[Ffmpeg-devel-irc] ffmpeg-devel.log.20150505

[burek]

[00:00:04 CEST] <BBB> Ill let it sit for a bit, its barely important
[00:00:14 CEST] <BBB> if it still annoys me tomorrow Ill send a patch
[00:00:15 CEST] <BBB> :-p
[00:00:21 CEST] <jamrial> ok :p
[00:04:46 CEST] <jamrial> nevcairiel: not sure if you saw my comment some days ago, but mingw-w64 now ships the hevc dxva header
[00:04:55 CEST] <drwx> Compn: well -avoid_negative_ts 0 works but shouldn't it be the default?
[00:04:56 CEST] <jamrial> it's not in any release so far (only trunk), but msys2 uses that so compiling your lavfilters ffmpeg fork fails
[00:05:37 CEST] <nevcairiel> i dont like msys2, nor do i like using mingw-w64 trunk, its managed by equally logic-challenged persons =p
[00:05:48 CEST] <jamrial> heh
[00:06:18 CEST] <nevcairiel> but i'll try to remember to change the condition next time i update it
[00:06:25 CEST] <jamrial> agree, sticking to mingw-w64 trunk made the gcc packages break more than once for me
[00:17:41 CEST] <wm4> nevcairiel: managing such a messed up combination of bad windows and unix parts probably requires someone who is challenged
[00:17:53 CEST] <wm4> or they would go insane or depressive
[00:18:45 CEST] <nevcairiel> maybe
[00:21:13 CEST] <jamrial> interestingly enough msys2 hasn't yet upgraded their gcc package to 5.1.0
[00:21:15 CEST] <nevcairiel> I would love if msvc would one day produce a faster ffmpeg than gcc, but the 10-20% difference is just too much to not care
[00:22:30 CEST] <cone-168> ffmpeg 03Andreas Cadhalpun 07master:ca234639ac49: avidec: avoid infinite loop due to negative ast->sample_size
[00:22:31 CEST] <cone-168> ffmpeg 03Michael Niedermayer 07master:c7369f3a4bd2: avformat/avidec: print a warning for negative sample_size
[00:50:11 CEST] <BBB> does GENERIC_INDEX duplicate index entries if I seek back to the beginning of a file?
[00:50:26 CEST] <BBB> I didnt check, but the code doesnt seem to be checking for stuff like that, it just unconditionally adds new index entries
[00:50:40 CEST] <BBB> I didnt s/check/test/
[01:05:59 CEST] <Compn> drwx : no idea, but if you have an account on the bug trac, is that answer already posted? 
[03:40:51 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:6ef3426d90a3: avcodec/x86/deinterlace: use INIT_MMX like other asm code does too
[04:58:44 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:7f45f7fc7f49: avcodec/mpeg12dec: use the correct dimensions for checking SAR
[06:07:53 CEST] <jamrial> Compn: http://pastebin.com/DB0Qvabx
[07:18:54 CEST] <cone-890> ffmpeg 03James Almer 07master:3f7986f98392: avfilter/Makefile: add missing lavfutils.c dependency to cover_rect and find_rect filters
[07:25:13 CEST] <jamrial> michaelni: f320f2f4c6398404a8653b4a78134ab42dda9ea7 broke msvc 2012 since it lacks round()
[07:26:38 CEST] <jamrial> for the libraries a libavutil/libm.h compat implementation is used instead. can that header be included in cmdutils.c?
[07:48:25 CEST] <cone-890> ffmpeg 03James Almer 07master:858f81fde2c1: cmdutils: move get_rotation() up in the file
[09:13:05 CEST] <drwx> Compn: the avoid_negative_ts answer has been posted indeed but the bug is still open, so i guess it means this should be the default behaviour, doesn't it?
[09:19:02 CEST] <wm4> clearly the user should be forced to use extremely obscure options to prevent the creation of broken files
[12:01:21 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:36993527dd3d: cmdutils: Add libm.h for round()
[12:12:27 CEST] <michaelni> "<wm4> clearly the user should be forced to use extremely obscure options to prevent the creation of broken files" <-- please talk with ubitux he added avoid_negative_ts = 1 in b08273c9ca8b91d03133f9fbef100bbd3cd8e271 citing the specification (... Blocks with a negative Raw  Timecode are not valid.")
[12:13:38 CEST] <wm4> does ffmpeg write the required codec delay elements in this case?
[12:13:42 CEST] <wm4> or read them?
[12:14:04 CEST] <wm4> because it's pointless if it doesn't, and from what I've seen, the timestamps even start _after_ 0
[13:27:56 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:7ac5c38ec5ba: avformat/mux: Add avoid_negative_ts_use_pts
[13:27:57 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:9d4fdfe24c73: avformat/matroskaenc: Use avoid_negative_ts_use_pts if no stream writes dts
[13:30:33 CEST] <Compn> wm4 : i dont disagree :P
[14:36:57 CEST] <drwx> Compn: shouldn't you close the other tickets as well?
[14:38:37 CEST] <drwx> ticket*
[14:41:50 CEST] <Compn> i havent tested it
[14:41:57 CEST] <Compn> maybe you are thinking of carl (cehoyos)
[14:42:57 CEST] <drwx> Compn: you're the one who committed the fix and closed ticket 487, right?
[14:43:05 CEST] <drwx> https://trac.ffmpeg.org/ticket/2975 is probably fixed too, isn't it?
[14:43:39 CEST] <wm4> I wonder if x264 will add a similar thing like x265 for bit depth selection
[14:43:56 CEST] <wm4> and I'm also wondering why they don't just do this internally
[14:44:49 CEST] <Daemon404> they == ?
[14:44:51 CEST] <Daemon404> x265?
[14:45:27 CEST] <wm4> at least
[14:45:51 CEST] <Daemon404> that would break at least one use case
[14:46:02 CEST] <Daemon404> i.e. outputting simultanious 10bit and 8bit streams from ffmpeg
[14:47:00 CEST] <kierank> Daemon404: what happened to your x265 interlaced patch?
[14:47:17 CEST] <Daemon404> kierank, it became apparent to me that with teh current API, it is not feasible.
[14:55:48 CEST] <drwx> Compn: i tested and it fixes delays when encoding: http://sprunge.us/PZjU (system ffmpeg first, then latest ffmpeg)
[16:25:59 CEST] <Daemon404> michaelni, ping
[16:29:55 CEST] <Daemon404> nvm
[16:37:06 CEST] <Daemon404> hmm
[16:37:07 CEST] <Daemon404> libavcodec/x86/hevc_mc.asm:1628: warning: label alone on a line without a colon might be in error
[16:37:11 CEST] <Daemon404> i get a buttload of these
[16:39:38 CEST] <kierank> I think because there's a loop label that isn't used
[16:39:48 CEST] <kierank> but meh nasm
[16:40:36 CEST] <Daemon404> ?
[16:40:38 CEST] <Daemon404> this is yasm
[16:41:06 CEST] <Daemon404> aw shit
[16:41:09 CEST] <Daemon404> i had nasm installed but not yasm
[16:41:15 CEST] <nevcairiel> hah
[16:41:33 CEST] <Daemon404> (because a diff project needed nasmO
[16:42:20 CEST] <nevcairiel> i know that there is stuff that builds with yasm that nasm doesnt like, is it the other way around as well in some cases?
[17:44:58 CEST] <wm4> is windows media center the thing that committed the wtv crime?
[17:45:20 CEST] <Daemon404> wtv was western digital wasnt it
[17:46:29 CEST] <wm4> WTV (Windows Recorded TV Show)
[17:46:45 CEST] <wm4> I only know it from ffmpeg issues scrolling by on this channel
[17:50:16 CEST] <Daemon404> o
[18:02:11 CEST] <BBB> wm4: well one is a product and one is a container
[18:02:21 CEST] <BBB> wm4: but it was introduced as part of, yes
[18:27:38 CEST] <Compn> drwx : no, michaelni fixed the commit and closed that ticket . thanks michaelni :)
[18:28:33 CEST] <Compn> drwx : i mostly just sit here and talk to people so they dont feel ignored. ask them to report or test bugs / fixes ... etc 
[18:30:06 CEST] <Compn> wm4 : windows media center edition created the wtv crime, yes. mpeg2 in asf.
[18:57:42 CEST] <nevcairiel> wtv is not asf Compn
[18:58:28 CEST] <nevcairiel> its a better format than dvr-ms was, which is just plain asf with some hacks =p
[19:12:16 CEST] <RiCON> <@Daemon404> wtv was western digital wasnt it <- you're thinking of wdtv
[19:18:05 CEST] <Compn> oh i was thinking of dvr-ms , not wtv
[19:18:06 CEST] <Compn> my bad
[19:28:16 CEST] <drwx> Compn: sorry, seeing the timing, i assumed you were michael
[20:12:32 CEST] <cone-890> ffmpeg 03Ronald S. Bultje 07master:ed68fccf76b4: vp9: error out if chroma subsampling changes between frames.
[20:50:21 CEST] <Compn> drwx : no problem. it all works for you now? its always nice to see a bug fix that quick :)
[20:56:35 CEST] <drwx> Compn: you mean two years after the bug report? :p
[20:56:59 CEST] <drwx> yeah it all works but only one (recent) bug was closed, the oldest wasn't
[20:57:33 CEST] <nevcairiel> but its the same bug, and it works now?
[20:58:31 CEST] <nevcairiel> sure reads like the same
[20:58:33 CEST] Action: nevcairiel closes
[20:59:49 CEST] <drwx> nevcairiel: yeah it works, one was -c:v copy and the other was encoding
[20:59:59 CEST] <drwx> cheers
[22:01:22 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:984f50deb2d4: diracdec: prevent overflow in data_unit_size check
[22:14:24 CEST] <emustafa> Hi all, I was playing around with the ffmpeg source code. I am having difficulty in understanding makefile for the programs (ffmpeg, fplay and etc.). I want to add a new program file with main function in it. However, I couldn't compile it. I tried to understand how makefile works for ffmpeg, fplay and etc. but I couldn't do replicate the same thin
[22:14:24 CEST] <emustafa> g for my program. I also couldn't find docs on this issue. Any help is appreciated. Thanks. Elshad.
[22:19:42 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:5a455dd01115: avcodec/diracdec: Move buf[] read after size check
[22:32:54 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:8fc8024ea56e: aasc: return correct buffer size from aasc_decode_frame
[22:32:55 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:54d985db847b: Merge commit '8fc8024ea56e814cd257d5fe27b21a865080782f'
[22:45:27 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:e191aaca44b9: alsdec: limit avctx->bits_per_raw_sample to 32
[22:45:28 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:df9528161845: Merge commit 'e191aaca44b986816695e3b7ecfae64697fd6631'
[22:53:17 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:110f7f35fb61: aacpsy: correct calculation of minath in psy_3gpp_init
[22:53:18 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:9810aa340842: Merge commit '110f7f35fb615b97d983b1c6c6a714fddd28bcbe'
[22:57:09 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:94bb1ce882a1: alsdec: check sample pointer range in revert_channel_correlation
[22:57:10 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:ff849e11e361: Merge commit '94bb1ce882a12b6d7a1fa32715a68121b39ee838'
[23:20:33 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:60f1cc4a1ffc: alsdec: only adapt order for positive max_order
[23:20:34 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:1ace789fb344: Merge commit '60f1cc4a1ffcbf24acbb543988ceeaec76b70818'
[23:53:00 CEST] <cone-890> ffmpeg 03Carl Eugen Hoyos 07master:e31cdb5cc808: lavf/riff: Add MultiScope II fourcc MSC2 as MJPEG.
[23:56:49 CEST] <Compn> msc2 hmmm sounds familiar
[00:00:00 CEST] --- Wed May  6 2015