[Ffmpeg-devel-irc] ffmpeg-devel.log.20140808

[burek]

[00:18] <Zeranoe> To whoever decided to allow wildcards in --enable and --disable options: Thank you, you just saved me a lot of time.
[00:20] <wm4> Author: Måns Rullgård <mans at mansr.com>  2009-08-06 17:45:22
[00:20] <wm4> that guy added it
[01:54] <cone-128> ffmpeg.git 03James Almer 07master:5c3c67126feb: lavf/avio: rename url_feof() to avio_feof()
[01:54] <cone-128> ffmpeg.git 03James Almer 07master:d34ec64a2290: replace calls to url_feof() with avio_feof()
[02:56] <Compn> ubitux : lwn still going? one reason i had to stop talking to diego
[07:17] <ubitux> wm4: you need to -show_streams to have the summary of the frame counting
[07:18] <ubitux> Compn: dunno, 4 more messages since last time i checked
[07:23] <ubitux> rcombs: yeah, the subtitle and data support in libavfilter is something we considered since years
[07:23] <ubitux> Nicolas (cigaes) mentioned the data part several time, and he might have submitted a PoC already somehow
[07:23] <rcombs> ubitux: have there been tickets tracking them that I'm duplicating?
[07:23] <ubitux> i was looking for it but i think it's mostly on the ml
[07:24] <ubitux> still, there is https://trac.ffmpeg.org/ticket/1305 and https://trac.ffmpeg.org/ticket/2067
[07:24] <rcombs> I looked around a bit, but it's kinda hard to search for, and I didn't turn anything up immediately
[07:25] <ubitux> note that adding subtitles support in libavfilter means moving the AVSubtitle & friends to libavutil
[07:25] <rcombs> shouldn't that second one be closed?
[07:25] <ubitux> #2067?
[07:26] <rcombs> yeah; there's a parameter in the filter to seek now, isn't there?
[07:26] <ubitux> well, it's kind of the "Support subtitles in libavfilter" ticket ;)
[07:26] <rcombs> or am I just thinking of Plex's
[07:27] <ubitux> no seek is not supported, and even if it was it would be a hack
[07:27] <ubitux> we need proper subtitles support
[07:27] <rcombs> yeah, agreed
[07:28] <rcombs> cehoyos seems to agree as well
[07:28] <ubitux> but yeah again we need to move AVSubtitles API to libavutil, and this is tricky because we wanted to take the opportunity to review the structure itself
[07:28] <ubitux> (of AVSubtitles, rects, etc)
[07:29] <wm4> ehhhh
[07:29] <ubitux> like question such as; do we need to support other pixel formats for rgb bitmap subs
[07:29] <wm4> I'm pretty skeptic against subtitles in lavfi
[07:29] <ubitux> or how are we supposed to store text subs, etc
[07:29] <rcombs> I'd like to see AVSubtitles stop including the timestamps in the text :3
[07:29] <ubitux> rcombs: i "fixed" that, somehow
[07:30] <ubitux> at least for matroska
[07:30] <wm4> you know, most things in audio/video can be modeled as uni-directional data flow in a graph
[07:30] <wm4> and lavfi looks like it's doing that
[07:30] <wm4> but lavfi is way too closely coupled and special-cased to what it's supposed to be to be useful as generic framework for such things
[07:31] <wm4> it's also too inflexible
[07:31] <rcombs> wm4: "input frames -> output frames", with some initialization and configuration functions, basically
[07:32] <wm4> it would be nice if we had a filter API that doesn't want to make you kill yourself, and that also allows exteral filters
[07:32] <wm4> the "killing yourself" part is mostly about format negotiation
[07:33] <wm4> and I think if we hack subs into lavfi in its current state, we will be very sad in a few years
[07:34] <ubitux> except that hardcoding will not require to specify files in the filtergraph
[07:35] <ubitux> and we'll be able to have one single blending process, in overlay
[07:35] <wm4> if you want to solve that in the context of ffmpeg.c, you might as well add a dirty hack to build an appropriate filter graph with the needed options
[07:35] <ubitux> i want it in ffplay too ;)
[07:35] <rcombs> wm4: whaddayaknow, that's how Plex did it
[07:35] <wm4> ubitux: nobody cares about ffplay
[07:36] <ubitux> like, i want to be able to hardsub the subtitles of my mkv which specifying the mkv twice, and have a weird caching behaviour
[07:36] <ubitux> wm4: well, ffplay == any similar user app
[07:36] <wm4> not really
[07:36] <rcombs> yeah, not really
[07:37] <rcombs> mpv, for instance, (at default settings) renders subtitles internally at display resolution and has OpenGL handle the overlaying
[07:37] <wm4> anyway, I really like the idea of a filter API that 1. allows external filters, and 2. allows much more things in the data flow than audio and video
[07:37] <rcombs> wm4: what stops lavfi from allowing external filters?
[07:38] <wm4> its API
[07:38] <rcombs> how so?
[07:38] <wm4> it even allowed external filters once, but it was so horrible and messy that it was made private
[07:38] <ubitux> libav deprecated the public symbols to ease internal changes
[07:38] <wm4> I mean, compare it with something like vapoursynth
[07:38] <ubitux> and never made them public again
[07:39] <wm4> anyway, if you want external apps to heavily rely on a ffmpeg filter graph thing, please don't make it lavfi, but something more flexible
[07:39] <rcombs> I might be thinking of something different than you when I think of "external filters"
[07:39] <wm4> lavfi is restricted to ffmpeg filters, ffmpeg formats, audio/video only (even if you add subs later)
[07:40] Action: ubitux still wants proper subtitles support in ffplay 
[07:40] <rcombs> ah, I was thinking of a hypothetical lavfi filter that passed input video through a vapoursynth script, or something similar
[07:40] <rcombs> ubitux: how does ffplay do scaling for display?
[07:40] <ubitux> badly
[07:40] <wm4> rcombs: that's a use-case, although that filter could just be part of lavfi
[07:41] <wm4> rcombs: it uses SDL
[07:41] <ubitux> not even sdl2 btw, which might fix a few problems we have with it
[07:41] <wm4> "problems"
[07:41] <ubitux> especially since sdl2 has a possible gl backend, right?
[07:42] <wm4> well they still have their own retarded video texture formats
[07:42] <wm4> hardcoded to bt.601 or whatever
[07:42] Action: rcombs points at FFDraw
[07:43] <ubitux> that's what we use to blend subtitles currently
[07:43] <rcombs> ubitux: yes
[07:43] <wm4> so what I envision is a ffmpeg-style directshow/gstreamer-like interface, except that it OF COURSE will not have the failure of those two
[07:43] <ubitux> but not in overlay; and i'd love to use overlay instead
[07:43] <wm4> or maybe mentioning these is not very constructive
[07:43] <rcombs> and last I checked, it converts RBG subtitle colors to YUV using BT.601, regardless of the video's color space
[07:44] <wm4> ubitux: wth is the point?
[07:44] <rcombs> *RGB
[07:44] <wm4> ubitux: overlay would just require blending the whole video surface, and it'd always force you to do video resolution rendering
[07:44] <ubitux> wm4: ffmpeg -i in.mkv -vf '[0:0][0:2]subtitles'
[07:44] <wm4> not to mention that vf_overlay format negotiation is fundamentally broken
[07:44] <wm4> uh
[07:44] <wm4> wqhat does that mean
[07:44] <ubitux> wm4: ffmpeg -i in.mkv -vf '[0:0][0:2]overlay'
[07:44] <ubitux> sorry
[07:45] <wm4> I see horrible syntax
[07:45] <ubitux> with [0:0] video stream, and [0:2] sub stream
[07:45] <rcombs> ubitux: so, you want the sub2video hack to support text subtitles?
[07:45] <ubitux> rcombs: exactly, but not as a hack
[07:45] <wm4> basing everything on lavfi will lead to misery
[07:45] <rcombs> ubitux: I'd rather that be in the "subtitles" filter tbh
[07:46] <rcombs> rather than making overlay into a hydra
[07:46] <ubitux> dunno
[07:46] <wm4> ubitux: the concept sounds nice, but the result will be not so good
[07:47] <ubitux> -vf rasterize @_@
[07:47] <wm4> mpv for example _pulls_ the subtitles from the subtitle renderer, as a list of surfaces that have to be blended
[07:47] <wm4> not pushing a giant alpha surface
[07:48] <ubitux> interesting
[07:49] <ubitux> but what do you suggestion to avoid -vf subtitles=/home/foo/bar/Sub... ?
[07:49] <ubitux> s/suggestion/suggest/
[07:49] <wm4> make ffmpeg/ffplay insert that filter?
[07:49] <wm4> *groan*
[07:49] <ubitux> haha, yeah right
[07:50] <wm4> I'm not very fond of all these hacks that are added to lavfi and lavf to work around ffmpeg deficiencies
[07:50] <wm4> like the concat demuxer...
[07:50] <ubitux> but that's just hiding shit under the carpet
[07:50] <wm4> isn't that what ffmpeg.c is best at
[07:50] <ubitux> o u
[07:50] <ubitux> :D
[07:50] <rcombs> ubitux: I'd suggest having the subtitles filter take a subtitles track as input, as you're suggesting, but just leaving the drawing internal rather than converting everything to full frame-sized bitmaps
[07:50] <wm4> seriously
[07:51] <ubitux> rcombs: alright
[07:51] <ubitux> so we agree that we need subtitles in lavfi anyway
[07:51] <rcombs> well, as wm4 says, I don't know if that'd be sane with the current API
[07:51] <ubitux> it will solve a bunch of things
[07:51] <wm4> I'm arguing that adding it to lavfi will lead to failure eventually
[07:51] <wm4> and there should be a light-weight filter API instead
[07:52] <ubitux> like ffmpeg -i bla.avi -ss 12.34 -i bla.srt -vf subtitles ...
[07:52] <wm4> that isn't as insanely closely-coupled and closed as lavfi
[07:52] <ubitux> (s/subtitles/hardsub/ if you prefer)
[07:53] <wm4> something that flexible could perhaps also include demuxers, decoders, bitstream filters, and such things
[07:54] <ubitux> btw, somehow related to light-weighting, i'm actually secretly working on a python binding, just to try if it's do-able
[07:55] <ubitux> you'll probably hate me for what i'll submit as a PoC but well..
[07:56] <wm4> probably
[07:57] <wm4> the current APIs don't allow for great things IMO
[07:57] <wm4> e.g. lavfi would need to be turned inside-out before it could have the flexibility of avisynth
[07:57] <wm4> because lavfi wants to do its own thing once the graph is setup
[11:32] <ubitux> so, hashes are highlighted in the trac, but "Repository '' not found"
[12:26] <ubitux> btw, we need to find a solution to test filters using floats
[12:29] <wm4> ubitux: what's hard about it?
[12:31] <ubitux> thresholding on the resulting frames
[12:31] <ubitux> i suppose
[12:31] <wm4> on that matter, I'm wondering how libass should be tested
[12:32] <wm4> because libass sure needs a testsuite
[12:32] <wm4> but there the problem is way worse
[12:32] <ubitux> you need a "bitexact" mode in freetype i suppose
[12:33] <ubitux> and force a freetype version
[12:33] <wm4> I'm not even sure if libass still uses freetype for rasterization
[12:33] <wm4> I think not
[12:34] <wm4> and how do you get bitexact fontconfig ;)
[12:34] <wm4> anyway, most calculations are done with floats
[12:47] <ubitux> i wonder if we could add a simple bitmap font in ffmpeg to test the drawtext filter
[13:08] <ubitux> so we don't have any filter test making of this fuzz thing?
[13:28] <cone-822> ffmpeg.git 03Justin Ruggles 07release/1.1:7997acee0542: Check if an mp3 header is using a reserved sample rate.
[13:28] <cone-822> ffmpeg.git 03Justin Ruggles 07release/1.1:d7dbc687e312: Check mp3 header before calling avpriv_mpegaudio_decode_header().
[13:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:244a58fff0ad: Merge commit '7997acee0542f6e0bb9ea42ff783f80b70878a2f' into release/1.1
[13:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:cf7f798984a9: Merge commit 'd7dbc687e312a91ef2ccf797d57b95c61d0e8a2f' into release/1.1
[13:46] <cone-822> ffmpeg.git 03Martin Storsjö 07release/1.1:744e7eea5d81: adpcm: Avoid reading out of bounds in the IMA QT trellis encoder
[13:46] <cone-822> ffmpeg.git 03Martin Storsjö 07release/1.1:21d3e0ac9e17: adpcm: Write the proper predictor in trellis mode in IMA QT
[13:46] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:5fa56e6e6239: Merge commit '744e7eea5d815efea777b6179d96e8d94b63ccfa' into release/1.1
[13:46] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:6333c6c17d1f: Merge commit '21d3e0ac9e1719d8444b3f5466983587ac0ad240' into release/1.1
[13:59] <cone-822> ffmpeg.git 03Reinhard Tartler 07release/1.1:5e8eaa26b227: Prepare for 9.14 Release
[13:59] <cone-822> ffmpeg.git 03Reinhard Tartler 07release/1.1:3ecbd911ff91: Update Changelog for v9.14
[13:59] <cone-822> ffmpeg.git 03Anton Khirnov 07release/1.1:e8ff79720646: eamad: use the bytestream2 API instead of AV_RL
[13:59] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:53c3abc1083d: Merge commit '3ecbd911ff9177097820e5d00401c9bf29e5d167' into release/1.1
[14:00] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:c074feed292e: Merge commit 'e8ff7972064631afbdf240ec6bfd9dec30cf2ce8' into release/1.1
[14:18] <cone-822> ffmpeg.git 03Diego Biurrun 07release/1.1:3a6bc3e38164: vf_select: Drop a debug av_log with an unchecked double to enum conversion
[14:18] <cone-822> ffmpeg.git 03Bernhard Übelacker 07release/1.1:d16515ae5fe7: video4linux2: Avoid a floating point exception
[14:18] <cone-822> ffmpeg.git 03Diego Biurrun 07release/1.1:58d7b835e3ce: fate: Add dependencies for dct/fft/mdct/rdft tests
[14:18] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:eaf64192d832: Merge commit '3a6bc3e381647bb4434317113f131f7e0ab5bf83' into release/1.1
[14:19] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:a4de70df20f0: Merge commit 'd16515ae5fe7daa6327d903cafb9a5ee43477b1e' into release/1.1
[14:19] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:6e83c266201a: Merge commit '58d7b835e3cec48ab5a2393405fe82dee72c06a0' into release/1.1
[14:30] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:00915d3cd2ce: pgssubdec: Check RLE size before copying
[14:30] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:01f954032027: h264_sei: check SEI size
[14:30] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:1ee5e2ce3d6d: Merge commit '00915d3cd2ce61db3d6dc11f63566630a9aff4ec' into release/1.1
[14:30] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:0bcf514198bf: Merge commit '01f9540320279954b2764645ab7136847d53d89f' into release/1.1
[14:43] <cone-822> ffmpeg.git 03Vittorio Giovara 07release/1.1:512354191328: h264: prevent theoretical infinite loop in SEI parsing
[14:43] <cone-822> ffmpeg.git 03Diego Biurrun 07release/1.1:43d676432740: huffyuv: Check and propagate function return values
[14:43] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:ae7ea2eabf35: Merge commit '512354191328c559fcff56070dab897ee2a1b4c1' into release/1.1
[14:43] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:451bc8ee2fce: Merge commit '43d676432740c6d5e5234ed343f13902909fd124' into release/1.1
[15:02] <cone-822> ffmpeg.git 03Luca Barbato 07release/1.1:146b187113e3: lavc: Check the image size before calling get_buffer
[15:03] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:36d8914f1b94: wmalosslessdec: fix mclms_coeffs* array size
[15:03] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:02018a359e05: Merge commit '146b187113e3cc20c2a97c5f264da13e701ca247' into release/1.1
[15:03] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:a88236f3d59c: Merge commit '36d8914f1b94e4731d2fc67162902839c106e72e' into release/1.1
[15:42] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:ede738880032: mmvideo: check horizontal coordinate too
[15:42] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:c53effc41b93: huffyuvdec: check width size for yuv422p
[15:42] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:0d6ed2f13faf: Merge commit 'ede738880032db62b7dc5b3712f769d3826f5974' into release/1.1
[15:42] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:91437631d76f: Merge commit 'c53effc41b9359261b17c8da3b7062369cafd686' into release/1.1
[16:01] <cone-822> ffmpeg.git 03Anton Khirnov 07release/1.1:8cd67ddde46a: cdgraphics: switch to bytestream2
[16:01] <cone-822> ffmpeg.git 03Anton Khirnov 07release/1.1:80c268eaaee4: cdgraphics: do not return 0 from the decode function
[16:01] <cone-822> ffmpeg.git 03Anton Khirnov 07release/1.1:af9b62654d5a: svq1: do not modify the input packet
[16:01] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:420f63984b39: Merge commit '8cd67ddde46a42a33149e7d42a2ab47852ff2a83' into release/1.1
[16:01] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:06f7e87e15a6: Merge commit '80c268eaaee402695a74d14acf76063100692a99' into release/1.1
[16:01] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:e4b1cffadef2: Merge commit 'af9b62654d5aa023a96906215365532d18541a09' into release/1.1
[16:02] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:09e3fe79fc09: avcodec/svq1dec: Fix multiple bugs from "svq1: do not modify the input packet"
[16:16] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:52254067b312: error_concealment: avoid using the picture if not fully setup
[16:16] <cone-822> ffmpeg.git 03Reinhard Tartler 07release/1.1:ecda9b90eccc: Update Changelog for v9.15
[16:17] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:b65c290f7fec: Merge commit '52254067b312e78d30bbe79fc33dbdf995b22b4e' into release/1.1
[16:17] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:b52952c6e96b: Merge commit 'ecda9b90eccc687202fe9fa20f7ca61d92d816b4' into release/1.1
[16:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:fe461238d33e: avformat/utils: do not wait for packets from discarded streams for genpts
[16:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:e5fcc16a1fa7: avcodec/dvdsub_parser: never return 0 when the input isnt 0
[16:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:1298aa83180e: avcodec/dvdsub_parser: Check buf_size before reading 32bit packet size
[16:28] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:a04bb8d6e7f7: avcodec/dvdsub_parser: print message if packet is smaller than the packet size field
[16:30] <michaelni> Timothy_Gu, do you want to write a changelog/release notes for 1.1.13 ?
[16:30] <michaelni> ill make the 1.1.13 release later today or tomorrow depening on when ive time
[17:23] <cone-822> ffmpeg.git 03Carl Eugen Hoyos 07master:e4d983e2db0e: configure: Check if libwebp compilation will succeed.
[17:23] <cone-822> ffmpeg.git 03Carl Eugen Hoyos 07master:f73d75384f1d: Autodetect webp files.
[18:42] <cone-822> ffmpeg.git 03Marek Fort 07master:d1e750cd512a: avformat/mov: Support reading Avid's metadata for DNXHD codec.
[19:00] <cone-822> ffmpeg.git 03Muhammad Faiz 07master:94494dab9101: avfilter/showcqt: add and extend tlength and volume options
[20:00] <cone-822> ffmpeg.git 03Clément BSsch 07master:aaf82dc0fa47: avfilter/dctdnoiz: make color [de]correlation less clumsy
[20:00] <cone-822> ffmpeg.git 03Clément BSsch 07master:eb16a6d2294c: avfilter/dctdnoiz: remove forward declarations after previous commit
[20:00] <cone-822> ffmpeg.git 03Clément BSsch 07master:1ba7c6ead24f: avfilter/dctdnoiz: use 32-bit (float) operations instead of 64 (double) for DCTs
[20:10] <ubitux> eval api is not thread safe, right?
[20:19] <michaelni> ubitux, iam not sure i understand, what can be (not) thread safe on it ?
[20:20] <ubitux> calling av_eval_expr() on the same context from different thread
[20:20] <ubitux> i think there might be concurrent access in state variables or something
[20:20] <ubitux> but maybe i'm wrong?
[20:22] <michaelni> dunno, id have to RTFS
[20:23] <michaelni> is it a problem to use seperate contexts ?
[20:43] <ubitux> michaelni: yeah i thought of this but don't remember why i dismissed it
[20:44] <ubitux> i think the idea that there was a state to keep between calls was the reason, but it's actually wrong when i think again
[21:17] <cone-822> ffmpeg.git 03Justin Jacobs 07master:87dc8b3af913: avformat/matroskadec: Check avpriv_new_chapter() for failure
[21:37] <cone-822> ffmpeg.git 03Diego Biurrun 07master:454697603e4e: mpegts: Use av_free() to free memory allocated by av_strdup()
[21:37] <cone-822> ffmpeg.git 03Luca Barbato 07master:e4c9e59a4547: mpeg: K&R formatting cosmetics
[21:37] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:fd1e4d66f183: Merge commit '454697603e4efdfc04fadec40518d56c7dc1e5dd'
[21:37] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:a698efbc9df3: Merge commit 'e4c9e59a4547adaaa0ce9f25b0d0c5b91ae15472'
[21:45] <cone-822> ffmpeg.git 03Diego Biurrun 07master:0026e356d044: configure: Die if gas is unavailable under aarch64 as well as ARM
[21:45] <cone-822> ffmpeg.git 03Diego Biurrun 07master:444c73583d28: configure: Only run gas checks on ARM and PowerPC
[21:45] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:0472c5f8a1f1: Merge commit '0026e356d044e72b6e743b234708b8b8af457ac0'
[21:46] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:eb2def0ff2c4: Merge commit '444c73583d2848a542330c03949e1f933ac68f53'
[22:00] <cone-822> ffmpeg.git 03Diego Biurrun 07master:6f1960ab71b4: idct: cosmetics: Drop one unnecessary if-block level
[22:01] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:5ff2b33401ec: Merge commit '6f1960ab71b4f18551243ce22d01913108265233'
[22:01] <Timothy_Gu> michaelni: not really. I'm not really interested in such an old branch.
[22:14] <michaelni> Timothy_Gu, ok
[22:26] <cone-822> ffmpeg.git 03Diego Biurrun 07master:84d173d3de97: xvididct: Ensure that the scantable permutation is always set correctly
[22:26] <cone-822> ffmpeg.git 03Michael Niedermayer 07master:0dcebb9f6345: Merge commit '84d173d3de97c753234ab0c0b50551d51413d663'
[23:01] <cone-822> ffmpeg.git 03Michael Niedermayer 07release/1.1:41e859151543: update for 1.1.13
[23:53] <cone-822> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n1.1.13': unknown revision or path not in the working tree.
[23:53] <cone-822> Use '--' to separate paths from revisions
[23:53] <cone-822> refs/tags/n1.1.13:HEAD: Merge commit '84d173d3de97c753234ab0c0b50551d51413d663'
[00:00] --- Sat Aug  9 2014