[Ffmpeg-devel-irc] ffmpeg-devel.log.20150413

[burek]

[00:08:18 CEST] <BBB> it is
[00:08:20 CEST] <BBB> but less
[00:08:46 CEST] <BBB> x86-64 allows you to use rip as a register for addressing, whereas on x86-32, you needed to load the ip in an explicit register per function
[00:08:56 CEST] <BBB> however, it still means you need the register in addresses
[00:09:23 CEST] <BBB> so if you have an expression [base+4*offset], rip doesnt fit in
[00:09:28 CEST] <BBB> and you still need an extra instruction
[00:09:42 CEST] <BBB> (re: to iive, rcombs and daemon404)
[00:13:58 CEST] <iive> isn't the whole point that the base is relative to the rip? and the linker calculates that?
[00:16:57 CEST] <iive> it seems like not.
[00:22:05 CEST] <BBB> when pic is enabled, yasm changes [base] into [base+rip]
[00:22:30 CEST] <BBB> but x86 (and that includes x86-64) allows only a base and an offset as registers in address statements like that
[00:22:54 CEST] <BBB> so if your statement has an offset as register, then it doesnt work
[01:26:20 CEST] <Timothy_Gu> BBB: the base -> base+rip transformation is `default rel` right?
[01:37:41 CEST] <jamrial> release candidate of gcc 5 is out
[01:43:20 CEST] <Timothy_Gu> yay?
[01:44:07 CEST] <cone-235> ffmpeg 03Himangi Saraogi 07master:0b72a47f5592: rtsp: Check a malloc return value
[01:44:08 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:4dd0c0712206: Merge commit '0b72a47f5592ecc035edd5bac57558ad9604612d'
[01:52:44 CEST] <cone-235> ffmpeg 03Himangi Saraogi 07master:4e5443216445: rtsp: Fix unchecked return value
[01:52:45 CEST] <cone-235> ffmpeg 03Michael Niedermayer 07master:818e8897094d: Merge commit '4e5443216445f3a9b8e6cb3fa4f448762e19006e'
[01:53:10 CEST] <BBB> Timothy_Gu: I believe so, but Im not 10000% sure
[04:11:53 CEST] <cone-235> ffmpeg 03Mark Reid 07master:c919cc61a155: tests/fate: added mxf opatom audio regression tests
[04:36:21 CEST] <cone-235> ffmpeg 03Rostislav Pehlivanov 07master:f7f71b5795d7: aacenc: Add support for Perceptual Noise Substitution energy values
[07:12:50 CEST] <rcombs> fun fact, ff_all_formats lies when it says "all"
[11:32:50 CEST] <cone-632> ffmpeg 03Rodger Combs 07master:a4d58c97d051: lavf/http: handle case where the server returns a redirect during a seek
[11:53:07 CEST] <cone-632> ffmpeg 03James Almer 07master:30ce065ffec3: fate: add tta encoder test
[12:05:46 CEST] <cone-632> ffmpeg 03Rodger Combs 07master:c69ff12dc3b4: lavf/mpeg: vobsub add an option to specify the .sub's URI
[12:26:14 CEST] <wm4> "- 1 x tablecloth: 81,61 ¬" wat
[12:26:36 CEST] <BtbN> huh?
[12:26:45 CEST] <wm4> must be some sort of tablecloth with a gold printed ffmpeg logo
[12:26:52 CEST] <BtbN> Or a hell lot of tables.
[12:27:16 CEST] <BtbN> If it was actualy printed with ffmpeg logo, the price doesn't seem too far off though.
[13:00:45 CEST] <wm4> spotted an apng in the wild, and ffmpeg is being useless: "apng: Blending with pixel format rgb24 is not implemented. Update your FFmpeg version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented."
[13:55:34 CEST] <kierank> so it looks like the web server will be the highest scoring proposal
[13:55:36 CEST] <kierank> :(
[14:03:16 CEST] <cone-632> ffmpeg 03Michael Niedermayer 07master:d5dcd94630ca: avfilter/vf_scale: Add warning for AVCOL_SPC_YCGCO
[14:03:17 CEST] <cone-632> ffmpeg 03wm4 07master:f4f3065c69ce: h264: remove bogus YCgCo warning message
[14:04:55 CEST] <kierank> lglinskih: how's it going?
[15:19:50 CEST] <kos2dk1> Looking someone for custom FFmpeg Modification, will pay. Contact me with PM
[15:22:19 CEST] <kierank> kos2dk1: very specific
[15:22:25 CEST] <kierank> you need to go into more detail
[15:36:32 CEST] <lglinskih> kierank: fine! dealing with the comments from ML 
[15:36:48 CEST] <kierank> ok
[15:36:56 CEST] <kierank> please let me know if oyu have problems
[15:38:55 CEST] <lglinskih> kierank: ok, thanks ^_^
[18:31:50 CEST] <loki_> can somebody explain me, if need set optimization options in ffmpeg configure stage?
[18:32:00 CEST] <loki_> or all we be configured best way without me?
[18:32:07 CEST] <loki_> for example -march=native
[18:32:11 CEST] <loki_> or -Ofast
[18:51:51 CEST] <wm4> loki_: no, the configure script adds more than enough rice by default
[18:52:22 CEST] <loki_> ok... because when i messed with optimization i recieved even worse result lol
[19:05:04 CEST] <jamrial> wm4: not true, configure forces -fno-tree-vectorize
[19:05:08 CEST] <jamrial> so that's a lot of rice lost
[19:06:26 CEST] <wm4> and a lot of codegen bugs?
[19:07:36 CEST] <ramiro> jamrial, gcc's tree vectorize sucks
[19:08:37 CEST] <ramiro> I remember massive slowdowns due to that (in 2010, but still =)
[19:10:17 CEST] <jamrial> well, gcc has come a long way since then
[19:10:25 CEST] <jamrial> i think it was disabled back in the gcc 4.3 days
[20:03:57 CEST] <t_than> Hello, all.
[20:04:31 CEST] <t_than> I am a student who has applied for Google Summer of Code at FFMpeg. Does anyone know if I can contact someone for some basic information?
[20:04:43 CEST] <t_than> Just to clarify some things
[20:16:58 CEST] <kierank> t_than: what would you like to know
[20:19:34 CEST] <cone-632> ffmpeg 03Vittorio Giovara 07master:73dacabfc9b9: ffv1: Check memory allocations
[20:19:35 CEST] <cone-632> ffmpeg 03Michael Niedermayer 07master:fb6dd77000f1: Merge commit '73dacabfc9b9ef1fd2c08105fdab6238ee29c2fc'
[20:19:36 CEST] <cone-632> ffmpeg 03Michael Niedermayer 07master:a3963cc8ec3b: avcodec/ffv1: use av_freep()
[20:19:37 CEST] <cone-632> ffmpeg 03Michael Niedermayer 07master:d19fc69f15c3: avcodec/ffv1: remove unneeded variable
[20:19:38 CEST] <cone-632> ffmpeg 03Michael Niedermayer 07master:4d74c8d55035: avcodec/ffv1: dont leave stale pointers in memory
[20:21:08 CEST] <t_than> hi! I was a late applicant, and I've seen that a lot of the projects/qualification tasks have been assigned to different students. I'd like to know if it's too late to start working on some qualification tasks if there are any vacancies.
[20:22:49 CEST] <t_than> You see, I got an email from another project today showing interest on my proposal, but truth be told my main interests are on video coding and I'd really prefer to be chosen by FFMpeg for GSoC. I wouldn't like my late application to seem like I wasn't interested enough.
[20:23:34 CEST] <Compn> t_than : we dont care about applications or dates, only qualifying task and the quality of your code and ability to work with us :P
[20:24:02 CEST] <Compn> but, i guess as you havent started on a task, its hard for us to see your work...
[20:24:16 CEST] <Compn> i dont know if its too late, wait around for someone more involed in gsoc
[20:24:18 CEST] <Compn> (not me)
[20:24:40 CEST] <Compn> and do not pick apng task lol
[20:26:23 CEST] <t_than> i think I've seen that on my ffmpeg-devel email subscription... :)
[20:26:38 CEST] <philipl> If you filed a GSoC application in time, then you're still eligible. But you need to finish a qualification task by the 27th, when final selections are mode.
[20:26:42 CEST] <philipl> So move fast.
[20:27:15 CEST] <michaelni> 27 is too late
[20:27:20 CEST] <philipl> is it?
[20:28:01 CEST] <t_than> but I should start on some qualification task that someone else hasn't picked yet, right? Or else, wouldn't there be a conflict?
[20:28:21 CEST] <michaelni> 24 April 07:00 UTC Student acceptance choice deadline
[20:28:42 CEST] <michaelni> that means if someone posts apatch on the 24th thats too late
[20:28:47 CEST] <philipl> Ah. It was wrong in my mind.
[20:29:03 CEST] <philipl> I hope I haven't been saying 27th to people in the past :-P
[20:30:20 CEST] <michaelni> 27 April: 19:00 UTC Accepted student proposals announced on the Google Summer of Code 2015 site.
[20:34:07 CEST] <wm4> <michaelni> 24 April 07:00 UTC Student acceptance choice deadline <- how nice, they're accounting for students working hard through the night to make the goal
[20:35:50 CEST] <BtbN> Isn't that just midnights at the google HQ?
[20:36:09 CEST] <wm4> that's a boring interpretation
[21:29:50 CEST] <kierank> Anyone know what happened to the FFmpeg dirac encoder gsoc
[21:30:09 CEST] <kierank> I seem to recall the student getting fed up with cosmetic reviews
[21:30:18 CEST] <pengvado> Timothy_Gu: (if you haven't already got your answer about PIC)
[21:30:19 CEST] <pengvado> GOT is only needed when library A calls a function that's implemented in library B.
[21:30:22 CEST] <pengvado> x86inc.asm doesn't implement GOT, because I have never used asm for that case.
[21:30:25 CEST] <pengvado> it is not needed for references within one library, since references within one library can be resolved at compile-time and thus don't leave any relocation at all in the compiled library.
[21:30:40 CEST] <Compn> kierank : theres a page that lists all of the gsoc projects and what happened to them
[21:30:42 CEST] <Compn> on the wiki
[21:32:31 CEST] <Compn> kierank : http://wiki.multimedia.cx/index.php?title=FFmpeg_/_Libav_Summer_Of_Code#Dirac_Encoder_and_Decoder
[21:32:38 CEST] <Compn> Code Status: Decoder committed to FFmpeg, the encoder still needs more work.
[21:33:20 CEST] <kierank> Thanks
[21:33:33 CEST] <kierank> I might rip the encoder out of ffmpeg
[21:34:05 CEST] <kierank> I think there was a second gsoc as well actually
[21:35:46 CEST] <Compn> people using dirac now ? :P
[21:35:49 CEST] <Compn> besides bbc
[21:35:58 CEST] <Compn> didnt bbc also abandon it
[21:50:32 CEST] <kierank> It is not abandoned any more
[21:52:54 CEST] <j-b> kierank: seriously?
[21:53:14 CEST] <kierank> Yes
[21:53:36 CEST] <j-b> kierank: they should use VLC to decode and encode :)
[21:53:41 CEST] <kierank> http://www.4rfv.com/VKIXRRF1HR2A/snell-and-quantel-demonstrate-realworld-ip-routing-workflow-interoperability-at-nab.htm
[21:53:49 CEST] <j-b> kierank: can you ask them to clarify their ogg/dirac mapping?
[21:54:01 CEST] <kierank> Yes but they are using a simple profile and encoding at a latency of a few lines
[21:54:17 CEST] <kierank> I can assure you zero people give a shit about ogg
[21:55:13 CEST] <kierank> If they don't use vc-2 they use tico or they use Sony low latency codec which are all patented
[21:55:14 CEST] <jamrial> xiph people do, i assume :p
[21:56:03 CEST] <jamrial> oh, you meant at bbc
[21:56:56 CEST] <kierank> Yes
[22:42:04 CEST] <kierank> Compn: any idea where the code is kept?
[23:01:59 CEST] <lglinskih> does decoder _have_ to respect the  request_channel_layout context field?
[23:02:32 CEST] <lglinskih> e.g. FLAC ignores it completely
[23:03:28 CEST] <kierank> lglinskih: it doesn't afaik
[23:15:16 CEST] <Compn> kierank : there was a git gsoc repo long ago...
[23:15:24 CEST] <Compn> lets see
[23:17:21 CEST] <Compn> no, no idea
[23:17:23 CEST] <Compn>  ask luca ?
[23:17:38 CEST] <Compn> or maybe michaelni has a copy of it 
[23:25:47 CEST] <Compn> kierank : you are looking for dirac encoder gsoc code, right ? just to make it easier on whomever has it and reading this.
[23:25:54 CEST] <kierank> yes
[23:34:47 CEST] <Compn> http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/2007-August/thread.html
[23:36:41 CEST] <Compn> it was a svn rpeo
[23:36:42 CEST] <Compn> repo
[23:43:17 CEST] <Compn> kierank : might be in here > svn://svn.mplayerhq.hu/soc
[23:43:38 CEST] <Compn> i make no promises :P
[00:00:00 CEST] --- Tue Apr 14 2015