[Ffmpeg-devel-irc] ffmpeg-devel.log.20130204

burek burek021 at gmail.com
Tue Feb 5 02:05:03 CET 2013


[00:02] <burek> michaelni, these powerpcs also need a different timeout value?
[00:04] <michaelni> burek, probably
[00:10] <burek> judging by this history log, it seems powerpcs are also stuck or something: http://fate.ffmpeg.org/history.cgi?slot=powerpc-linux-gnu-gcc-4.3.5
[00:11] <burek> they used to provied results every couple of hours
[00:11] <burek> provide*
[01:30] <michaelni> burek, maybe, i have no access to them so i cant tell ...
[04:19] <cone-298> ffmpeg.git 03Michael Niedermayer 07master:91debec976d2: swr/resample: fix filter rounding and cliping for s32
[04:19] <cone-298> ffmpeg.git 03Michael Niedermayer 07master:b8c55590d54d: swr/resample: fix integer overflow, add missing cast
[04:19] <cone-298> ffmpeg.git 03Michael Niedermayer 07master:24244cec2fdd: swr-test: Fix clip to 32bit
[04:19] <cone-298> ffmpeg.git 03Michael Niedermayer 07master:be989f923678: swr: reorder init code to make rematrix status available earlier
[04:19] <cone-298> ffmpeg.git 03Michael Niedermayer 07master:b5f544a0f998: swr: Dont use floats for S32->S32 when possible
[11:11] <elk> I'm trying to get ffmpeg to read from a file which is growing (ongoing recording) but to not exit when it reaches the files EOF, I have tried retuning AVERROR(EAGAIN) instead of a AVERROR_EOF when it reaches the EOF.. 
[11:12] <elk> Because I'm reading from a mov-file i looked at mov.c in libavformat but I can find where the actual file is read
[11:12] <elk> I was thinking of peeking into stream and wait if the next data is a eof
[11:58] <ubitux> JEEB: arg @captioning :)
[12:02] <mateo`> ubitux: \o/
[12:02] <ubitux> :)
[12:16] <JEEB> ubitux, ARIB captions are quite fabulous, are they not 8)
[13:04] <cone-831> ffmpeg.git 03Anton Khirnov 07master:d9df93efbf59: mpegvideo: simplify REBASE_PICTURE
[13:04] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:a2bff49e26a9: Merge remote-tracking branch 'qatar/master'
[13:31] <cone-831> ffmpeg.git 03Michaël Cinquin 07master:60f06f986f81: libopenjpegenc: make dci compliant j2c
[13:31] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:c67ddb56da55: Merge remote-tracking branch 'mjbshaw/master'
[13:32] <ubitux> JEEB: well, it's not surprising to see such thing
[13:38] <JEEB> it's been around for quite a while but I finally found a kind of open source renderer for it
[13:39] <JEEB> caption2ass has been around for a while, but the conversion is shoddy at best
[14:37] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:63c61f0be036: vc1dec: Fix 2 "warning: may be used uninitialized in this function"
[14:37] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:353bd158f5d4: sws: Fix "warning: comparison of distinct pointer types lacks a cast"
[14:41] <cone-831> ffmpeg.git 03Rudolf Polzer 07master:81f2549ec9ab: lavfi/showspectrum: display multiple channels in separate row
[15:01] <durandal_1707> michaelni: what patch is supposed to fix #2227
[15:05] <ubitux> divVerent: you should add yourself in the showspectrum filter copyright
[15:26] <ubitux> showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt  i e this
[15:26] <ubitux> thanks divVerent :)
[15:27] <ubitux> we should find a friendly way of making the playback fluid&
[15:49] <michaelni> durandal_1707, [FFmpeg-devel] [PATCH 2/2] ffmpeg: Avoid matching attachment pics to video on wildcard maps.
[15:57] <durandal_1707> what is policy about pathes that quatar folks just forget to apply?
[15:57] <durandal_1707> *patches
[17:07] <michaelni> durandal_1707, i dont think theres a policy about that, but which patch ?
[17:43] <burek> can anyone please help me to properly answer this question about security vulnerabilities in ffmpeg 1.0.2: http://ffmpeg.gusari.org/viewtopic.php?f=16&t=818
[17:47] <burek> also, if someone knows what is a NEON simd, I would appreciate any help on that too: http://ffmpeg.gusari.org/viewtopic.php?f=8&t=819
[17:48] <JEEB> NEON simd is just simd for armv7 cpus
[17:48] <iive> burek: simd - single instruction multiple data. NEON is used on arm cpu's
[17:48] <JEEB> he is expecting some kind of hw H.264 encoding/decoding or whatever, and the reality hits him :P
[17:48] <JEEB> yay armv7
[17:48] <burek> i see, how can someone check if ffmpeg is using simd or main cpu for encoding
[17:49] <burek> that's assentially what the guy is asking
[17:49] <JEEB> SIMD is main cpu
[17:49] <burek> oh
[17:49] <JEEB> it's just instructions that deal with "multiple data" at the same time (like SSE/MMX and their derivatives)
[17:50] <burek> yes, i understand, i guess the guy thought it was some kind of additional gpu or something
[17:50] <JEEB> yeah
[17:50] <iive> MMX and SSE are SIMD
[17:50] <JEEB> yup
[17:50] <JEEB> for IA32 and x86_64
[17:50] <burek> ok, at least now i know what to answer there :) thank you both :)
[17:52] <wm4> burek: didn't we have this question here on this channel recently
[17:52] <JEEB> yeah
[17:52] <JEEB> I remember a similar guy coming around here
[17:53] <burek> probably, i didnt follow the channel in detail, but this one was from the forum, so it might be the same guy perhaps
[18:04] <wm4> which mp3 decoder is actually preferred, "mp3" or "mp3float"?
[18:08] <wm4> I would have thought it's mp3float, but both ffplay and ffmpeg seem to select "mp3" by default
[18:09] <wm4> avcodec_find_decoder(AV_CODEC_ID_MP3) returns "mp3" too
[18:12] <burek> if anyone knows this thing about security vuln. in ffmpeg 1.0.2 ( http://ffmpeg.gusari.org/viewtopic.php?f=16&t=818 ), just ping me and I'll see the message later :) and thanks in advance :)
[18:23] <burek> one more thing :) is there any relation between "FFserver version SVN-r25767" and some git version currently hosted on ffmpeg.org ?
[18:24] <burek> i.e. is there a way to find out what is the git commit hash of that svn version?
[18:26] <wm4> burek: it was 47d2ddca802f4c1bc4b454c5ac40f06f79b740a0
[18:27] <wm4> the svn commits imported in git always have this in their log message: "Originally committed as revision 25767 to svn://svn.ffmpeg.org/ffmpeg/trunk"
[18:27] <wm4> so you can just do git log and search by text
[18:27] <burek> oh great
[18:27] <wm4> and that version is from 2010 (lol)
[18:27] <burek> thanks
[20:31] <michaelni> burek, ffmpeg 1.0.2 security fixes are listed at ffmpeg.org/security.html 
[21:13] <llogan> michaelni: i'd like to remove ffserver-user today: rmlist ffserver-user
[21:13] <llogan> that should keep the archives
[21:13] <llogan> actually i'll send out one more notice beforehand...
[21:15] <michaelni> llogan, assuming theres consensus on this, is there anything i will need to do ?
[21:16] <Compn> move all ffserver-user subscribers to ffmpeg-user , but turn off messages, so they are allowed to post, but dont recieve ffmpeg-user mails
[21:16] <Compn> ?
[21:17] <llogan> michaelni: i asked on -devel if there were no objections. only saste replied saying he supported this. i made an [announce] on ffserver-user describing the situation on jan 27.
[21:18] <llogan> Compn: i am not migrating users due to the volume of messages in ffmpeg-user. they will have to do that themselves as mentioned in the announce.
[21:19] <michaelni> so i dont have to do anything ? :)
[21:19] <llogan> michaelni: you will have to do the actual list removal with "rmlist".
[21:19] <michaelni> work :(
[21:19] <llogan> i know...i would do it if i could.
[21:20] <llogan> you should check first with "rmlist --help"
[21:20] <llogan> in case the defaults/syntax differs from what i refered to
[21:21] <llogan> i also already updated contact
[21:22] <michaelni> isnt it better to leave the list and forward mails to ffmpeg-user ?
[21:23] <michaelni> that is if someone from somewhere uses the ML address to post a mail
[21:23] <michaelni> or reply to an old mail
[21:23] <llogan> if you prefer that method you can do it that way
[21:23] <michaelni> you can do that yourself i think :)
[21:23] <llogan> i didn't see an option to set that up in the admin panel, but maybe i missed it
[21:24] <michaelni> remove all subscribers, put the ffmpeg-user list as only subscriber to ffserver-user then it would get mails sent to it
[21:27] <llogan> there did not seem to be a trivial method to disable subscribers.
[21:28] <llogan> and considering the low-volume of ffserver i thought rmlist was the best tradeoff of work vs potential user convienence
[21:28] <llogan> *disable subscribing
[21:29] <michaelni> well i can rmlist it if thats what the ffserver-user people want
[21:35] <llogan> alternatively, you can setup an alias in postfix to forward ffserver-user mail to ffmpeg-user.
[21:36] <llogan> but truthfully i don't see much "reply to old mail" happening.
[23:05] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:96a08d862730: wmavoice: silence may be used uninitialized warnings
[23:05] <cone-831> ffmpeg.git 03Michael Niedermayer 07master:fc0d069feb69: flacdec: silence several "warning: X may be used uninitialized in this function"
[00:00] --- Tue Feb  5 2013


More information about the Ffmpeg-devel-irc mailing list