[Ffmpeg-devel-irc] ffmpeg-devel.log.20120925

[burek]

[00:40] <ubitux> no comment on "[PATCH 1/2] Move subrip/text API change info from Changelog to doc/APIchanges."?
[00:41] <ubitux> i'm a bit uneasy with the bump
[00:41] <ubitux> and anyway, it imo doesn't belong in the changelog
[00:41] <ubitux> philipl ^
[00:42] <philipl> Hmm.
[00:43] <philipl> Did the bump happen already?
[00:43] <ubitux> yeah i had to bump lavf recently
[00:43] <philipl> When I discussed this with michaelni, he wanted to avoid a major bump.
[00:43] <philipl> but yeah, now that there is one, then it's the right place for it.
[00:43] <ubitux> a minor bump is harmless
[00:43] <ubitux> a micro bump was appropriate in that case
[00:44] <philipl> Well, if the goal of the bump is to avoid silently breaking behaviour, only a major bump will work.
[00:44] <philipl> michaelni rationalised it as a bug fix :-)
[00:45] <ubitux> in that case we could as well just drop the entry
[00:48] <Daemon404> \o/ i finally got a good fix for mingw/brpint
[00:49] Action: Daemon404 tests on msvc too and then sends
[01:32] <philipl> ubitux: Well, there's a real behavioural change that clients have to account for.
[01:32] <philipl> It's a "bug" that everyone accounted for.
[01:38] <CIA-15> ffmpeg: 03Alexander Strasser 07master * rf70122dd57 10ffmpeg/ffprobe.c: 
[01:38] <CIA-15> ffmpeg: ffprobe: Simplify CSV field quoting code
[01:38] <CIA-15> ffmpeg: Sync with bac1b31 .
[01:38] <CIA-15> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[01:38] <CIA-15> ffmpeg: 03Alexander Strasser 07master * rdd83028374 10ffmpeg/ffprobe.c: 
[01:38] <CIA-15> ffmpeg: ffprobe: Rename variable in csv_escape_str()
[01:38] <CIA-15> ffmpeg: Sync to 3245be9 .
[01:38] <CIA-15> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[01:38] <CIA-15> ffmpeg: 03Alexander Strasser 07master * rdde8068856 10ffmpeg/ffprobe.c: 
[01:38] <CIA-15> ffmpeg: ffprobe: Consider the chosen separator when doing CSV escaping
[01:38] <CIA-15> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[02:01] <Daemon404> what are these 'sync to <hash>' commits?
[02:41] <michaelni> which formats can contain id3 / apics ?
[02:44] <michaelni> or better said do contain in the wild 
[02:45] <michaelni> we need a list because few demuxers handle it correctly the extra attachment streams crash them
[02:51] <CIA-15> ffmpeg: 03Michael Niedermayer 07master * r6bcbe768bb 10ffmpeg/libavformat/utils.c: 
[02:51] <CIA-15> ffmpeg: lavf: dont add apic streams for demuxers that dont support it.
[02:51] <CIA-15> ffmpeg: Fixes crash
[02:51] <CIA-15> ffmpeg: Fixes ticket1730
[02:51] <CIA-15> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:34] <michaelni> fflogger, thank you i know its out of diskspace
[03:35] <michaelni> it will be fine on the next run again
[03:49] <CIA-15> ffmpeg: 03Michael Niedermayer 07master * rd8676140de 10ffmpeg/libavcodec/ffv1.c: 
[03:49] <CIA-15> ffmpeg: ffv1: inject sentinels for RC->VLC transitions begining with 1.3
[03:49] <CIA-15> ffmpeg: This breaks bitstream compatibility of 1.3 VLC streams
[03:49] <CIA-15> ffmpeg: This fixes very rare errors in the RC stream at the transition point
[03:49] <CIA-15> ffmpeg: to VLC.
[03:49] <CIA-15> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:49] <CIA-15> ffmpeg: 03Michael Niedermayer 07master * rad9371ef24 10ffmpeg/libavcodec/ffv1.c: 
[03:49] <CIA-15> ffmpeg: ffv1enc: fix RC slice sentinel
[03:49] <CIA-15> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:49] <CIA-15> ffmpeg: 03Michael Niedermayer 07master * raea15df3b4 10ffmpeg/libavcodec/ffv1.c: 
[03:49] <CIA-15> ffmpeg: ffv1dec: change bitstream end check to a exact check instead of a +-1 check
[03:49] <CIA-15> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:30] <ubitux> michaelni: i saw some in mp3; maybe with aiff too
[07:30] <ubitux> <@Daemon404> what are these 'sync to <hash>' commits? // s/Sync to/consistent with/ ?
[07:32] <ubitux> michaelni: oh m4a too
[07:32] <ubitux> aiff needs confirmation; mateo` likely has a large db of these, maybe he can confirm
[07:37] <CIA-15> ffmpeg: 03Clément BSsch 07master * r4390f8c28d 10ffmpeg/ (ffmpeg.c libavformat/utils.c tools/patcheck): Fix two "skiping" typo.
[07:38] <ubitux> well maybe in m4a it should belong in a particular box
[07:38] <ubitux> like pdin or something
[07:38] <ubitux> not id3
[08:52] <ubitux> i meant pnot* sorry
[08:53] <ubitux> seems to be only in mov so maybe no
[09:25] <CIA-15> ffmpeg: 03Stefano Sabatini 07master * r4b370d6126 10ffmpeg/ffprobe.c: (log message trimmed)
[09:25] <CIA-15> ffmpeg: ffprobe: rework checking logic in writer_print_time() and writer_print_rational()
[09:25] <CIA-15> ffmpeg: Make writer_print_time() and writer_print_rational() always call
[09:25] <CIA-15> ffmpeg: writer_print_int() or writer_print_string(). This way the checks for
[09:25] <CIA-15> ffmpeg: determining if the value should be printed or not are consistently
[09:25] <CIA-15> ffmpeg: performed in the low level functions.
[09:25] <CIA-15> ffmpeg: writer_print_rational() is moved downward in order to avoid a forward
[09:25] <CIA-15> ffmpeg: 03Stefano Sabatini 07master * r58e9025993 10ffmpeg/ffprobe.c: ffprobe: reindent after previous commit
[10:18] <saste> ubitux: CGA fonts are cool
[11:04] <CIA-35> ffmpeg: 03Stefano Sabatini 07master * r2089f93300 10ffmpeg/ffprobe.c: 
[11:04] <CIA-35> ffmpeg: ffprobe: drop pointless opaque parameter in writer init callback
[11:04] <CIA-35> ffmpeg: The parameter was never used.
[11:58] <c10ud> Hello, repost from #ffmpeg:
[11:58] <c10ud> Hello, i was looking into ffserver and filters and i read it's not possible to pass a filter through ffserver. I wondered if it was a quite easy task adding support for filters but after digging a bit into ffmpeg_opt code it is still unclear to me how filter options are applied to the stream. I guess it's the starting point for such implementation (parsing in ffserver and passing to ffmpeg shouldn't be hard from what i've seen). Hints?
[11:59] <ubitux> c10ud: there are some more obvious examples of filtering than ffmpeg.c in doc/examples
[11:59] <ubitux> you should have a look there
[11:59] <c10ud> oh ok, i'll give it a look
[11:59] <c10ud> thanks
[11:59] <ubitux> np
[11:59] <ubitux> the ffserver development is mostly stalled unfortunately, but a new maintainer will be more than welcome
[12:01] <c10ud> heh, i'm not that good, i just wanted to have time on the stream :p
[12:02] <ubitux> well working on it is welcome ;)
[12:03] <ubitux> saste: i'll review your ffprobe patch asap :)
[12:38] <burek> I hope someone with more knowledge will update that article
[12:39] <burek> adding precise terminology and stuff
[12:42] <michaelni> ubitux, the generic ID3 code + m4a or mov definitly doesnt work
[12:42] <michaelni> i mean the APIC part of of the generic code
[12:43] <michaelni> mov expects its AVStreams to be streams created by mov and have the expected private struct ...
[13:52] <burek> michaelni, it seems ffmpeg.org is down, can you please check
[14:02] <burek> iive, who is to be pinged when the ffmpeg.org website is down
[14:03] <iive> arpi, michaelni, reimar.
[14:04] <burek> arpu?
[14:04] <iive> btw, it works for me atm.
[14:04] <burek> try ctrl+f5
[14:04] <iive> still works.
[14:05] <burek> it doesn't even reply to pings to me
[14:05] <burek> Pinging 192.190.173.45
[14:05] <arpu> 192.190.173.45 yes no pings 
[14:05] <iive> 20ms
[14:05] <burek> what ip are you pinging iive
[14:06] <iive> ping 192.190.173.45
[14:06] <iive> burek: can you run a traceroute?
[14:06] <burek> could be some serious isp linkage problem
[14:07] <burek> sure
[14:07] <burek> just a sec
[14:07] <burek> http://pastebin.com/6DtCjcad
[14:08] <iive> 192.168.!!!!
[14:09] <burek> so?
[14:09] <iive> that's private network.
[14:09] <burek> well, yes it is, so? :)
[14:10] <av500> 127.0.0.1 is down!
[14:10] <burek> :)
[14:10] <burek> well, it works when I ping google :)
[14:10] <burek> so that obviously is not a problem
[14:10] <iive> what does it do in the middle of your traceroute?
[14:10] <burek> and it used to work with ffmpeg before too :)
[14:10] <burek> it traverses over private network addresses
[14:10] <burek> gets nat-ed and gets out normally
[14:11] <burek> for ex. 89.216.69.10 is a public ip
[14:11] <iive> but you can't have public ip routed over private ip into public again.
[14:11] <michaelni> iive can you confirm that you can access ffmpeg.org ?
[14:11] <iive> yes, do you want pastebin?
[14:11] <michaelni> i cant access it either no http, no ssh 
[14:11] <michaelni> yes
[14:12] <burek> probably some backbone links between isps have broken
[14:12] <burek> and yes, iive, you can: http://pastebin.com/usdwaQvY
[14:13] <burek> if you correctly setup all your routes :)
[14:13] <iive> http://pastebin.com/Dzq2NL8h
[14:13] <iive> first 2 hops are in my own isp.
[14:13] <ubitux> < michaelni> ubitux, the generic ID3 code + m4a or mov definitly doesnt work // i realized that later; i'm too fast to talk as usual
[14:13] <ubitux> ffmpeg.org doesn't work here
[14:14] <michaelni> http://pastebin.com/0NeWzKam
[14:14] <burek> http://tracert.com/trace_exe.html
[14:14] <burek> try that
[14:15] <burek> trace from diff parts of the world
[14:26] <burek> this one might be better: http://www.locaping.com/
[14:29] <michaelni> burek, that page says "100% packet loss." everywhere :(
[14:30] <burek> yes, it seems some really serious link went down
[14:30] <burek> maybe you could call your isp and ask them to check
[14:30] <burek> ffmpeg.org's hoser's isp
[14:30] <burek> hoster's*
[14:31] <michaelni> arpi will have to do that
[14:31] <michaelni> i dont even know who to call
[14:31] <michaelni> also i dont speak hungarian ...
[14:32] <Compn> whoa
[14:32] <Compn> did half of the internet just go down ?
[14:32] <Compn> http://internethealthreport.com/
[14:32] <burek> I could also setup fflogger to ping ffmpeg.org for this kind of cases, but I don't know if it matters that much, since I believe it doesn't happen that often
[14:33] <Compn> looks like sprint and verizon link having trouble
[14:34] <iive> burek: that site is crap, it uses lookups that are no longer working...
[14:34] <burek> iive, it pinged me without problems
[14:35] <burek> I believe you are on the "other side of the internet" split that just happened :)
[14:35] <iive> it gives me page not found on 90% of the sites. there are some sites that redirect to other sites, like yahoo.
[14:35] <burek> iive, can you ping ffmpeg.gusari.org
[14:36] <iive> 8 hops 80ms, want traceroute log?
[14:36] <burek> no need
[14:37] <burek> im just checking if you are in deed on the "other side" :)
[14:38] <iive> i guess somebody is installing new anti-terrorist, anti-cporn, anti-pirate equipment.
[14:39] Action: michaelni sent arpi mail & sms
[14:41] <iive> can you open piratebay?
[14:41] <michaelni> yes
[14:42] <iive> ok, so no conspiracy.
[14:43] <Compn> surprised they havent banned piratebay in all countries
[14:43] <iive> they are working on it.
[14:45] <burek> it's useless anyway, since they are now hosting hashes only
[14:46] <burek> no more torrents..
[14:46] <iive> well, they managed to make this illegal too.
[15:19] <CIA-35> ffmpeg: 03Anton Khirnov 07master * rc084a975aa 10ffmpeg/libavcodec/utils.c: 
[15:19] <CIA-35> ffmpeg: lavc: use av_mallocz to allocate AVFrames.
[15:19] <CIA-35> ffmpeg: Otherwise the frame is uninitialized, so avcodec_get_frame_defaults()
[15:19] <CIA-35> ffmpeg: cannot determine whether to free extended_data.
[15:19] <CIA-35> ffmpeg: 03Anton Khirnov 07master * ra83499b13b 10ffmpeg/libavcodec/ (avcodec.h utils.c): 
[15:19] <CIA-35> ffmpeg: lavc: rename the argument of avcodec_alloc_frame/get_frame_defaults
[15:19] <CIA-35> ffmpeg: AVFrame is used for both audio and video, so calling the argument 'pic'
[15:19] <CIA-35> ffmpeg: is misleading.
[15:19] <CIA-35> ffmpeg: 03Anton Khirnov 07master * r2bc0de3858 10ffmpeg/libavcodec/utils.c: lavc: initialize AVFrame.extended_data in avcodec_get_frame_defaults()
[15:19] <CIA-35> ffmpeg: 03Anton Khirnov 07master * rb437cec143 10ffmpeg/libavcodec/utils.c: lavc: ensure extended_data is set properly on decoding
[15:19] <CIA-35> ffmpeg: 03Anton Khirnov 07master * ra42aadabc6 10ffmpeg/ (4 files in 2 dirs): 
[15:19] <CIA-35> ffmpeg: lavc: add avcodec_free_frame().
[15:19] <CIA-35> ffmpeg: Since an AVFrame now has malloced members (extended_data), it must have
[15:19] <CIA-35> ffmpeg: a destructor.
[15:19] <CIA-35> ffmpeg: ogg: check that the expected number of headers had been parsed
[15:19] <CIA-35> ffmpeg: Not having the header for a codec is a tell-tale of a broken file.
[15:19] <CIA-35> ffmpeg: 03John Van Sickle 07master * ra716006a7d 10ffmpeg/libavcodec/libx264.c: 
[15:19] <CIA-35> ffmpeg: libx264: change default to closed gop to match x264cli
[15:19] <CIA-35> ffmpeg: open-gop can be enabled with "-flags -cgop"
[15:19] <CIA-35> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[15:30] <michaelni> Domain of sender address michaelni at gmx.at does not resolve <-- was my mail to arpi
[15:32] <burek> probably dns resolving doesn't work right now
[15:39] <michaelni> got reply from arpi, he says: "I know :( issue at the net provider, they re working on it"
[15:41] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * rcc4deafed1 10ffmpeg/libavformat/oggdec.c: 
[15:41] <CIA-35> ffmpeg: oggdec: print error when headers couldnt be parses successfully
[15:41] <CIA-35> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:41] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * r4c273eb64c 10ffmpeg/libavformat/oggdec.c: 
[15:41] <CIA-35> ffmpeg: oggdec: print a warning if the number of headers mismatch expectations.
[15:41] <CIA-35> ffmpeg: This is based on:
[15:41] <CIA-35> ffmpeg:  commit 7751e4693dd10ec98c20fbd9887233b575034272
[15:41] <CIA-35> ffmpeg:  Author: Luca Barbato <lu_zero at gentoo.org>
[15:42] <CIA-35> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:55] <ubitux> http://lucy.pkh.me/bbb.gif  any idea how the colors & size could be improved?
[15:55] <ubitux> i got this with ./ffmpeg -i ~/samples/big_buck_bunny_1080p_h264.mov -ss 45 -vf 'scale=320:160,format=rgb24' -r 20 -frames:v 50 -y bbb.gif
[15:57] <ubitux> mmh maybe the palette could be made dynamic
[15:58] <ubitux> i get an interesting result with scale=320:160,format=rgb8,format=rgb24
[15:59] <ubitux> i wonder if the gif encoder shouldn't request some rgb8 instead
[16:02] <ubitux> since the pic is almost 3MB, i also wonder if it's not in intra only mode :p
[16:03] <CIA-35> ffmpeg: 03Luca Barbato 07master * rd1f05dd183 10ffmpeg/libavformat/ (oggdec.c oggdec.h oggparseskeleton.c): 
[16:03] <CIA-35> ffmpeg: ogg: calculate the start position once all the headers are parsed
[16:03] <CIA-35> ffmpeg: The fisbone packets can be muxed in any order as long the last one
[16:03] <CIA-35> ffmpeg: comes before the first data packet.
[16:03] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * r0842e32f99 10ffmpeg/: (log message trimmed)
[16:03] <CIA-35> ffmpeg: Merge commit 'd1f05dd18375f2f8e68372edee11436927e43ba8'
[16:03] <CIA-35> ffmpeg: * commit 'd1f05dd18375f2f8e68372edee11436927e43ba8':
[16:03] <CIA-35> ffmpeg:  ogg: calculate the start position once all the headers are parsed
[16:03] <CIA-35> ffmpeg: Conflicts:
[16:03] <CIA-35> ffmpeg:  libavformat/oggdec.c
[16:03] <CIA-35> ffmpeg:  libavformat/oggparseskeleton.c
[16:04] <ubitux> michaelni: so do you want mov faststart in the next days or you want me to wait for the release?
[16:06] <michaelni> ubitux, upto you, you know the code better to decide if its likey safe or likely might need follow up bugfixes
[16:07] <michaelni> also we surely will have bugs and we surely will have a .1 release
[16:07] <ubitux> ok
[16:07] <ubitux> ffplay doesn't support animated gif :(
[16:09] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * rd3ef2da59d 10ffmpeg/libavformat/oggdec.h: 
[16:09] <CIA-35> ffmpeg: oggdec: add missing () for OGG_NOGRANULE_VALUE
[16:09] <CIA-35> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:10] <mateo`> michaelni: i can confirm that aiff files use id3 covers
[16:11] <mateo`> this reminds me that id3 at the end of aiff files are still not parsed by ffmpeg :)
[16:11] <ubitux> (oh it looks like we actually don't have a gif demuxer)
[16:12] <michaelni> mateo`, do you have a sample id3-aiff so i can test before enabling it ?
[16:13] <mateo`> michaelni: yes, however this sample won't be public since i have bought the file from a commercial site
[16:14] <michaelni> mateo`, then add aiff to that mp3 strcmp in utils and test if it works
[16:15] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * rbb146bb57b 10ffmpeg/libavformat/oggparsetheora.c: 
[16:15] <CIA-35> ffmpeg: ogg: prevent NULL pointer deference in theora gptopts
[16:15] <CIA-35> ffmpeg: Additional safety in case a special ogg stream is crafted
[16:15] <CIA-35> ffmpeg: with the proper number of
[16:15] <CIA-35> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:15] <CIA-35> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:15] <CIA-35> ffmpeg: 03Luca Barbato 07master * r0336dea2ef 10ffmpeg/libavformat/oggparsetheora.c: oggparsetheora: make it more robust
[16:15] <CIA-35> ffmpeg: 03Derek Buitenhuis 07master * r2768b71798 10ffmpeg/libavcodec/cbrt_tablegen.c: 
[16:15] <CIA-35> ffmpeg: cbrt_tablegen: Include libm.h
[16:15] <CIA-35> ffmpeg: Needed for cbrtf fallback on systems which lack it.
[16:15] <CIA-35> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[16:15] <CIA-35> ffmpeg: 03Nathan Caldwell 07master * r94a69dee64 10ffmpeg/libavcodec/libfdk-aacenc.c: 
[16:15] <CIA-35> ffmpeg: libfdk-aac: Limit to supported sample rates.
[16:15] <CIA-35> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:15] <CIA-35> ffmpeg: 03Nathan Caldwell 07master * rc8ba8be27e 10ffmpeg/libavcodec/libfdk-aacenc.c: 
[16:15] <CIA-35> ffmpeg: libfdk-aac: reindent after last commit
[16:15] <CIA-35> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:15] <CIA-35> ffmpeg: 03Anton Khirnov 07master * r7a1a9dd56c 10ffmpeg/libavcodec/pthread.c: 
[16:15] <CIA-35> ffmpeg: pthread: make sure AVFrame.extended_data is set properly.
[16:15] <CIA-35> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[16:16] <CIA-35> ffmpeg: * qatar/master:
[16:16] <CIA-35> ffmpeg:  pthread: make sure AVFrame.extended_data is set properly.
[16:16] <CIA-35> ffmpeg:  libfdk-aac: reindent after last commit
[16:16] <CIA-35> ffmpeg:  libfdk-aac: Limit to supported sample rates.
[16:16] <CIA-35> ffmpeg:  cbrt_tablegen: Include libm.h
[17:00] <burek> how can someone open a raw bgra32 stream from file, using ffmpeg
[17:00] <burek> According to output, we have 8-bit BGRA raw video. What does it mean? It is just a sequence of one-byte values for each of four channels (Blue, Green, Red, Alpha) for each pixel for each frame. If we open that file with mplayer, we’ll see that our image flickers.
[17:00] <burek> test.raw
[17:05] <Compn> burek : ffmpeg -vcodec raw -i input ?
[17:05] <burek> rawvideo?
[17:05] <Compn> check ffmpeg -formats
[17:05] <burek> it recognizes the input format as image2.. :S http://pastebin.com/E6wq6Wpa
[17:06] <burek> oh there is -f rawvideo too
[17:06] <Daemon404> -f rawvideo
[17:06] <Daemon404> not vcodec
[17:06] <burek> ok
[17:07] <Compn> whoops
[17:07] <Compn> yeha
[17:11] <Compn> it work ?
[17:12] <burek> thanks guys :)
[17:12] <burek> it worked :)
[17:12] <burek> (#ffmpeg)
[17:16] <burek> btw, is it possible to integrate ffplay into ffmpeg, using some -f display option or something
[17:16] <burek> they probably share a decent amount of code base already I guess
[17:16] <burek> so why not just integrate those two
[17:17] <burek> because this way, there are some options that ffplay doesn't support which ffmpeg does, which makes things kinda weird sometimes
[17:17] <Compn> you could split out common code
[17:17] <Compn> like what mplayer / mencoder and mpcommon.c do 
[17:18] <Compn> if thar is any common code...
[17:19] <burek> or that, yes
[18:11] <CIA-35> ffmpeg: 03Michael Niedermayer 07master * rd9d56712e9 10ffmpeg/libavformat/oggparsetheora.c: 
[18:11] <CIA-35> ffmpeg: oggparsetheora: print error when failing on unknown header
[18:11] <CIA-35> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:40] <c10ud> so, i think i found out the right place to hack in order to have {a,v}filters in the ffserver stream
[18:40] <c10ud> however now i am thinking about the way ffserver could give the filter information to ffmpeg
[18:40] <c10ud> i thought of a new AVFilterOption <filter> <args>
[18:41] <c10ud> however i am digging into the ffmpeg data structures but i don't see a clean way to pass this information from ffserver->ffmpeg
[18:42] <c10ud> i thought of using some unused member of some stream struct, but it seems too much of a hack to me
[18:42] <c10ud> suggestions? thoughts?
[18:43] <c10ud> for the record, filters in the ffserver stream looks like a matter of setting the string in ost->avfilter in ffmpeg_opt
[18:51] <ubitux> what do you mean by "filter information"?
[18:51] <c10ud> ubitux, i think a simple string is enough
[18:51] <ubitux> you mean in the configuration?
[18:52] <ubitux> i really have no idea how ffserver works though
[18:52] <c10ud> from what i see, through some hacks :P
[18:52] <ubitux> :)
[18:52] <c10ud> the goal is, set eg. AVFilterOption drawtext text=whatever:fontfile=/wherever/
[18:52] <c10ud> the information is parsed
[18:52] <ubitux> so in the configuration file, right?
[18:53] <c10ud> yes
[18:53] <c10ud> and passed to ffmpeg_opt
[18:53] <c10ud> where it's set in ost->avfilter
[18:53] <c10ud> in the right place
[18:53] <ubitux> "Filter" should be enough
[18:54] <c10ud> however, it looks like there's no direct way for passing such string if not by hacking the stream struct 
[18:54] <c10ud> cfr. read_ffserver_streams function in ffmpeg_opt.c
[18:56] <ubitux> wait, you want ffmpeg to do the filtering?
[18:56] <ubitux> why not make it in ffserver?
[18:56] <c10ud> that's how ffserver works (?)
[18:57] <ubitux> mmh indeed there is no decode
[18:57] <c10ud> ubitux, it looks the right place to do it, especially since filters are now forced to "null"
[18:57] <ubitux> i really need to get a clue on how that stuff works
[18:58] <j-b> Daemon404: I do not give a shit about troll
[18:58] <Daemon404> kudos
[18:58] Action: Daemon404 +1s j-b 
[18:58] <j-b> I need a fix, because some of my users complain.
[18:58] <c10ud> tl;dr: ffmpeg asks for info about what to do with the stream to ffserver, then it sends ready data to ffserver which streams to clients
[19:01] <ubitux> oh ok fun
[19:01] <ubitux> wasn't aware of that
[19:02] <ubitux> c10ud: and well yeah it looks pretty appropriate to change that strdup [a]null code
[19:03] <ubitux> but you'll likely be limited to simple filters
[19:03] <c10ud> well it would be a start
[19:03] <ubitux> yes sure
[19:03] <c10ud> but still figuring out a "clean" way to pass a simple string seems complicated
[19:04] <c10ud> i'm asking just to be sure i'm not missing some obvious way
[21:17] <saste> ubitux: subsection field selection in ffprobe is getting longer than 200 LOC
[21:17] <ubitux> :(
[21:18] <ubitux> i'm reviewing your patch right now
[21:18] <ubitux> give me 20-30 minutes
[21:18] <saste> yes, I believe the patch in its current form is good enough, that is I didn't need to do more design changes in the current patch
[21:19] <saste> I just changed section type -> id, but that's only cosmetics
[21:20] <saste> the only thing i really dislike is the packets_and_frames code, which is specific of ffprobe
[21:20] <saste> but that can be generalized (with a bit of effort) later
[21:21] <saste> I hope the added generality won't make the code (much) slower
[22:19] <ubitux> oh fun the protocol bug
[22:19] <ubitux> ([FFmpeg-user] X-Files, The.mpg: Protocol not found)
[22:27] <saste> why is "," treated specially in "resource" names?
[22:27] <Daemon404> i hate when people do Name, The
[22:27] <saste> Daemon404, Beatles, The
[22:28] Action: Daemon404 stabs
[22:28] <saste> complain with musicians which always choose names starting with "The"
[22:28] <Daemon404> i see no practical reason to print it as Name, The
[22:28] <Daemon404> other than shitty sortign algos
[22:29] <saste> apparently people think that "The Beatles" is harder to find that "Beatles, The"
[22:29] <Daemon404> cant say i agree
[22:29] <saste> because otherwise the "The" region will be filled with too many entries
[22:30] <gnafu> The, The
[22:30] <Daemon404> [16:28] <@Daemon404> other than shitty sortign algos
[22:30] <Daemon404> ^
[22:30] <Daemon404> s/^The //
[22:30] <Daemon404> then sort
[22:30] <gnafu> Or would it be ", The The"?
[22:31] <gnafu> http://en.wikipedia.org/wiki/The_The
[23:01] <ubitux> <@saste> why is "," treated specially in "resource" names? // maybe it's related to the extension list matching
[23:01] <ubitux> separated by ','
[00:00] --- Wed Sep 26 2012