[Ffmpeg-devel-irc] ffmpeg-devel.log.20140925

burek burek021 at gmail.com
Fri Sep 26 02:05:02 CEST 2014


[00:10] <cone-600> ffmpeg.git 03Martin Storsjö 07master:a2efbecc4ed1: libavformat: Move avc mp4 startcode parsing to a shared file
[00:10] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:b8d3f7fc21b9: Merge commit 'a2efbecc4ed12d287cf29856418c4da4a7648d95'
[00:16] <jamrial> michaelni: if i change the prototypes for these functions, do i have to do it for every arch?
[00:16] <michaelni> yes but it should be trivial for 32bit based archs
[00:52] <cone-600> ffmpeg.git 03Thomas Volkert 07master:ddf5fb71ee9c: rtpenc: HEVC/H.265 support
[00:52] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:d8ddac363e77: Merge commit 'ddf5fb71ee9c8b2d9a23c0f661a84896cd7050fc'
[01:09] <cone-600> ffmpeg.git 03Martin Storsjö 07master:2f172f1ae984: rtsp: Clear the session id on redirects
[01:09] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:6441d522e9fe: Merge commit '2f172f1ae984b763c06069adb51d0053b3834f4b'
[02:43] <cone-600> ffmpeg.git 03James Almer 07master:c99a88281489: avcodec/idctdsp: change {put,add}_pixels_clamped to ptrdiff_t line_size
[02:51] <cone-600> ffmpeg.git 03James Almer 07master:4f4f08e6f00f: x86/idctdsp: port {put,add}_pixels_clamped to yasm
[05:43] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:303fc9b36833: avcodec/mpegvideo_enc: write a log message at verbose level if frame re-encoding occurs due to VBV
[09:19] <Ko_deZ> ubitux: Thanks for the answer. I saw the time on my post, and realized I had 2 minutes to catch my train. I failed :-/ Anyway, I will look into that, thank you.
[09:41] <durandal_1707> how many opw slots we have?
[09:43] <ubitux> durandal_1707: afaik we only have one
[10:24] <nevcairiel> afaik you have to come up with your own funding if you want more
[12:06] <michaelni> ubitux, some companies donate to OPW wthout specifing a project, and their donation is used for whichever applications+applicants+project looks best so if there are alot of applicants wanting to work on ffmpeg and their applications and abilities are great we likely would get more slots, thats at least how i understand it
[12:27] <Jul13t> michaelni, I'm working on my proposal for the MPEG, I've not yet been assigned my qualification task yet
[12:34] <Daemon404> michaelni, did anything change recently in how we probe images?
[12:34] <Daemon404> i can bisect, but now suddenly it doesnt work at all for me.
[12:37] <michaelni> Daemon404, yes there where a few changes related to image2/3(pipe)
[12:37] <Daemon404> ok
[12:37] <Daemon404> pretty much all image probing fails for me now
[12:37] <Daemon404> as an API user i mean
[12:38] <michaelni> Jul13t, did you ask paul / thilo ?
[12:38] <Jul13t> michaelni, he's just given me what I need to to. Thanks :)
[12:38] <michaelni> Jul13t, ok great then
[12:39] Action: Jul13t doing Thilo's assignment
[12:55] <Daemon404> hmm actually its not probing
[12:55] <Daemon404> it fails at decode... hmm
[12:55] Action: Daemon404 checks
[12:59] <J_Darnley> WTF?  Does that ALS things really put floating-point and lossless together?
[13:00] <Daemon404> dts-hd-ma does too
[13:00] <Daemon404> as does lagarith
[13:00] <Daemon404> as such, we have an x86 fp emu in the codebase
[13:00] <J_Darnley> :O
[13:01] <J_Darnley> I knew lagarith did but that's cause it is pants-on-head but the other thing is for bluray isn't it?
[13:02] <Daemon404> yes ok the problem is probing...
[13:02] <Daemon404> avformat_find_stream_info succeeds, no streams are found or set
[13:03] <Daemon404> video_codec_id also not set
[13:03] <Daemon404> :|
[13:05] <Daemon404> seriously broken
[13:05] Action: Daemon404 guess bisect time
[13:09] <cone-763> ffmpeg.git 03Martin Storsjö 07master:d07ddc93e29a: avcodec: Fix a doxy comment to refer to the right function
[13:09] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:a6fd685fedb1: Merge commit 'd07ddc93e29a4fdf12cd426953a7ecd5abe5e36d'
[13:22] <Jul13t> michaelni, while working on my qualification task, is there any format applications should follow apart from the one on the Gnome wiki?
[13:23] <Jul13t> I mean perculiar to ffmpeg may be
[13:25] <Daemon404> 'format applications' ?
[13:30] <J_Darnley> I think that mean guides or examples to be followed, like the gnome one.
[13:30] <kierank> possibly Jul13t means code style?
[13:30] <Jul13t> Daemon404, i'm talking of the the application format
[13:30] <Jul13t> submitting the application to the gnome student app portal
[13:31] <Daemon404> oh ok
[13:31] <Jul13t> as for the coding styles i  think there's a wiki for that on code conventions
[13:31] <Daemon404> ffmpeg has no application format
[13:31] <Jul13t> and stuff, and it's very clear
[13:31] <Daemon404> afaict
[13:31] <Jul13t> Daemon404, guess i'll just follow the format given by Gnome,
[13:31] Action: Jul13t hopes its safe ;)
[13:31] <ubitux> Jul13t: see http://ffmpeg.org/developer.html#Contributing
[13:32] <Jul13t> ubitux, thanks
[13:32] <Daemon404> michaelni, found the offending commit
[13:33] <Daemon404> michaelni, b3fd2b175c90b4766034095e74e1f5112b1547ef breaks all image stream probing
[13:37] <michaelni> Daemon404, how can i reproduce it ?
[13:37] <Daemon404> avformat_open_input(&ctx->afctx, file_path, in_format, NULL);
[13:37] <Daemon404> avformat_find_stream_info(ctx->afctx, NULL);
[13:38] <Daemon404> then inspect the avformatcontext
[13:38] <Daemon404> in_format here is NULL
[13:39] <michaelni> do you have a testfile & command line ?
[13:39] <Daemon404> command line is no applicate
[13:39] <Daemon404> it's API.
[13:39] <Daemon404> not*
[13:39] <Daemon404> ffmpeg.c doesnt reproduce it. you only broke it for API users.
[13:40] <Daemon404> http://chromashift.org/img/ef1.d2v_001_0.png is teh url i opened
[13:40] <Daemon404> to test
[13:40] <Daemon404> but any will fail.
[13:40] <michaelni> why does it differ between ffmpeg and API users ?
[13:41] <Daemon404> because ffmpeg has a bunch of hacks
[13:41] <Daemon404> very liekly
[13:41] <Daemon404> its not my problem to investigate ffmpeg.c
[13:42] <Daemon404> i can certainly write a tiny test c file and submit it, but it will only be a few api calls.
[13:44] <michaelni> "./demuxing_decoding ef1.d2v_001_0.png video audio" works too
[13:45] <cone-763> ffmpeg.git 03Katerina Barone-Adesi 07master:28f5cd312c9d: fate: Switch ra4-288 test from framecrc() to pcm()
[13:45] <Daemon404> unsurprising as it is mostly copypasted
[13:45] <Daemon404> i will write a test file.
[13:45] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:9b53691f7318: Merge commit '28f5cd312c9da9072108edf8b7685d009374ea96'
[13:46] <michaelni> Daemon404, thx
[13:51] <cone-763> ffmpeg.git 03Gabriel Dume 07master:95d312d6c82e: concat: K&R formatting cosmetics
[13:51] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:8f3d3e5f8373: Merge commit '95d312d6c82e7485f43c3ddda0f369af0e41c83b'
[14:00] <cone-763> ffmpeg.git 03Gabriel Dume 07master:86a361081d3e: lmlm4: K&R formatting cosmetics
[14:00] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:59f49667569b: Merge commit '86a361081d3eb4e999d75bc2a462f01d88cbd2e5'
[14:08] <cone-763> ffmpeg.git 03Gabriel Dume 07master:4620affa24ee: m4vdec: K&R formatting cosmetics
[14:08] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:454f98b881fc: Merge commit '4620affa24eedb167482198aa04126bcedd05620'
[14:24] <Daemon404> michaelni, msg'd you a code test
[14:24] <Daemon404> reverting b3fd2b175c90b4766034095e74e1f5112b1547ef causes the 'good' output.
[15:31] <michaelni> Daemon404, fixed
[15:31] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:1dbdcb4a8c76: avformat/img2dec: fix error code at EOF for pipes
[15:33] <Daemon404> michaelni, thanks!
[15:54] <Daemon404> michaelni, you missed one EOF
[15:55] <Daemon404> -        return AVERROR(EIO); /* signal EOF */
[15:55] <Daemon404> +        return AVERROR_EOF; /* signal EOF */
[15:55] <Daemon404> ^^
[15:58] <Daemon404> michaelni, ping after that one is fixed because there is another bug which is not related... also in img2...
[16:09] <Daemon404> michaelni, ok so attempting to seek first (to see if it is a seekable file) leaves it in a fucked up state
[16:10] <Daemon404> seeking to 0 causes av_read_image to never ready any packets ever again
[16:10] <Daemon404> this i only after b3fd2b175c90b4766034095e74e1f5112b1547ef
[16:11] <Daemon404> i dont see how this can possibly be correct behavior
[16:22] <Daemon404> afctx->pb->seekable is true too
[16:22] <Daemon404> sorry but this fix just seems buggy as fuck..
[16:28] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:2497914a1846: avformat/img2dec: pass error code and signal EOF
[17:48] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:6d35aba167a2: avformat/img2dec: initialize pkt->pos for image pipes
[17:48] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:ce6e46be7229: avformat/img2dec: enable generic seeking for image pipes
[17:57] <michaelni> Daemon404, maybe fixed, can you find more issues in it ?
[17:57] <ubitux> "please hurt me"
[17:57] <Daemon404> lol
[17:57] <Daemon404> let me try.
[17:58] <Daemon404> michaelni, nope it seems to work now
[18:01] <Daemon404> and many thanks of course
[18:02] <reynaldo> guess no one would object removing the "copied straight from gsoc" banner at https://trac.ffmpeg.org/wiki/SponsoringPrograms/OPW/2014-12 ?
[18:02] <reynaldo> michaelni ^
[18:03] <michaelni> reynaldo, sure, remove it
[18:04] <michaelni> but then IMHO it doesnt matter, we could also leave it ;)
[18:04] <reynaldo> ppl will actually look at that for reference, I think it matters
[18:05] <reynaldo> the amount of info is overwelming enough to make it start with "this shite is c&p-ed crap, info needs to be updated"
[18:05] <reynaldo> 8)
[18:09] <reynaldo> michaelni: is anyone chasing mentors/backups to update that page already? Other than the discussion on MPEG ALS I see nothing more on the ml
[18:14] <michaelni> reynaldo, no idea, i just ask on a "as needed" basis ATM
[18:15] <cone-763> ffmpeg.git 03Thilo Borgmann 07master:e113692c899f: lavu/ffmpeg_opt: Check return value of avcodec_find_encoder().
[18:48] <kierank> michaelni: where did all the opw money come from
[19:20] <wm4> kierank: *chirping crickets*
[19:46] <reynaldo> llogan: there are two msgs from me to ffmpeg-devel@ been held due to a too-many-recipients situation, please drop the first one and let the second pass
[19:46] <reynaldo> I CCed the mentors whos emails I have on file, that's why they got held
[19:58] <michaelni> kierank, see reynaldos mail 
[20:00] <kierank> ah
[20:04] <wm4> if someone picks up the vdpau opw task, I could probably provide some guidance for that
[20:07] <llogan> reynaldo: i can't access mailman right now. it's 404.
[20:07] <reynaldo> llogan: noticed, tried to drop the first one by myself but went 404 too
[20:08] Action: llogan attempts to summon beastd
[21:02] <cone-763> ffmpeg.git 03James Almer 07master:4b892e469bc9: x86/cavsdsp: fix buffer alignment in cavs_idct8_add_mmx()
[21:02] <cone-763> ffmpeg.git 03James Almer 07master:a829870b2f4d: avcodec/svq1enc: align buffer used by simd functions
[21:36] <wm4> haha
[21:37] <wm4> another image opening regression
[21:37] <wm4> what's even going on
[21:39] <Compn> blame carl :P
[21:39] <Compn> ehe
[21:48] <gnafu> "But Caaaaaarl, that /kills/ people."
[22:20] <cone-763> ffmpeg.git 03Carl Eugen Hoyos 07master:c0f9df30dd02: lavc/x86/idctdsp.h: Fix make checkheaders.
[22:27] <ubitux> Andreas hasn't packaged 2.4.1?
[22:27] <ubitux> or maybe that's what the warning is about
[22:51] <ubitux> https://lists.ffmpeg.org/mailman/listinfo/ffmpeg-devel/ ?
[22:51] <ubitux> something went wrong?
[23:04] <bryno> but... where's the mail? :(
[23:06] <wm4> the ML got accidentally deleted!
[23:22] <J_Darnley> wm4 what?
[23:58] <cone-763> ffmpeg.git 03Michael Niedermayer 07master:a43bcaafe20d: avcodec/dxtory: Support dxtory-2.0.127.avi
[00:00] --- Fri Sep 26 2014


More information about the Ffmpeg-devel-irc mailing list