 CET] <Mavrik> ubitux, there's actually a guess_frame_rate call or some such
[15:43:52 CET] <Mavrik> avg_frame_rate isn't always filled out even when it should be :)
[15:43:54 CET] <Mavrik> sec
[15:44:11 CET] <Mavrik> yep, av_guess_frame_rate
[15:44:57 CET] <wm4> first time I hear about this function
[15:45:00 CET] Action: ubitux wonders why all these 3 params aren't const
[15:45:03 CET] <TimNich> wm4:  http://www.datafilehost.com/d/722c7406
[15:45:38 CET] <Mavrik> the function was private for a long time :)
[15:45:54 CET] <wm4> TimNich: ah, thanks
[17:05:19 CET] <kierank> 2:35 PM <"Daemon404> recommended
[17:05:19 CET] <kierank> 2:35 PM <"Daemon404> due to pcr overhead
[17:05:20 CET] <kierank> ??
[17:05:30 CET] <kierank> pts is only a few bytes
[17:05:45 CET] <Daemon404> akamai has one practically every frame
[17:05:49 CET] <Daemon404> for example.
[17:14:25 CET] <Daemon404> "Don't Miss Our Free Webcast Tomorrow: IMF Application 2 and Beyond"
[17:14:30 CET] <Daemon404> thanks SMPTE, but ill give it a msis
[17:14:31 CET] <Daemon404> miss8
[17:14:32 CET] <Daemon404> *
[17:19:25 CET] <cone-112> ffmpeg 03Claudio Freire 07master:374ec68139d1: tests/fate-run: Print more details on failure of stddev / filesize compares
[17:51:00 CET] <ninten> durandal_1707,  can you please tell me what information is thr in structure AVCodecContext ??
[17:51:54 CET] <ninten> I cant find the initialization in alsdec.c
[17:53:32 CET] <Rodeo> hi! your nvenc code is marked as nonfree, is that because of the code itself or because you dlopen a nonfree library?
[17:53:55 CET] <Rodeo> or something else I missed?
[17:54:37 CET] <Daemon404> i think it is the header Rodeo 
[17:54:37 CET] <Daemon404> iirc
[17:54:45 CET] <Daemon404> the header's license is derpy iirc
[17:54:51 CET] <Daemon404> (and has some impl in it maybe?)
[17:55:59 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:1f4088b28540: avcodec/options_table: remove extradata_size from the AVOptions table
[17:57:35 CET] <durandal_1707> ninten: where you see thr in AVCodecContext?
[17:59:05 CET] <ninten> durandal_1707,  inside ALSDecContext struct AVCodecContext is used
[17:59:29 CET] <durandal_1707> yes but there is nowhere thr
[18:02:22 CET] <ninten> durandal_1707,  http://pastebin.com/225gq6sP   line 3,5  this one is from alsdec.c
[18:03:05 CET] <Daemon404> durandal_1707, he typo'd
[18:03:07 CET] <Daemon404> (or she)
[18:03:22 CET] <Daemon404> question was what info is there in avcodeccontext, and why is it not initialized in alsdec.c
[18:03:42 CET] <Daemon404> answer: avcodeccontext is initialized elsewhere and passed to the functions in alsdec.c
[18:04:33 CET] <durandal_1707> aw
[18:05:51 CET] <ninten> Daemon404,  ok i will see how we are accessing it, but what information is thr in it ??
[18:06:17 CET] <Daemon404> ... why dont you look at the struct definition or how it is used?
[18:06:24 CET] <Daemon404> it's commented and documented
[18:06:26 CET] <Daemon404> (mostly)
[18:13:29 CET] <ninten> Daemon404,  same with GetBitContext
[18:14:33 CET] <durandal_1707> you do not need to know what is in GetBitContext, it is just bit reader
[18:17:35 CET] <ninten> ok
[18:31:06 CET] <cone-112> ffmpeg 03zhaoxiu.zeng 07master:b3a56e60be30: avcodec/hevc_parser: use avpriv_find_start_code in hevc_split()
[18:55:52 CET] <ninten> durandal_1707, ping !!
[18:56:05 CET] <durandal_1707> ninten: pong
[18:56:39 CET] <ninten> durandal_1707,  what is the use of num_bytes_diff_float !!
[18:58:02 CET] <durandal_1707> ninten: do you have reference implementation code?
[18:58:23 CET] <durandal_1707> use that instead of pdf
[19:00:35 CET] <ninten> durandal_1707,  yeah i have that pdf but the question is num_bytes_diff_float is a element of frame_data, and i from code it seems ALSDecContext is the struct for frame data now there is no mention of num_byte_diff_float !! 
[19:02:37 CET] <ninten> durandal_1707,  ref code !! right now i am trying to implement it on alsdec.c 
[19:15:16 CET] <ninten> durandal_1707,  you were talking about some ref code !!
[19:19:28 CET] <durandal_1707> ninten:  http://www.nue.tu-berlin.de/fileadmin/fg97/Forschung/Projekte/Beendete_Projekte/MPEG4_ALS/mp4alsRM22rev2.zip
[19:20:45 CET] <durandal_1707> ninten: there is even newer: http://www.nue.tu-berlin.de/fileadmin/fg97/Forschung/Projekte/Beendete_Projekte/MPEG4_ALS/mp4alsRM23.zip
[19:35:43 CET] <andrei_il> hi 
[19:36:27 CET] <andrei_il> i need some help on graphedit for the dshow video capture qual task 
[19:50:52 CET] <bigpet> andrei_il I don't think there's many people here that know much about it, if you have specific questions try stackoverflow.com or somewhere where people that program with directshow hang out (whereever that is)
[19:51:44 CET] <bigpet> anyways, if you have a question it doesn't hurt to ask (although asking to ask is not very useful)
[19:52:29 CET] <iive> i think ffmpeg already have dshow capture (probably for analog signal, aka raw images). maybe looking for the developer and asking him would be helpful.
[19:53:16 CET] <andrei_il> thanks
[20:16:40 CET] <cone-112> ffmpeg 03Vittorio Giovara 07master:bfeb83a8b7d3: rtpdec_hevc: Drop extra sanity check for size of input packet
[20:16:41 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:d898482ae34f: Merge commit 'bfeb83a8b7d3fcf09a54d8dbc9c521e10bb17530'
[20:28:59 CET] <Compn> Rodeo : yes, nvenc header license was incompatable with gpl iirc. thats why nonfree
[20:29:48 CET] <Rodeo> aah
[20:29:54 CET] <Rodeo> the header, OK
[20:30:20 CET] <Rodeo> but so if nvidia fixed the header, that would fix it?
[20:30:31 CET] <j-b> yes
[20:31:05 CET] <Rodeo> aha, well, of course thaz's probably easier said than done
[20:31:09 CET] <Compn> http://ffmpeg.org/pipermail/ffmpeg-devel/2014-November/165940.html
[20:31:14 CET] <Rodeo> but it's not too bad
[20:31:15 CET] <Rodeo> thanks
[20:31:20 CET] <Compn> is the original post with the link to the header we are tlaking about
[20:31:44 CET] <Rodeo> thanks
[20:31:48 CET] <Compn> i think timo (author of patch) or someone else was talking to nvidia about it
[20:32:01 CET] <bigpet> well can't you, like get the symbols from the *.dll without the header?
[20:32:58 CET] <Compn> heres the nvidia version of the same feature > https://github.com/agathah/ffmpeg_libnvenc
[20:33:30 CET] <Compn> bigpet : no clue , i guess the objection was posed by carl to ask fsf to issue an opinion about the incompatability
[20:35:36 CET] <Compn> its 2800 lines and has lots of secret nooks and crannies.
[20:35:43 CET] <Compn> https://github.com/jp9000/OBS/blob/master/ObsNvenc/inc/nvEncodeAPI.h
[20:36:05 CET] <BtbN> I'm not going to re-implement that header. Specialy not after i worked that much with it.
[20:36:59 CET] <Compn> ah yes BtbN is timo :)
[20:37:16 CET] <Compn> BtbN : what happened to the nvidia version of nvenc for ffmpeg ?
[20:37:21 CET] <Compn> i forgot what happened
[20:37:55 CET] <BtbN> Compn, it is still around, just not even close to a mergeable state.
[20:38:14 CET] <BtbN> No idea if they are still working on it.
[20:38:21 CET] <j-b> bigpet: the header is big.
[20:38:39 CET] <BtbN> The dll does only export a single symbol
[20:38:45 CET] <BtbN> so that's not too usefull
[20:38:53 CET] <bigpet> alright, that's just mean
[20:39:02 CET] <BtbN> You call that function with a struct from the nvenc header, and it fills in the actual functions.
[20:45:24 CET] <cone-112> ffmpeg 03Vittorio Giovara 07master:9f25a109922d: matroskaenc: Also validate chapter end time
[20:45:25 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:2e8020c66ccb: Merge commit '9f25a109922da43c1f81273a431d3b40cb5a785a'
[21:15:56 CET] <cone-112> ffmpeg 03Vittorio Giovara 07master:ad94c6ca0b86: siff: Use the correct type for packet size variables
[21:15:57 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:8f92c0898d2b: Merge commit 'ad94c6ca0b86c463f476b26606259a2041dcddc9'
[21:15:58 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:e71dce5769b2: avformat/siff: Fix checks and variable types
[21:35:43 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.4:491c4bbb38ad: Revert "avutil/opencl: is_compiled flag not being cleared in av_opencl_uninit"
[21:35:45 CET] <cone-112> ffmpeg 03Michael Niedermayer 07release/2.5:79c9d9b13432: Revert "avutil/opencl: is_compiled flag not being cleared in av_opencl_uninit"
[21:44:10 CET] <cone-112> ffmpeg 03Vittorio Giovara 07master:93f7948136fc: libvpx: Fix mixed use of av_malloc() and av_reallocp()
[21:44:11 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:a11440c1857b: Merge commit '93f7948136fcda8ddbbc44a6c24418f11ca829b8'
[21:44:12 CET] <cone-112> ffmpeg 03Michael Niedermayer 07master:cf714205bdd0: avcodec/libvpxenc: Clear twopass_stats.sz on deallocation
[21:56:46 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07master:cf82c426fadf: roqvideoenc: set enc->avctx in roq_encode_init
[22:11:31 CET] <cone-112> ffmpeg 03Andreas Cadhalpun 07master:2c8cff2be4a0: asfenc: fix leaking asf->index_ptr on error
[22:14:48 CET] <bigpet> which is the oldest windows version ffmpeg is supporting?
[22:14:57 CET] <bigpet> (please don't let it be older than XP)
[22:16:14 CET] <Compn> building ffmpeg for win2k makes it faster though
[22:16:25 CET] <Compn> because xp's libs suck
[22:16:31 CET] <Compn> (dont hate the player)
[22:16:51 CET] <Compn> also you know ffmpeg supports DOS right ?
[22:17:22 CET] <Compn> and os/2 and amiga and ....
[22:18:48 CET] <bigpet> well yeah, I guess I should have been more specific, because I'm still on that shitty stdin problem and the code I'm interested in is ifdef'd 
[22:19:03 CET] <bigpet> so I guess I should look up when the buildscript defs it
[22:21:54 CET] <bigpet> alright, it's in Windows 2000 too, I don't wanna install another VM :(
[22:22:23 CET] <bigpet> I simply don't know what to do to make this line execute: https://github.com/FFmpeg/FFmpeg/blob/master/ffmpeg.c#L395
[22:22:41 CET] <bigpet> nothing seems to make the condition true
[22:25:45 CET] <Compn> what stdin problem again
[22:26:33 CET] <bigpet> I searched around the web for other programs that did that
[22:27:21 CET] <Compn> whats the problem, i mean ?
[22:27:52 CET] <bigpet> stdin->_cnt relies on FILE having a member named _cnt which it no longer does on the new C runtime
[22:28:57 CET] <bigpet> I found this https://github.com/bagder/curl/commit/49b63cf30d4242b51bd49295f9f20cfbfb641a7d#diff-3f8cd1135d62394a5519c6de12de1efbR655 
[22:29:14 CET] <bigpet> maybe I should try redirecting nework traffic
[22:29:24 CET] <nevcairiel> bigpet: we dont care for anything below xp
[22:30:05 CET] <bigpet> to see if that does anything, but redirecting files with "test.exe<a.txt" did not do anything
[22:30:27 CET] <nevcairiel> but FILE->_cnt should have never been used in the first place, private undocumented unstable API
[22:33:22 CET] <Compn> what are you tring to do on the command line ?
[22:33:38 CET] <bigpet> well, my issue is kind of, I need to figure out what that code was there for, because as of now I can not figure out what triggers it
[22:33:43 CET] <Compn> and why arent you using rxvt or cygwin shell on windows if you are running into weird windows shell problems
[22:33:48 CET] <bigpet> I can't observe a difference from removing it
[22:34:04 CET] <Compn> did you go back in the history and find out where it was committed ?
[22:34:10 CET] <bigpet> yes
[22:34:21 CET] <Compn> let me guess, original commit by bellard ?
[22:34:46 CET] <Compn> in the r1000~ range
[22:35:10 CET] <nevcairiel> no
[22:35:11 CET] <bigpet> it was this: https://github.com/FFmpeg/FFmpeg/commit/ca4d71b149ebe32aeaf617ffccf362624b9aafb1
[22:35:28 CET] <bigpet> I mean the named Pipe thing I understand
[22:35:47 CET] <bigpet> and can trigger with "echo somehting|test.exe"
[22:36:11 CET] <Compn> ah windows pipe
[22:36:17 CET] <Compn> \\.\namedpiper
[22:37:27 CET] <Compn> thats all under the  # if HAVE_PEEKNAMEDPIPE
[22:38:55 CET] <Compn> well have to ask michaelni about it then
[22:39:12 CET] <Compn> not many windows devs though
[22:40:47 CET] Action: michaelni doesnt really remember about this commit ... so i dont think i can help
[22:41:42 CET] <bigpet> anyone brave enough to just remove those lines and see if someone complains :P
[22:42:56 CET] <michaelni> I cant test on windows atm so i wont make any changes that havnt been tested by someone else
[22:43:37 CET] <bigpet> I've been up and down that thing with mintty.exe cmd.exe powershell on windows 8, windows 10 and windows XP
[22:44:18 CET] <michaelni> if you have a patch that you tested and think is correct i can apply it
[22:44:46 CET] <michaelni> or you can post it to ffmpeg-dev and ill apply it in 24h or so, is probably beter
[22:45:01 CET] <bigpet> yeah, that sounds more reasonable
[22:47:14 CET] <bigpet> hm, this sounds like a reason: http://www.open-chess.org/viewtopic.php?f=7&t=947
[22:47:58 CET] <bigpet> "After a call to such functions, the input waiting in the pipe will be copied to the buffer, and the call to PeekNamedPipe can indicate no input available. Setting stdin to unbuffered was not enough."
[22:48:56 CET] <bigpet> ok, so it's input that was comming from a pipe, but was copied from the pipe to the stdin buffer from the windows C runtime
[22:49:09 CET] <bigpet> when calling some very specific c functions
[22:50:18 CET] <bigpet> at least it explains why my tests never triggered it
[22:56:13 CET] <bigpet> man this is fucking stupid, so to make sure that this 100% does not happen anymore I need to disassemle the new C runtime?
[22:56:50 CET] <bigpet> I think I'm gonna ask on some microsoft sites
[22:56:53 CET] <Daemon404> ... you realize they ship CRT sources
[22:57:03 CET] <bigpet> they do?
[22:57:07 CET] <Daemon404> they always have.
[22:57:29 CET] <Daemon404> also you should not be doing anything that relies on anything ot officially documented....
[22:57:41 CET] <Daemon404> i lack context, but if you need to read the crt sources, this doesnt sound kosher.
[22:58:36 CET] <bigpet> well the thing is, the reason for the use of the undocumented function is some weird behavior in the C runtime, to get rid of that hack I need to make sure the behavior does not happen in the new C runtime
[22:59:30 CET] <Daemon404> that depends if the "feature" that is hitting thi behavior is even useful IMO
[22:59:54 CET] <Compn> this is an awful lot of work for 3 lines in ffmpeg.c
[22:59:56 CET] <Compn> :D
[23:00:08 CET] <bigpet> I'm probably just making this to complicated
[23:00:35 CET] <bigpet> anyway what it does is that if it uses a GUI or other pipes to provide input it uses pipes
[23:00:42 CET] <nevcairiel> just make sure the simple use-cases work, we can try to fix the crazy ones later
[23:00:49 CET] <bigpet> we check the pipes with "PeekNamedPipe"
[23:00:56 CET] <bigpet> which normally works
[23:01:28 CET] <bigpet> BUT for compatibility reasons some C runtime functions used copy things from the named pipe into the stdin file
[23:02:14 CET] <bigpet> so while there is input from the pipe we would not detect it just with a check of the pipe and it would potentially not react on input
[23:03:23 CET] <nevcairiel> should use a thread. =)
[23:03:55 CET] <bigpet> not sure how that would help in this case
[23:04:16 CET] <bigpet> oh, right just read blockingly
[23:04:49 CET] <bigpet> I guess I can do that, I just really didn't want to
[23:05:07 CET] <nevcairiel> yeah that wouldnt really be liked
[23:05:25 CET] <Daemon404> its better than abusing undocumented CRT internals
[23:05:33 CET] <Daemon404> keep this up and youll end up on the old new thing
[23:06:06 CET] <nevcairiel> the point is to get rid of the old hack, since its going away in the  2015 crt =)
[23:06:27 CET] <Daemon404> shouldnt we, you know
[23:06:30 CET] <Daemon404> wait until 2015 is
[23:06:31 CET] <Daemon404> you know
[23:06:32 CET] <Daemon404> released?
[23:06:55 CET] <nevcairiel> its in a stage final enough
[23:08:05 CET] <bigpet> well the current "solution" is a hack anyway, I think getting rid of it is almost reward enough onto itself
[23:08:11 CET] <nevcairiel> besides this isnt about that as such, using the proper APIs in the first place wouldnt have required any changes, so going to the proper API now isnt exactly reliant on the new version
[23:08:21 CET] <Daemon404> yes
[23:11:08 CET] <nevcairiel> although non-blocking IO is kinda annoying on windows, so ugly code isnt unexpected...
[23:12:24 CET] <Daemon404> it isnt annoying with the new AIO apis
[23:12:25 CET] <Daemon404> ;)
[23:12:26 CET] Action: Daemon404 runs
[23:12:33 CET] <nevcairiel> well, async, sure
[23:12:51 CET] <Daemon404> anyway, aio, or nonblocking io is crap everywhere
[23:13:00 CET] <Daemon404> the only portable way is with threads
[23:13:02 CET] <Daemon404> manually.
[23:13:31 CET] <Daemon404> (or events...)
[23:15:32 CET] <bigpet> I wonder, is the threads one even a solution, since I imagine there could be other functions reading from stdin
[23:15:32 CET] <nevcairiel> yeah ffmpeg transcodes on the main thread and then wants to p oll the stdin for commands..
[23:19:11 CET] <wm4> maybe it shouldn't do this?
[23:22:23 CET] <bigpet> not sure something else would be better (especially since all of this is just because there's no select() on Windows)
[23:23:45 CET] <bigpet> anyway, I'm off to ask on technet or some other windows forums, maybe some 30 year Windows veteran knows something I can't find online
[23:24:40 CET] <wm4> I meant it could start a thread instead of polling, but yeah, maybe it's overkill
[23:25:27 CET] <bigpet> yeah, but then you have to deal with all the synchronization
[23:27:25 CET] <bigpet> especially since don't imagine doing something like that for just for one platform is gonna be ugly
[23:28:14 CET] <bigpet> and you guys listed all kinds of platforms that would probably prefer not to have a thread introduced just to query the user
[00:00:00 CET] --- Tue Mar 10 2015


More information about the Ffmpeg-devel-irc mailing list