[Ffmpeg-devel-irc] ffmpeg-devel.log.20130703

[burek]

[00:03] <saste> any MP4 experts around?
[00:04] <saste> I'm getting "[iso file] Box "avcC" size 8 invalid (read 909)" with MP4Box
[00:07] <llogan> saste: i recall you wanting a PPC. i have acquired a powerbook G4. still interested?
[00:08] <llogan> shipping cost may be prohibitive. i haven't checked.
[00:11] <durandal_1707> saste: you asked that already, i lost my magic wand
[00:12] <durandal_1707> first thing is to get mp4box source and locate that printf...
[00:12] <cone-60> ffmpeg.git 03Tudor Suciu 07master:820ffaed0f64: avformat/matroskaenc: Allow VP9 and Opus in webm
[00:45] <mark4o> llogan: http://ffmpeg.org/download.html now links to a non-functional site http://ffmpegmac.net/ ?
[00:47] <llogan> mark4o: non-functional? worksforme
[00:47] <llogan> or am i missing something?
[00:47] <mark4o> says "This site is currently under construction.
[00:47] <mark4o> Please try again later."
[00:48] <llogan> i can't duplicate the issue. anyone else?
[00:54] <michaelni> bcoudurier, do you have time to review the remaining gxf patches ? (0617 22:19 Reuben Martin   (1.5K) [FFmpeg-devel] [RFC] GXF Patches - HD)
[00:55] <mark4o> llogan: aha, it works on IPv4.  It seems ffmpegmac.net has an IPv6 address and an IPv4 address, and the IPv4 address works and the IPv6 address doesn't.
[01:00] <llogan> mark4o: ah, i see. (reminds me to check my upcoming new server for IPv6 stuff soon).
[01:04] <Compn> mark4o : what web browser do you have that defaults to ipv6 first ?
[01:05] <mark4o> I tried chrome, firefox, and safari and they all use ipv6 first if you have it.
[01:05] <mark4o> In firefox if I set network.dns.disableIPv6 in about:config it will use ipv4 only and then ffmpegmac.net works.
[01:08] <Compn> wow thats terrible
[01:08] <Compn> ipv6 first ? so many broken sites
[01:09] <Compn> good to know, thanks
[01:09] <mark4o> broken sites?  that is the first broken site i've encountered since enabling ipv6 in my router a few months ago
[01:10] <Compn> sometimes i make stuff up. 
[01:10] <Compn> :)
[01:10] <mark4o> most sites do not have an ipv6 address, or wouldn't publish one if it didn't work
[02:32] <cone-60> ffmpeg.git 03Michael Niedermayer 07master:e0be3cbd2ba7: MAINTAINERS: add address to contact "AvxSynth Team"
[09:43] <khali> hi
[09:43] <khali> can anyone explain to me what "av_cold" means/does?
[09:44] <nevcairiel> it informs the compiler about functions that are unlikely to be executed
[09:44] <ubitux> http://gcc.gnu.org/onlinedocs/gcc/Function-Attributes.html
[09:45] <nevcairiel> which means its used for functions that are rarely called
[09:45] <nevcairiel> like init/close
[09:46] <khali> ah, so "av_cold" resolves to "cold" if the compiler supports it, and nothing if it doesn't?
[09:46] <nevcairiel> __attribute__((cold)) if  you want to be precise, but sure
[09:47] <khali> oh, and there is "hot" as well
[09:48] <khali> but apparently we don't use that
[09:48] <nevcairiel> i havent seen that in use anywhere
[09:48] <khali> nevcairiel, ubitux: thanks!
[10:43] <cone-5> ffmpeg.git 03Stefano Sabatini 07master:b2498dcf5ca9: lavc/avcodec.h: fix prolems -> problems typo
[11:25] <cone-5> ffmpeg.git 03Luca Barbato 07master:5650e331a733: jpeg2000: Validate resolution levels
[11:25] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:37a970bbb668: Merge commit '5650e331a733711b809bc27f968500f0f5530126'
[11:28] <cone-5> ffmpeg.git 03Stefano Sabatini 07master:628ceac65267: doc: add script to compute texi files dependencies
[11:28] <cone-5> ffmpeg.git 03Timothy Gu 07master:4ec46b1160eb: doc/muxers: Add AIFF doc
[11:33] <cone-5> ffmpeg.git 03Stefano Sabatini 07master:759dfd56845d: examples/muxing: make more compact audio/video_time computation
[11:46] <michaelni> saste, Cannot open file 'doc/avtools-common-opts.texi': No such file or directory at /home/michael/ffmpeg-git/ffmpeg/doc/texidep.pl line 12, <$fh> line 217.
[11:47] <michaelni> mkdir whatever ; cd whatever ; ../configure --cc='ccache gcc'  && make -j12
[11:50] <saste> michaelni, revert
[11:51] <saste> wait
[12:02] <nevcairiel> quite a bunch of fate boxes are also failing to build now
[12:03] Action: michaelni suspects all will fail
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:fbcc03db8f39: jpeg2000: Check code-block size
[13:08] <cone-5> ffmpeg.git 03Luca Barbato 07master:be3271009ef8: jpeg2000: Define the maximum decomposition levels
[13:08] <cone-5> ffmpeg.git 03Luca Barbato 07master:5efadcb8cdef: jpeg2000: Clean up return paths and error messages
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:1a3598aae768: jpeg2000: Use bytestream2
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:3f0364327c32: Merge commit 'fbcc03db8f3919e2ea46d6ad11a00906b1c0ef0e'
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:b0448e512357: Merge commit 'be3271009ef893fbed9640e8edbd771685fd957b'
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:26f6acc66b66: Merge commit '5efadcb8cdef6fc5ca5f1f72e45ef3b23016b1c4'
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:57bc64e235fd: Merge commit '1a3598aae768465a8efc8475b6df5a8261bc62fc'
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:ffed7227c331: Revert "doc: add script to compute texi files dependencies"
[13:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:7c71ef4f1727: Merge commit '57bc64e'
[13:09] Action: michaelni looks around and sees no saste
[13:09] <michaelni> saste, ive reverted it
[13:10] <durandal_1707> why?
[13:10] <michaelni> durandal_1707, because it broke build
[13:11] <durandal_1707> i read that in commit log
[13:12] <durandal_1707> it broke normal build, or something special?
[13:13] <durandal_1707> aw, it made fate red
[13:14] <saste> michaelni, ok thanks
[13:23] <cone-5> ffmpeg.git 03Stefano Sabatini 07master:5efbeae38cdc: configure: rename have_incompatible_fork_abi -> have_incompatible_libav_abi
[13:23] <cone-5> ffmpeg.git 03Stefano Sabatini 07master:838bd731393a: lavfi: create Libav-API compatibility layer for avfilter_graph_parse() at the next bump
[13:24] <nevcairiel> does it really make all that much sense to try to restore compat month (if not years) later? :d
[13:26] <ubitux> allow apps to build with both projects with a single code path
[13:26] <saste> http://thread.gmane.org/gmane.comp.video.ffmpeg.libav.user/11317
[13:26] <saste> michaelni, do you have any hint?
[13:27] <saste> why the profile information is not propagated?
[13:27] <khali> when working with yuv420p pixel format, each Y, U and V sample is 8-bit, isn't it?
[13:27] <saste> or better, how should I copy the profile information?
[13:27] <nevcairiel> khali: yes
[13:33] <saste> michaelni, plans for the next release?
[13:33] <durandal_1707> yes, really, this becoming tedious
[13:37] <michaelni> durandal_1707, help fix some bugs
[13:39] <michaelni> saste, yes but my todo list is long
[13:39] <michaelni> redoing merges because someone commited in the meantime
[13:40] <michaelni> reverting commits that broke build
[13:41] <michaelni> and above all, merging my own commits that libav decided to rebase and randomly change
[13:41] <Compn> how many breakages have we had in the past month? ;\
[13:41] <michaelni> too many ...
[13:41] <michaelni> i mean 1 is too many already IMHO
[13:42] <Compn> is there anyway to have a pre-commit hook that tries to build ffmpeg? :P
[13:43] <michaelni> yes we have (sort of), just dont commit to the main tree and instead commit to your private one and ask me to merge, i test it before push
[13:43] <Compn> pre-commit hook aka michael :)
[13:44] <michaelni> exactly :)
[13:47] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:d3cb302b8850: jpeg2000: Validate SOT parsing
[13:47] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:2e85737b0e0b: Merge commit 'd3cb302b88503c3111e25add196622110c056188'
[13:51] <wm4> uh so how does avfilter_graph_parse work
[13:52] <wm4> is there an API that actually works on both libs?
[13:52] <nevcairiel> in this case probably not since libav changed theirs and ffmpeg kept the old one, or something like that
[13:53] <nevcairiel> not sure libav has the "2" variant
[13:53] <wm4> I think it was actually ffmpeg which changed it
[13:54] <wm4> ah libav has the 2 variant
[13:56] <xlinkz0> why are there no examples of using av_write_frame?
[13:56] <xlinkz0> i found a seemingly pretty well written example on the web but it's from 0.49 :(
[13:57] <wm4> so I guess users can actually use the 2 variant with no problems... except that I don't quite understand what the 2 variant does
[13:57] Action: wm4 looks it up
[14:06] <wm4> ok avfilter_graph_parse2 is oddly different
[14:20] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:278a923c5118: jpeg2000: Validate SIZ parsing
[14:20] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:f0358dc1d30c: Merge commit '278a923c51187d37445d88a6b21082036ec9568d'
[14:29] <cone-5> ffmpeg.git 03Luca Barbato 07master:7e201d575dc4: jpeg2000: Validate block lengthinc
[14:30] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:db5c93a1a97d: Merge commit '7e201d575dc4385eb67314b0419d4d77185e65f4'
[14:36] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:17e5d614a864: jpeg2000: Check zero bit-plane validity
[14:36] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:2cad1ea29e4d: Merge commit '17e5d614a8647d51b9795cb8bccf97ee33ca2d58'
[14:37] <superware> michaelni: hi. I couldn't manage to build ffmpeg, spent more than a few hours on this, too advanced for me. Is there any other way to test the mpegts patchset? do you know of someone that might be able to build related dll for me to try w/ a udp stream?
[14:38] <michaelni> superware, i suspect zeranoe could do that for you but probably not for free
[14:39] <wm4> superware: building ffmpeg is quite simply on other OSes... now I don't want to say "install Linux", but...
[14:39] <superware> wm4: can you do a cross build?
[14:41] <wm4> cross compiling from Linux to Windows isn't that hard either, but requires the use of mingw
[14:43] <superware> wm4: if I'll send the UDP stream to your IP, can you please try the patchset?
[14:44] <wm4> no, I'm behind a router anyway
[14:45] <superware> I see
[14:45] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:eae63e3c156f: jpeg2000: Check component number in get_coc() and get_qcc()
[14:45] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:fd54dd028bc9: jpeg2000: check len before parsing header
[14:45] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:a4a16821d732: Merge commit 'fd54dd028bc9f7bfb80ebf823a533dc84b73f936'
[14:48] <michaelni> superware, btw if you manage to build ffmpeg yourself you also would have learnt alot and this may be usefull to you in the future as you then can build your own custom versions and dont depend on what binaries you find on the web
[14:49] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:ac921338a498: jpeg2000: Correctly calculate sgnd
[14:49] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:86a2602e1617: jpeg2000: Drop unused and writeonly fields
[14:49] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:9dfa5ea5dcc8: Merge commit '86a2602e16171ddc10c3ec655c23b16ffa595059'
[14:51] <superware> michaelni: it's not that big advantage. It's was just pure luck that I happend to stumble upon such a bug, I mean, it's quite severe IMHO.
[14:52] <superware> or at least should have been spotted by other users a long time ago
[14:52] <michaelni> it probably was but i guess they didnt report it
[14:56] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:78962d3df49a: jpeg2000: Check that nreslevels2decode has been initialized before use
[14:56] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:ef35d6dbc6c3: jpeg2000: Propagate error code from get_cox()
[14:56] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:ac3ca3d311bb: Merge commit 'ef35d6dbc6c3b7ba6b13ac13fc8e797cc1268c8f'
[14:56] <superware> well, I'll be here if you have a few minutes to try the stream out. thanks.
[15:01] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:2c3901b2c34b: jpeg2000: Remove unneeded returns
[15:01] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:b564784a207b: jpeg2000: Check that there is a SOT before SOD
[15:01] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:273f600b037c: Merge commit 'b564784a207b1395d2b5a41e580539df04651096'
[15:13] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:cf04af2086be: jpeg2000: Check that we have enough components for MCT
[15:13] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:b44925ae6b4b: jpeg2000: Initialize code blocks structures in precincts to 0
[15:13] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:4cbd5ed11b01: jpeg2000: Fix compute precedence error in lut_gain index
[15:13] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:c7a092cb2d57: Merge commit '4cbd5ed11b0161a6a192174934b241e0cc0599a9'
[15:26] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:4e11b1553451: jpeg2000: Calculate code-block coord in ff_jpeg2000_init_component()
[15:26] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:952f7ed3c0b0: jpeg2000: Compute quantization for 'scalar derived' in the correct case.
[15:26] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:f9581f1414ec: jpeg2000: Improve reduced resolution decoding
[15:26] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:11cf4b72475f: Merge commit 'f9581f1414ec8e3e7d0868793cda4f2e4fa113ea'
[15:39] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:a458b91cf435: jpeg2000: Split int/float codepaths depending on the DWT
[15:39] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:5bf208f65970: jpeg2000: Use separate fields for int and float codepaths
[15:39] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:2328e7e86521: Merge commit '5bf208f659703895df7926238dcfa8a8175de36b'
[15:43] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:690c8af97bb1: jpeg2000: Remove unneeded variable initialization
[15:43] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:db13e02c6f03: jpeg2000: Compute corrections of sub-band coordinates
[15:43] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:44741113257b: Merge commit 'db13e02c6f0355fabd540e77b61ade127d6411ff'
[16:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:33e665a6abf1: jpeg2000: Update pixel format support
[16:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:5b73916d3452: jpeg2000: Simplify init_tile()
[16:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:589e5b52f634: jpeg2000: Use the correct sizeof in memset for T1 data
[16:08] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:ee7da5baccc3: Merge commit '589e5b52f634f6b2d307a167a19eef7e7328cb08'
[16:18] <cone-5> ffmpeg.git 03Luca Barbato 07master:daeb4e3042f2: jpeg2000: Proper cleanup on failure in decode_frame()
[16:18] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:1abb875c980a: Merge commit 'daeb4e3042f2ecae2d41aaa4cae0bed932539788'
[16:32] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:aa16bbaf9bca: jpeg2000: Refactor SOT marker parsing
[16:32] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:09d5929f3721: jpeg2000: Do not crash on NULL node in tag_tree_decode
[16:32] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:d57c737ac30c: jpeg2000: Simplify jpeg2000_decode_packets()
[16:32] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:19ff6f55e800: Merge commit 'd57c737ac30cf079a342fa649fd4888c2d059dd9'
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:64f6570c6e2c: jpeg2000: Use EBCOT's CAUSAL and BYPASS mode in decode_cblk()
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:53d5d89c1b75: jpeg2000: Speed up jpeg2000_decode_tile()
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:c1dcbc590d90: jpeg2000: Do not assume a single tile
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:e11099db203c: jpeg2000: Optimize dequantization
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:f0552e63a669: jpeg2000: Reset s->numX/Ytiles on tile deallocation
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:22e18ea39e37: jpeg2000: Optimize output sample conversion
[16:42] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:5eb23e6b0ae2: Merge commit '22e18ea39e371030cc78973d1b46aae45a7ea215'
[16:48] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:c4604b8792ca: jpeg2000: Merge rescaling with interleaving in 9/7 IDWT
[16:48] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:95a57d26d865: jpeg2000: Use the matching coding style struct for transform selection
[16:48] <cone-5> ffmpeg.git 03Michael Niedermayer 07master:13f86baa1d98: Merge remote-tracking branch 'qatar/master'
[16:50] <durandal_1707> michaelni: why you merge stuff?
[16:52] <Compn> why not ?
[16:54] <durandal_1707> then why complaining?
[16:55] <durandal_1707> michaelni: what bugs needs to be fixed?
[16:55] <durandal_1707> i posted patches for two bug fix and nobody cares
[16:55] <michaelni> the ones marked as imortant on trac should be an approximate list
[16:55] <michaelni> which patches ?
[16:57] <Compn> there arent many people reviewing patches lately, remember 2 weeks is the time limit so far :)
[17:03] <durandal_1707> looks like it extending to 2 years
[17:04] <durandal_1707> michaelni: i think more than half of important tickets are not that
[17:05] <michaelni> thats why i said approximate :)
[17:06] <michaelni> also dont hesitate to update priority where its bad
[17:07] <durandal_1707> i hestiate to touch trac because cehoyos may undo it at random
[17:22] <cone-5> ffmpeg.git 03Matthieu Bouron 07master:4943eff2cd28: lavf/movenc: fix writing of some iTunes metadata tag
[17:29] <wm4> what defines AV_HAVE_INCOMPATIBLE_LIBAV_ABI? I get warnings like include/libavfilter/avfilter.h:1314:5: warning: "HAVE_INCOMPATIBLE_LIBAV_ABI" is not defined [-Wundef]
[17:30] <wm4> installation problem?
[17:30] <nevcairiel> configure should set it
[17:30] <nevcairiel> but config.h is usually not included in public API
[17:30] <nevcairiel> so someone messed up
[17:32] <wm4> also I see both HAVE_INCOMPATIBLE_LIBAV_ABI and AV_HAVE_INCOMPATIBLE_LIBAV_ABI in the ffmpeg sources
[18:57] <Daemon404> lul
[21:50] <durandal_1707> michaelni: what's your opinion on adpcm ima wav in separate file?
[22:03] <durandal_1707> no more comments for "default" opt patch?
[22:04] <durandal_1707> michaelni: are you forcing another policy? that only you commit stuff directly?
[22:07] <michaelni> no that would be really wrong
[22:07] <michaelni> IMHO best is if maintainers commit in their repository and that  gets merged
[22:08] <Daemon404> the only annoyance with that is 9000 merge commit messages
[22:08] <Daemon404> (even more than now)
[22:08] <michaelni> yes
[22:09] <michaelni> that problem is a bit specific to merging 2 year diverged trees though
[22:09] <Daemon404> well it would be nice of it the branch could be rebased on master before merge
[22:10] <Daemon404> and thus no crappy message
[22:10] <michaelni> about spliting adpcm, i have no objections if that makes technical sense
[22:11] <michaelni> rebases can lead to "darn i tested that and it worked why does it now fail on every single checkout i try from the past"
[22:12] <Daemon404> heh
[22:12] <michaelni> iam not 100% sure but possibly that avisynth 2.5.? bug was of that category or at least someone said it worked and now it doesnt
[22:13] <Daemon404> it's probably another biproduct of support for the peice of crap avxsynth
[22:13] <Daemon404> <_<
[22:13] <durandal_1707> michaelni: its not splitting. it needs little endian bitreader and one already used in adpcm.c is bigendian
[22:14] <michaelni> durandal_1707, ah, thats a mess, i guess if puting it in a seperate files cleanly solves it then that should be fine
[22:17] <durandal_1707> and why redoing merges is that hard?
[22:19] <durandal_1707> alternatively i could move swf adpcm to separate file
[22:23] <michaelni> i have no real preferrace on the adpcm issue, its hard to guess ahead of time which would be better
[22:29] <durandal_1707> but why one can't use different bitreaders at same time?
[22:30] <durandal_1707> can i push something now?
[22:31] <durandal_1707> michaelni: i would like to avoid adding new files with copyright spam
[22:33] <michaelni> sure you can push if you have some commit that you want to push
[22:34] <michaelni> fixing the bitreader so 2 can be used at the same time is of course also an option
[22:37] <durandal_1707> how?
[22:38] <durandal_1707> adding get_bits_le/be ?
[22:38] <durandal_1707> or get_be_bits/get_le_bits ?
[22:42] <michaelni> yes, something like that
[22:47] <cone-568> ffmpeg.git 03Paul B Mahol 07master:a64248d1ec15: lavu/opt: show default values
[23:06] <mateo`> michaelni: i'm not familiar with chapters at all. Is there an option to create them ?
[23:06] <durandal_1707> yes, .ffmeta
[23:07] <durandal_1707> with a text editor
[23:07] <Daemon404> do we not support ogg-style chapter text files?
[23:08] <Daemon404> thats what basically every other app on earth uses
[23:08] <durandal_1707> i know nothing about ogg
[23:08] <mateo`> durandal_1707: thanks
[23:09] <Daemon404> http://wiki.xiph.org/Chapter_Extension#Examples
[23:09] <Daemon404> it's exactly what the example shows
[23:09] <Daemon404> super simple
[23:09] <Daemon404> (and why its popular)
[23:09] <durandal_1707> i think there is open bug report for that
[23:19] <cone-568> ffmpeg.git 03Jean Delvare 07master:99f1d7493349: lavfi/delogo: take SAR into account
[23:19] <cone-568> ffmpeg.git 03Jean Delvare 07master:f0bcb13aedcf: lavfi/delogo: avoid propagation of rounding errors in chroma planes
[00:00] --- Thu Jul  4 2013