[Ffmpeg-devel-irc] ffmpeg-devel.log.20120905

[burek]

[00:00] <ubitux> the .mak files?
[00:00] <saste> yes
[00:00] <saste> i was looking for the equivalent of a global hash, but seems missing right now
[00:00] <saste> i could just keep using the old script
[00:02] <ubitux> afaict, you would define a lavfi or something in tests/fate-run.sh
[00:02] <ubitux> and use CMP = lavfi
[00:02] <ubitux> which would do all the work
[00:02] <ubitux> no?
[00:02] <ubitux> sounds simple when said like this sure..
[00:04] <ubitux> heh, there is a "lavfitest"
[00:05] <saste> regtest lavfi lavfi tests/vsynth1
[00:07] <ubitux> mmh the fate/filter.mak looks particularly cluttered
[00:07] <ubitux> i don't understand why sometimes the FATE-FILTER-$(CONFIG_XXXX_FILTER) += fate-filter-xxxx is before the variables
[00:07] <ubitux> and sometimes after
[00:08] <ubitux> and why sometimes using the config and sometimes no
[00:12] <saste> well, a lavfi-regression.sh test is better than nothing
[00:12] <saste> and adding it should be a matter of a few minutes
[00:12] <llogan> what say y'all about fflogger announcing new bugs from trac here in this channel?
[00:13] <ubitux> i wouldn't mind
[00:14] <llogan> is there a particular format you prefer?
[00:14] <llogan> [trac] #1717: Concat filter can not join more than two video files
[00:14] <llogan> for example?
[00:15] <ubitux> (and even trac edit given that the activity is quite limited; it could kind of make us review the edit at times)
[00:15] <ubitux> llogan: sounds good to me
[00:15] <llogan> what about a url on the next line?
[00:15] <ubitux> llogan: only new ticket or any activity?
[00:15] <ubitux> i don't care about the url i have an alias
[00:15] <ubitux> fftrac() { firefox https://ffmpeg.org/trac/ffmpeg/ticket/$1 }
[00:15] <ubitux> :))
[00:16] <llogan> i was thinking just new ticket due to potential noisyness, but i'd probably be fine with either
[00:17] <ubitux> don't forget the colors!
[00:17] <ubitux> 3[5trac3] 4#1717: 7Concat filter can not join more than two video files
[00:18] <ubitux> (totally random)
[00:18] <llogan> ohh...fancy
[00:19] <llogan> i'll spam the ML with a RFC too
[00:19] <ubitux> \x03 followed by a num for the colors
[00:19] <ubitux> 12, 15, etc
[00:20] <llogan> actually burek will be working on it i imagine since i think it is his bot
[00:20] <ubitux> ok :)
[00:21] <ubitux> llogan: i don't think it would hurt much to have a message for each activity
[00:21] <ubitux> it gives some visibility, i personally like it
[00:22] <ubitux> in a perfect world we would just !shutup when too much spam and there is an "important" discussion here
[00:44] Action: michaelni likes the idea of our KGB agent reporting trac activity to us
[00:44] <ubitux> ffbi like saste suggested recently
[11:56] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r0e28e3cf82 10ffmpeg/doc/filters.texi: doc/filters: remove reference to mp=hue filter, which was dropped
[11:56] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r8a7b82ccab 10ffmpeg/doc/filters.texi: doc/filters: itemize example for mp filter
[11:56] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * rfc15f58adb 10ffmpeg/doc/filters.texi: doc/filters: add mp=noise example
[13:52] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r2c340596ca 10ffmpeg/libavcodec/elbg.c: 
[13:52] <CIA-56> ffmpeg: elbg: Fix an assert
[13:52] <CIA-56> ffmpeg: It seems the condition was flipped from what was intended.
[13:52] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[13:52] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r19000122a4 10ffmpeg/libavcodec/mpegvideo_enc.c: 
[13:52] <CIA-56> ffmpeg: mpegvideo_enc: remove assert that has become obsolete with the new API
[13:52] <CIA-56> ffmpeg: It now just checks uninitialized and unused data.
[13:52] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[13:53] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r2d7d91f06d 10ffmpeg/libavcodec/svq1enc.c: 
[13:53] <CIA-56> ffmpeg: svq1enc: Set picture_structure correctly
[13:53] <CIA-56> ffmpeg: This fixes assert failures when running in debug mode.
[13:53] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[13:53] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r91672504a4 10ffmpeg/libavcodec/mpegvideo.c: 
[13:53] <CIA-56> ffmpeg: mpegvideo: remove last_picture_ptr / h264 assert.
[13:53] <CIA-56> ffmpeg: This assert is no longer true since h264 error concealment needs
[13:53] <CIA-56> ffmpeg: last_picture_ptr to be set.
[13:55] <Daemon404> how does pp and spp differ from libpp?
[14:02] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rd5f65e9d40 10ffmpeg/: (log message trimmed)
[14:02] <CIA-56> ffmpeg: Merge commit '124134e42455763b28cc346fed1d07017a76e84e'
[14:02] <CIA-56> ffmpeg: * commit '124134e42455763b28cc346fed1d07017a76e84e':
[14:02] <CIA-56> ffmpeg:  avopt: Store defaults for AV_OPT_TYPE_CONST in the i64 union member
[14:02] <CIA-56> ffmpeg: Conflicts:
[14:02] <CIA-56> ffmpeg:  libavcodec/aacenc.c
[14:02] <CIA-56> ffmpeg:  libavcodec/libopenjpegenc.c
[14:02] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r124134e424 10ffmpeg/ (31 files in 7 dirs): 
[14:02] <CIA-56> ffmpeg: avopt: Store defaults for AV_OPT_TYPE_CONST in the i64 union member
[14:02] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[14:14] <CIA-56> ffmpeg: 03Martin Storsjö 07master * rd58dd4b5b5 10ffmpeg/ (15 files in 6 dirs): 
[14:14] <CIA-56> ffmpeg: avopt: Store defaults for AV_OPT_TYPE_FLAGS in the i64 union member
[14:14] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[14:14] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r99de3df3d3 10ffmpeg/: (log message trimmed)
[14:14] <CIA-56> ffmpeg: Merge commit 'd58dd4b5b5d31cfd4092e38a5f2c894eee2ab078'
[14:14] <CIA-56> ffmpeg: * commit 'd58dd4b5b5d31cfd4092e38a5f2c894eee2ab078':
[14:14] <CIA-56> ffmpeg:  avopt: Store defaults for AV_OPT_TYPE_FLAGS in the i64 union member
[14:14] <CIA-56> ffmpeg: Conflicts:
[14:14] <CIA-56> ffmpeg:  libavcodec/libvpxenc.c
[14:14] <CIA-56> ffmpeg:  libavcodec/options_table.h
[14:15] <Daemon404> michaelni, i hope you merge that correctly ;)
[14:15] <Daemon404> that particular commit is important for msvc support
[14:16] <saste> Daemon404: can you elaborate on it?
[14:16] Action: saste wishes people would make more auto-explanatory commit messages
[14:16] <michaelni> Daemon404, the pp/spp i meant are the wrapers that connect libpp to libmpcodecs
[14:17] <Daemon404> https://github.com/mstorsjo/c99-to-c89/commit/953db19b3c9ee12c74d639ec78b26561e0114b64
[14:17] <Daemon404> ^ related
[14:17] <Daemon404> its to ease c89 conversion
[14:17] <Daemon404> and in general, just less ugly.
[14:17] <saste> C99->c89
[14:17] <saste> a downgrade?
[14:17] <Daemon404> yes
[14:17] <Daemon404> on-the-fly conversino with that wrapper
[14:18] <Daemon404> --cc="c99wrap cl" --ld="link:
[14:19] <Daemon404> (and i -do- have ffmpeg building with msvc this way)
[14:19] <Daemon404> few things remain to be merged
[14:20] <saste> Daemon404: I see thanks
[14:20] <JEEBsv> Daemon404: oh so it works like a cc now?
[14:20] <JEEBsv> nice
[14:20] <Daemon404> chrome guys plan to use ffmpeg this way
[14:20] <JEEBsv> \automation/
[14:22] <Daemon404> the latest ffms2 builds have been using it
[14:25] <Tjoppen> neato
[14:25] <Daemon404> chrome using it also means things using chromium will
[14:25] <Daemon404> like steam.
[14:25] <Daemon404> and some games
[14:27] <michaelni> Daemon404, we will _need_ a msvc fate box
[14:27] <Daemon404> i have one sitting in my office
[14:27] <Daemon404> unhooked-up
[14:27] <Daemon404> it would be hooked up if i didnt get my flights to paris cancelled last minute
[14:28] <Daemon404> both you and diego are on my case... so there is likely little chance i will stray off the path of setting it up without someone letting me know
[14:28] <Daemon404> >.>
[14:33] <Tjoppen> am I right in understanding the DV decoder considers everything to be bottom-field-first?
[14:34] <Tjoppen> s->picture.top_field_first  = 0;  in dvdec.c
[14:37] <CIA-56> ffmpeg: 03Martin Storsjö 07master * re6153f173a 10ffmpeg/ (74 files in 7 dirs): 
[14:37] <CIA-56> ffmpeg: avopt: Store defaults for AV_OPT_TYPE_INT in the i64 union member
[14:37] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[14:37] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rd46c1c72e4 10ffmpeg/: (log message trimmed)
[14:37] <CIA-56> ffmpeg: Merge commit 'e6153f173a49e5bfa70b0c04d2f82930533597b9'
[14:37] <CIA-56> ffmpeg: * commit 'e6153f173a49e5bfa70b0c04d2f82930533597b9':
[14:37] <CIA-56> ffmpeg:  avopt: Store defaults for AV_OPT_TYPE_INT in the i64 union member
[14:37] <CIA-56> ffmpeg: Conflicts:
[14:37] <CIA-56> ffmpeg:  libavcodec/libopenjpegdec.c
[14:37] <CIA-56> ffmpeg:  libavcodec/libopenjpegenc.c
[14:48] <CIA-56> ffmpeg: 03Martin Storsjö 07master * rc7b610aa0b 10ffmpeg/ (5 files in 3 dirs): 
[14:48] <CIA-56> ffmpeg: avopt: Explicitly store float/double option defaults in .dbl
[14:48] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[14:48] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r98840ee017 10ffmpeg/: (log message trimmed)
[14:48] <CIA-56> ffmpeg: Merge commit 'c7b610aa0b1bac47eea0056b13fe6e982b85844a'
[14:48] <CIA-56> ffmpeg: * commit 'c7b610aa0b1bac47eea0056b13fe6e982b85844a':
[14:48] <CIA-56> ffmpeg:  avopt: Explicitly store float/double option defaults in .dbl
[14:48] <CIA-56> ffmpeg: Conflicts:
[14:48] <CIA-56> ffmpeg:  libavcodec/ac3dec.c
[14:48] <CIA-56> ffmpeg:  libavcodec/libx264.c
[15:01] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r21bc440384 10ffmpeg/ (3 files in 2 dirs): 
[15:01] <CIA-56> ffmpeg: avopt: Explicitly store rational option defaults in .dbl
[15:01] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:01] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r8510150f48 10ffmpeg/: (log message trimmed)
[15:01] <CIA-56> ffmpeg: Merge commit '21bc44038491e525335ad763883b0288402ff6a0'
[15:01] <CIA-56> ffmpeg: * commit '21bc44038491e525335ad763883b0288402ff6a0':
[15:01] <CIA-56> ffmpeg:  avopt: Explicitly store rational option defaults in .dbl
[15:01] <CIA-56> ffmpeg: Conflicts:
[15:01] <CIA-56> ffmpeg:  libavfilter/buffersrc.c
[15:01] <CIA-56> ffmpeg:  libavfilter/vsrc_testsrc.c
[15:06] <CIA-56> ffmpeg: 03Diego Biurrun 07master * raa66111365 10ffmpeg/configure: build: utvideoenc: Add missing dependency on Huffman code
[15:06] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r686a329395 10ffmpeg/ (doc/APIchanges libavutil/opt.h libavutil/version.h): 
[15:06] <CIA-56> ffmpeg: avopt: Reorder the default_val struct, making i64 the first field
[15:06] <CIA-56> ffmpeg: Also mention this change in APIchanges.
[15:06] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:06] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rb21b5b04cc 10ffmpeg/: (log message trimmed)
[15:06] <CIA-56> ffmpeg: Merge remote-tracking branch 'qatar/master'
[15:06] <CIA-56> ffmpeg: * qatar/master:
[15:06] <CIA-56> ffmpeg:  build: eamad: Add missing dependency on mpegvideo code
[15:06] <CIA-56> ffmpeg:  build: utvideoenc: Add missing dependency on Huffman code
[15:06] <CIA-56> ffmpeg:  avopt: Reorder the default_val struct, making i64 the first field
[15:06] <CIA-56> ffmpeg: Conflicts:
[15:06] <CIA-56> ffmpeg: 03Diego Biurrun 07master * r6980453569 10ffmpeg/configure: build: eamad: Add missing dependency on mpegvideo code
[16:08] <Compn> hmmmmmm
[16:08] <Compn> wiki bot ?
[16:08] <burek> trac bot
[16:08] <burek> wiki + tickets
[16:08] <Compn> ah
[16:08] <Compn> now we can watch carl do his thing on the trac :)
[16:09] <burek> :)
[16:09] <burek> oops, was that a notice?
[16:10] <Compn> yes
[16:10] <Compn> channel notice
[16:10] <Compn> would be better as a message
[16:10] <burek> what a mistake-a to make-a :)
[16:10] <Compn> can you edit what info it puts out ?
[16:10] <Compn> e.g. drop the date and time , we know the date and time ...
[16:10] <burek> sure
[16:11] <Compn> but have the first sentence of the change or something ?
[16:11] <av500> md5sum of the message
[16:11] <burek> let me see what I can do
[16:11] <Compn> make it highlight av500 every time as well
[16:11] <burek> :D
[16:11] <av500> svakako!
[16:12] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r5666a1b270 10ffmpeg/libavutil/ (opt.c version.h): 
[16:12] <CIA-56> ffmpeg: lavu/opt: support NULL and special "none" values for size and pixel format options
[16:12] <CIA-56> ffmpeg: Allow to specify NULL values explicitly, thus overriding the default
[16:12] <CIA-56> ffmpeg: values set in the context.
[16:20] Action: Compn afk
[16:21] <Compn> would be cool if we could get ciabot to drop the 'ffmpeg:' from its messages. we know its ffmpeg, thats the only commits we see here
[16:21] <Compn> hah
[16:21] <Compn> burek : get rid of the md5, thats terrible
[16:21] <burek> :)
[16:21] <av500> lol
[16:22] <burek> well, it looks cool :D
[16:22] <Compn> you cant get a summary eh ?
[16:22] <burek> yes I can, but it'll hurt :)
[16:22] <av500> you want to spam all the people out of this channel it seems :)
[16:22] <Compn> i just want one line, not whole messages
[16:23] <Compn> but if its too much work, dont worry
[16:23] <burek> I have to calc max_line_len - current_len
[16:23] <burek> to see how much of descr can be put into
[16:23] <burek> and i dont know what's the max line len here on freenode
[16:26] <av500> burek: shorten the descriction: att|new|edit
[16:26] <av500> and wiki
[16:26] <av500> also Ticket is not needed
[16:27] <av500>  [edit] patters: #1457
[16:31] <burek> not that simple
[16:31] <burek> it merges info from wiki and trac and milestones
[16:31] <burek> all activities from trac
[16:31] <burek> so just "edit" will be confusive :)
[16:31] <av500> no
[16:31] <av500> one has a ticket #
[16:31] <av500> the other has not
[16:31] <burek> also, description is in encoded html
[16:32] <av500> also, you could make it expand on mouse hover :)
[16:32] <burek> I'm not reading all the fields from the database
[16:32] <burek> but rather from rss
[16:32] <burek> which does not contain ticket id
[16:33] <burek> just a summary of title, desc, date, etc
[16:33] <burek> :)
[16:33] <burek> and do a google search for the possible solutions, I guess? :))
[16:33] <ubitux> yes please remove the md5
[16:34] <ubitux> if you want some green change one of the other colors :)
[16:34] <av500> burek: indeed
[16:34] <ubitux> (IMO the description could be the default color)
[16:35] <ubitux> you should switch the action & author style: the action is more important than the author, so it should be bold instead of the author
[16:35] <burek> there is no bold
[16:36] <av500> could we use a different font too?
[16:36] <burek> at least I didnt put any bold
[16:36] <burek> :)))))
[16:36] <burek> no, descr is definitely not a good idea
[16:36] <ubitux> there is some bold
[16:36] <ubitux> the "light" version of the colors are bold
[16:37] <divVerent> Hi
[16:38] <ubitux> burek: see: http://ubitux.fr/pub/pics/_fflogger.png
[16:38] <burek> o.O
[16:38] <ubitux> hey, hi divVerent :)
[16:38] <divVerent> I need a bit of help with pts... basically, I am hitting the message from this change by my own code
[16:38] <divVerent> https://lists.ffmpeg.org/pipermail/ffmpeg-cvslog/2012-March/048382.html
[16:38] <divVerent> what I am doing, is sending a zero pts value for the first audio frame
[16:38] <Daemon404> why do tickets have md5s
[16:38] <divVerent> which kinda makes sense, as zero IS the time when my stream begins...
[16:38] <divVerent> so where am I wrong? is pts 0 forbidden? What pts to use instead then?
[16:39] <burek> no they dont Daemon404 :D
[16:39] <divVerent> my output file is "perfect", as this code basically maps pts 0 to 0 because this is the very first frame
[16:40] <ubitux> divVerent: maybe dts should be set as well?
[16:40] <ubitux> (just a guess)
[16:40] <divVerent> ubitux: no
[16:40] <divVerent> then this code is still hit
[16:40] <divVerent> oh wait, it is not
[16:40] <divVerent> but how to get dts, other than by receiving it from lavc
[16:40] <divVerent> which does not yield one
[16:41] <divVerent> if it did, I'd be providing it
[16:41] <burek> is it ok now?
[16:43] <ubitux> burek: looks broken
[16:43] <ubitux> (remove the "Descr:"
[16:43] <ubitux> )
[16:43] <divVerent> ubitux: my guess is that there is some stupid requirement that pts must have a positive offset, but that'd be... annoying
[16:44] <JEEBsv> yeah
[16:44] <divVerent> either that, or that audio pts are meant to contain the time of the LAST sample, not the FIRST
[16:44] <JEEBsv> starts at one at has a delay of one :S
[16:44] <burek> ubitux, attachments have no descr :)
[16:44] <burek> https://ffmpeg.org/trac/ffmpeg/timeline?from=09%2F05%2F12&daysback=10&authors=&wiki=on&update=Update
[16:44] <ubitux> burek: yeah i know; we will guess when it's the description, so just remove the "Descr: " string
[16:44] <burek> take a look at the RSS link at the bottom
[16:44] <burek> and you'll see what am I working with...
[16:45] <burek> sorry, this: https://ffmpeg.org/trac/ffmpeg/timeline
[16:46] <ubitux> divVerent: dunno :(
[16:47] <ubitux> burek: so can you 1) fallback to "default" color for the description, remove the "Descr:" string (but not the description), and switch the type & author styles? :)
[16:47] <ubitux> (oups forgot numbering)
[16:47] <divVerent> ubitux: okay, alternate question, do you know a tool to show a file's pts values?
[16:47] <ubitux> -af ashowinfo
[16:47] <ubitux> :p
[16:47] <burek> what is "type and author styles"
[16:47] <ubitux> the issue i raised earlier
[16:47] <ubitux> about bold
[16:48] <burek> give me colors you want
[16:48] <divVerent> ubitux: mplayer shows this:
[16:48] <burek> 1-15
[16:48] <divVerent> DEMUX: Append packet to d_video, len=27441  pts=0.000  pos=32816  [packs: A=0 V=1]
[16:48] <divVerent> DEMUX: Append packet to d_audio, len=140  pts=0.000  pos=32816  [packs: A=1 V=1]
[16:48] <divVerent> so pts 0 is already used for audio in ffmpeg's own output file
[16:48] <divVerent> while using the ffmpeg binary does not show the warning
[16:48] <ubitux> burek: just switch the style between "[wiki]" and "burek"
[16:48] <divVerent> [ashowinfo @ 0x1ae8920] n:0 pts:0 pts_time:0 pos:1105510 fmt:flt chlayout:stereo nb_samples:1024 rate:48000 checksum:AE09006D plane_checksum[AE09006D]
[16:49] <divVerent> ashowinfo on the input file
[16:49] <ubitux> (you also have -debugts in ffmpeg)
[16:49] <divVerent> how to use -debugts?
[16:49] <divVerent> it's not in the manpage and wants an argument
[16:50] <nevcairiel> burek: the main text of the line should really be in the default color, meaning not colored at all, it looks weird now
[16:50] <burek> ok, I'll show it now
[16:50] <burek> and just tell me what colors you'd like
[16:50] <divVerent> oh wait, -debugts does not exist
[16:50] <ubitux> divVerent: -debug_ts sorry
[16:51] <divVerent> the encoder sees:
[16:51] <divVerent> encoder -> type:audio pkt_pts:-1024 pkt_pts_time:-0.0213333 pkt_dts:-1024 pkt_dts_time:-0.0213333
[16:51] <divVerent> encoder -> type:audio pkt_pts:0 pkt_pts_time:0 pkt_dts:0 pkt_dts_time:0
[16:51] <divVerent> for the first two audio frames
[16:51] <ubitux> burek: i meant switching the styles, not the position ;)
[16:51] <divVerent> negative audio pts...
[16:51] <nevcairiel> the html tags are particularly useless :p
[16:51] <divVerent> but where does it get dts from
[16:51] <divVerent> because the codec yields no dts
[16:52] <ubitux> i have no idea sorry
[16:55] <Prottey> How can I contribute specifically for G.729-compatible encoder development?
[16:56] <divVerent> ubitux: is debug_ts a feature of the command line client, or of the library?
[16:56] <Compn> Prottey : do you mean contribute code or money ?
[16:57] <divVerent> looks like a feature of the binary
[16:57] <ubitux> yes it's ffmpeg-program specific
[16:57] <Prottey> code
[16:58] <ubitux> Prottey: by editing libavcodec/g729* and libavformat/bit.c files and sending your git-format-patches to ffmpeg-devel at ffmpeg.org?
[16:58] <Prottey> I'd like to give money as well, if I had enough
[16:58] <Compn> was there ever a g729 patch ?
[16:58] Action: Compn checks wiki
[16:59] <Prottey> there is only decoder afaik
[16:59] <Compn> seems like decoder only
[17:00] <av500> so write an encoder
[17:00] <Compn> ffmpeg can be compiled against libg729 tho
[17:00] <burek> let's see now
[17:01] <Prottey> libg729 is an EasyG729A wrapper, isn't it?
[17:01] <Compn> i dont know which lib it uses 
[17:01] <ubitux> please avoid posting the content of the comment burek 
[17:01] <ubitux> burek: some comments are very large
[17:01] <Compn> actually, i dont know if the libg729 is a decoder only ?
[17:02] <ubitux> burek: fortunately the content is truncated by the 512 chars limit, but better not paste it anyway
[17:02] <burek> ubitux, earlier here someone did say to put descr/comments
[17:02] <Prottey> EasyG729A is not free
[17:02] <burek> maybe it would be wise that you guys first agree on what you want? :)
[17:02] <Compn> ah
[17:02] <burek> yes, I was hoping it will get truncated
[17:02] Action: av500 wants a burek right now
[17:03] <ubitux> burek: well, there is not much meaning in having a big-partial content of the description, right?
[17:03] <burek> :)
[17:03] <ubitux> burek: "description" was most likely confused with the "title"
[17:03] <ubitux> (that's what i thought actually)
[17:03] <ubitux> so let me correct myself: the title style should be in the current description style (so no style), and you should remove the description
[17:03] <ubitux> are you ok with that? :)
[17:04] <Compn> Prottey : do you need help with creating an encoder ? you can ask if anyone else is interested on the ffmpeg-devel mailing list. i think the preferred way to start, is to create a ffmpeg git branch repo. then work on your encoder there. when its finished, it can be merged back into ffmpeg master
[17:04] <burek> it's no problem for me to remove descr
[17:04] <Prottey> Thanks Compn
[17:04] <burek> I'm just asking if you all agree on it, so that I don't have to do things twice :)
[17:04] <Prottey> Well, it seems Linphone has its own G.729 implementation
[17:04] <Compn> Prottey : g729 is amr right ?
[17:05] <ubitux> burek: i believe it will be fine after the changes, at least for me ;)
[17:05] <Compn> arent there some free decoders part of opencore project ?
[17:05] <Prottey> g729 is celp afaik
[17:05] <saste> Prottey: I remember someone was working on g.729 encoding
[17:05] <Compn> er maybe no encoder...
[17:05] Action: Compn dumb today
[17:05] <burek> 4[newticket] 5aurelien: Ticket #1719 (MPEG-TS demuxer : don't parse a PAT (and clear programs) if its version ...) created 12https://ffmpeg.org/trac/ffmpeg/ticket/1719
[17:05] <burek> like that?
[17:05] <ubitux> burek: LGTM
[17:06] <burek> ok
[17:06] <Prottey> git://git.linphone.org/bcg729.git
[17:06] <ubitux> burek: thanks :)
[17:06] <burek> :beer: :)
[17:06] <av500> +1
[17:06] <saste> Prottey: Vladimirir Voroshilov
[17:06] <saste> my memory is still good
[17:06] <saste> https://gitorious.org/~voroshil/ffmpeg/ffmpeg-voroshil/commits/g729enc-20111107
[17:06] <Compn> oo nice 
[17:07] <saste> you could try to contact the author and ask what's going on
[17:07] <Compn> saste : well done
[17:07] <Compn> yes, contacting vlad , see if you can help him finish it and submit it to ffmpeg :)
[17:07] <Prottey> nice
[17:09] <burek> that's final I hope :)
[17:10] <ubitux> < divVerent> what I am doing, is sending a zero pts value for the first audio frame // why are you doing that btw?
[17:11] <ubitux> i mean, if it's an audio demuxer, don't you just need to set the sample rate, and raise samples?
[17:13] <Prottey> Nov 06 2011 is the latest commit
[18:06] <av500> time to /ingore fflogger 
[18:06] <Daemon404> indeed
[18:08] <divVerent> 17:11:05    @ubitux | i mean, if it's an audio demuxer, don't you just need to set the sample rate, and raise samples?
[18:08] <divVerent> ubitux: that's how it was in the past
[18:09] <divVerent> but I found the cause
[18:09] <divVerent> the pts handling of lavc changed a lot
[18:09] <divVerent> in the past, the audio codec got no pts
[18:09] <divVerent> now they do, from an AVFrame
[18:09] <divVerent> so, in the past, you first encoded, then calcuated pts from the encoder
[18:09] <divVerent> now you first calculate pts, then let lavc do its thing, then lavf
[18:09] <divVerent> like with video
[18:12] <divVerent> which has the side effect that you "automatically" get dts, and then the warning in question does not occur
[18:17] <divVerent> ubitux: so thanks for the help, -debug_ts was what led me to notice there ARE dts in ffmpeg.c, and from there it was simple tracing of where they come from :)
[18:17] Action: ubitux is happy to be useful to something he doesn't understand
[18:19] <ubitux> question not related, is there some particular place to set the AVStream language, long and short name?
[18:20] <ubitux> example, in case of matroska you could have the english audio stream identified by "eng", with a long name such as "English track" or something
[18:20] <ubitux> does all of this belong to metadata?
[18:25] <ubitux> it seems matroska uses "language" metadata for "eng" and "title" meta for the long name, ok.
[18:34] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r8267f0f7ae 10ffmpeg/libavfilter/vf_scale.c: 
[18:34] <CIA-56> ffmpeg: vf_scale: fix incompatible pointer type warning
[18:34] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:48] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r7666828c44 10ffmpeg/libavutil/opt.h: 
[18:48] <CIA-56> ffmpeg: avopt: fix examples to match the same style about default values as the actual code.
[18:48] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:48] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r9afd55922a 10ffmpeg/libavcodec/imgconvert.c: 
[18:48] <CIA-56> ffmpeg: imgconvert: silence some const incompatible pointer warnings.
[18:48] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:53] <j-b> good morning
[18:53] <ubitux> ohai
[18:55] <ubitux> will avformat_find_stream_info() return an error if no stream are found or i can assume there will be at least one stream in case of no error?
[18:55] <ubitux> mmh, bad phrasing
[18:56] <ubitux> will avformat_find_stream_info() return an error if no stream are found or i can need to add a check myself?
[18:56] <ubitux> -can
[18:57] <michaelni> without checking id say if there are 0 streams it shouldnt return an error
[18:58] <michaelni> i mean a file with just metadata for example or something odd liek that
[18:58] <ubitux> ok
[20:26] <maister> FFmpeg is written in C99, right? Some ISO C90 warnings keep popping up :p
[20:28] <JEEB> yes, C99 features are used
[20:29] <maister> Any reason why -std=c99 or -std=gnu99 isn't used then?
[20:29] <nevcairiel> -std=c99 is used
[20:30] <nevcairiel> at least my config.mak has it in CFLAGS
[20:30] <maister> hm, I keep getting: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]
[20:30] <maister> I'll see if there's any -std=c99
[20:31] <nevcairiel> was that relaxed in C99? :p
[20:31] <maister> well, yes
[20:31] <ubitux> we use a limited subset of c99
[20:31] <maister> oh, i see
[20:31] <maister> it's -std=c99
[20:31] <ubitux> we avoid mixed declarations and code
[20:31] <maister> but that warning is explicitly added in
[20:31] <ubitux> even if it's technically not really a problem
[20:32] <nevcairiel> well, it is if you try to use the msvc compiler support that was just added in libav
[20:32] <ubitux> maister: yes, it seems to still cause problem with ancient and weird compilers so we avoid them
[20:32] <maister> ubitux, ok, fair enough.
[20:32] <nevcairiel> wonder how many things i have to fix before ffmpeg compiles with msvc, once michael merges the latest libav
[20:32] <maister> nevcairiel, msvc? lol. Dropped all sane struct declarations? :p
[20:33] <nevcairiel> maister: nah, there is a pre-processor tool that converts code to c89, but stuff like moving variable declarations to the top of the block it doesnt handle
[20:33] <maister> I see
[20:33] <maister> nevcairiel, alternatively compile a subset of c99 as c++ but I guess that's a sin punishable by death? :p
[20:34] <nevcairiel> that wouldnt work out that well either
[20:34] <ubitux> "sin" is too good
[20:34] <nevcairiel> C++ requires explicit casts in many places where C doesnt
[20:34] <nevcairiel> that would be a pain to add
[20:34] <maister> ah, right
[20:34] <nevcairiel> like, malloc returns
[20:34] <nevcairiel> o.O
[20:34] <maister> I maintain a codebase that has to be compatible with both c99 and c++ at same time
[20:34] <maister> ...
[20:35] <ubitux> afaik c++ doesn't even allow trailing coma in enum and stuff
[20:35] <maister> ubitux, correct
[20:35] <ubitux> and various other limitations over C
[20:35] <maister> goto rules are stricter
[20:35] <maister> for one
[20:35] <nevcairiel> that can only be a good thing =p
[20:35] <maister> since you don't allow mixed code and declarations it doesn't change much
[20:36] <maister> :)
[20:37] <CIA-56> ffmpeg: 03Nedeljko Babic 07master * rb3fdfc8c4e 10ffmpeg/ (13 files in 3 dirs): (log message trimmed)
[20:37] <CIA-56> ffmpeg: Optimization of AC3 floating point decoder for MIPS
[20:37] <CIA-56> ffmpeg: FFT in MIPS implementation is working iteratively instead
[20:37] <CIA-56> ffmpeg:  of "recursively" calling functions for smaller FFT sizes.
[20:37] <CIA-56> ffmpeg: Some of DSP and format convert utils functions are also optimized.
[20:37] <CIA-56> ffmpeg: Signed-off-by: Nedeljko Babic <nbabic at mips.com>
[20:37] <CIA-56> ffmpeg: Reviewed-by: Vitor Sessak <vitor1001 at gmail.com>
[20:43] <maister> hm, trying to write an video encoder, and when I try to mux it into mkv, it complains about "No bmp codec ID found.". ff_get_codec_tag() seems to fail. Any ideas?
[20:45] <michaelni> maister, add a entry to ff_codec_bmp_tags[] for your encoder
[20:46] <michaelni> btw, does it make sense to post all wiki edits here ?
[20:47] <maister> michaelni, thanks
[20:47] <ubitux> yay, vobsub demuxer "almost" done
[20:47] <ubitux> just needs a way to hijack the pts
[20:50] Action: ubitux loves this voyeurism
[21:34] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r397ebc1d9b 10ffmpeg/libavcodec/proresenc_anatoliy.c: 
[21:34] <CIA-56> ffmpeg: proresenc_anatoliy: add 2 const to silence 2 warnings
[21:34] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:34] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r269e6212ee 10ffmpeg/libavcodec/g723_1.c: 
[21:34] <CIA-56> ffmpeg: g723_1: add some const for more correctness
[21:34] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:13] <ubitux> http://pastie.org/4670006  sscanf question; any idea how to avoid the "15" hardcode and use sizeof(s)-1 instead?
[22:13] <ubitux> "%*[^,]", 15, s doesn't seem to do the trick
[22:15] <ubitux> the only trick i have in mind is a define of the value and use s[BS+1]; sscanf(p, "%" BSIZE "[^,]", ..
[22:16] <ubitux> s/BS/BSIZE/
[22:18] <ubitux> (+stringified)
[22:18] <ubitux> oh i can stringify the sizeof
[22:21] <ubitux> my bad, it doesn't do what i want obviously
[22:22] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r750cc3c0e5 10ffmpeg/libavcodec/acelp_pitch_delay.c: 
[22:22] <CIA-56> ffmpeg: acelp_pitch_delay: add missing libm.h include for exp2()
[22:22] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:22] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * re60fcdb989 10ffmpeg/libavfilter/vf_hue.c: 
[22:22] <CIA-56> ffmpeg: vf_hue: fix AVOption defaults
[22:22] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:23] <ubitux> http://pastie.org/4670060  can't find anything better
[22:23] <ubitux> ideas welcome
[22:59] <ubitux> AVInputFormat->read_seek callback is deprecated over generic seeking using the stream index entries, or read_seek2(), right?
[22:59] <nevcairiel> it was deprecated, but read_seek2 was never implemented for anything, so in reality its still used
[23:02] <ubitux> meh, indeed i don't see lot of read_seek2 haha
[23:18] <CIA-56> ffmpeg: 03Clément BSsch 07master * r9d9f34a3c1 10ffmpeg/libavformat/utils.c: 
[23:18] <CIA-56> ffmpeg: lavf: remove a few useless AV_NOWARN_DEPRECATED().
[23:18] <CIA-56> ffmpeg: read_seek was undeprecated in e5c708151.
[23:18] <CIA-56> ffmpeg: 03Clément BSsch 07master * r5620088494 10ffmpeg/libavformat/utils.c: lavf: fix a few typo in some comments.
[23:21] <saste> thread.gmane.org/gmane.comp.video.ffmpeg.devel/79667/
[23:21] <saste> that's why I felt a dejavu
[23:22] <ubitux> hehe
[23:23] <ubitux> saste: yeah, prefer char* for the return value
[23:23] <ubitux> iirc in mplayer we have mp_basename() returning a const char*
[23:23] <ubitux> and we basically have to cast it all the time :p
[23:25] <saste> ubitux: *and* strptime returns (non const) char *
[23:25] <saste> so that's the way to go
[23:40] <ubitux> mpegps_read_pes_header() has some checks for ppos, but how can it be NULL?
[23:40] <ubitux> seems not possible to me, can anyone confirm?
[23:41] <saste> michaelni: you ok with char *av_small_strptime(...)?
[23:42] <saste> i think you are based on your comment, but i prefer to be sure rather than to change it post-commit
[23:42] Action: michaelni is ok with char*
[23:43] <saste> ok will push it soon
[23:43] <llogan> burek: that was quick.
[23:59] <burek> llogan :beer: :)
[00:00] --- Thu Sep  6 2012