[Ffmpeg-devel-irc] ffmpeg-devel.log.20130110

[burek]

[00:34] <J_Darnley> Having seen that Libav committed a patch to move yadif's filter code from inline to yasm I want to ask:  Will this also be committed into FFmpeg?
[00:35] <J_Darnley> I ask because I have some patches and I want to know if I should be converting them from inline asm to yasm.
[00:35] <Compn> J_Darnley : as long as it passes review here by michaelni (author of yadif)
[00:35] <Compn> you should ask michaelni :)
[00:38] <J_Darnley> Well, I think he may have just been highlighted so I will wait to see if he responds.
[01:15] <michaelni> J_Darnley, yes if the commit applies and the code works after applying then yes it will be commited to ffmpeg too
[01:26] <Compn> hope its still binary identical output ...
[01:43] <J_Darnley> Right, I will see what happens tomorrow and then perhaps volunteer.
[01:44] <J_Darnley> Or rather, today.
[01:46] <mfg> michaelni are you free to talk about how my changes were tested?
[02:02] <cone-149> ffmpeg.git 03Michael Niedermayer 07master:ead3a2a37e67: swr: add noise_scale field to seperate it from the noise shaping scales
[02:02] <cone-149> ffmpeg.git 03Michael Niedermayer 07master:0b60fc94aab6: swr: reorder the operations in the noise shaping code
[02:02] <cone-149> ffmpeg.git 03Michael Niedermayer 07master:2672b2c1d2d4: swr: scale data down in noise shaping to avoid cliping
[02:04] <michaelni> mfg, ill look at it now, sorry i forgot again ...
[02:08] <mfg> np
[02:08] <mfg> i'll stick around in case you have questions
[02:25] <michaelni> mfg can you fix tha av_free/av_freep() case ? and split it in a http and a hls patch?
[02:26] <michaelni> the http code is independant of the hls code i think
[02:27] <mfg> yeah, np... let me just see the av_free/av_freep you are referring to
[02:27] <michaelni> theres a av_free or 2 that are followed  by a =NULL
[02:27] <michaelni> av_freep() does that in one
[02:27] <mfg> ah, nice. its in hls
[02:28] <michaelni> also the interface between http and hls should be documented i think
[02:28] <mfg> how/where?
[02:28] <mfg> just in hls.c?
[02:30] <michaelni> currently the cookie field has a comment "Store information sent in the response"
[02:30] <michaelni> and the AVOption lacks any
[02:30] <michaelni> a reader just looking at that would not know how to use them
[02:32] <mfg> heh... woops
[02:32] <mfg> ok
[02:33] <mfg> ok, if thats it, I'll resubmit two patches tonight :)
[02:36] <michaelni> its all i spoted now, maybe wbs wants to take a look too ? but if not then the patches should be ok
[02:36] <mfg> excellent!
[02:49] <mfg> k, posted back to the list -- i'll check back in a while if I dont hear anything
[04:58] <llogan> the gmane descriptions for the mailing lists are dumb. i requested an update.
[04:58] <llogan> http://dir.gmane.org/search.php?match=ffmpeg
[05:01] <llogan> less graybeard and more neckbeard
[05:01] <llogan> although "Photos of donkeys" or something would have been lolesque.
[05:56] <cone-444> ffmpeg.git 03Michael Niedermayer 07master:2b12d1ffd841: qdm2: fix out of stack array read.
[05:56] <cone-444> ffmpeg.git 03Michael Niedermayer 07master:7cb46b519103: targa: use checked bytestream read
[05:56] <cone-444> ffmpeg.git 03Michael Niedermayer 07master:6471f63da2e9: wmalosslessdec: make arrays indexed by ch large enough for maximum number of channels
[10:05] <cone-671> ffmpeg.git 03Carl Eugen Hoyos 07master:f023003ce610: Fix compilation with --disable-everything.
[10:32] <cone-671> ffmpeg.git 03Martin Storsjö 07master:71194ef6a8ff: rtpdec_vp8: Mark broken packets with AV_PKT_FLAG_CORRUPT
[10:32] <cone-671> ffmpeg.git 03Justin Ruggles 07master:3f98848d6e04: au: validate bits-per-sample separately from codec tag
[10:32] <cone-671> ffmpeg.git 03Justin Ruggles 07master:9a7b56883d13: au: set bit rate
[10:32] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:70a65ecabff0: Merge commit '9a7b56883d1333cdfcdf0fa7584a333841b86114'
[10:56] <cone-671> ffmpeg.git 03Justin Ruggles 07master:bd4cdef5a8ca: au: set block_align and use it in au_read_packet()
[10:56] <cone-671> ffmpeg.git 03Justin Ruggles 07master:2613de880518: au: do not set pkt->size directly
[10:56] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:4cd89022e7ab: Merge commit '2613de88051818abafccb98646394341887acb3a'
[11:03] <cone-671> ffmpeg.git 03André Pankratz 07master:0287eea91430: lavfi/yadif: fix shorthand/option mismatch
[11:13] <ubitux> maybe we should make a 1.1.1 "soon"
[11:19] <cone-671> ffmpeg.git 03Justin Ruggles 07master:fb48f825e33c: au: do not arbitrarily limit channel count
[11:19] <cone-671> ffmpeg.git 03Justin Ruggles 07master:c837b38dd33a: au: move skipping of unused data to before parameter validation
[11:19] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:be894d6caeaf: Merge commit 'c837b38dd33a11c3810e988a60193a858eb4f58c'
[11:29] <cone-671> ffmpeg.git 03Justin Ruggles 07master:47d029a4c1b9: au: validate sample rate
[11:29] <cone-671> ffmpeg.git 03Justin Ruggles 07master:af68a2baae67: au: use %u when printing id and channels since they are unsigned
[11:29] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:452f63290379: Merge commit 'af68a2baae6761044cbed95575e8bcfebf55c6f1'
[12:10] <cone-671> ffmpeg.git 03Justin Ruggles 07master:bdd00e2d1be3: au: set stream start time and packet durations
[12:10] <cone-671> ffmpeg.git 03Justin Ruggles 07master:c88d245c9866: au: use ff_raw_write_packet()
[12:10] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:dc5767ff0a2a: Merge commit 'c88d245c9866e48cb8a238b7564964c1fcf3315f'
[12:23] <cone-671> ffmpeg.git 03Justin Ruggles 07master:fd9147f11456: au: cosmetics: pretty-print and remove pointless comments
[12:23] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:8c7de73e20eb: Merge commit 'fd9147f11456a7e39a998d7270684922a2a46e6d'
[12:32] <cone-671> ffmpeg.git 03Justin Ruggles 07master:2f8207b1c605: au: return AVERROR codes instead of -1
[12:32] <cone-671> ffmpeg.git 03Justin Ruggles 07master:f7a3c540c54c: au: remove unnecessary casts
[12:32] <cone-671> ffmpeg.git 03Daniel Kang 07master:899157b30800: yadif: Port inline assembly to yasm
[12:32] <cone-671> ffmpeg.git 03Justin Ruggles 07master:f7bf72a4a114: idcinvideo: correctly set AVFrame defaults
[12:32] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:65b8527993a1: Merge commit 'f7bf72a4a1146a7583577c9bdc066767e1ba3c6a'
[12:40] <cone-671> ffmpeg.git 03Justin Ruggles 07master:b0c96e06134d: idcin: validate header parameters
[12:40] <cone-671> ffmpeg.git 03Justin Ruggles 07master:12c2530b1d87: idcin: fix check for presence of an audio stream
[12:40] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:8d691f877769: Merge commit '12c2530b1d87fa94f81ea97df575b77c825e6f4f'
[12:45] <cone-671> ffmpeg.git 03Justin Ruggles 07master:4b840930da7c: idcin: set channel_layout
[12:45] <cone-671> ffmpeg.git 03Justin Ruggles 07master:ccc0ffb1ba3f: idcin: set start_time and packet duration instead of manually tracking pts.
[12:45] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:6e44662b0521: Merge commit 'ccc0ffb1ba3fc1adb05a9f56dfc26131e61db3fb'
[12:50] <cone-671> ffmpeg.git 03Justin Ruggles 07master:49543373f3e5: idcin: set AV_PKT_FLAG_KEY for video packets with a palette
[12:50] <cone-671> ffmpeg.git 03Justin Ruggles 07master:7040e479a153: idcin: allow seeking back to the first packet
[12:50] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:a0dbe2a58cf5: Merge commit '7040e479a1530b2eda4b89a182d5eb50a77bd907'
[12:56] <cone-671> ffmpeg.git 03Justin Ruggles 07master:33f58c3616d2: idcin: check for integer overflow when calling av_get_packet()
[12:56] <cone-671> ffmpeg.git 03Justin Ruggles 07master:5d0450461ff7: idcin: better error handling
[12:56] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:56ca871fb3d6: Merge commit '5d0450461ff729be5f531d333d29754155e406c5'
[13:02] <cone-671> ffmpeg.git 03Justin Ruggles 07master:06deaf8ad33a: idcin: return 0 from idcin_read_packet() on success.
[13:02] <cone-671> ffmpeg.git 03Luca Barbato 07master:fba8e5b60857: oggdec: fix faulty cleanup prototype
[13:02] <cone-671> ffmpeg.git 03Luca Barbato 07master:a800fd5fc797: yuv4mpeg: do not use deprecated functions
[13:02] <cone-671> ffmpeg.git 03Martin Storsjö 07master:ed79093222ce: rtpdec: Add a terminating null byte at the end of the SDES/CNAME
[13:02] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:ac6e074fb711: Merge commit 'ed79093222ceb42f0c3a39095a69af0b32be5450'
[13:03] <cone-671> ffmpeg.git 03Carl Eugen Hoyos 07master:b563afe6fa93: Do not allow arbitrary packet size in aac decoder.
[13:12] <cone-671> ffmpeg.git 03Martin Storsjö 07master:3b366c3aa080: rtpdec_vp8: Simplify code by using an existing helper function
[13:12] <cone-671> ffmpeg.git 03Martin Storsjö 07master:76c40fbef06a: rtpdec_vp8: Don't trim too much data from broken frames
[13:12] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:91da6b97c7fb: Merge remote-tracking branch 'qatar/master'
[14:04] <cbsrobot> wut - a ffmpeg book ? I wonder how long the examples will work &.
[14:15] <JEEB> oh boy
[14:15] <JEEB> wonder what version of the APIs it was written for :D
[14:20] <pross-au> Edition 2. $$
[15:26] <divVerent> this is fun, there's so many different implementations of no_proxy and no standard
[15:26] <divVerent> I am leaning towards suggesting to not support it at all, which is still better than the broken one we have now. Or using a very simple one like curl or lynx.
[17:25] <wm4> <wm4> well, sometimes ffmpeg marks functions as deprecated, even if there are no stable replacements yet
[17:25] <wm4> lol why do I have to explain this to people
[17:25] <ubitux> you are refering to the url thing?
[17:27] <wm4> that's just in general, but I remember similar occasions
[17:27] <ubitux> please be specific if you expect a solution :P
[17:30] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:36055aa87645: idcin: fix memleak
[18:56] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:fca51256d4f4: swr: rename dither_pos to noise_pos
[18:56] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:558aa6cab7ff: swr: add seperate in/out pointers to the noise shaping code
[18:56] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:2eec98125ead: swr: mark noise data for noise shaping as const
[18:56] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:e5a736261baf: swr: Use a temporary buffer for dither/Noise shaping when the input is read only
[19:21] <wm4> ubitux: is the reason still not known for this ticket?
[19:24] <ubitux> yes
[19:24] <ubitux> looks like a build issue
[19:24] <wm4> ubitux: I think it might be the fontconfig configuration
[19:24] <wm4> this seems to be often an issue on osx
[19:25] <wm4> I have it on wine too ;D
[19:25] <wm4> (with mpv)
[19:25] <wm4> fontconfig probably can't find a font or so
[19:25] <ubitux> :)
[19:25] <ubitux> i can't really tell :p
[19:25] <ubitux> feel free to comment
[19:25] <wm4> just look at this crap that recently made it in: https://github.com/mpv-player/mpv/blob/master/TOOLS/osxbundle/mpv.app/Contents/Resources/fonts.conf
[19:26] <wm4> though that was a different issue
[19:26] <ubitux> heh
[19:26] <wm4> so maybe try to make them check their fontconfig configuration
[20:16] <cbsrobot> wm4: if you have any hints on how to debug #2100 - feel free to share
[20:18] <cbsrobot> I tried a few things with help of ubitux ( I may call him soon subitux ), but I didn't find a solution yet
[20:33] <cone-420> ffmpeg.git 03Michael Niedermayer 07master:a74548521b23: swr: check that allocation of in/out_convert succeeded
[20:45] Action: cbsrobot wonders since when mov files always have 12800 tbn
[20:48] <michaelni> does it cause some problem ?
[20:48] <michaelni> it was needed for perfect AV sync IIRC
[20:49] <cbsrobot> it causes problems for the concat demuxer
[20:50] <cbsrobot> or lets say I guess it causes problems for it
[21:47] <cbsrobot> so --disable-everything leavs aformat, crop null, anull, format and setpts enabled ...
[21:52] <Zeranoe> I'm trying to come up with a changlog for the windows builds that will display all the changes from the previous builds to latest. I'm trying "git log --date=iso --pretty=fuller 2672b~.." for example, which seems to cut the log, but I'm still getting commits that are a day or so older than 2672b. Any ideas why this might be happening or how I could go about creating a build by build changelog?
[21:55] <cone-420> ffmpeg.git 03Nicolas George 07master:7e5c5fa5fc8f: lavfi/vf_overlay: check that filter_frame does not return EAGAIN.
[22:09] <durandal_1707> michaelni: why you reverted au.c change by me?
[22:09] <durandal_1707> that code for reading packet is overkill
[22:10] <durandal_1707> and does nonsense duplication
[22:10] <durandal_1707> also qatar idcin changes are nonsense, like limiting number of channels to 2
[22:11] <michaelni> durandal_1707, justin added duration setting code in au read packet, to merge this it was not possibly to easily keep using pcm read packet
[22:14] <durandal_1707> that is not needed at all
[22:14] <durandal_1707> and can be trivially added to pcm read packet
[22:19] <michaelni> i thought so too, but such changes should be followup commits to merges not hidden in a merge because thats not what the commit that is merged does
[22:20] <michaelni> even more so because its possible to make mistakes and if droped the duration addition and it actually was needed i would have added a bug
[22:21] <durandal_1707> yes, i never wanted that you do it in merges
[22:24] <durandal_1707> btw, that lswr work is amazing
[23:21] <wm4> wtv seems to be the most popular thing these days...
[23:48] <llogan> michaelni: looks like a single, custom black shirt will be $33.25. Expensive.
[23:49] <llogan> that includes "economy" shipping for $5.25 USD.
[00:00] --- Fri Jan 11 2013