[Ffmpeg-devel-irc] ffmpeg-devel.log.20171030

Tue Oct 31 03:05:04 EET 2017

[01:16:22 CET] <cone-501> ffmpeg 03Lukas Stabe 07master:d3f1b0d3d87f: movenc: add m4b to ipod format extensions
[01:16:23 CET] <cone-501> ffmpeg 03Jonas Licht 07master:59ad50469695: libavformat/mov.c: use calculated dts offset when seeking in streams
[01:29:40 CET] <cone-501> ffmpeg 03wm4 07master:195dd5c59c89: lavc/avrndec: remove AV_CODEC_CAP_DR1, as it's broken
[01:29:41 CET] <cone-501> ffmpeg 03Anton Khirnov 07master:4776c61424fa: decode: avoid leaks on failure in ff_get_buffer()
[02:22:29 CET] <cone-501> ffmpeg 03James Almer 07master:783535a4cd26: x86/bswapdsp: add missing preprocessor wrappers for AVX2 functions
[10:02:15 CET] <RiCON> could someone let jamrial know about the dangling parenthesis he left in libavdevice/{libndi_newtek_dec.c,decklink_dec.cpp}?
[14:17:49 CET] <fx159159> Hey, I'm still developing an indev for android devices, got it working nice so far, it does receive frames by the android camera2 api, now my question: is there an api to rotate single avpackets/avframes?
[14:18:36 CET] <fx159159> I can get the rotation of the sensor and optionally of the device and though about "normalizing" the image rotation based on the sensor orientation?
[14:18:48 CET] <fx159159> *thought
[14:20:16 CET] <fx159159> Or should I leave that out of the indev and up to the user?
[14:20:22 CET] <JEEB> usually you'd pass the frame through avfilter's rotate filter or so after the fact as an API user
[14:22:45 CET] <fx159159> JEEB: do you think I should keep that out of the indev? And apply filters afterwardsß
[14:23:34 CET] <JEEB> in general yes, unless you can specifically request the indev itself to do some "normalization" without you having to implement it
[14:23:43 CET] <JEEB> the indev should be an interface for the indev
[14:29:28 CET] <fx159159> JEEB: I can get a sensor orientation, a picture has to be rotated by it so it is displayed upright... without any "normalization" the user would have to try some parameters
[14:29:58 CET] <fx159159> JEEB: But for YUV images I can't ask the api to pre rotate the images... so... I should leave it out then I guess
[14:30:13 CET] <JEEB> I think but am not sure if we have AVFrame-specific rotation flag
[14:30:27 CET] <JEEB> because that would make sense if you get the rotation for each picture
[14:32:26 CET] <JEEB> fx159159: it would make sense to expose that option for RGB capture I guess?
[14:32:31 CET] <JEEB> if it is somehow optimized
[14:33:22 CET] <fx159159> JEEB: As far as I understood only precompressed JPEG images can be rotated by the api, but that's not what I want
[14:34:01 CET] <JEEB> heh
[14:34:19 CET] <JEEB> yea that sounds not exactly nice. You noted YCbCr and thus I thought the other alternative was RGB
[14:35:28 CET] <fx159159> Yeah, that would have been nice, but RGB unfortunately also has no rotation
[14:36:00 CET] <fx159159> I decided to stay with YUV for now, as the api states all devices must support YUV, all other formats are "optional"
[14:36:57 CET] <jkqxz> Rotation information can be attached to a packet as side data - it's AV_PKT_DATA_DISPLAYMATRIX.
[14:37:10 CET] <JEEB> yea but that's packets
[14:37:19 CET] <JEEB> AVFrames are for raw images, no?
[14:37:46 CET] <jkqxz> AV_FRAME_DATA_DISPLAYMATRIX, then.
[14:37:58 CET] <JEEB> yup
[14:38:29 CET] <fx159159> Currently I'm working with AVPackets, is it av_packet_new_side_data then?
[14:38:30 CET] <jkqxz> (Indevs still have to give you packets, though.)
[14:38:36 CET] <JEEB> jkqxz: :<
[14:38:46 CET] <JEEB> excuse me while I go weep on the side
[14:39:45 CET] <JEEB> although it makes sense since it's similar to lavf :P
[14:40:25 CET] <jkqxz> fx159159:  Should be.  There is av_display_rotation_set() as well for simple rotation by angles.
[14:40:46 CET] <jkqxz> JEEB:  s/'s similar to/ is/
[14:40:55 CET] <JEEB> well yea
[15:07:11 CET] <fx159159> Is it up to the player to interpret the AV_PKT_DATA_DISPLAYMATRIX?
[15:08:42 CET] <jkqxz> Yes, though ffmpeg also tries to apply the rotation automatically in transcode cases.
[15:11:50 CET] <fx159159> jkqxz: Does that look right? https://pastebin.com/Kq46F9cT
[15:49:21 CET] <cone-233> ffmpeg 03James Almer 07master:5c819f433e85: avdevice/decklink_dec: remove extra bracket
[15:49:22 CET] <cone-233> ffmpeg 03James Almer 07master:da71351dd36a: avdevice/libndi_newtek_dec: remove extra bracket
[17:28:23 CET] <cone-233> ffmpeg 03Diego Biurrun 07master:0fd0d4fd0a51: swscale-test: const correctness
[17:28:24 CET] <cone-233> ffmpeg 03Diego Biurrun 07master:f25609ff06b0: doc/examples/filter_audio: const correctness
[17:28:25 CET] <cone-233> ffmpeg 03Diego Biurrun 07master:ecc5c4db2dd3: doc/examples/output: Cast pointer to the right (const) type
[17:28:26 CET] <cone-233> ffmpeg 03James Almer 07master:087e9ab1b357: Merge commit '0fd0d4fd0a518e30ff23972828ad7cf7f35cfb9d'
[17:28:27 CET] <cone-233> ffmpeg 03James Almer 07master:4c410a58d783: Merge commit 'f25609ff06b093d82548c92fc2031cd2e66c20d3'
[17:28:28 CET] <cone-233> ffmpeg 03James Almer 07master:91b4d109175e: Merge commit 'ecc5c4db2dd3a0f328d95df89daa59f78b4b2810'
[17:30:41 CET] <cone-233> ffmpeg 03Sean McGovern 07master:2e1ab6a67dd5: vc1dec: raise an error if sprite picture data is missing
[17:30:42 CET] <cone-233> ffmpeg 03James Almer 07master:29e1247d4132: Merge commit '2e1ab6a67dd5d9a30558d008aa6e541c68d76221'
[17:54:01 CET] <cone-233> ffmpeg 03Anton Khirnov 07master:3889dfde9c02: dca: Move the downmix request check outside the loop
[17:54:02 CET] <cone-233> ffmpeg 03James Almer 07master:032129cabd50: Merge commit '3889dfde9c0284b008c5cfcbc921c1d137fb81e2'
[18:10:32 CET] <cone-233> ffmpeg 03Luca Barbato 07master:44aafa3325ec: configure: Properly escape arguments in filter/filter_out helper functions
[18:10:33 CET] <cone-233> ffmpeg 03James Almer 07master:9b0ddd1d8767: Merge commit '44aafa3325ec09a120bdb81b8579b8dfd65b1693'
[18:12:19 CET] <cone-233> ffmpeg 03Luca Barbato 07master:41262498ff89: configure: Fix sem_timedwait probe
[18:12:20 CET] <cone-233> ffmpeg 03James Almer 07master:85f1896cffb9: Merge commit '41262498ff89ef48e5ce4823b2fc2dc514a3eb0c'
[18:18:19 CET] <cone-233> ffmpeg 03Luca Barbato 07master:8c0cadd17e98: avplay: Do not try to allocate new frames when the player is closing
[18:18:21 CET] <cone-233> ffmpeg 03James Almer 07master:ba45ca0e62e2: Merge commit '8c0cadd17e98547d84e82111550caca4fb40ff8d'
[18:26:47 CET] <cone-233> ffmpeg 03Paul B Mahol 07master:7c25523cc8e6: utvideodec: Fix decoding odd sizes with interlaced video with some formats
[18:26:48 CET] <cone-233> ffmpeg 03James Almer 07master:b428d445a817: Merge commit '7c25523cc8e618e77dc84d960e41e9644eaf8c33'
[18:34:08 CET] <cone-233> ffmpeg 03Paul B Mahol 07master:b812db66af16: utvideodec: Fix gradient prediction when stride does not match width
[18:34:09 CET] <cone-233> ffmpeg 03James Almer 07master:dbfc31b86ec9: Merge commit 'b812db66af160bfac2f64aef2ade4aed7eb76b12'
[18:41:27 CET] <cone-233> ffmpeg 03James Almer 07master:39e9a7a80fd3: avcodec/avcodec.h: remove doxy from the old bsf API functions
[18:59:25 CET] <cone-233> ffmpeg 03Luca Barbato 07master:c55f2ae2b8d4: avprobe: Handle unknown values for the color description
[18:59:26 CET] <cone-233> ffmpeg 03James Almer 07master:162989521ce2: Merge commit 'c55f2ae2b8d4de473f945aebd6a7a89b6b83e87a'
[19:18:26 CET] <jamrial> michaelni: 189157c3fc merges a decoder we already have, with basically no differences outside of cosmetic, some missing bug fixes, and the addition of an init_threadsafe flag
[19:19:13 CET] <jamrial> it's the flag that i'm not sure if it's correct. init() fills some vlc tables without using ff_thread_once
[19:19:45 CET] <jamrial> should i skip adding that flag, or is it indeed safe?
[19:21:26 CET] <jamrial> AVCodec.init_static_data() should really be called from somewhere else than avcodec_register() tbh for cases like this
[19:24:21 CET] <durandal_1707> ported ty demuxer
[19:30:07 CET] <jamrial> durandal_1707: what do you think about the flag, for that matter?
[19:30:22 CET] <jamrial> since you commited the decoder on our end back then
[19:33:18 CET] <nevcairiel> jamrial: the vlc init seems to be things in the context, not global stuff, so the flag seems ok? unless i'm missing something
[19:34:43 CET] <jamrial> nevcairiel: you're probably not, but since derek not too long ago sent a patch to add that flag everywhere, seeing it missing here made me doubt
[19:47:04 CET] <michaelni> jamrial, all static tables seem to have const, iam not sure why this would be no thread safe
[19:47:13 CET] <michaelni> noT
[19:47:32 CET] <jamrial> michaelni: ok, will add the flag then. thanks
[19:50:41 CET] <cone-233> ffmpeg 03Kostya Shishkov 07master:189157c3fc8e: Add ClearVideo decoder
[19:50:42 CET] <cone-233> ffmpeg 03James Almer 07master:269ed71e93a4: Merge commit '189157c3fc8eeb691e3684b09d971eb5ddb47d5b'
[19:52:34 CET] <cone-233> ffmpeg 03Diego Biurrun 07master:d0ce0634e094: clearvideo: Convert to the new bitstream reader
[19:52:35 CET] <cone-233> ffmpeg 03James Almer 07master:63f82ab3897b: Merge commit 'd0ce0634e09403eecaea4283cf82a2c5cea7d1e1'
[20:00:45 CET] <cone-233> ffmpeg 03Seppo Tomperi 07master:358adef03056: hevc: Add NEON IDCT DC functions for bitdepth 8
[20:00:46 CET] <cone-233> ffmpeg 03James Almer 07master:cc8c2d3609b2: Merge commit '358adef0305618219522858e471edf7e0cb4043e'
[20:05:13 CET] <cone-233> ffmpeg 03Alexandra Hájková 07master:3d69dd65c677: hevc: Add support for bitdepth 10 for IDCT DC
[20:05:14 CET] <cone-233> ffmpeg 03James Almer 07master:f9c3fbc00cd6: Merge commit '3d69dd65c6771c28d3bf4e8e53a905aa8cd01fd9'
[20:35:50 CET] <cone-233> ffmpeg 03Anton Khirnov 07master:04b0f0e371ff: mem: uninline av_malloc(z)_array()
[20:35:51 CET] <cone-233> ffmpeg 03James Almer 07master:4959f18a8e11: Merge commit '04b0f0e371ff81b682274b574fb465ba4395c09f'
[20:36:55 CET] <cone-233> ffmpeg 03James Almer 07master:bee89ed3ec61: matroskaenc: don't warn about unknown spherical metadata when there isn't any
[20:36:56 CET] <cone-233> ffmpeg 03James Almer 07master:21525e6d0d70: Merge commit 'bee89ed3ec61a4a0e72255ae6f4a4274d25b5ccd'
[20:47:10 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:1f6e5dd4d4dd: mlp: Rename the channel layout variable
[20:47:11 CET] <cone-233> ffmpeg 03James Almer 07master:8b63acaeee24: Merge commit '1f6e5dd4d4dd9c02371ad211049d27edb6d95286'
[20:56:16 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:cea7fd9afb84: mlp: Drop ff_ prefix from a static function
[20:56:17 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:94c54d97e7f4: mlp: Factor out channel layout subset checks
[20:56:18 CET] <cone-233> ffmpeg 03James Almer 07master:1242a6170e93: Merge commit 'cea7fd9afb8488a6c48f7d7ee38602e1fd3dd425'
[20:56:19 CET] <cone-233> ffmpeg 03James Almer 07master:e3ebc293c624: Merge commit '94c54d97e7f4fe90570c323803f2bdf6246c1010'
[21:04:27 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:85c2bf9d2cc7: avfoundation: update to use AVCodecParameters
[21:04:28 CET] <cone-233> ffmpeg 03Mark Thompson 07master:6af014f40282: vaapi_encode: Use gop_size consistently in RC parameters
[21:04:29 CET] <cone-233> ffmpeg 03Mark Thompson 07master:64a5260c695d: lavc: Add flag to allow profile mismatch with hardware decoding
[21:04:30 CET] <cone-233> ffmpeg 03Mark Thompson 07master:7acb90333a18: vaapi: Add external control of allow-profile-mismatch
[21:04:31 CET] <cone-233> ffmpeg 03James Almer 07master:644dd659c5c2: Merge commit '85c2bf9d2cc7dad3d5c6dc79f802a74efa8d8fce'
[21:04:32 CET] <cone-233> ffmpeg 03James Almer 07master:623a0526282f: Merge commit '7acb90333a187b0e847b66f9d3511245423dc0ce'
[21:16:09 CET] <cone-233> ffmpeg 03James Almer 07master:ad52eef997dc: spherical: add functions to retrieve and request projection names
[21:16:10 CET] <cone-233> ffmpeg 03James Almer 07master:498864fe800a: dump: use av_spherical_projection_name() to print spherical projection names
[21:16:11 CET] <cone-233> ffmpeg 03James Almer 07master:c7bf98d4372e: avprobe: use av_spherical_projection_name() to print spherical projection names
[21:16:12 CET] <cone-233> ffmpeg 03James Almer 07master:f3ccc5e22acd: Merge commit 'c7bf98d4372e9217c8d75d2ca1f298385909e4da'
[21:31:15 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:dd3aa85b68c0: aac_ac3_parser: Drop in-parser downmix functionality
[21:31:16 CET] <cone-233> ffmpeg 03Vittorio Giovara 07master:970c76f32283: mlp_parser: Drop in-parser downmix functionality
[21:31:17 CET] <cone-233> ffmpeg 03James Almer 07master:9484b9f4ba66: Merge commit 'dd3aa85b68c017c419acb0c39ff6aa890ce89e87'
[21:31:18 CET] <cone-233> ffmpeg 03James Almer 07master:34542ac033f8: Merge commit '970c76f32283bddf3a5afd24fe52db7a96186244'
[21:37:04 CET] <cone-233> ffmpeg 03Michael Niedermayer 07master:f1251a6b9cca: ffmpeg: Fix stored encoder metadata with -bitexact
[21:37:05 CET] <cone-233> ffmpeg 03Michael Niedermayer 07master:097675242070: avcodec/aacdec_fixed: Fix integer overflow in predict()
[21:37:06 CET] <cone-233> ffmpeg 03Michael Niedermayer 07master:41d96af2a74c: avcodec/aacdec_fixed: Fix integer overflow in apply_dependent_coupling_fixed()
[21:37:07 CET] <cone-233> ffmpeg 03Kaustubh Raste 07master:1181d93231e9: avcodec/mips: Improve avc avg mc 22, 11, 31, 13 and 33 msa functions
[21:37:08 CET] <cone-233> ffmpeg 03Kaustubh Raste 07master:93218c223496: avcodec/mips: Improve avc chroma avg hv mc msa functions
[21:40:27 CET] <cone-233> ffmpeg 03Seppo Tomperi 07master:0d4d43513786: hevc: Add NEON add_residual for bitdepth 8
[21:40:28 CET] <cone-233> ffmpeg 03James Almer 07master:999c2271a59a: Merge commit '0d4d43513786f1df4d561e1fac924fb0722c6700'
[21:42:59 CET] <cone-233> ffmpeg 03Martin Storsjö 07master:e1c2453a4fac: arm: hevc_idct: Tune the add_res_8x8 and add_res_32x32 functions
[21:43:00 CET] <cone-233> ffmpeg 03James Almer 07master:05beee44c6b2: Merge commit 'e1c2453a4fac1f7116244d0d05310935c20887e6'
[21:55:36 CET] <cone-233> ffmpeg 03Anton Khirnov 07master:831018b0bbe2: mpeg4audio: Make avpriv_copy_pce_data() inline
[21:55:37 CET] <cone-233> ffmpeg 03James Almer 07master:9ea06d48c272: Merge commit '831018b0bbe26a603802a9022472f714a59293be'
[22:03:37 CET] <cone-233> ffmpeg 03Mark Thompson 07master:b43b95f4789b: vp9_raw_reorder_bsf: Remove a redundant allocation
[22:03:38 CET] <cone-233> ffmpeg 03Mark Thompson 07master:9203aac22874: avconv_hw: Add implicit device creation with default parameters
[22:03:39 CET] <cone-233> ffmpeg 03Mark Thompson 07master:e669db76108d: avconv: Support setting the hardware device to use when filtering
[22:03:40 CET] <cone-233> ffmpeg 03Mark Thompson 07master:aa51bb3d2756: hwcontext_qsv: Support derivation from child devices
[22:03:41 CET] <cone-233> ffmpeg 03Mark Thompson 07master:27978155bc66: hwcontext: Add frame context mapping for nontrivial contexts
[22:03:42 CET] <cone-233> ffmpeg 03Mark Thompson 07master:eaa5e0710496: hwcontext_qsv: Implement mapping frames from the child device type
[22:03:43 CET] <cone-233> ffmpeg 03Mark Thompson 07master:e1c5d56b18b8: hwcontext_qsv: Implement mapping frames to the child device type
[22:03:44 CET] <cone-233> ffmpeg 03Mark Thompson 07master:c5714b51aad4: hwcontext: Improve allocation in derived contexts
[22:03:45 CET] <cone-233> ffmpeg 03Mark Thompson 07master:38cb05f1c89c: vf_hwmap: Add device derivation
[22:03:46 CET] <cone-233> ffmpeg 03Mark Thompson 07master:81a4cb8e5863: vf_hwmap: Add reverse mapping for hardware frames
[22:03:47 CET] <cone-233> ffmpeg 03James Almer 07master:550e5d660253: Merge commit '81a4cb8e58636d4efd200c2b4fec786a7e948d8b'
[22:08:42 CET] <cone-233> ffmpeg 03Alexandra Hájková 07master:edbf0fffb15d: hevc: Add NEON add_residual for bitdepth 10
[22:08:43 CET] <cone-233> ffmpeg 03James Almer 07master:31a411293619: Merge commit 'edbf0fffb15dde7a1de70b05855529d5fc769f14'
[22:11:11 CET] <cone-233> ffmpeg 03Luca Barbato 07master:73fc82f3434d: vlc: Add header #include when the types are used
[22:11:12 CET] <cone-233> ffmpeg 03Luca Barbato 07master:0ac1fec1c3da: bitstream: Move VLC reading functions into the vlc.h header
[22:11:13 CET] <cone-233> ffmpeg 03James Almer 07master:ff173ce8f1a8: Merge commit '0ac1fec1c3dacedabbf3dd4122ef4bf8523e688c'
[22:25:04 CET] <BBB> re: bug 6775, why flush the decoder after each discontinuity?
[22:25:10 CET] <BBB> that seems overlay aggressive and unnecessary
[22:26:51 CET] <nevcairiel> it is
[22:28:18 CET] <nevcairiel> its perfectly expected for decoding to stop working after draining unless you call flush buffers
[22:28:25 CET] <nevcairiel> i think its even documented that its required
[23:13:33 CET] <cone-233> ffmpeg 03Diego Biurrun 07master:b5f19f747849: aac: Split function to parse ADTS header data into public and private part
[23:13:34 CET] <cone-233> ffmpeg 03James Almer 07master:b9d3def9b2cb: Merge commit 'b5f19f7478492307e4b4763aeac3180faf50e17f'
[23:26:22 CET] <kierank> what tools are good for debugging OOM memory leaks
[23:27:07 CET] <atomnuker> malloc wrappers?
[23:28:00 CET] <jkqxz> Normal memory leak tools + exit()?
[23:31:24 CET] <kierank> valgrind just exits
[23:31:32 CET] <kierank> after pruning
[23:31:53 CET] <kierank> I can clearly reproduce that h264 leak at least
[23:32:11 CET] <kierank> dunno how that guy managed to record that
[23:36:31 CET] <kierank> something related to ff_h2645_extract_rbsp
[23:48:52 CET] <iive> strange, valgrind has an option to report unfreed memory on exit
[23:51:36 CET] <cone-233> ffmpeg 03Alex Converse 07master:504403ab2868: aacsbr: Turnoff in the event of over read.
[23:51:37 CET] <cone-233> ffmpeg 03James Almer 07master:989254fcff82: Merge commit '504403ab28688588fb85817ad58964482f67f29f'
[00:00:00 CET] --- Tue Oct 31 2017