[Ffmpeg-devel-irc] ffmpeg-devel.log.20150514

[burek]

[01:11:49 CEST] <cone-959> ffmpeg 03Michael Niedermayer 07master:93cfa7d1692c: avcodec/mpeg4audio: add some padding/alignment to MAX_PCE_SIZE
[01:11:50 CEST] <cone-959> ffmpeg 03Michael Niedermayer 07master:22f15f573538: avcodec/on2avc: Check run more carefully
[01:31:49 CEST] <BBB> Compn: binary codec loader isnt interesting
[01:34:58 CEST] <cone-959> ffmpeg 03Michael Niedermayer 07master:50d878d9303d: avcodec/msrledec: Ask for sample that uses a branch with a unused byte
[02:22:55 CEST] <cone-959> ffmpeg 03Timothy Gu 07master:eaeb63219872: nutdec: Remove unused label
[02:36:04 CEST] <cone-959> ffmpeg 03Michael Niedermayer 07master:88f29406b7ae: avformat/nutdec: Remove unused variable
[02:36:05 CEST] <cone-959> ffmpeg 03Shivraj Patil 07master:8252f63d1b98: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC uni hv mc functions
[03:02:38 CEST] <yvear> I'm getting [Parsed_pan_0 @ 0000000002fcbf20] Syntax error near "*1+FR*0+"   with    -af "pan=stereo|FL=FL*1+FR*0+FCE*0.7071+LFE*1+BL*0.7071+BR*0+FCL*0.7071+FCR*0|FR=FL*0+FR*1+FCE*0.7071+LFE*1+BL*0+BR*0.7071+FCL*0+FCR*0.7071"   anyone know why?
[04:07:52 CEST] <cone-959> ffmpeg 03Clément BSsch 07release/2.2:1f3ca53b8ae3: tests: drop bc dependency
[04:17:11 CEST] <cone-959> ffmpeg 03Clément BSsch 07release/2.4:07256a7a6ec8: tests: drop bc dependency
[04:26:20 CEST] <cone-959> ffmpeg 03Clément BSsch 07release/2.5:346c4f0ad0bd: tests: drop bc dependency
[11:28:09 CEST] <cone-732> ffmpeg 03Rong Yan 07master:a2cd07d22a5e: avcodec/ppc/h264dsp: POWER LE support in h264_idct_dc_add_internal() fix vec_lvsl bug
[12:15:22 CEST] <cone-732> ffmpeg 03Clément BSsch 07master:56bc0a6736cd: avcodec/ass: make default playback resolution available to decoders
[12:15:23 CEST] <cone-732> ffmpeg 03Clément BSsch 07master:5c219e289e7f: avcodec/srtdec: attempt to correct SubRip positioning
[12:16:56 CEST] <ubitux> rcombs: i'm going back for good on that subtitles stuff, but i'd like to extend the ASS & mkv tests before
[12:17:00 CEST] <ubitux> do you wish to help?
[12:19:21 CEST] <ubitux> one thing that could help is sending me ass/mkv files with weird cases of extradata & sub entries
[12:29:24 CEST] <rcombs> ubitux: that I can do
[12:30:26 CEST] <rcombs> http://puu.sh/hMPd3/4edc3878f7.mks and http://puu.sh/hMPds/1ededcbd7b.ass are small extradata samples
[12:33:40 CEST] <ubitux> thanks :)
[12:34:01 CEST] <ubitux> so what's special about the second .ass?
[12:34:19 CEST] <ubitux> the last section that ends up in the extradata?
[12:34:22 CEST] <rcombs> yup
[12:34:56 CEST] <ubitux> that's what your last patches were fixing?
[12:35:01 CEST] <rcombs> yup
[12:35:14 CEST] <rcombs> oh, also, there are some bugs in libavcodec/ass_split.c
[12:35:41 CEST] <rcombs> 1. it doesn't handle `\h` (non-breaking space)
[12:36:27 CEST] <rcombs> 2. it passes through the content of a {\q1}<content>{\q0} block as text
[12:36:49 CEST] <rcombs> (that <content> would be parsed by an ASS renderer as a drawing; it should probably just be dropped)
[12:37:38 CEST] <ubitux> feel free to work on that, i'm going to focus on the ass <-> mk* remux & transcode
[12:37:53 CEST] <rcombs> couple other things as well
[12:38:10 CEST] <rcombs> I've got a patch for it sitting around that really just needs review
[12:38:38 CEST] <ubitux> so 4edc3878f7.mks is generated from 1ededcbd7b.ass with mkvmerge, or that .ass is actually something you extracted from the mks?
[12:38:47 CEST] <ubitux> which patch?
[12:39:20 CEST] <ubitux> i streamcopied the sub stream into a .ass and it gave me the exact same content as 1ededcbd7b.ass
[12:40:56 CEST] <rcombs> haven't uploaded it yet, doing some cleanup real quick
[12:41:31 CEST] <rcombs> uh& I forget which file came from which, to be honest
[12:41:39 CEST] <rcombs> but I'm reasonably sure they're both correct
[12:42:09 CEST] <ubitux> ok
[12:44:37 CEST] <cone-732> ffmpeg 03Michael Niedermayer 07master:5c8e4bf7c4f4: avcodec/qdrw: another try at skipping the first 512 bytes
[12:44:38 CEST] <cone-732> ffmpeg 03Michael Niedermayer 07master:019daa077597: avcodec/qdrw: Fix the code which asks for version 1 samples
[13:40:13 CEST] <wm4> ubitux: you're planning on moving AVSubtitle to lavu, right?
[13:40:24 CEST] <wm4> ubitux: can we get rid of the AVPicture field too?
[13:40:31 CEST] <wm4> I think AVPicture should be killed off
[13:40:39 CEST] <ubitux> yes, but i want to add more tests before
[13:40:46 CEST] <wm4> and AVSubtitle is the only API that still inherently requires AVPicture
[13:40:47 CEST] <ubitux> because that's a bit sensitive
[13:40:49 CEST] <ubitux> :p
[13:40:52 CEST] <wm4> ok
[13:41:11 CEST] <wm4> (other AVPicture uses are some freestanding functions)
[13:42:40 CEST] <ubitux> https://ffmpeg.org/pipermail/ffmpeg-devel/2014-December/166625.html ?
[13:43:01 CEST] <ubitux> this one was ok-ed already, i could push it
[13:43:12 CEST] <ubitux> but i guess you prefer to kill it off
[13:43:25 CEST] <ubitux> which i'm not sure about
[13:45:03 CEST] <wm4> it's a worse AVFrame
[13:45:07 CEST] <wm4> and users constantrly abuse it
[13:45:12 CEST] <wm4> and run into trouble because of it
[13:45:31 CEST] <wm4> https://ffmpeg.org/pipermail/libav-user/2015-May/008144.html
[13:45:35 CEST] <wm4> is this correct? who knows...
[13:46:04 CEST] <wm4> AVPicture can explicitly alias AVFrame, but that's a crap idea at most
[13:46:13 CEST] <ubitux> yes i know, but if i start by changing this, i'm never going to be able to solve the subtitles problem
[13:46:50 CEST] <wm4> doesn't moving it to lavu require a major bump?
[13:46:54 CEST] <ubitux> it already pulls a cascade of api problems i'm having a hard time to fix :P
[13:46:58 CEST] <ubitux> not really
[13:47:04 CEST] <ubitux> because lavc depends on lavu
[13:47:17 CEST] <wm4> hmk, well if it does, then consider dropping AVPicture for AVframe
[13:47:43 CEST] <ubitux> you want an AVFrame in an AVSubtitles? :)
[13:47:53 CEST] <ubitux> i mean, N AVFrames
[13:49:43 CEST] <wm4> there are many things which would be considered; for now I'd just replace AVSubtitleRect.pict
[13:50:58 CEST] <ubitux> which would mean N AVFrames in AVSubtitles
[13:51:25 CEST] <wm4> effectively yes
[14:27:51 CEST] <ubitux> rcombs: just made a quick list to make sure i don't forget anything: http://pastie.org/pastes/10188491/text
[14:27:55 CEST] <ubitux> do you see anything else?
[14:28:35 CEST] <ubitux> if you want to suggest special dialogue lines or whatever for the test cases i'm going to build, now is the time
[14:28:45 CEST] <rcombs> ubitux: ASS->other stuff
[14:29:27 CEST] <ubitux> this can be covered by the encoders of other stuff themselves
[14:29:30 CEST] <rcombs> https://gist.github.com/f562ee1e07a529cfe44e I just rewrote a decent chunk of what I already had for this patch
[14:29:43 CEST] <rcombs> because I realized some components were wrong
[14:29:56 CEST] <rcombs> so it needs more testing
[14:30:37 CEST] <ubitux> if you can craft a .ass with all the special dialogue that would be nice
[14:30:50 CEST] <ubitux> so i can re-use it as a template with the different tweak listed
[14:31:10 CEST] <ubitux> (like, i will create different sample with or without [Events] header etc)
[14:31:45 CEST] <ubitux> then it can be use to encode it to .srt or whatever encoder that needs to split the ass to cover also that code
[14:32:15 CEST] <rcombs> alright, I just need to put together a file with all the known edge cases, then
[14:32:29 CEST] <ubitux> i'll complete them with the weird cases i found as well
[14:32:36 CEST] <rcombs> I was poking at this line:
[14:32:36 CEST] <rcombs> Dialogue: 6,0:00:05.35,0:00:10.31,Typsetting 1,,0,0,0,,{\an7\bord6.75\alpha&H40&\p1\3c&H4D4556&\c&HF4EBE8&}m 0 0 l 750 0 750 810 0 810 c{\p0}test\h{test
[14:32:52 CEST] <ubitux> sounds beautiful
[14:32:54 CEST] <rcombs> (i.e. I had a slightly weird line and I edited it to be weirder)
[14:32:54 CEST] <ubitux> :')
[14:33:33 CEST] <rcombs> in current ass_split.c that would give AVERROR_INVALIDDATA
[14:33:56 CEST] <ubitux> Dialoue: Marked=0,0:03:49.03,0:03:50.79,Style1,Comment,0000,0000,0000,,He's nobody suspicious, so rest assured.
[14:33:58 CEST] <ubitux> Dilogue: Marked=0,0:06:37.29,0:06:37.85,Style1,Comment,0000,0000,0000,,Yo!
[14:34:09 CEST] <ubitux> here are some funny things i found too, in the wild
[14:34:11 CEST] <rcombs> with that patch, it gives the "{test" as regular text, because I'm pretty sure that's what's supposed to happen
[14:34:15 CEST] <ubitux> (read carefully the first word)
[14:34:26 CEST] <rcombs> heh, Dilogue
[14:34:40 CEST] <rcombs> libass would treat that as a comment
[14:34:46 CEST] <rcombs> (I think)
[14:34:51 CEST] <rcombs> (and I think vsfilter would do the same)
[14:34:55 CEST] <ubitux> they basically end up in extradata
[14:35:06 CEST] <rcombs> like other comments
[14:35:07 CEST] <ubitux> i mean, in the header
[14:36:16 CEST] <ubitux> isn't "{test" considered a comment?
[14:36:24 CEST] <ubitux> iirc "{foo}" is inlined comment
[14:37:11 CEST] <wm4> ubitux: yes, vsfilter doesn't care a lot for closing tags of anything
[14:38:03 CEST] <rcombs> I'll double-check, but I'm pretty sure "{test" is the one exception to that rule
[14:40:21 CEST] <rcombs> https://github.com/Cyberbeing/xy-VSFilter/blob/17472325c39613389673341a905c219cc0e9a8b0/src/subtitles/RTS.cpp#L3033
[14:41:28 CEST] <wm4> oh true
[14:41:31 CEST] <wm4> I think I remember now
[14:41:55 CEST] <ubitux> yeah i needs to be "valid" basically
[14:41:56 CEST] <ubitux> ok
[14:42:13 CEST] <rcombs> but anything in a {} that isn't a valid tag is ignored
[14:42:17 CEST] <wm4> http://git.srsfckn.biz/mpv/tree/sub/sd_ass.c#n284
[14:42:25 CEST] <rcombs> and if there's a { without a } then it's treated as plain text
[14:42:53 CEST] <ubitux> heh, ok
[14:43:07 CEST] <ubitux> and of course you're going to tell me some users rely on this?
[14:43:27 CEST] <wm4> always
[14:43:55 CEST] <wm4> also, when is ffmpeg going to support microdvd tags in srt?
[14:44:02 CEST] <ubitux> lol
[14:44:18 CEST] <rcombs> does anyone use html tags in ass
[14:44:35 CEST] <rcombs> (reading that vsfilter code, I think it's supported there)
[14:44:39 CEST] <rcombs> (kill me now)
[14:44:46 CEST] <ubitux> :D
[14:45:05 CEST] <wm4> rcombs: haven't seen that yet, but ass in srt of course exists
[14:45:14 CEST] <wm4> and microdvd in srt also exists
[14:45:16 CEST] <Daemon404> ive seen it in chinese .ass
[14:45:19 CEST] <Daemon404> years ago
[14:45:23 CEST] <wm4> whoo
[14:45:27 CEST] <rcombs> but yeah, ass->* is probably the least of our concerns right now
[14:45:32 CEST] <ubitux> Daemon404: in utf16 ofc?
[14:45:43 CEST] <Daemon404> beats me
[14:45:44 CEST] <wm4> chinese like utf16 ass
[14:45:51 CEST] <wm4> even though it's less space-efficient than utf-8
[14:45:58 CEST] <rcombs> utf16 stinx
[14:45:59 CEST] <wm4> (most text in ass is ASCII due to overhead)
[16:47:42 CEST] <Daemon404> people still use that thing?
[17:04:05 CEST] <wm4> people will use anything, except if it's good
[17:05:00 CEST] <wm4> ok I found out that I wasted 1 or 2 hours of time just because my freshly updated msys doesn't have some hevc related header updates yet
[17:05:09 CEST] <wm4> of course this is my fault, but I still blame msys
[17:14:25 CEST] <cone-345> ffmpeg 03Nedeljko Babic 07master:729466dc68f9: libavutil/softfloat: Add test case for av_add_sf
[18:19:58 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07master:139e1c8009df: avcodec/cavsdec: Check esc_code
[18:19:59 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07master:c5c06e392b56: avcodec/cavsdec: Check frame_rate_code
[18:20:00 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07master:f6b8b966076f: avcodec/cavsdec: Use ff_set_dimensions()
[19:09:45 CEST] <cone-345> ffmpeg 03wm4 07master:cc5e4bb48476: hevc: make avcodec_decode_video2() fail if get_format() fails
[20:58:45 CEST] <cone-345> ffmpeg 03Martin Storsjö 07release/2.6:c3b1261afa31: rtpenc_mpegts: Free the right ->pb in the error path in the init function
[20:58:46 CEST] <cone-345> ffmpeg 03Martin Storsjö 07release/2.6:692fd5635f9f: rtpenc_mpegts: Set chain->rtp_ctx only after avformat_write_header succeeded
[20:58:47 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:76ee9fdb61db: mxfenc: fix memleaks in mxf_write_footer
[20:58:48 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:763ab41f771c: mxfenc: don't try to write footer without header
[20:58:49 CEST] <cone-345> ffmpeg 03wm4 07release/2.6:262c678357f5: matroskadec: export cover art correctly
[20:58:50 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:4ea7ff435427: msrledec: use signed pixel_ptr in msrle_decode_pal4
[20:58:51 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:aebafed24fd3: aacdec: consistently use avctx for logging in decode_eld_specific_config
[20:58:52 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:7b13aef5d2f0: aasc: return correct buffer size from aasc_decode_frame
[20:58:53 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:4aa4c78daeae: aacpsy: avoid psy_band->threshold becoming NaN
[20:58:54 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:9c826d8d51d0: ac3: validate end in ff_ac3_bit_alloc_calc_mask
[20:58:55 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:50fb69c73751: alsdec: ensure channel reordering is reversible
[20:58:56 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:a298e13c2caa: alsdec: limit avctx->bits_per_raw_sample to 32
[20:58:57 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:84cd276d0e82: alsdec: validate time diff index
[20:58:58 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:3258e12d8c7b: aacpsy: correct calculation of minath in psy_3gpp_init
[20:58:59 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:f77cb3d4a61f: alsdec: check sample pointer range in revert_channel_correlation
[20:59:00 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:2f8f4351b857: mpeg4videodec: only allow a positive length
[20:59:01 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:787e094ed0b0: bink: check vst->index_entries before using it
[20:59:02 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:01291b4e2f4f: alsdec: only adapt order for positive max_order
[20:59:03 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:47f5f6b2303b: alac: reject rice_limit 0 if compression is used
[20:59:04 CEST] <cone-345> ffmpeg 03Rong Yan 07release/2.6:be1b665dec58: swscale/ppc/swscale_altivec.c: POWER LE support in yuv2planeX_8() delete macro GET_VF() it was wrong
[20:59:05 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:d35086d71506: apedec: set s->samples only when init_frame_decoder succeeded
[20:59:06 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:9f2c8734b9c3: apedec: prevent out of array writes in decode_array_0000
[20:59:07 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:a13a288ec810: nutdec: check for negative frame rate in decode_info_header
[20:59:08 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:9b87d15ca80c: nutdec: check chapter creation in decode_info_header
[20:59:09 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:2523bdcd6702: nutdec: fix memleaks on error in nut_read_header
[20:59:10 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:2f290cf8815b: nutdec: fix illegal count check in decode_main_header
[20:59:11 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:835037506bc3: pngdec: check s->last_picture.f->data[0] before using it
[20:59:12 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:99e737a7c787: pngdec: return correct error code from decode_frame_common
[20:59:13 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:e6a5023d1f2f: pngdec: don't use AV_PIX_FMT_MONOBLACK for apng
[20:59:14 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:33d77bc38467: matroskadec: use uint64_t instead of int for index_scale
[20:59:15 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:35013fa23aba: matroskadec: check s->streams[k] before using it
[20:59:16 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07release/2.6:c74846388bf7: avformat/matroskadec: Use tracks[k]->stream instead of s->streams[k]
[20:59:17 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:0cb8d786f29c: avidec: check for valid bit_rate range
[20:59:18 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:f78d7e6a0314: diracdec: prevent overflow in data_unit_size check
[20:59:19 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:c6418be04314: diracdec: avoid overflow of bytes*8 in decode_lowdelay
[20:59:20 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:c6f6be93ac28: diracdec: check if reference could not be allocated
[20:59:21 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:04f80ed3f81a: diracdec: check that block length is valid
[20:59:22 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:0856eea7707f: aacsbr: break infinite loop in sbr_hf_calc_npatches
[20:59:23 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:0f7e67be3a56: avi: Validate sample_size
[20:59:24 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:66b95ee4df80: id3v2: catch avio_read errors in check_tag
[20:59:25 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:d9249b558285: aacdec: don't return frames without data
[20:59:26 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:9bcaf9037813: imgutils: initialize palette padding bytes in av_image_alloc
[20:59:27 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:ac8339928111: cafdec: check avio_read return value
[20:59:28 CEST] <cone-345> ffmpeg 03Andreas Cadhalpun 07release/2.6:73cf5d9a2853: cafdec: free extradata before allocating it
[21:32:18 CEST] <jamrial> michaelni: c0d847e should also be backported to maintained branches
[21:42:11 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07master:a6a45774d045: avcodec/dcadec: Check nchans
[21:42:12 CEST] <cone-345> ffmpeg 03Michael Niedermayer 07master:a9bf628bfdad: avcodec/dcadec: Check subsubframes
[21:42:51 CEST] <jamrial> Also 7ac5c38 and 9d4fdfe (they apply cleanly to 2.6, not sure about the rest, but that can be checked later)
[21:57:25 CEST] <wm4> is there a tool like ffprobe, but that actually has an understandable command line syntax?
[21:57:51 CEST] <kierank> mediainfo
[21:58:17 CEST] <wm4> I want to do something like "show me the timestamps of the decoded frames of the video stream"
[22:12:31 CEST] <Compn> for what kind of video
[22:12:35 CEST] <Compn> maybe special mpegts tools
[22:12:53 CEST] <Compn> some other timestamp tools for windows too , but i dont remember anything else
[22:13:52 CEST] <Compn> wm4 : what are you trying to do, i mean what are you using the timestamps for ?
[22:16:51 CEST] <wm4> inspecting files?
[22:30:07 CEST] <Daemon404> [20:58] < wm4> I want to do something like "show me the timestamps of the decoded frames of the video stream"
[22:30:18 CEST] <Daemon404> ffprobe -of json -show_frames file.ext
[22:30:30 CEST] <Daemon404> i consume that in scripts all the time
[22:30:32 CEST] <Daemon404> it is convenient
[22:33:41 CEST] <wm4> I thought about this too
[22:33:50 CEST] <wm4> but it requires a minimum of effort
[22:58:27 CEST] <RiCON> wm4: "mkvextract timecodes_v2 movie.mkv 1:timecodes.txt"?
[22:59:11 CEST] <wm4> RiCON: I tend to want to do other things too, like listing other fields per frame
[00:00:00 CEST] --- Fri May 15 2015