[Ffmpeg-devel-irc] ffmpeg-devel.log.20131014

[burek]

[00:07] <BBB> ubitux: no, you can mathematically prove that it's the same thing - try it :)
[00:08] <ubitux> ok
[00:08] <BBB> a*b+c>>d is the same as a*(b<<1)+(c<<1)>>d>>1
[00:09] <BBB> with brackets where appropriate, but you know what I mean
[00:09] <BBB> do you want me to upload my current working version of 32x32?
[00:09] <ubitux> no, don't spoil me the fun
[00:09] <ubitux> :)
[00:09] <BBB> lol
[00:09] <BBB> glad you're liking it
[00:09] <BBB> it's certainly more fun to figure out yourself
[00:10] <ubitux> i want to struggle a bit more before i get more hints
[00:10] <BBB> feel free to send me code if there's something you'd like me to review or comment on
[00:10] <ubitux> ok
[00:10] <BBB> sgtm
[00:11] <cone-735> ffmpeg.git 03Derek Buitenhuis 07master:eb90a2091ffb: pthread: Fix deadlock during thread initialization
[00:11] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:005200887b2c: Merge commit 'eb90a2091ffb94d8c29aaa5ff50f4192520254fc'
[00:22] <ubitux> BBB: pmaddwd is cool, but there is no pmsubwd :(
[00:36] <cone-735> ffmpeg.git 03Martin Storsjö 07master:eb8b05a3824a: http: Add an option for forcing basic authentication
[00:36] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:5dee3e8bb45d: Merge commit 'eb8b05a3824a9fa85e20d603595ac8a3b83505d4'
[00:42] <BBB> ubitux: make the multiplier negative
[00:42] <BBB> ubitux: the multiplication is signed
[00:42] <BBB> (awh I just gave away the trick, now you didn't have to suffer)
[00:42] <BBB> (anyway...)
[01:41] <cone-735> ffmpeg.git 03Martin Storsjö 07master:71549a857b13: http: Support auth method detection for POST
[01:41] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:6b1b63df85ff: Merge commit '71549a857b13edf4c4f95037de6ed5bb4c4bd4af'
[01:42] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:5970f4bb027c: avformat/http: check the auth string contents not the pointer which cannot be NULL
[01:58] <michaelni> wbs, if you improve or change a patch taken from ffmpeg-devel, please post or CC changed versions to ffmpeg-devel, this decreases the chance of bugs being missed and gives all a chance to comment 
[02:09] <cone-735> ffmpeg.git 03Luca Barbato 07master:14ddbb477fae: cavs: K&R formatting cosmetics
[02:10] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:fb8a10db5d77: Merge commit '14ddbb477faef359983151b763fd8b20e578651b'
[02:16] <cone-735> ffmpeg.git 03Luca Barbato 07master:1b20d0f581f0: cavs: Return meaningful error values
[02:16] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:d794b7db1453: Merge commit '1b20d0f581f01f2df601c9e68d0d321672d97af7'
[11:25] <cone-360> ffmpeg.git 03Luca Barbato 07master:39185ec4faa9: cavs: Check for negative cbp
[11:25] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:8d9f08ef32f1: Merge remote-tracking branch 'qatar/master'
[13:22] <cone-360> ffmpeg.git 03James Almer 07master:aae8975ffbb5: oggparseopus: use ff_alloc_extradata()
[13:22] <cone-360> ffmpeg.git 03James Almer 07master:00408f95e7fa: oggparsecelt: use ff_alloc_extradata()
[13:22] <cone-360> ffmpeg.git 03James Almer 07master:1d4476d5dab1: movenc: use ff_alloc_extradata()
[13:41] <smjd> ubitux: http://trac.ffmpeg.org/ticket/2936#comment:9 it's a regression because I can't use an external tool like Gifsicle to join single GIF images to an animation, to get an optimal palette instead of the generic one
[13:42] <ubitux> ok
[13:51] <ubitux> smjd: and gifsicle can't take png as input?
[13:53] <smjd> nope
[13:55] <ubitux> only gif?
[13:56] <smjd> "gifsicle: image.png: file not in GIF format"
[13:56] <Daemon404> eh
[13:56] <Daemon404> use somthing better
[13:56] <Daemon404> like imagemagick
[13:57] <smjd> I tried ImageMagicks layer optimization on a file ffmpeg created and output was about 50% larger
[13:58] <ubitux> :)
[13:58] <Daemon404> ive only every use it via its C api
[13:58] <Daemon404> it could also be the difference in dither quality
[13:58] <ubitux> smjd: i won't have time to make that feature soon, are you willing to work on it?
[14:01] <smjd> I don't think I have good enough programming skills yet
[14:02] <smjd> but I'll take a look
[15:05] <Compn> where did i see those unroll every loop instruction
[15:05] <Compn> seems like some programming guidelines
[15:05] <Daemon404> maybe from 1997
[15:06] <Daemon404> you dont need to manually unroll too many things in 2013
[15:07] <Compn> good point
[15:07] <Compn> seems like an old cpu problem
[15:07] <Compn> now no one cares about bloated operations :P
[15:08] <Daemon404> Compn, more like compilers arent as shit
[15:08] <Daemon404> still shit
[15:08] <Daemon404> but not as much.
[15:09] <Compn> talking about gcc 2.95? :)
[15:09] <Compn> ehe
[15:10] <Daemon404> the holy 2.95
[15:11] <Compn> maybe we've gone too far in a few places
[15:11] <Compn> with the 2.95
[15:12] <Daemon404> ive had changes rejected because they broke 2.95 before
[15:12] <Daemon404> ;)
[15:29] <michaelni> only 2.95 ?
[15:29] <Daemon404> yeah
[15:29] <Daemon404> it was a 2.95 bug
[15:29] <Daemon404> it was ages ago, and mru flamed me
[15:36] <cone-360> ffmpeg.git 03Martin Storsjö 07master:84a125c4c28f: rtmp: Allocate the prev_pkt arrays dynamically
[15:36] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:953dd72321b7: Merge commit '84a125c4c28f3e3e215d2e6c32f7f0ec43bbc04c'
[15:48] <cone-360> ffmpeg.git 03Derek Buitenhuis 07master:15748773bf33: avresample/x86: Switch operand order for mulps
[15:48] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:12ca544aa0fc: Merge commit '15748773bf33c110e6e2e9526c7ba5478274c74c'
[15:50] <Compn> of course, you could say, devotion to 2.95 ensured our compliance and easy transistion to newer compilers (icc, clang, etc)
[15:50] <Compn> and of course, breaking those newer compilers :)
[15:53] <cone-360> ffmpeg.git 03Henrik Gramner 07master:c108ba0175d4: x86inc: Use VEX-encoded instructions in AVX functions
[15:53] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:12e4493f9c14: Merge commit 'c108ba0175d4fc3a3253a8b0f782fbfb96ba5098'
[16:05] <cone-360> ffmpeg.git 03Derek Buitenhuis 07master:206895708ea2: x86inc: Remove our FMA4 support
[16:05] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:9ac124c889a6: Merge commit '206895708ea2b464755d340e44501daf9a07c310'
[16:08] <saste> what's the name of the function which counts bits in a number?
[16:09] <saste> av_popcount
[16:12] <cone-360> ffmpeg.git 03Jason Garrett-Glaser 07master:c6908d6b4b37: x86inc: FMA3/4 Support
[16:12] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:e3e0e3d0c913: Merge commit 'c6908d6b4b377a04a5d055ba874bdbcf06c80497'
[16:21] <cone-360> ffmpeg.git 03Jason Garrett-Glaser 07master:a3fabc6cb389: x86: more AVX2 framework
[16:22] <cone-360> ffmpeg.git 03Michael Niedermayer 07master:f9bef2bec9dc: Merge remote-tracking branch 'qatar/master'
[16:31] <cone-360> ffmpeg.git 03Carl Eugen Hoyos 07master:f7ed044eeaf3: Support H.264 fourcc UMSV.
[16:42] <Daemon404> oh
[16:42] <Daemon404> TheRyuu, 
[16:42] <Daemon404> see [15:42] < fflogger> [editedticket] cehoyos: Ticket #3048 (ICL and HAVE_INLINE_ASM) updated https://trac.ffmpeg.org/ticket/3048#comment:1
[16:43] <Daemon404> and of course
[16:43] <Daemon404> carl's response is as uninformed as usual.
[18:54] <cone-360> ffmpeg.git 03Vignesh Venkatasubramanian 07master:0f99aad80f65: lavc: Adding seek_preroll to AVCodecContext
[18:57] <TheRyuu> Daemon404: that patch doesn't look complete to me
[19:45] <Daemon404> TheRyuu, its not
[19:50] <beastd> hi
[19:52] <durandal_1707> good morning
[20:31] <cone-360> ffmpeg.git 03Paul B Mahol 07master:42a8d8aee858: avformat/rpl: use avpriv_report_missing_feature/avpriv_request_sample
[21:58] <cone-360> ffmpeg.git 03Vignesh Venkatasubramanian 07master:d6f86d74edfa: matroskadec: Demux support for SeekPreRoll and CodecDelay
[22:11] <michaelni> ubitux, do you want to review "[FFmpeg-devel] [PATCH 4/6] lavc: support multiple frames in a singe AVPacket in avcodec_decode_subtitle2"
[22:11] <michaelni> and the other subtitle related patches in that thread ?
[22:35] <llogan> durandal11707: "Broken." Providing no comment would have been better than that.
[22:41] <michaelni> mraulet, did you get a chance to talk with elenril and could you find a consensus ?
[22:47] <mraulet> well I think smarter did answer for us
[22:54] <michaelni> mraulet, ohh, ok, understood
[22:55] <durandal11707> llogan: i'm guilty
[23:10] <saste> BBB: That makes my hobbyist life a lot easier
[23:10] <saste> :D
[23:12] <cone-360> ffmpeg.git 03Lou Logan 07master:a06dcde5073b: doc: make x11grab examples consistent with option names
[23:22] <llogan> saste: thanks
[23:22] <saste> llogan, you lazy ;)
[23:22] <llogan> and terrible at git
[23:32] <iive> so, what is the fate of the hevc decoder?
[23:53] <saste> iive, we wait
[23:53] <saste> we lost the penis waving contest
[23:53] <iive> we wait for what? libav to finally merge it?
[23:54] <saste> iive, don't listen to me, i'm just saying silly things
[23:54] <michaelni> mraulet, smarter, I understand you are not opposed to the merge, and i thank you for that :) but id like to do exactly what you/ the authors prefer, its your code and i dont want to do something that you guys dont want, thus a clear "merge it" vs. "wait" would be ideal. 
[23:58] <mraulet> there is no "clear" wait no "clear" merge
[00:00] --- Tue Oct 15 2013