[Ffmpeg-devel-irc] ffmpeg-devel.log.20140908

[burek]

[00:38] Action: Daemon404 wonders if nicholas has ever had a non-terrible design idea ever
[00:45] <cone-977> ffmpeg.git 03Martin Storsjö 07master:803e82276b37: libavformat: Check mkdir return error codes
[00:45] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:aecd064744c8: Merge commit '803e82276b3716bf6012ec69e8854dae14a4fd2b'
[00:52] <cone-977> ffmpeg.git 03Martin Storsjö 07master:fcf597625c7a: ismindex: Avoid writing ismf files if no base name has been specified
[00:52] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:1a59a288380a: Merge commit 'fcf597625c7a991ca389f3a9b8ff4f5e383301c0'
[01:25] <cone-977> ffmpeg.git 03Diego Biurrun 07master:1274ea8dbaec: Split off floating point AAN (I)DCT into separate components
[01:25] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:f044fc03727d: Merge commit '1274ea8dbaec44355bde07f6bb31fec0c4e6fd2d'
[05:25] <cone-312> ffmpeg.git 03Hendrik Leppkes 07master:e9c08e32d6b9: avformat/utils: Exclude the header size when computing duration from bit_rate
[12:29] <ao2> hi, I fixed a problem I had with the x11grab input device. Is it OK if I send the fix to both ffmpeg and libav? The two patches are slightly different, that's why I am asking.
[12:34] <J_Darnley> Probably.
[12:34] <J_Darnley> I think it would save michael from having to do some work merging it.
[12:39] <ao2> J_Darnley, OK, thanks.
[13:18] <ubitux> ao2: since you're working on x11grab, would you be interested in implementing http://trac.ffmpeg.org/ticket/2563 ?
[13:23] <michaelni> nevcairiel, ok to cherry pick de45959d and 63324443 or is there some reason not to / wait ?
[13:25] <ao2> ubitux, I don't think so, sorry. JFYI GStreamer's ximagesrc provides an xid option already to capture a specific window; some fixes are needed there too, tho (https://bugzilla.gnome.org/show_bug.cgi?id=690646).
[13:26] <ubitux> ok :)
[14:33] <ubitux> so libpostproc is like 6k LoC
[14:33] <ubitux> is it really a problem?
[14:34] <Compn> ubitux : depends if you are a deletionist or not
[14:34] <Compn> :p
[14:46] <Compn> curious why debian wants it seperate ? is it because its called 'lib*' ?
[14:47] <Compn> theres some talk on debian somewhere about ffmpeg that i'm not in the loop of.
[15:03] <BtbN> Debian has all ffmpeg libs in seperate packages as far as i know
[15:04] <Compn> blurgh
[15:06] <BtbN> But it was allways like that, even for libav
[15:06] <Compn> guess it makes some sense to do it that way , just a lot more work
[15:24] <cone-636> ffmpeg.git 03Stefano Sabatini 07master:f51de9cac2fd: lavf/segment: abort in case of invalid segment format options
[15:24] <cone-636> ffmpeg.git 03Stefano Sabatini 07master:24db666d1549: MAINTAINERS: add myself as segment maintainer
[17:18] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:951f6fa3f31b: avformat/rtpdec_qt: Fix 'warning: passing argument 2 of ffio_init_context discards const qualifier from pointer target type'
[17:18] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:5a6dd925cb3f: avcodec/ac3dec_fixed: remove unused arrays
[18:09] <durandal_1707> how you escape :
[19:30] <someone-noone> Hi. Perhaps someone knows a possibility (fork, pull-request or work-around) to get a microphone as input in OS X?
[19:31] <someone-noone> Or someone know how to fix "malloc checksum" errors in OpenAL described here: https://trac.ffmpeg.org/ticket/314#no1
[19:34] <Compn> someone-noone : paste me output of ffmpeg -list_devices true 
[19:37] <someone-noone> Compn, http://pastebin.com/0HjDzaRS
[19:37] <Compn> oops, that option is gone 
[19:37] <Compn> lol thats what i get for looking at old bugreports
[19:38] <Compn> someone-noone : do ffmpeg -devices
[19:38] <someone-noone> Compn,  D  avfoundation    AVFoundation input device
[19:38] <someone-noone>  D  lavfi           Libavfilter virtual input device
[19:38] <someone-noone>  D  qtkit           QTKit input device
[19:39] <someone-noone> ./ffmpeg -f qtkit -list_devices true -i ""  \\ produces:
[19:39] <someone-noone> [QTKit input device @ 0x7fb822c10c60] QTKit video devices:
[19:39] <someone-noone> [QTKit input device @ 0x7fb822c10c60] [0] AB@>5==0O iSight
[19:39] <someone-noone> [0] is my webcam
[19:40] <someone-noone> ./ffmpeg -f avfoundation -list_devices true -i ""  \\ produces:
[19:40] <someone-noone> [AVFoundation input device @ 0x7f88f3c11d20] AVFoundation video devices:
[19:40] <someone-noone> [AVFoundation input device @ 0x7f88f3c11d20] [0] AB@>5==0O iSight
[19:40] <someone-noone> same 
[19:40] <Compn> humm
[19:42] <Compn> try ffmpeg -f avfoundation -list_devices true -i "" or ffmpeg -f avfoundation -devices true -i ""
[19:42] <Compn> probably the latter.
[19:43] <someone-noone> first, I did it, result is posted above
[19:43] <someone-noone> second, produces same result as ffmpeg -devices
[19:43] <Compn> you did -f qtkit
[19:43] <Compn> er never mind
[19:43] <Compn> i see :)
[19:43] <Compn> humm
[19:47] <Compn> someone-noone : what osx ver are you running ?
[19:49] <Compn> someone-noone : so ffmpeg -f openal -list_devices crashes ?
[19:52] <Compn> someone-noone : you dont happen to have any bluetooth microphones that could be used? or maybe a webcam with mic or anything like that ?
[19:52] <Compn> usb mic
[20:15] <cone-636> ffmpeg.git 03Carl Eugen Hoyos 07master:2b1f6a351638: Print the actual bit depth for audio if it is different from the sample_fmt bit depth.
[20:15] <cone-636> ffmpeg.git 03Carl Eugen Hoyos 07master:42b4da75b73d: Print bits_per_raw_sample for ffprobe -show_streams.
[20:15] <cone-636> ffmpeg.git 03Carl Eugen Hoyos 07master:552ce6874c8a: Revert the default for ffplay to -noautoexit.
[20:26] <someone-noone> Compn, I can't install openal. But I can do like here https://trac.ffmpeg.org/ticket/314#no1 and it will produce "malloc checksum" error. Do you know how to fix it? I have a built-in micro
[20:54] <ubitux> there is still no ass specs?
[20:54] <ubitux> even markdown got one recently based on a lost perl script full of regex
[20:55] Action: ubitux wonders if we'll see vp9 specs before ass ones
[20:57] <wm4> ubitux: there is
[20:57] <wm4> the vsfilter source code
[20:57] <ubitux> right
[20:57] <ubitux> libvpx are the vp9 specs as well
[20:58] <Daemon404> webvtt is the future obviously
[20:58] <wm4> unlike libvpx, vsfilter is immutable, though
[20:58] <Daemon404> wm4, \blur disagrees
[20:58] <ubitux> vsfilter is dead?
[20:59] <wm4> it's kept alive by chinese intrinsic grinders
[21:09] <ubitux> what was the meaning of "Marked"?
[21:11] <J_Darnley> In a particular context or just in English?
[21:12] <ubitux> in the ass context
[21:12] <ubitux> or more likely, ssa
[21:12] <ubitux> it was later replaced with a layer field
[21:13] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:baee313aa9a5: avcodec/diracdec: Fix "warning: comparison of constant 4 with expression of type enum dirac_subband is always true [-Wtautological-constant-out-of-range-compare]"
[21:13] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:a67b6c954154: avcodec/cabac_functions: fix "warning: UNCHECKED_BITSTREAM_READER is not defined, evaluates to 0"
[21:14] <J_Darnley> What a warning option name!
[21:15] <wm4> ubitux: maybe gui state
[21:16] <ubitux> GUI state? oO
[21:17] <wm4> ubitux: ass was never meant for playback
[21:17] <wm4> it was the file format of some subtitle editor written in visual basic
[21:17] <wm4> aegisub still stores gui state in .ass files btw.
[21:18] <wm4> lately I've even got a file that had 20MB of such state
[21:18] <wm4> I didn't check, but the subtitles probably took up more file size than the audio track
[21:19] <ubitux> lol
[21:20] <ubitux> wm4: so it would be a GUI state such that the editor would still keep the selected sub events in memory?
[21:20] <wm4> these people must suffer from brain damage
[21:20] <wm4> ubitux: probably
[21:26] <ubitux> http://web.archive.org/web/20030603235926/http://www.eswat.demon.co.uk/substation.html
[21:26] <ubitux> blast from the past
[21:26] <ubitux> Compatibility with other subtitling programs
[21:26] <ubitux> Reads JACOsub, ZeroG, Subsonic, MacSub, RealText and Plain Text files.
[21:26] <ubitux> omg
[21:52] <ubitux> michaelni: when merging 1ec335513f95bf2441f81d761ea127325ecd81a0, please mention ebef9f5a56d7df91e010a177a80cfc8dbe394305
[21:52] <ubitux> ebef9f5a56d7df91e010a177a80cfc8dbe394305 was truncated to 56d7df91e010a177a80cfc8dbe394305 instead of ebef9f5a for some random reason
[21:53] <ubitux> maybe a joke with "leftover" or something
[22:04] <ubitux> wm4: <&jfs> iirc it was whether an editor had marked the line for attention
[22:04] <ubitux> seems you were right
[22:11] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:1360db7e888e: avcodec/mpeg4videodec: fix 'libavcodec/mpeg4videodec.c:2762:22: warning: unused variable mpeg4_vdpau_class [-Wunused-const-variable]'
[22:29] <cone-636> ffmpeg.git 03Gabriel Dume 07master:1ec335513f95: time: Add missing zero
[22:29] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:6daff3bd8f30: Merge commit '1ec335513f95bf2441f81d761ea127325ecd81a0'
[23:19] <wm4> I get stuff like "Channel layout '5.1(side)' with 6 channels does not match specified number of channels 2: ignoring specified channel layout"
[23:19] <wm4> it seems this happens because I set request_channel_layout (for downmix), but not request_channels
[23:19] <wm4> but request_channels is deprecated
[23:19] <wm4> ????
[23:22] <wm4> oh, actually setting request_channels doesn't help
[23:27] <wm4> oh looked at the wrong code
[23:28] <wm4> anyway, this looks like at least the ac3 decoder fails to set the channel layout correctly on downmix
[23:28] <wm4> this is just because ffmpeg moronically decided to keep the redundant channel count...
[23:41] <Plorkyeran> the 20 MB of crap is undo data for the motion tracking script
[23:41] <Plorkyeran> i.e. entirely useless garbage
[23:42] <wm4> nice
[23:43] <Plorkyeran> I guess I should get around to releasing 3.2.1, which deduplicates that data
[23:43] <Plorkyeran> but I was kinda hoping people would just make their release scripts strip garbage
[23:51] <J_Darnley> Can I ask if this does what I think it does...
[23:51] <J_Darnley> pinsrb     m3,  byte r5m, 0
[23:52] <J_Darnley> load 1 byte from the memory at r5m into the lowest byte of m3
[23:52] <J_Darnley> ?
[00:00] --- Tue Sep  9 2014