[Ffmpeg-devel-irc] ffmpeg-devel.log.20140723

[burek]

[00:26] <ePirat> how to add a codec?
[00:26] <ePirat> *protocol
[00:28] <cone-382> ffmpeg.git 03Diego Biurrun 07master:57f1b1dcc77a: fft-test: Drop unnecessary #ifdefs around header includes
[00:28] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:77304cf3523c: avcodec: dct/rdft only support float mode, skip their contents for fixed point
[00:28] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:9eeedf876998: Merge commit '57f1b1dcc77a93c2a5c503d4e47fe2f567cf9db5'
[00:34] <cone-382> ffmpeg.git 03Diego Biurrun 07master:5be7d8597d02: fft-test: Comment some #endifs
[00:34] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:9515d7c629cd: Merge commit '5be7d8597d02bcaba2de2503c0e59cf865c746dc'
[00:34] <Compn> ePirat : probably copy a current protocol and change it from there
[00:38] <wm4> ePirat: didn't you just send a patch that adds a protocol
[00:38] <ePirat> already figured out
[00:39] <ePirat> i just meant how to add it in terms of in which files to reference it
[00:39] <ePirat> so that it compiles
[00:45] <Compn> yes i thought thats what you wanted
[00:45] <Compn>  configure and probably libavformat/allformats
[00:45] <Compn> maybe some others
[01:05] <llogan> michaelni: are you envisioning the wiki projects page to be open to all, or marked as read-only and users continue asking on devel so potential projects can be reviewed?
[01:05] <llogan> laziness says to be open, but nitiness says to review to help reduce crap
[01:09] <relaxed> I nominate llogan as wiki sheriff
[01:19] <michaelni> llogan, imo we shouldnt try to fix an issue before tere is one
[01:20] <michaelni> that is why restrict the access to a wiki page before anyone tries to do something nasty
[01:20] <michaelni> or rather why add more work on our sholders before we need to
[01:21] <ePirat> ubitux, done https://ffmpeg.org/pipermail/ffmpeg-devel/2014-July/160172.html
[01:22] <cone-382> ffmpeg.git 03Martin Storsjö 07master:06a1d1323bc8: configure: Extend the neon intrinsics check
[01:23] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:7e5d75d53f9b: Merge commit '06a1d1323bc8afd03f085ae98d9b74e431b3534e'
[01:24] <wm4> so I'm seeing this debian guy on the mailing list
[01:25] <wm4> apparently he's working on ffmpeg packages?
[01:25] <wm4> no such packages in debian unstable or experimental, though
[01:27] <wm4> ePirat: your patch lacks some files, I think
[01:30] <ePirat> fail
[01:30] <ePirat> forgot to add the actual protocol&
[01:35] <iive> wm4: last time i've heard about debian packages, they were waiting approval from ftp masters.
[01:36] <ePirat> wm4, fixed https://ffmpeg.org/pipermail/ffmpeg-devel/2014-July/160173.html
[01:36] <llogan> If you grew a beard during the whole Debian process, by the time it was accepted you would look like Compn
[01:37] <J_Darnley> ePirat: It looks like you've sent it to the wrong project.  This is FFmpeg not Libav
[01:45] <ePirat> J_Darnley, it was requested that I send it to ffmpeg too
[01:46] <llogan> ePirat: he's referring to the "This file is part of Libav."
[01:47] <llogan> etc, etc, etc
[01:48] <ePirat> llogan, anything else then the license/header that I should change?
[01:49] <llogan> just look at the source of the other stuff using the same license in libavformat
[01:49] <llogan> for examples
[01:55] <J_Darnley> Also: the micro version starts at 100 in ffmpeg
[01:55] <J_Darnley> and there's a cosmetic alignment issue with your addition to the makefile.
[01:56] <ePirat> ?
[01:56] <wm4> I think it's bullshit to make contributors prepare two patches
[01:56] <wm4> one for ffmpeg and one for libav
[01:57] <J_Darnley> I agree.
[01:57] <wm4> the way it works is that libav will apply it, and ffmpeg will merge the libav commit
[01:57] <wm4> so there's absolutely no need to make the contributor send a ffmpeg specific patch
[01:57] <wm4> the value in having the contributor send a patch to ffmpeg at all is so that ffmpeg can review the patch
[01:57] <wm4> and give suggestions for improvement
[01:58] <wm4> and review can be done on a patch even if it contains the evil "Libav" string
[01:59] <ePirat> J_Darnley, whats wrong in the makefile?
[02:00] <J_Darnley> I'll send an email
[02:01] <ePirat> oh i think i found it, I indented with tabs accidentally 
[02:01] <J_Darnley> OMG!  TABS!  THEY'RE DEVIL CHARACTERS!
[02:02] <wm4> yes they are
[02:03] <llogan> ePirat: tools/patcheck will discover those for you
[02:03] <wm4> they have the ability to incite endless bikesheds and flames
[02:04] <J_Darnley> I don't even like them in a word processor.
[02:06] <ePirat> does patcheck expects a patch file?
[02:06] <ePirat> I get somewhat messy output
[02:07] <J_Darnley> A patch ot diff, yes
[02:08] <J_Darnley> *or
[02:08] <J_Darnley> You can also pipe into it.
[02:13] <ePirat> J_Darnley, fixed I think https://ffmpeg.org/pipermail/ffmpeg-devel/2014-July/160177.html
[02:16] <J_Darnley> I don't see any more obvious problems.
[02:46] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:a61325798a9a: Changelog: add "version <next>:" so people dont add new things to the last release
[03:04] <Orix> hey guys, I'm trying to get the bink codec to load: http://pastebin.com/azvFGhLJ but I keep getting the following error: [bindvideo @ <memory>] Extradata missing or too short
[03:08] <michaelni> ePirat, ive posted a few minor commets, btw, how can i test it ?
[03:10] <michaelni> Orix, you need a demuxer which will set extradata from the file header(s)
[03:11] <Orix> michaelni: would you be able to elaborate or link me to the documentation?
[03:11] <Orix> i've been going off of https://ffmpeg.org/doxygen/trunk/api-example_8c-source.html
[03:12] <michaelni> Orix, doc/examples/demuxing_decoding.c might work
[03:13] <michaelni> or well, ffplay.c /ffmpeg but they are big
[03:13] <Orix> yeah, tried to dig at them a bit
[03:13] <Orix> i'll check out demuxing
[03:13] <Orix> thanks
[03:47] <jamrial> [cpuflags @ 000000000031f188] [Eval @ 000000000031e960] Invalid chars 'now' at the end of expression '3dnow'
[03:47] <jamrial> [cpuflags @ 000000000031f188] Unable to parse option value "3dnow"
[03:48] <jamrial> Failed to set value 'mmx+mmxext+3dnow' for option 'cpuflags': Error number -22 occurred
[03:48] <jamrial> http://fate.ffmpeg.org/history.cgi?slot=x86_64-athlon64-gentoo-gcc
[03:49] <jamrial> http://fate.ffmpeg.org/history.cgi?slot=x86_32-athlon64-gentoo-gcc
[03:49] <jamrial> http://fate.ffmpeg.org/history.cgi?slot=x86_32-athlon64-gentoo-gcc-3dnow
[04:27] <jamrial> 265dadbe introduced this regression
[08:00] <ubitux> llogan: you should tweet about the recent release
[08:02] <ubitux> (michaelni: i'll test the icecast thing so please don't apply in a hurry ;))
[08:07] <jamrial> speaking of recent release, it needs to be added to trac
[08:50] <ePirat> michaelni, you need to have an icecast server or alternatively just something that will accept a PUT request
[08:51] <ePirat> michaelni, and then just do icecast://username:pass@mycoolserver:8000/mountpoint.mp3 or so
[08:56] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:bbc5adfb9428: avutil/opt: Fix parsing regression with constants starting with numbers
[08:57] <michaelni> ePirat, ok, thx
[08:57] <ePirat> how do I use av_freep?
[08:59] <michaelni> av_free(x) -> av_freep(&x)
[08:59] <michaelni> jamrial, 2.3 added
[09:01] <ePirat> michaelni, thanks
[09:02] <ePirat> michaelni, and I wonder if there is a function that could replace my cat_header stuff
[09:14] <michaelni> ePirat, av_bprint_init+ n  * av_bprintf+av_bprint_finalize or something based on AV_DICT_APPEND but cat_header() is fine too 
[09:15] <ePirat> ok
[09:15] <ePirat> is it fine to freep it inside the function and return NULL?
[09:19] <ePirat> michaelni, if I just use one av_log it looks ugly:
[09:19] <ePirat> [icecast @ 0x7fc63be015c0] Streaming ogg but appropriated content type NOT set!
[09:19] <ePirat> Set it with -content_type application/ogg
[09:20] <ePirat> the second missing the icecast @ & part, causing it to look weird
[09:21] <ePirat> and the line will get too long and it's awful to split it to two lines
[09:25] <michaelni> ePirat, hmm, ok sure, leave it as 2 calls then
[09:26] <ePirat> would work if I somehow could suppress the  icecast @ & part tough 
[09:26] <michaelni> that would be a NULL as first parameter but its probably better to know where a message came from 
[09:26] <ePirat> ok
[10:33] <cone-382> ffmpeg.git 03Carl Eugen Hoyos 07master:c75fdee74717: avcodec/x86/hevc_deblock: Fix compilation with nasm.
[10:51] <cone-382> ffmpeg.git 03Vittorio Giovara 07release/2.2:68fd80ee1ca2: g2meet: allow size changes within original sizes
[10:51] <cone-382> ffmpeg.git 03Michael Niedermayer 07release/2.2:3e78f8689167: Merge commit '68fd80ee1ca22c39b6ef4e6641b5b2e0d4d89a14' into release/2.2
[10:51] <cone-382> ffmpeg.git 03Alessandro Ghedini 07release/2.2:4ddac7199b24: vc1: Do not return an error when skipping b frames
[14:26] <cone-382> ffmpeg.git 03Diego Biurrun 07master:2b6ab3a2bd7e: mpegvideo: Move QUANT_BIAS_SHIFT define to the only place it is used
[14:26] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:d0cb6caf691c: Merge commit '2b6ab3a2bd7e7cee5e7a55dd2e48b8feb4a826bb'
[14:36] <cone-382> ffmpeg.git 03Diego Biurrun 07master:20ec0daaa1cd: fft-test: Make global exptab variable static
[14:36] <cone-382> ffmpeg.git 03Michael Niedermayer 07master:54256482cbc0: Merge commit '20ec0daaa1cdd3996ac6ed0198de961129474044'
[18:05] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:1bf371b24ebd: avcodec/h264: factor is_extra() out
[18:05] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:601c2388549f: avcodec/h264: support AV_PKT_DATA_NEW_EXTRADATA
[18:05] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:01b236b70438: avformat/flvdec: use AV_PKT_DATA_NEW_EXTRADATA for h264 too
[18:57] <cone-109> ffmpeg.git 03Andrey Utkin 07master:8a4c0866dc7d: avio: Introduce avio_read_to_bprint(avioctx, bp, max_size)
[18:57] <cone-109> ffmpeg.git 03Andrey Utkin 07master:b6a6459a24c6: lavd/lavfi: work with non-mappable files for graph_file option
[19:45] <kurosu_> If I had a new threading in lavc/thread.h, what do I need to do documentation-wise ?
[19:45] <kurosu_> is it a public api? (ie api bump?)
[19:53] <jamrial> that header is not installed, and all the functions have the ff_ prefix, so it's internal to lavc
[20:01] <kurosu_> I see there's a doc/multithreading.txt
[20:39] <cone-109> ffmpeg.git 03Nidhi Makhijani 07master:eac77fcd56fc: mpegenccontext: Remove unused dct_unquantize_h261 function pointers
[20:39] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:97d2b8d6f5a4: Merge commit 'eac77fcd56fc2a3391f0d86faf54302afb368ff7'
[20:46] <cone-109> ffmpeg.git 03Diego Biurrun 07master:d445c865f87e: fft-test: Reduce variable scope where sensibly possible
[20:46] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:09f5b3750f2d: Merge commit 'd445c865f87e7f13230e4dac3f77f5af21da95cb'
[20:53] <cone-109> ffmpeg.git 03Diego Biurrun 07master:7288b3458507: fft-test: Drop unnecessary pointer indirection for context structs
[20:53] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:088875b6b855: Merge commit '7288b345850792430302a8f85a4b29140b770497'
[21:04] <cone-109> ffmpeg.git 03Christophe Gisquet 07master:ca081217cda8: hevcdsp: change types of SAO parameters
[21:42] <cone-109> ffmpeg.git 03Christophe Gisquet 07master:373677f95816: hevc_mvs: avoid deriving tmvp in amvp
[22:18] <llogan> ubitux: done
[22:18] <ubitux> thx
[22:19] <llogan> damn, i should have added an image from mandelbrot filter
[22:25] <cone-109> ffmpeg.git 03Christophe Gisquet 07master:670b7f203ad2: x86: hevcdsp: align
[23:35] <BBB> kurosu_: so & Im wondering how & portable this new threading function really is
[23:35] <BBB> kurosu_: do you epect other codecs to use it?
[23:35] <BBB> kurosu_: like, realistically, not theoretically
[23:52] <Orix> hey guys, i have a decent movie loader rendering right now, but I'm currently experiencing an "issue". I noticed that one of the movies i'm trying to render is grayscale, how do i properly convert that into RGB32? I currently see a green overlay
[23:54] <michaelni> Orix, libswscale or libavfilters scale filter (which uses libswscale)
[23:55] <michaelni> or well, if you just need gray ->rgb32 you can also write your own converter
[23:55] <Orix> ah i see, and in swscale swap the pixel format for the source depending on what the file is?
[23:55] <michaelni> yes
[23:55] <Orix> makes sense
[23:55] <Orix> thanks
[23:59] <Orix> imageConvertContext = sws_getCachedContext(nullptr, mCodecContext->width, mCodecContext->height, mCodecContext->pix_fmt, mWidth, mHeight, AV_PIX_FMT_RGB32, SWS_FAST_BILINEAR | SWS_CPU_CAPS_MMX2, nullptr, nullptr, nullptr);
[23:59] <Orix> that is what I currently have, but I'm still seeing the green overlay on the grayscale video
[00:00] --- Thu Jul 24 2014