[Ffmpeg-devel-irc] ffmpeg-devel.log.20111215

burek burek021 at gmail.com
Fri Dec 16 02:05:03 CET 2011


[00:02] <cbsrobot> lol - in mtv.c: /* Calculate width and height if missing from header */
[00:03] <cbsrobot> should have read /* SIGFPE if width and height missing from header */
[00:19] <CIA-49> ffmpeg: 03Diego Biurrun 07master * r0d9bd2d584 10ffmpeg/configure: 
[00:19] <CIA-49> ffmpeg: configure: do not manually enable protocols
[00:19] <CIA-49> ffmpeg: Protocols are handled just like all other components groups such as
[00:19] <CIA-49> ffmpeg: muxers and are automatically enabled.
[00:19] <CIA-49> ffmpeg: 03Diego Biurrun 07master * r144904e901 10ffmpeg/ (Makefile subdir.mak): build: rename subdir.mak ---> library.mak
[00:19] <CIA-49> ffmpeg: 03Derek Buitenhuis 07master * r86ae512453 10ffmpeg/doc/developer.texi: 
[00:19] <CIA-49> ffmpeg: doc: make developer.texi point to correct fate doc
[00:19] <CIA-49> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[00:19] <CIA-49> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[00:19] <CIA-49> ffmpeg: 03Derek Buitenhuis 07master * r80f6c0bb95 10ffmpeg/ (10 files in 4 dirs): 
[00:19] <CIA-49> ffmpeg: v410 encoder and decoder
[00:19] <CIA-49> ffmpeg: v410 is a packed 10-bit 4:4:4 YCbCr format used in
[00:19] <CIA-49> ffmpeg: QuickTime.
[00:19] <CIA-49> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[00:19] <CIA-49> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[00:19] <CIA-49> ffmpeg: 03Diego Biurrun 07master * r88b9735753 10ffmpeg/ (3 files in 2 dirs): build: conditionally compile x86 H.264 chroma optimizations
[00:20] <CIA-49> ffmpeg: 03Diego Biurrun 07master * r52de07e1f1 10ffmpeg/libswscale/x86/ (scale.asm swscale_mmx.c): swscale: Mark yuv2planeX_8_mmx as MMX2; it contains MMX2 instructions.
[00:20] <CIA-49> ffmpeg: 03Diego Biurrun 07master * r3c62a71486 10ffmpeg/libswscale/x86/swscale_mmx.c: swscale_mmx: drop no longer required parameters from VSCALEX macros
[00:20] <CIA-49> ffmpeg: 03Andrey Utkin 07master * r4f820131fa 10ffmpeg/libavcodec/ (h261dec.c h263dec.c mpegvideo.c mpegvideo_enc.c vc1dec.c): 
[00:20] <CIA-49> ffmpeg: mpegvideo: remove abort() in ff_find_unused_picture()
[00:20] <CIA-49> (32 lines omitted)
[00:34] <Diana_Muscalu> michaelni: i have a question, when using zzuf the ffmpeg_g crashes sending a signal 11, but when using valgrind it does not signal any leaks in memory. is that possible ?
[00:37] <michaelni> sure
[00:37] <michaelni> a memory leak exists when memory is allocated and not freed
[00:38] <michaelni> a sig11 can happen when an access is made outside the allocated memory, stack, ...
[00:38] <Diana_Muscalu> michaelni: and if i submit a file which in zzuf gives signal 11 and in valgrind no problems, is a valid bug?
[00:38] <michaelni> if i can reproduce the crash its valid
[00:39] <Diana_Muscalu> michaelni: i generate the corrupt file but when i test it with ffmpeg_g it gives invalid data, and no segmentation, but if i put it in zzuf it gives sig11
[00:40] <Diana_Muscalu> michaelni: am i missing something here?
[00:40] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r8b9b6332df 10ffmpeg/libavformat/mtv.c: 
[00:40] <CIA-49> ffmpeg: mtv: Fix FPE with 0 dimensions.
[00:40] <CIA-49> ffmpeg: Fixes Ticket755
[00:40] <CIA-49> ffmpeg: Bug Found by: Diana Elena Muscalu
[00:40] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:41] <michaelni> Diana_Muscalu, id like to see the file
[00:42] <Diana_Muscalu> michaelni: where should i put it?
[00:42] <michaelni> our bug tracker for example
[00:42] <michaelni> but it doesnt matter where, elsewhere is fine too
[00:42] <Diana_Muscalu> michaelni: okei 1 min and you will have the original file and the corrupt one and the command to generate it
[00:59] <Diana_Muscalu> michaelni: http://ptp-roman.org/ffmpeg/  are the 2 files, original is cubes.mov and the corrupt one. to generate the sig11 use zzuf -c -s1:2000 -r0.001:1 ./ffmpeg_g -i cubes.mov -f null -
[01:00] <Diana_Muscalu> michaelni: maybe my linux is messed up, or dunno
[01:17] <michaelni> Diana_Muscalu, report is ok i can reproduce the crash
[01:18] <Diana_Muscalu> michaelni: i have deleted the files, am reporting another bug now. with valgrind errors too, all the required files 
[01:26] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7172289 please review if correct. 10x
[01:30] <michaelni> Diana_Muscalu, good work!
[01:32] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7185247 accept the claim, 10x
[01:33] <michaelni> Diana_Muscalu, done :)
[01:34] <Diana_Muscalu> michaelni: 10x, a question why when i look at the leaderboard i see that i have only 1 point and 1 task. don`t i have 4 tasks and 4 points because my task are only these bugs findings. and when i click "show my tasks" i see al 4 of them. is there a problem at GCI?
[01:36] <michaelni> maybe, i dont know
[01:37] <michaelni> maybe theres some delay 
[01:37] <Diana_Muscalu> michaelni: thx anyway, how long will you be here? how many hours from now on?
[01:39] <michaelni> 4h or so probably then ill need to go to bed
[01:40] <Diana_Muscalu> michaelni: ok, i think i`ll submit another bug and i`ll go to sleep too. what timezone are you?
[01:40] <michaelni> something middle/central europe
[01:40] <michaelni> i think :)
[01:41] <Diana_Muscalu> i`m gmt+2
[01:42] <michaelni> seems iam 1 hour ahead of gmt
[01:43] <iive> Diana_Muscalu: just a wild guess, are you from Greece?
[01:44] <Diana_Muscalu> Romania 
[01:44] Action: iive fail
[01:49] <Diana_Muscalu> michaelni: i see there are some tasks to close ffmpeg bugs, if i take one of them will i get some support from you?
[01:50] <Compn> yes michaelni is head devel of ffmpeg :)
[01:50] <Compn> he will help closing ffmpeg bugs 
[01:51] <michaelni> most of the bugs we have open are pretty hard though ...
[01:51] <Diana_Muscalu> i have some experience in c++ but not at high level
[01:52] <pasteeater> documentation improvement is an always open "bug"
[01:52] <michaelni> pasteeater, true
[01:52] <Compn> ffmpeg does not have c++ really :)
[01:52] <Compn> mostly C and asm
[01:52] <Diana_Muscalu> so i won`t have any chance 
[01:53] <Compn> yes, i think its pretty hard to fix ffmpeg bugs if you dont have good coding skills
[01:54] <michaelni> the problem is, all the easy bugs have been fixed already
[01:54] <pasteeater> i think my first patch was s/gop/g or something like that
[01:54] <pasteeater> so you don't always need to know c
[01:55] <Diana_Muscalu> i have a question, me and my father we are building a robot rover which has on IP camera on it, an Arecont camera, and i want to know how can i get video from it. the encoding is MJPEG and i want to get some video from it and store it on some media. can i acomplish this using ffmpeg?
[01:56] <Compn> if its supported by v4l, yes
[01:56] <Compn> or if it has some http/rtsp/rtmp stream, yes
[01:57] <Compn> ip camera sounds like it has a stream :)
[01:58] <ubitux> michaelni: we expected an easy one at work today, if you are willing to fix one simple
[01:58] <ubitux> i didn't took the time to have a look or report it yet
[01:59] <ubitux> basically trying to muxing only audio in mxf made it crash
[01:59] <ubitux> mux*
[01:59] <ubitux> we expected it with a ffmpeg -i foo.mxf -map 0:a bar.mxf (and not -map 0 because it tried to map the data stream and failed to it)
[02:00] <ubitux> we got a floating exception in a av_round_* iirc, because the timebase wasn't defined (no video)
[02:00] <Diana_Muscalu> Compn: http://www.arecontvision.com/uploads/product_images/16Arecont_Vision_AV3130_1209pdf.pdf this is the camera which i have . it says there that it has TFTP and HTTP image transmission protocols
[02:02] <ubitux> well then maybe ffmpeg -i http://... 
[02:03] <Diana_Muscalu> i want to run the ffmpeg on a computer with 256 ram and 1ghz CPU, a Vortex CPU. the camera i at 3MP, will it be able to run?
[02:04] <ubitux> it should, depending on what you need
[02:04] <Diana_Muscalu> my goal is to create an output file like a movie or something
[02:04] <ubitux> doing h264 realtime encoding might be difficule for instance :p
[02:05] <ubitux> difficult*
[02:05] <ubitux> yes sure, it may just take a while to encode :)
[02:05] <ubitux> ffmpeg runs on ARM and pretty small machines
[02:07] <ubitux> anyway, got to go, good night ppl in the internet box
[02:08] <Diana_Muscalu> good night
[02:41] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7185247 marked as completed, awaiting review
[02:45] <Diana_Muscalu> michaelni: is it good?
[02:47] <michaelni> yes, good i can reproduce the crash
[02:48] <michaelni> also if the number of counted tasks google shows doesnt reach the amount it should in a day or 2, tell me and ill see if theres someone at google i can report that to
[02:49] <Diana_Muscalu> michaelni: okei, i will do so. meanwhile please accept http://google-melange.appspot.com/gci/task/view/google/gci2011/7196353
[02:50] <michaelni> done
[02:52] <Diana_Muscalu> michaelni: thx, i`m going to sleep. see you 2morrow. good night
[02:53] <michaelni> night
[03:02] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rcd645c15d8 10ffmpeg/libavcodec/indeo3.c: 
[03:02] <CIA-49> ffmpeg: indeo3: check return values of allocate_frame_buffers()
[03:02] <CIA-49> ffmpeg: Bug Found by: Diana Elena Muscalu
[03:02] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:02] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r7fff64e00d 10ffmpeg/libavcodec/h264_ps.c: 
[03:02] <CIA-49> ffmpeg: h264: check chroma_format_idc range.
[03:02] <CIA-49> ffmpeg: Fixes Ticket758
[03:02] <CIA-49> ffmpeg: Bug found by: Diana Elena Muscalu
[03:02] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:02] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7196353 marked as complete, awaiting review
[03:08] <michaelni> Daemon404, Ticket760 and 761 both crash in libavcodec/aacsbr.c:629 so these seem to point to the same bug
[03:08] <michaelni> s/ Daemon404 /Diana_Muscalu / 
[03:08] <michaelni> sorry
[03:09] <Diana_Muscalu> michaelni: need to submit another one?
[03:10] <michaelni> valgrind also points to read_sbr_grid.isra.17 (aacsbr.c:629) for both
[03:11] <michaelni> Diana_Muscalu, yes, i think its the same bug, we need seperate ones
[03:11] <Diana_Muscalu> ok
[03:11] <Diana_Muscalu> plesase return the task as needs more work
[03:11] <michaelni> Diana_Muscalu, but the files you deleted would have been valid ...
[03:11] <michaelni> for the indeo3 bug
[03:12] <Diana_Muscalu> i will make new ones
[03:12] <michaelni> done, thanks, and sorry
[03:12] <Diana_Muscalu> befone i got these tasks i researched many bugs, have a list of 40 files wich create a bug, now i must see which of them make different bugs 
[03:27] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7196353 uploaded corruptfile2 and valgrind2.txt new files. check if you can reproduce the bug.
[03:33] <michaelni> reproduced, thanks
[03:35] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7197332 please accept, and that`s all for the night. see ya
[03:36] <michaelni> Diana_Muscalu, done, and good night :)
[03:56] <Daemon404> michaelni, can you explain to me how the .comp member of AVPixFmtDescriptor works
[03:56] <Daemon404> it isn't documented
[03:56] <Daemon404> oh wait, i forgot to scroll down
[03:56] <Daemon404> ignore me.
[04:49] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r30b996d443 10ffmpeg/libavcodec/flicvideo.c: 
[04:49] <CIA-49> ffmpeg: flicvideo: input buffer pointer checks.
[04:49] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:49] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rafb2bac48d 10ffmpeg/libavcodec/flicvideo.c: 
[04:49] <CIA-49> ffmpeg: flicvideo: fix overread.
[04:49] <CIA-49> ffmpeg: Bug Found by: Diana Elena Muscalu
[04:49] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:49] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r944f5b2779 10ffmpeg/libavcodec/aacsbr.c: 
[04:49] <CIA-49> ffmpeg: aacsbr: Fix memory corruption.
[04:49] <CIA-49> ffmpeg: Fixes Ticket760 and Ticket761
[04:49] <CIA-49> ffmpeg: Bug Found by: Diana Elena Muscalu
[04:49] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:10] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r21270cffae 10ffmpeg/libavcodec/mpegvideo.c: 
[05:10] <CIA-49> ffmpeg: h263dec: Fix regression / crash with lowres.
[05:10] <CIA-49> ffmpeg: Fixes Ticket757
[05:10] <CIA-49> ffmpeg: Bug Found by: ami_stuff
[05:10] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:41] <Compn> hyc laying the smackdown on rtmpdump mailing list :D
[06:32] <hyc> just sick of the scriptkiddies thinking they have any valid basis for discussing development
[06:34] <hyc> the mailing list's usefulness has kinda dissolved. any actual development work is done in anonymous/private email anyway
[06:35] <hyc> and I still don't see how rtmpsuck is any use for a big video server
[06:35] <Daemon404> sure is rage
[06:35] Action: Daemon404 is tempted to read
[06:36] <hyc> eh. mostly meaningless junk
[06:37] <Daemon404> seems silly to do dev work in private email tbh
[06:38] <hyc> unfortunately, the peanut gallery on the list would just generate too much noise
[06:40] <Daemon404> make it a moderated list?
[11:43] <Tjoppen> whoever thought of the way essence maps to tracks in mxf ought to be shot, or at least be given a stern talking to
[11:45] <Tjoppen> it seems it's impossible to figure out where the essence is located (MOV style) without looking at it
[12:04] <Tjoppen> ok, I think I've got it
[12:08] <Tjoppen> seems like a hybrid-ish solution is in order
[12:09] <Tjoppen> we might want to only use the new demuxing code for opatom in 0.9 for now
[14:16] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rd998a6ddee 10ffmpeg/libavcodec/h264.c: 
[14:16] <CIA-49> ffmpeg: h264: Fix build with --disable-everything --enable-decoder='h264' --disable-optimizations
[14:16] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:40] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r484b1cdd53 10ffmpeg/libavcodec/mjpegdec.c: 
[15:40] <CIA-49> ffmpeg: jpegdec: check return value of mjpeg_decode_dc()
[15:40] <CIA-49> ffmpeg: Fixes Ticket754
[15:40] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:31] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * ra4755df523 10ffmpeg/libavcodec/h264.c: 
[17:31] <CIA-49> ffmpeg: h264: fix midstream extardata handling with dump.f4v
[17:31] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:31] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r62b5197f97 10ffmpeg/libavcodec/arm/asm.S: 
[17:31] <CIA-49> ffmpeg: arm: add back ldr_dpren macro that was lost.
[17:31] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:06] <Tjoppen> seeks past EOF should fail unless AVSEEK_FLAG_BACKWARD, right?
[18:16] <Tjoppen> hah, of course. every demuxer treats this differently
[18:32] <Compn> http://americancensorship.org/?ak_proof=1
[18:32] <Compn> today its going for a vote...
[18:33] <michaelni> Tjoppen, yep, maybe there should be a flag to switch this behavior
[18:37] <iive> Compn: i've heard it is only is some commission, not in the house/senate.
[18:38] <Compn> iive : well, its good idea to call now, calling later means not much time to change opinion
[18:38] <Compn> if it fails one of these comissions, thats a good blow
[18:38] <iive> of course.
[18:39] <Tjoppen> michaelni: yeah, some demuxers are OK with seeking past the end or before the start, but others aren't
[18:41] <Tjoppen>  ret: 0         st:-1 flags:1  ts: 1.894167
[18:41] <Tjoppen> -ret:-1
[18:41] <Tjoppen> +ret: 0         st: 0 flags:1 dts: 0.960000 pts: 0.960000 pos:5117952 size:150000
[18:41] <Tjoppen> this is OK, right?
[18:41] <Tjoppen> mxf_d10 in this case
[18:42] <Tjoppen> if I understand correctly it used to be OK with seeking to 1.894167 but the subsequent read would fail. now it gives the last packet isntead
[18:43] <Tjoppen> if so then I have a solution the fixes #746, supports proper seeking and opatom
[18:44] <Tjoppen> this also reuses most of the old demuxing code, now that I realize how it's actually supposed to work
[18:46] <CIA-49> ffmpeg: 03Nicolas George 07master * r2c82ab946b 10ffmpeg/ (7 files in 3 dirs): lavf: add SBG demuxer.
[18:46] <CIA-49> ffmpeg: 03Nicolas George 07master * rb33fd66f46 10ffmpeg/ (6 files in 2 dirs): lavc: add ffwavesynth pseudo-codec.
[18:49] <Tjoppen> woop, make fate passes. time to head home. will clean up and post tomorrow
[19:04] <Diana_Muscalu> michaelni: can a signal9(memory exceeded) be reported as a bug?
[19:06] <michaelni> Diana_Muscalu, yes, unless you are testing on a system with very little free memory
[19:07] <Diana_Muscalu> michaelni: testing on a virtuale machine ubuntu with 4gb ram and 8 gb hdd free space
[19:07] <av500> ubuntu even boots on that? :)
[19:08] <Diana_Muscalu> av500: needed fast for zzufing the files and testing them 
[19:08] <Diana_Muscalu> michaelni: can i report it?
[19:08] <michaelni> yes, 4gb is alot
[19:08] <Diana_Muscalu> michaelni: okei
[19:09] <av500> michaelni: imagine, all in one machine :)
[19:12] <Diana_Muscalu> my system is a macbookpro with 8gb ram, thats why
[19:15] <michaelni> Diana_Muscalu, i need to go to supermarket, ill review your task when iam back (40min or so)
[19:15] <Diana_Muscalu> michaelni: okei, perfect
[19:15] <Diana_Muscalu> michaelni: pm here when you are back
[19:25] <CIA-49> ffmpeg: 03Carl Eugen Hoyos 07master * r6f2054c589 10ffmpeg/libavcodec/bmp.c: 
[19:25] <CIA-49> ffmpeg: Fix linesize for bmp with bitdepth < 8.
[19:25] <CIA-49> ffmpeg: Fixes ticket #751.
[19:49] <ubitux> hah!
[19:49] <ubitux> i finally get why all the fate tests where not run locally
[19:49] <ubitux> make fate without SAMPLES defined runs anyway
[19:49] <ubitux> but for a subset of tests, those who don't need samples
[19:54] <michaelni> Diana_Muscalu, iam back
[19:54] <pasteeater> what did you buy for me?
[19:54] <av500> mozart balls
[19:55] <Diana_Muscalu> michaelni: i have marked the task as completed
[19:55] <pasteeater> and damn, michaelni, your estimate of 40 minutes (or so) was exact
[19:56] <pasteeater> av500: mmmmm
[19:56] <Diana_Muscalu> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7197332 here is it
[20:00] <Diana_Muscalu> michaelni: my points and number of tasks are correct now, is was a timing issue i think
[20:02] Action: michaelni gives pasteeater some virtual mozart balls
[20:02] <ubitux> oo
[20:08] <michaelni> Diana_Muscalu, task ok & closed, sorry took me a little to reproduce that 1gb alloc
[20:08] <Diana_Muscalu> michaelni: it`s nothing, happy that the tasks is good 
[20:10] <Diana_Muscalu> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7172287 please accept claim
[20:11] <michaelni> done
[20:11] <Diana_Muscalu> michaelni: 10x
[20:30] <Diana_Muscalu> michaelni: i am zzufing a file and it`s stopped at a phase for like 20 minutes in the ffmpeg_g program. can it be reported as i bug? or only if it gives memory exceeded?
[20:31] <michaelni> infinite loops are valid bugs
[20:32] <Diana_Muscalu> michaelni: i must send the file and the command to generate the loop no?
[20:32] <michaelni> yes, i need to be able to reproduce it
[20:33] <Diana_Muscalu> michaelni: okei, post it on trac?
[20:33] <michaelni> yes
[20:38] <Diana_Muscalu> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7172287
[20:38] <Diana_Muscalu> michaelni: check if it is good
[20:44] <michaelni> good
[20:44] <Diana_Muscalu> michaelni: 10x
[20:50] <Diana_Muscalu> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7185248 , please accept
[20:52] <michaelni> done
[20:53] <Diana_Muscalu> michaelni: question, in the timetable it said that on the 16th new tasks will be given by the organisations, what kind of tasks especially ? more like these that i am doing?
[20:55] <michaelni> organizations can add tasks any time it seems but google hides them from students until the 16th AFAIK
[20:55] <michaelni> ive added a few more bugfix tasks yesterday and 4 draw a logo with the theme of spring/winter/... longer ago
[20:56] <michaelni> s/bugfix/bugfind/ :)
[21:00] <Diana_Muscalu> michaelni: code cleaning tasks like deleting tabs and stuff like that have you added?
[21:07] <michaelni> there are no tabs in our code :)
[21:44] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r4255d6d96c 10ffmpeg/ (libavformat/westwood.c tests/ref/fate/vqa-cc): 
[21:44] <CIA-49> ffmpeg: westwooddemuxer: use av_get_packet()
[21:44] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:44] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rd19ca77f42 10ffmpeg/libavformat/utils.c: 
[21:44] <CIA-49> ffmpeg: lavf: Fix fps misdetection when codec_info_duration is 0 and only a few frames are used to estimate fps.
[21:44] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:44] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r7f9369b743 10ffmpeg/libavformat/utils.c: 
[21:44] <CIA-49> ffmpeg: lavf: update rfps debug code.
[21:44] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:44] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r5f95c130a0 10ffmpeg/libavformat/mov.c: 
[21:44] <CIA-49> ffmpeg: mov: Fix near infinite loop in stsd parsing.
[21:44] <CIA-49> ffmpeg: Fixes Ticket768
[21:44] <CIA-49> ffmpeg: Bug found by: Diana Elena Muscalu
[21:44] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:45] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rf68b19fc28 10ffmpeg/libavformat/westwood.c: 
[21:45] <CIA-49> ffmpeg: westwooddemuxer: Fix 1gb alloc
[21:45] <CIA-49> ffmpeg: Fixes Ticket765
[21:45] <CIA-49> ffmpeg: Bug Found by: Diana Elena Muscalu
[21:45] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:08] <michaelni> btw, Diana_Muscalu if you have some ideas for tasks that you want to work on, i can add them, they must be usefull for us of course and they must require only little of my time ...
[22:10] <michaelni> note, though i dont know when the deadline is for adding tasks
[22:10] <Diana_Muscalu> michaelni: for the moment i do not know, bugfixing is hard for me. finding bugs is very good for you as i see, you have closed some of them.
[22:11] <michaelni> yes :)
[22:12] <Diana_Muscalu> michaelni: as i see many of the ffmpeg bugs are the same with the libav bugs, what components they have in common?
[22:14] <michaelni> libav is a fork of ffmpeg, so its more or less the same code.
[22:14] <michaelni> we have quite a few more features and i think fewer bugs
[22:14] <Diana_Muscalu> michaelni: yes, you are right about the few bugs, there are some that have not applied at ffmpeg
[22:19] <ubitux> michaelni: from what i understand the fate tests stderr outputs (and others) are kept when the test fails
[22:19] <ubitux> test $err = 0 && rm -f $outfile $errfile $cmpfile $cleanfiles
[22:20] <ubitux> also,
[22:20] <ubitux> just above we have: echo "${test}:${sig:-$err}:$($base64 <$cmpfile):$($base64 <$errfile)" >$repfile
[22:20] <ubitux> so this means these output are reported
[22:21] <ubitux> and so, the main fate interface should be able to show them
[22:23] <ubitux> maybe the main fate box will need some updates from upstream
[22:25] <michaelni> ubitux, stderr can be viewed on the server but when you run make fate locally a failure is quite uninformative
[22:25] <ubitux> the .err should be kept in the data dir for the failing test
[22:27] <ubitux> it seems the tests/data/fate/fate-<test>.err is indeed kept
[22:27] <michaelni> ubitux, true, but it seems you are the first to find it 
[22:28] <ubitux> it's lacking documentations i agree :p
[22:28] <michaelni> yes but not only that
[22:28] <ubitux> i'm just sharing my discoveries :p
[22:28] <michaelni> it should just print the file contents if it fails IMHO
[22:29] <michaelni> the .err contents
[22:29] <ubitux> that might be easily done, but we need to be careful because of the dup
[22:30] <ubitux> if the stdout file get the stderr or similar, things might get ugly
[22:30] <michaelni> or at least print a "the stderr is in file X"
[22:30] <ubitux> yep good idea
[22:30] <ubitux> anyway
[22:30] <ubitux> the "valgrind box" i have should raise the necessary information
[22:32] <michaelni> is it already setup to send data to the server ?
[22:39] <ubitux> it should, but the behaviour is weird after the first run
[22:39] <Diana_Muscalu> michaelni: i have a file that is blocked for 5 minutes in a loop, can i file it ?
[22:39] <ubitux> but well the report file is available and i should be able to send it 
[22:40] <ubitux> the script rsync & run fate in a loop with a sleep
[22:40] <ubitux> maybe i'll set up a fake fate server to re-check 
[22:40] <durandal_1707> Diana_Muscalu: yes
[22:40] <michaelni> Diana_Muscalu, yes
[22:41] <michaelni> durandal_1707, do you want to mentor GCI tasks for ffmpeg ? :)
[22:42] <ubitux> :D
[22:42] <durandal_1707> michaelni: i'm busy with writing decoders and encoders, want to improve my c skills
[22:42] <michaelni> thats very good too :)
[22:42] Action: Daemon404 thinks durandal_1707 got scared of by lioneesh or w/e
[22:42] <Daemon404> off*
[22:52] <pasteeater> michaelni: i can help mentor the logo designs
[22:57] <michaelni> pasteeater, official ? if yes then i think the page to join was: http://www.google-melange.com/gci/profile/mentor/google/gci2011
[22:58] <michaelni> then you need to request to become mentor for videolan and send a mail to jb so he approves it
[23:02] <CIA-49> ffmpeg: 03Tom Evans 07master * rffa1f541b3 10ffmpeg/ (configure libavdevice/v4l2.c): 
[23:02] <CIA-49> ffmpeg: libavdevice: Fix v4l2 compilation under FreeBSD
[23:02] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:02] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rd0f7927177 10ffmpeg/libavcodec/smacker.c: 
[23:02] <CIA-49> ffmpeg: smacker: Check for overread in smka_decode_frame()
[23:02] <CIA-49> ffmpeg: Fixes a segfault with wetlog_fail.smk
[23:02] <CIA-49> ffmpeg: Bug found by: Shitiz Garg
[23:02] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:10] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7185248 , use the command i specified to reproduce the crash. in valgrind there is no problem. please check
[23:14] <michaelni> Diana_Muscalu, good work, successfully reproduced
[23:15] <Diana_Muscalu> michaelni: have you valgrind the corrupted file?
[23:16] <michaelni> i get a null ptr crash in valgrind too
[23:16] <michaelni> well close to 0, more correct 0x40
[23:16] <Diana_Muscalu> i do no get nothing in valgrind, seems my machine is running wild :)
[23:17] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7183251 please accept
[23:18] <michaelni> done
[23:18] <Diana_Muscalu> 10x
[23:21] <Diana_Muscalu> michaelni: when ffmpeg_g says "operation not permitted" what does it mean?
[23:25] <Diana_Muscalu> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7183251 marked as completed, awaiting review
[23:29] <ubitux> it likely means the module return -1 instead of the appropriate error
[23:29] <ubitux> so well, the error is handled, but not explicit
[23:29] <Diana_Muscalu> ubitux: so it`s not a crash
[23:30] <michaelni> 7183251 is ok (signal 9)
[23:33] <ubitux> Diana_Muscalu: no, not a crash
[23:34] <ubitux> not worth reporting it except if you're sure the error should not happen
[23:45] <CIA-49> ffmpeg: 03Nicolas George 07master * r1509bd3e9e 10ffmpeg/libavdevice/lavfi.c: lavd/lavfi: remove a duplicated line.
[00:00] --- Fri Dec 16 2011


More information about the Ffmpeg-devel-irc mailing list