[Ffmpeg-devel-irc] ffmpeg-devel.log.20140620

burek burek021 at gmail.com
Sat Jun 21 02:05:02 CEST 2014


[00:02] <cone-797> ffmpeg.git 03Vittorio Giovara 07master:440842c4eb1d: stereo3d: add missing include guards
[00:02] <cone-797> ffmpeg.git 03Michael Niedermayer 07master:b05d50898f5c: Merge commit '440842c4eb1d7709654ec97cd687663d11ef499c'
[00:05] <J_Darnley> Oh my.  I didn't realise doing a frequency transform would be this complicated.
[00:06] <BBB> I love compilers
[00:06] <BBB> ffmpeg_g[0x10082e87c]:  movss  %xmm0, -0x4(%rsp)
[00:06] <BBB> ffmpeg_g[0x10082e882]:  movss  %xmm2, -0x8(%rsp)
[00:06] <BBB> ffmpeg_g[0x10082e888]:  movss  -0x8(%rsp), %xmm0
[00:06] <BBB> ffmpeg_g[0x10082e88e]:  movss  -0x4(%rsp), %xmm1
[00:06] <BBB> ...
[00:13] <ubitux> must check if the stack is readable and writable regularly
[00:14] <nevcairiel> it does kinda flip the values around a bit
[00:14] <nevcairiel> for whatever reason
[00:14] <cone-797> ffmpeg.git 03Anton Khirnov 07release/2.2:6672f672d98c: avconv: make -shortest work with streamcopy
[00:14] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/2.2:220bbc44c2cf: Merge commit '6672f672d98cf4887e13321b6f67682cfd1273b4' into release/2.2
[00:14] Action: ubitux missed that
[00:59] <cone-797> ffmpeg.git 03Anton Khirnov 07release/1.1:9455a023be9f: matroskaenc: do not write negative timestamps
[00:59] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/1.1:6af1793dd5a2: Merge commit '9455a023be9f3915ccf5511a0b8fdb5b8897b2b6' into release/1.1
[01:10] <cone-797> ffmpeg.git 03Anton Khirnov 07release/1.1:e0fcad77618a: avconv: make -shortest work with streamcopy
[01:10] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/1.1:45283653a30e: Merge commit 'e0fcad77618a0455ca9c2451ea0aa538597a08c0' into release/1.1
[01:21] <cone-797> ffmpeg.git 03Anton Khirnov 07release/1.1:ba8ab4e7ae01: avconv: do not send non-monotonous DTS to the muxers.
[01:22] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/1.1:62899d79a380: Merge commit 'ba8ab4e7ae016cf970b0f335a7933f2db53784bc' into release/1.1
[01:28] <BBB> now lets kill inline avx
[01:31] <Daemon404> 30
[01:32] <cone-797> ffmpeg.git 03Sean McGovern 07release/1.1:4310ba273d3b: sgidec: fix an incorrect backport
[01:32] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/1.1:724deeb7a8dc: Merge commit '4310ba273d3bcb7f19a4e40b55b9e832568e56e5' into release/1.1
[01:36] <J_Darnley> BBB: is inline avx "more bad" than other asm?
[01:37] <J_Darnley> inline asm I mean
[01:38] <cone-797> ffmpeg.git 03Sean McGovern 07release/0.10:9c7321e2b898: sgidec: fix an incorrect backport
[01:38] <cone-797> ffmpeg.git 03Michael Niedermayer 07release/0.10:c1ac5896ef9a: Merge commit '9c7321e2b8981ec867294309e9cf3833055df78f' into release/0.10
[01:54] <BBB> J_Darnley: yes!
[02:04] <cone-797> ffmpeg.git 03Carl Eugen Hoyos 07release/1.2:3ac85d70f09d: Revert two backports to release/1.2 that broke fate.
[02:41] <BBB> I guess I gotta find a host to test speed now
[02:41] <BBB> or does someone else want to do that?
[02:41] <BBB> jamrial: maybe you can test speed?
[02:42] <jamrial> alright. although last time my results were very different than michaelni's
[02:43] <jamrial> so one of us did something wrong
[02:43] <jamrial> probably me
[02:43] <jamrial> so i'd recommend you testing on your own as well to cross check resutls :P
[02:50] <BBB> ok
[02:50] <BBB> just make sure you check a long music sample and you do some per-sample correction or so
[02:50] <BBB> otherwise frame length becomes the defining factor
[02:50] <BBB> anyway
[02:50] <BBB> just do what you feel is right
[03:04] <jamrial> i'll add timer.h macros around the c->dsp.resample_{common,linear} calls (no need to bench swri_resample() as a whole since the call overhead didn't change with this patch)
[03:04] <jamrial> how do i do per-sample correction, though?
[03:18] <BBB> ignore for now
[03:20] <BBB> reason is that if you resample a few frames of audio, the number of samples processed per resample call is not the same
[03:20] <BBB> so cycle counts vary hugely
[03:20] <BBB> but thats the case for before and after so i guess itsn not a major issue
[03:20] <BBB> just make sure number of skips didnt change too much
[03:22] <jamrial> ok
[03:26] <BBB> ty btw :)
[03:29] <Timothy_Gu> #3649 what the heck?!
[03:29] <Timothy_Gu> ./version.sh: !: not found
[03:57] <jamrial> i'm getting a sigsegv on win64 with "movss         [dstq], xm0" (end of the horizonal sum). This is sse, not avx
[03:57] <jamrial> let me check on a linux x64 vm
[04:10] <BBB> linux64 works for me
[04:11] <BBB> mac64, same
[04:11] <BBB> win64 probably has bugs, I couldnt test that one :)
[04:15] <jamrial> yeah, works on linux64
[04:19] <BBB> weird
[04:20] <BBB> does the first loop cycle crash, or only after a couple?
[04:47] <jamrial> after a couple .loop cycles
[04:55] <jamrial> or several, rather
[06:01] <cone-232> ffmpeg.git 03Timothy Gu 07master:cc0057a31c70: tests/fate.sh: report different status for different errors
[06:10] <llogan> almost done encoding ~1000 hrs of video
[06:11] <drv> nyan cat ultimate edition?
[06:12] <llogan> a boring government contract
[07:21] <llogan> AMD A10-5800K is slow
[07:22] <llogan> just an idle server that i conscripted into encoding
[07:53] <ubitux> should i wait for reviews on hqx?
[08:03] <llogan> you already did
[09:11] <ubitux> http://lav.io/2014/06/videogrep-automatic-supercuts-with-python/ cool usage of ffmpeg
[09:12] <plepere> the only feedback for my avx2 patch is the SECTION_RODATA 32 ?
[09:13] <ubitux> check if that's necessary for the text section as well maybe
[09:14] <plepere> well things work for me without it. how do I force a crash if it's not there ?
[09:15] <ubitux> no idea :)
[09:15] <plepere> so I need SECTION .text 32 ?
[09:17] <ubitux> there is a SECTION_TEXT in one of the x264 header
[09:20] <plepere> I'll simply put this in the ML and if there's a problem, someone  will say it
[10:16] <zfox> with ffmpeg 2.0, converting a AAC audio to PCM produced will product a lot of  noise, when resampling from AV_SAMPLE_FMT_FLTP to AV_SAMPLE_FMT_S16 . my code such as http://codepad.org/gsJU4Ydp
[11:40] <zfox> when i use avresample_convert or swr_convert to convert AV_SAMPLE_FMT_FLTP to AV_SAMPLE_FMT_S16 of a 2 channels,44100 audio,it always has some noise. how to resolve this problem ? the code http://codepad.org/qoknuYAB
[12:11] <anshul> is subtitle pts are in 1/1000000 timebase  ?
[12:23] <cone-296> ffmpeg.git 03Stefano Sabatini 07master:8c443767995d: doc/codecs: extend documentation for thread_type option
[12:23] <cone-296> ffmpeg.git 03Stefano Sabatini 07master:d381109ce8ac: lavc/h264_slice: clarify informational message in case of disabled slice multi-threaded decoding
[12:57] <J_Darnley> So, I've got a few questions about using libavcodec's avfft.h frequency transforms...
[12:58] <J_Darnley> Are all 3 (fft, rdft, dct) suitable for use on audio?
[12:59] <J_Darnley> Which one is fastest?  I don't need a lot of quality when I put it into 8-bit samples.
[13:01] <J_Darnley> How can I know the output range of values?
[13:11] <BBB> plepere: alignment is random multiple of 16
[13:11] <BBB> plepere: so in one of 2 cases, its 32
[13:11] <BBB> plepere: so its not a surprise it doesnt crash for 1 of 2 people here
[13:11] <BBB> (who have avx2 capable machines)
[13:13] <plepere> I don't have an AVX2 capable machine. I'm using the intel SDE to emulate it
[13:13] <BBB> well, same thing I guess
[13:13] <plepere> maybe, yes
[13:35] <plepere> ubitux, how do you do a !avx-enabled in preprocessor ?
[13:37] <plepere> to do %if (%1 > 8 && !avx_enabled) || (%1 > 16)
[13:47] <BBB> notcpuflag(avx)
[13:48] <plepere> thanks BBB
[13:48] <michaelni> J_Darnley, all the transforms are fft based so their spee should be similar
[13:49] <plepere> I did an ugly if to assign
[13:50] <plepere> BBB : but does notcpuflag(avx) return 1 if the CPU is avx capable but the function is in a INIT_XMM sse2 ?
[13:50] <BBB> plepere: those are all preprocessor checks
[13:51] <BBB> INIT_XMM sse2 means notcpuflag(avx) returns 1 and cpuflag(avx) returns 0
[13:51] <BBB> INIT_XMM avx is the inverse
[13:51] <plepere> ok thanks
[13:51] <BBB> they are not runtime checks, I mean, its a preprocessor after all
[13:53] <plepere> ok thanks. just making sure. :)
[13:54] <plepere> it could be at compile check.
[13:57] <BBB> is someone interested in fixing my swr patch for win64?
[13:57] <BBB> I dont have a win64 machine so cant test
[14:01] <J_Darnley> BBB: I can look unless you also need avx
[14:12] <Daemon404> ffs that bug hilights me
[14:13] <BBB> J_Darnley: no, sse is fine
[14:13] <BBB> Daemon404: happens to me all the time when people talk about big buck bunny
[14:16] <Daemon404> lol
[14:19] <kierank> lol
[14:23] <ubitux> BBB: and beaglebone black? :)
[14:41] <ubitux> hey saste 
[14:41] <ubitux> saste: interested in having a look to hqx before i push?
[14:41] <saste> ubitux: yes
[14:41] <ubitux> ok, i'll give you about 3 minutes
[14:41] <ubitux> please hurry up
[14:41] <saste> give me 24 hours, if I won't review it feel free to push
[14:42] <ubitux> ok :)
[14:42] <saste> i'm debugging a curious issue involving several udp sources
[14:43] <ubitux> saste: you're back with us? :)
[14:43] <ubitux> it's been a long time :)
[14:43] <saste> ubitux, i hope so
[14:43] <ubitux> cool
[14:47] <cone-296> ffmpeg.git 03Timothy Gu 07master:6fdd19cbb26c: doc: remove RELEASE_NOTES
[15:25] Action: Daemon404 loves it when thunderbird screws up and says he has 23k new mails in ffmpeg-devel
[15:27] <ubitux> "oh btw, i forgot to tell you"
[15:27] <ubitux> "here are thousands of mails"
[15:34] <wm4> is there a non-shitty GUI mail client yet
[15:34] <Daemon404> is there a nonshitty mail client full stop?
[15:34] <Daemon404> ubitux, more like it reindexes half the folder on my drive
[15:34] <Daemon404> because it screws up
[15:35] <Daemon404> so it thinks theyre all unread
[15:35] <nevcairiel> i've seen that, it claims all mails in a folder are new because its local thing got messed up
[15:35] <Daemon404> yeah
[15:35] <J_Darnley> The quality of thunderbird is assisted by gmail's IMAP server(s)
[15:35] <Daemon404> lol
[15:35] <Daemon404> its still the least bad client ive used (after configuring it to hell)
[15:36] <nevcairiel> gmail maps how their interface works with tags to imap folders, thats just not perfect =p
[15:36] <Daemon404> my main complain about gmail is its dedup
[15:36] <J_Darnley> I'm not surprised people are always touting the next "email killer"
[15:36] <Daemon404> i have to bcc myself on ML mails or they dont show up in the folder
[15:36] <Daemon404> well it will, but only on its web interface
[15:36] <Daemon404> not useful for a mail client
[15:37] <Daemon404> there was one other client i liked, but they refuse to add a threaded view
[15:37] <nevcairiel> luckily i only use the web interface, given up on actual mail clients a long time ago
[15:37] <Daemon404> which is super shitty
[15:37] <Daemon404> all threads look flat
[15:37] <Daemon404> (copying Mail.app no doubt)
[15:38] <J_Darnley> What I dislike most about the web interface is that you need to be logged into google to use it.
[15:39] <nevcairiel> you need to be logged into the webmail you're using, shocker :D
[15:39] <Daemon404> no he means all of google
[15:39] <Daemon404> youtube, search, g+
[15:39] <Daemon404> just lets them track you even better
[15:39] <nevcairiel> youtube is separate, fwiw
[15:40] <Timothy_Gu> i dont think so
[15:40] <Daemon404> it is not
[15:40] <Daemon404> nto anymore
[15:40] <Daemon404> they merged
[15:40] <nevcairiel> i'm logged into gmail but not youtube
[15:40] <nevcairiel> so yes.
[15:40] <Daemon404> there was a whole shitstorm
[15:40] <Daemon404> about g+
[15:40] <Daemon404> i see
[15:40] <nevcairiel> they share accounts sure, but not sso
[15:40] <Daemon404> right ok
[15:40] <Daemon404> search shares
[15:40] <Daemon404> as does g+
[15:40] <Daemon404> i know that for sure
[15:41] <nevcairiel> everything officially on a google.com domain does
[15:41] <Daemon404> i would just switch to my own mail server, but i have too much history using my gmail account
[15:41] <Daemon404> in git
[15:41] <J_Darnley> At least they still (or did last time I had to use it) the classic HTML only interface
[15:41] <Daemon404> and stuff
[15:42] <nevcairiel> running your own mail server properly is also a serious PITA
[15:42] <Daemon404> it is
[15:42] <Daemon404> except there's an easy solution now
[15:42] <nevcairiel> i still maintain one, but i've been thinking of giving it up
[15:42] <Daemon404> (zimbra)
[15:42] <J_Darnley> I have been tempted to try Fastmail and finally leave the googleplex behind
[15:43] <nevcairiel> zimbra looks like an entire social-crapload-solution
[15:43] <Daemon404> or you could always set up your domain to use gmail
[15:43] <nevcairiel> unless there is multiple things with that name
[15:43] <Timothy_Gu> speaking of email, does anyone know how x264 changed all the "Jason Garrett-Glaser" to "Fiona Glaser" in the history and in the changes themselves, without changing hashes?
[15:43] <Daemon404> nevcairiel, its basically a vm image
[15:43] <Daemon404> with a web ui
[15:43] <kierank> 2:34 PM <"Daemon404> is there a nonshitty mail client full stop? --> outlook
[15:43] <nevcairiel> Timothy_Gu: they changed hashes
[15:43] <J_Darnley> They did change hashes
[15:43] <Daemon404> kierank, outlook has its own problems
[15:43] <kierank> such as being too good perhaps
[15:43] <Daemon404> i could not figure out how to get outlook to send non-flowed text emails
[15:44] <Daemon404> or "running on not-windows"
[15:44] <Daemon404> i run thunderbird on 3 oses
[15:44] <kierank> outlook is the reason i use windows
[15:44] <kierank> Daemon404: unified calendar and email
[15:44] <kierank> nothing does it better
[15:44] <Daemon404> i dont use the calende
[15:44] <Daemon404> ri fucking hate calendar shit
[15:44] <nevcairiel> i've always kinda liked outlook, but i still gave up on mail clients at some point
[15:44] <Daemon404> i*
[15:44] <kierank> 99% of people do
[15:44] <kierank> gmail can't do BST properly
[15:44] <Daemon404> kierank, linux desktop is the reason i use windows for day-to-day
[15:45] <Daemon404> windows desktop is the reason i use linux desktop for dve work
[15:45] <Daemon404> :V
[15:45] <Timothy_Gu> really? i found a hash on an old website and it still works
[15:45] <Daemon404> kierank, rhel fucked up BST too until i reconfigured by tz
[15:45] <nevcairiel> Timothy_Gu: does it also show the new name?
[15:45] <Daemon404> via tzconfig
[15:45] <Daemon404> or w/e
[15:45] <Timothy_Gu> yeah
[15:45] <nevcairiel> did it silently redirect you? :p
[15:45] <Timothy_Gu> also why does he change the name
[15:46] <Timothy_Gu> maybe ;)
[15:46] <Daemon404> kierank, UK customs (via parcel force) is satan
[15:46] <Timothy_Gu> kierank: even my mom gave up on Outlook :)
[15:46] <Daemon404> worst customs ive dealt with
[15:47] <kierank> Daemon404: yeah parcelforce are shit
[15:47] <kierank> i have a parcel stuck somewhere
[15:47] <kierank> right now
[15:47] <kierank> ups/dhl is better
[15:47] <Daemon404> i have 2 parcels stuck
[15:47] <Daemon404> :/
[15:47] <Daemon404> i would use ups or dhl or fedex
[15:47] <Daemon404> but the seller only uses usps
[15:47] <Daemon404> due to being in rural idaho
[15:48] <Daemon404> and that ends up with parcel force
[15:48] <Daemon404> rural idaho -> london = 3days
[15:48] <Daemon404> london/customs -> me = inb4 3 weeks
[15:48] <kierank> have you paid the monies?
[15:48] <Daemon404> no
[15:48] <Daemon404> they havent sent me a letter yet
[15:48] <kierank> lol
[15:49] <Daemon404> it sits i the customs backlog, then they have to physically post me a letter
[15:49] <kierank> then they are deliberating how much it's cost
[15:49] <Daemon404> with a new separate reference number
[15:49] <Daemon404> which is not teh tracking number
[15:49] <Daemon404> which i enter online, and pay
[15:49] <Daemon404> why the shit cant i pay via tracking no?!
[15:49] <kierank> you need a parcel reference number
[15:49] <kierank> becaus eit's not the same as the tracking number
[15:49] <Daemon404> but *WHY*
[15:50] <Daemon404> wouldnt it make more sense to let me pay ahead of time via tracking no
[15:50] <Daemon404> canada post always let me do that
[15:51] <kierank> because you don't know what the customs fees are ahead of time
[15:51] <kierank> ebay has a service now though that pays for you and charges you iirc
[15:52] <Daemon404> kierank, so update teh db
[15:52] <Daemon404> and support it via tracking number
[15:52] <Daemon404> there is no reason it needs a new number
[15:52] <Daemon404> it's a political problem i guess
[15:54] <Daemon404> kierank, my main gripe is: £13.50 "processing fee"
[15:54] <Daemon404> from parcel force
[15:54] <Daemon404> in adition to VAT
[15:54] <Daemon404> which is just bs
[15:54] <kierank> yeah that's fucking annoying because i can't claim 13.50 back
[15:54] <Daemon404> yeah
[15:55] <kierank> the fedex service is slick
[15:55] <kierank> it automatically gets your vat number
[15:56] <Daemon404> fedex let me pay after delivery
[15:56] <Daemon404> so there was no wait
[15:57] <Daemon404> but at teh same time i had to phone them and yell at them
[15:57] <Daemon404> because they charged me VAT on a package clearly labelled personal belongings
[15:57] <Daemon404> when i moved
[15:57] <Daemon404> in the end they refused to believe me
[15:58] <kierank> yeah the equipment I get sent clearly states that it belongs to the sender
[15:58] <kierank> but i still get charged fat
[15:58] <kierank> vat
[15:58] <nevcairiel> they are stupid like that
[15:59] <nevcairiel> i did a bunch of technical reviews for a book a couple years back and got send like 3 copies after it was published, and they charged me full vat .. for free books!
[16:00] <JEEB> yeah, here too gifts are taxed by the selling price
[16:01] <ubitux> anyone has a pdf or sth of http://www.amazon.com/JPEG-Compression-Standard-Multimedia-Standards/dp/0442012721 ?
[16:01] <ubitux> the price is so indecent it makes me really mad
[16:01] <JEEB> library genesis?
[16:01] <ubitux> more than $170 for a book written in 1992, wth seriously
[16:01] <nevcairiel> and thats already marked down from list price 229!
[16:02] <Daemon404> ubitux, i can tell it was from 1992 -- look at that sexy cover art
[16:02] <nevcairiel> ubitux: german amazon has a bunch of "used" copies of that for as low as ¬16
[16:03] <nevcairiel> guess US amazon does as well
[16:03] <JEEBsv> ubitux: library genesis /seems/ to have that one
[16:04] <JEEBsv> oh, not exactly that one
[16:04] <Daemon404> http://www.openisbn.com/preview/0442012721/
[16:04] <Daemon404> whole book is readbale online here
[16:04] <Daemon404> although as jpegs.
[16:05] <ubitux> ah, thanks
[16:07] <Daemon404> ubitux, entire think is on google books too
[16:07] <Daemon404> which has a slightly less bad interface
[16:08] <Daemon404> books.google.co.uk/books?isbn=0442012721
[16:08] <ubitux> it looks like a google book wrapper
[16:08] <Daemon404> yea
[16:08] <J_Darnley> BBB: I the first thing I see that might be a problem are that you probably need to sign or zero extend some of the integers you are loading
[16:08] <Daemon404> ubitux, i dont think it was ever digitized
[16:08] <Daemon404> as in OCRd
[16:08] <Daemon404> why are you looking at a book about jpeg anyway
[16:08] <ubitux> that's fine
[16:09] <ubitux> i'm looking for a particular section
[16:09] <Daemon404> anything interesting?
[16:09] <ubitux> i don't know yet
[16:09] <Daemon404> o ok
[16:12] <ubitux> Daemon404: it's not actually complete
[16:13] <ubitux> there are only a few pages
[16:13] <Daemon404> damn
[16:38] <cone-296> ffmpeg.git 03James Almer 07master:e93028790087: configure: remove superfluous check for AVX/FMA3 support in Yasm
[16:59] <cone-296> ffmpeg.git 03Anton Khirnov 07master:874390e16342: lavc: add a convenience function for rescaling timestamps in a packet
[16:59] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:4a993335812f: Merge commit '874390e163427c1fe7682ab27924a7843780dbb3'
[17:13] <cone-296> ffmpeg.git 03Anton Khirnov 07master:452860d7e056: Use av_packet_rescale_ts() to simplify code.
[17:13] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:afa7ab557a80: Merge commit '452860d7e056f9a894cac2fe52cdedeae4c38251'
[17:58] <cone-296> ffmpeg.git 03Anton Khirnov 07master:778111592bf5: dvenc: initialize the profile only once, at init
[17:59] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:c2b1225c16f1: Merge commit '778111592bf5f38630858ee6dfcfd097cd6c6da9'
[18:10] <cone-296> ffmpeg.git 03Anton Khirnov 07master:650dee63c8b1: dv: get rid of global non-const tables
[18:10] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:b0fdc76600a1: Merge commit '650dee63c8b1e6693c6cf5983f4a5ed3f571379f'
[18:15] <cone-296> ffmpeg.git 03Anton Khirnov 07master:d349afb07bac: dv: cosmetics, reindent
[18:15] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:0a995311ceeb: Merge commit 'd349afb07bacccb62eb5369c38d6406d2909d792'
[18:34] <jamrial> BBB: speed seems to be much worse here. timer.h macros show it's almost two times slower, and decoding time goes from 11 to 13 seconds using the same 22 minutes long song
[19:23] <llogan> J_Darnley: i moved from my own mail server to fastmail and i've been happy with it so far
[19:42] <kriegerod> if anybody is willing to help or consult on #3302 please let me know
[20:16] <llogan> kriegerod: patches get more love if you send them to ffmpeg-devel (but i'm not sure if you did or not).
[20:23] <cone-296> ffmpeg.git 03Moritz Barsnick 07master:10b8481a2634: fix several typos
[21:52] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:40cbad9ff7e0: avcodec/wavpackenc: fix () in macros
[21:52] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:33f83a215774: avcodec/x86/rv40dsp_init: fix () in macros
[22:00] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:329898aa45f5: avcodec/bitstream: try to workaround internal compiler bug in gcc 4.2
[22:11] <jamrial> BBB: nevermind, just tested win32 on that same machine (although native instead of a vm like the linux x64 above), and linux x64 on another machine and in both cases it's only slightly slower than inline
[22:11] <jamrial> no idea wtf happened with the results above
[23:17] <cone-296> ffmpeg.git 03Justin Ruggles 07master:9e500efdbe0d: Add av_image_check_sar() and use it to validate SAR
[23:17] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:0dceefc5fa81: Merge commit '9e500efdbe0deeff1602500ebc229a0a6b6bb1a2'
[23:17] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:fe43facd659f: avcodec/utils: check sar only when width/height are set
[23:33] <cone-296> ffmpeg.git 03Vittorio Giovara 07master:ca1e36a8e4cd: h264: fix build when error resilience is disabled
[23:33] <cone-296> ffmpeg.git 03Michael Niedermayer 07master:4667116c394b: Merge commit 'ca1e36a8e4cd416142487071dbca734567bdaddf'
[23:46] <llogan> i'm off IRC for a month (but i'll have email access once a day or so). going to do some non-computer work. see you guys later.
[23:49] <Compnn> non computer work?
[23:50] <Compnn> lou is working on a farm?
[23:50] <Compnn> green acres !
[00:00] --- Sat Jun 21 2014


More information about the Ffmpeg-devel-irc mailing list