[Ffmpeg-devel-irc] ffmpeg-devel.log.20170921

burek burek021 at gmail.com
Fri Sep 22 03:05:03 EEST 2017


[00:00:05 CEST] <j-b> and I'll soon move to doing Javascript and web development
[00:00:19 CEST] <Compn> no not javascript
[00:00:24 CEST] <JEEB> j-b: sounds like something that will pay the bills
[00:00:58 CEST] <j-b> JEEB: Javascript is awesome!
[00:01:12 CEST] <JEEB> ecmascript is bearable
[00:01:29 CEST] <JEEB> (as there really isn't anything better on the web)
[00:02:03 CEST] <JEEB> and there are things like TS (thank you visual studio for registering that extension from my video players) that add the "notice mistakes before you go into production"
[00:02:06 CEST] <JEEB> layer
[00:02:29 CEST] <j-b> oh yes
[00:03:04 CEST] <Compn> JEEB : you could make a new cross-plat browser language to replace javascript which is showing its age
[00:03:10 CEST] <Compn> and make a lot of money doing that...
[00:03:14 CEST] <Compn> there, another $1m idea.
[00:03:26 CEST] <iive> firefox support webassembly or something like that
[00:03:27 CEST] <JEEB> nah, I'll rather just watch mobile game footage https://www.youtube.com/watch?v=mSxrvUsssGE
[00:03:30 CEST] <j-b> yep
[00:03:34 CEST] <j-b> port Qt to wasm
[00:03:36 CEST] <JEEB> yea, webasm is the thing coming
[00:03:59 CEST] Action: Compn waiting for a real web browser to come along :(
[00:04:41 CEST] <j-b> vivaldi?
[00:04:42 CEST] <j-b> brave?
[00:05:05 CEST] <Compn> vivaldi just sent me an email
[00:05:17 CEST] <Compn> guess i should read how much they really want to change chrome but end up not changing it
[00:05:17 CEST] <JEEB> brave seemed fun. also mozilla making the tab containers a freely available add-on instead of a registration-needed experiment
[00:06:19 CEST] <Compn> containers lol
[00:06:32 CEST] <JEEB> hey, I like the idea
[00:06:33 CEST] Action: Compn just has portable browsers in different locations
[00:06:40 CEST] <Compn> guess i'm ahead of the currrrrrve
[00:06:47 CEST] <JEEB> https://addons.mozilla.org/en-US/firefox/addon/multi-account-containers/
[00:07:00 CEST] <JEEB> so you can have tab groups that have 100% separate local storage
[00:07:03 CEST] <Compn> right
[00:07:12 CEST] <wm4> isn't firefox breaking addons
[00:07:14 CEST] <Compn> its called portable browsers in different locations.
[00:07:25 CEST] <JEEB> wm4: yes, which is why these are based on the new APIs
[00:07:40 CEST] <JEEB> the old API extensions will have a lot of extinction
[00:07:41 CEST] <wm4> oh so there is a new API
[00:07:44 CEST] <JEEB> after 57 hits
[00:08:01 CEST] <JEEB> wm4: and they recommend people who have missing functionality to make experiments which let you draw the interfaces
[00:08:17 CEST] <JEEB> so that they can be evaluated and implemented if people feel it's a useful addition to the WE API
[00:08:32 CEST] <JEEB> but as I've already noted, the death of the XUL add-ons *will* kill a lot of extensions
[00:08:41 CEST] <JEEB> because people were making them for fun
[00:08:43 CEST] <Compn> LOL brave inserts its own ads
[00:08:50 CEST] <Compn> also based on chromium
[00:08:51 CEST] <JEEB> and people don't want to do random rewrites
[00:08:59 CEST] <JEEB> yea, everything and their dog has chromium there
[00:09:01 CEST] <Compn> how is it faster than chrome if its using chrome ?
[00:09:03 CEST] <graphitemaster> I've been using firefox-nightly on Windows and Linux (now uses the new stylo by default and electrosys (their multi threaded browser renderer)), they're working on an interruptible html/css rewrite too to be threaded which is in beta, the performance has been nothing short of amazing. In terms of user experience alone, I'd argue it's 2 to 4 times nicer feeling.
[00:09:22 CEST] <JEEB> graphitemaster: yea I'll get on that train as 57 hits beta
[00:09:32 CEST] <JEEB> also it will also show me how much I will miss tab groups
[00:09:38 CEST] <Compn> multi threaded just means its going to lock up more threads :P
[00:09:40 CEST] <JEEB> so I can have my por^Wprivate content in another group
[00:09:58 CEST] <Compn> opens browser ... 50 porn tabs start loading
[00:09:59 CEST] <Compn> hmmmmmmmmmm
[00:10:22 CEST] <JEEB> and that is why you have them in multiple groups ;)
[00:10:24 CEST] <wm4> I wonder if slack in firefox will finally be usable
[00:10:38 CEST] <graphitemaster> por^Wprivate content is best suited to the por^Wprivate hub app on Android with a VR headset, you don't even need VR content, just being able to lay back and het a nice "large screen TV" feel ;-)
[00:10:46 CEST] <JEEB> pfft
[00:10:51 CEST] <JEEB> I mostly read that sort of stuff
[00:11:02 CEST] <JEEB> so a screen and an easily operable reader is all I need 8)
[00:11:04 CEST] <graphitemaster> slack isn't usable in any context.
[00:12:36 CEST] <iive> what is that slack you are talking about?
[00:13:11 CEST] <graphitemaster> slack is a webapp based chat thing like IRC but with more features, like inline images and emojis and videos, very media rich
[00:13:51 CEST] <graphitemaster> also slow as crap since it's a billion lines of javascript apparently that takes 95 hours for your browser to interpret unless you have a JIT
[00:14:47 CEST] <Compn> i think the web as we know it is outdated
[00:14:51 CEST] <wm4> yes, it's like IRC but worse and with web shit
[00:14:59 CEST] <Compn> in the future all websites will just be streamed video with interactive content
[00:15:08 CEST] <Compn> thus getting rid of html/javascript 99%
[00:15:20 CEST] <Compn> j-b : you going to make this a reality? ^^^
[00:15:24 CEST] <Compn> you can have my idea for $1m
[00:15:57 CEST] <Compn> the only thing the client gets is html5 or flash video
[00:16:06 CEST] <Compn> website is all server side again :)
[00:17:19 CEST] Action: Compn wonders why he gives all of his great ideas away for free
[00:18:15 CEST] <iive> Compn: there will actually be only one site. facebook.
[00:18:23 CEST] <j-b> very more likely
[00:19:41 CEST] <iive> does facebook have its own Alexa like device?
[00:21:07 CEST] <Compn> good question 
[00:21:18 CEST] <Compn> google, apple siri, amazon alexa...
[00:21:51 CEST] <Compn> iive : the internet says fb is working on a voice speaker/video chat device
[00:22:22 CEST] <j-b> iive: every android device?
[00:22:41 CEST] <iive> j-b: android is google
[00:22:58 CEST] <j-b> fb is preinstalled in all android devices
[00:23:04 CEST] <j-b> with all the permissions set
[00:23:33 CEST] <iive> well, Alexa is a tube that you put in your house and that has 8 microphones that work at all time.
[00:23:41 CEST] Action: Compn checks and sees that there is no facebook app on his phone
[00:23:42 CEST] <Compn> wew
[00:23:47 CEST] <iive> phones turn off some stuff to preserve power.
[00:24:28 CEST] <Compn> facebook is built into the accounts thing in android 
[00:24:35 CEST] <Compn> if you add your account there, that is
[00:26:07 CEST] <iive> also, Alexa like devices cost about $50, they are basically giving them away for free.
[00:31:44 CEST] <Compn> wm4 : whats your plan for ffmpeg future ?
[00:34:42 CEST] <wm4> nothing
[00:36:08 CEST] <graphitemaster> what's your plan for ffmpeg past?
[00:36:49 CEST] <Compn> j-b : what about the proposed plan to destroy the git history and re-make it without the merge commits ?
[00:37:00 CEST] <Compn> or to modify git so a checkout does not land on a merge ?
[00:37:08 CEST] <Compn> and that merges are silenced ...
[00:37:09 CEST] <j-b> Compn: I have that, personnally
[00:37:44 CEST] <Compn> j-b : i was under the impression that a lot of devs would want such a repo... if you want to share it with them
[00:37:52 CEST] <j-b> no.
[00:37:54 CEST] <j-b> pay me.
[00:38:01 CEST] <Compn> how much
[00:38:04 CEST] <j-b> Compn: 1million
[00:39:12 CEST] <wm4> chance missed I guess
[00:39:27 CEST] <Compn> wm4 : didnt you want such a repo or no ?
[00:39:46 CEST] <j-b> Compn: seriously, now is not the time to discuss all this.
[00:39:48 CEST] <wm4> whatever would fix the dumb fork situation
[00:39:50 CEST] <Compn> oh ok :P
[00:40:02 CEST] <Compn> j-b : have fun at vdd! :)
[00:40:36 CEST] <j-b> Compn: we have the money (in the million $ range), we have the capacities, we have the skillz. So the issue is mostly the will.
[00:41:15 CEST] <j-b> and now, bed it is.
[00:42:09 CEST] <Compn> wm4 : come to vdd and see the fork situation in person , face to face :)
[00:43:45 CEST] <wm4> probably not
[00:45:59 CEST] <graphitemaster> there's an ffmpeg fork?
[00:46:17 CEST] <Compn> there are quite a few ffmpeg forks
[00:46:22 CEST] <wm4> for 6 years or so
[00:46:35 CEST] <graphitemaster> is it the whole libav vs ffmpeg thing
[00:46:42 CEST] <graphitemaster> I thought that was finally over
[00:46:45 CEST] <Compn> yes, wm4 wont let it go 
[00:46:50 CEST] <Compn> :P
[00:47:42 CEST] <wm4> graphitemaster: why would you think it's over
[00:50:43 CEST] <graphitemaster> because fighting over superficial shit isn't engineering?
[00:51:27 CEST] <JEEB> I don't think people fight, both camps are happily camping in their own corner
[00:52:48 CEST] <Compn> i've been trying to teach people that forks are natural (see xfree86 xorg or redhat / debian or even freebsd / openbsd )
[00:54:36 CEST] <Compn> and trying to reunite ffmpeg and libav would be like trying to get redhat and debian to merge
[00:55:08 CEST] <Compn> or debian/ubuntu if you prefer 
[00:55:34 CEST] <Compn> but i think everyone hates me now
[00:55:40 CEST] <iive> hum... ubuntu is way too similar to debian
[00:55:41 CEST] <Compn> and most people have me on /ignore
[00:59:17 CEST] <Loriker> compn: :-D
[01:01:44 CEST] <wm4> Compn: please stop spewing bullshit
[01:01:53 CEST] <wm4> why are you always making things harder than they are
[01:02:11 CEST] <Compn> you imply that i have any power or say in the matter, wm4
[01:02:30 CEST] <Compn> but i do not.
[01:03:31 CEST] <Compn> these are just my opinions
[01:20:22 CEST] <graphitemaster> reuniting ffmpeg and libav would be like trying to get north korea and the united states to stop instigating each other
[01:24:50 CEST] <wm4> no it wouldn't
[02:23:13 CEST] <rcombs> less "north and south korea" and more "north and south dakota" at this point, I think
[02:27:54 CEST] <wm4> on that topic: I bet 1988 nobody thought west/east germany unification would have been possible
[03:11:52 CEST] <cone-847> ffmpeg 03Tatsuyuki Ishi 07master:598e41684066: GnuTLS: eat PREMATURE_TERMINATION error
[03:11:52 CEST] <cone-847> ffmpeg 03Kaustubh Raste 07master:b5da07d4340a: avcodec/mips: Unrolled loops and expanded functions in avc put mc 10 & 30 msa functions
[03:11:52 CEST] <cone-847> ffmpeg 03Kaustubh Raste 07master:bba9c1c6bb17: avcodec/mips: Reduced conditional cases in avc inter lpf msa functions
[06:08:18 CEST] <atomnuker> kierank: which train are you getting to paris?
[06:13:22 CEST] <kierank> atomnuker: 12:24
[06:14:05 CEST] <atomnuker> I'm on the 9:24
[06:18:02 CEST] <kierank> I might be too sick for disneyland
[06:19:15 CEST] <atomnuker> did you go to dirtyburger?
[07:55:52 CEST] <kurosu> So, is this an exercise in futility to use movntq/movntdqa in checkasm's randomize buffers stuff ? I don't like the speed aspect of checkasm to be essentially reading from known addresses (thus cached) of very recent data (thus L1 or L2 at worst)
[07:56:43 CEST] <kurosu> plus everyone does rnd() and reads 16bits of it: is this being entropy-conscious or cargo-culting?
[08:01:26 CEST] <kurosu> can't find an equivalent for ARM, but caches are likely smaller
[08:02:15 CEST] <kurosu> hum, scratch the movntq/... stuff, you'd need to do it for every call of the tested functions
[09:46:20 CEST] <JEEB> was the result of something like (x > y) safe to put into an int, or is having a !! there preferred?
[10:08:16 CEST] <rcombs> JEEB: comparisons return 1 or 0, so that's fine
[10:08:48 CEST] <JEEB> ok
[10:20:05 CEST] <JEEB> rcombs: you can guess what sort of damage is being done in write_traf/trun >_>
[10:20:18 CEST] <JEEB> (and I'm trying to make it as presentable/sane as possible)
[10:20:35 CEST] <rcombs> my condolences
[10:41:13 CEST] <JEEB> there most likely is a much better way of doing this, but I want to get this to some state at the very least >_>
[13:35:06 CEST] <cone-208> ffmpeg 03Vittorio Giovara 07master:6f15f1cdc853: pixdesc: Add API to map color property names to enum values
[13:47:47 CEST] <bidiko> hi, I'm using FFmpeg library in my ios project. But I'm getting memory leak error how can i solve this problem ?
[13:50:15 CEST] <JEEB> valgrind is your friend
[13:50:39 CEST] <JEEB> also make as much of the code run'able outside of iOS so you don't cry when you have to debug your multimedia code, if only possible
[13:54:03 CEST] <bidiko> valgrind  ? I don't know him or her
[13:55:23 CEST] <bidiko> if you want to help maybe you can shut up
[13:55:38 CEST] <durandal_1707> bidiko: google for valgrind
[13:55:42 CEST] <JEEB> those both were very serious recommendations
[13:55:48 CEST] <JEEB> a) valgrind is awesome
[13:56:00 CEST] <JEEB> b) the more code you can run on your actual workstation, the better for any debugging
[13:56:09 CEST] <JEEB> since you don't have to be running it in your iOS emulator
[13:56:17 CEST] <JEEB> esp. since I'm not sure if you can run valgrind within your emulator
[13:56:37 CEST] <JEEB> http://valgrind.org/docs/manual/quick-start.html
[13:56:44 CEST] <JEEB> for all the spoon-feeding
[13:57:10 CEST] <JEEB> also this is not -devel stuff, you're an API user so off to #ffmpeg you go
[14:06:42 CEST] <atomnuker> who's arriving today?
[14:07:43 CEST] <thardin> JEEB: -fsanitize is pretty cool too. not as thorough as valgrind, but much faster
[14:07:56 CEST] <thardin> so not excuse not to run it on say FATE
[14:08:03 CEST] <thardin> no*
[14:57:24 CEST] <TD-Linux> atomnuker, some of use are chilling in the moz office
[16:01:59 CEST] <Gramner_> so where is everyone? I heard something about dinner&drinks at le bloc, but dunno what time
[17:12:34 CEST] <cone-762> ffmpeg 03Steven Liu 07master:7e9cdd3f49e5: avformat/hlsenc: fix CID 1418106
[17:12:34 CEST] <cone-762> ffmpeg 03Steven Liu 07master:02bf023afe00: MAINTAINERS: modify the hlsenc description
[17:12:34 CEST] <cone-762> ffmpeg 03Steven Liu 07master:c34c0e3a649e: avformat/hlsenc: support http method for hls fmp4 init file
[20:58:02 CEST] <atomnuker> kierank / Gramner_ / TD-Linux / anyone here: want to hang out
[20:58:34 CEST] <kierank> We are at the place
[20:58:39 CEST] <atomnuker> the place?
[20:59:01 CEST] <kierank> Le bloc
[20:59:38 CEST] <atomnuker> the cafe just next to the hotel?
[21:01:22 CEST] <atomnuker> kierank: guess its not that one, where is it?
[21:01:47 CEST] <kierank> No
[21:01:53 CEST] <kierank> It's 10 mins walk
[21:02:38 CEST] <atomnuker> ah, the one next to brochant?
[21:03:05 CEST] <atomnuker> k, leaving now, do wait for me
[22:59:47 CEST] <durandal_1707> dont get drunk and vomit here
[23:09:59 CEST] <jvanriper> Hullo, all.  I want to provide a small update to support captioning in DirecTV, but I also want to ensure it fits the code.
[23:10:15 CEST] <Compn> sure
[23:10:17 CEST] <Compn> whats up jvanriper
[23:11:11 CEST] <jvanriper> Currently, ffmpeg supports 'a53_cc', and the ff_alloc_a53_sei function sets up the header accordingly.
[23:11:45 CEST] <jvanriper> DirecTV, however, didn't wait for the standard, and went forward with a slightly different header.  It's close, but different enough to cause problems for some folks who are trying to send to their receivers.
[23:12:16 CEST] <jvanriper> Ergo, I want to provide another function, say, ff_alloc_directv_sei, that would hopefully set the headers according to what their receivers need.
[23:12:49 CEST] <jvanriper> So, nice and clean... separate functions... but we need a way to select the appropriate function based on some kind of argument.
[23:13:21 CEST] <jvanriper> I noticed that you have an a53cc argument that sets this up.  It uses OFFSET(a53_cc), etc.
[23:14:12 CEST] <jvanriper> But, it's expected to be a BOOL, where for my needs an INT might be better... when set to 0, it's disabled, 1 for a53_cc, 2 for directv_cc.
[23:15:17 CEST] <jvanriper> So, perhaps a directvcc argument would be set, which would lead to this (currently) a53_cc being set to 2.  But a53cc would set a53_cc to 1.
[23:15:41 CEST] <jvanriper> I'd need to ensure all the functions currently relying upon these settings work appropriately, of course.
[23:16:13 CEST] <jvanriper> (and, unfortunately, I haven't considered yet *reading* this instead of *encoding* it, as my focus has been more on encoding it).
[23:16:50 CEST] <jvanriper> Anyway, I don't want to move forward with an idea like this, if it doesn't quite fit in with things.  Or if someone might have a better approach that fits the project better (I'm certainly not familiar with it).
[23:17:04 CEST] <Compn> sure , someone should have a better idea
[23:17:08 CEST] <Compn> i'm not familiar with that part of the code
[23:17:14 CEST] <Compn> maybe ask ubitux ?
[23:17:42 CEST] <jvanriper> Okay.  Might I find him/her through email?
[23:17:57 CEST] <jvanriper> (or perhaps just go through the e-mail list?)
[23:18:02 CEST] <Compn> yes
[23:18:10 CEST] <jvanriper> Okay.
[23:18:22 CEST] <Compn> sure asking on ffmpeg-devel would also be good
[23:18:32 CEST] <Compn> or you can wait around here if you want
[23:18:36 CEST] Action: Compn going afk
[23:18:42 CEST] <jvanriper> kk.  Just thought I should check here first, just in case.  kk.
[23:18:47 CEST] <Compn> no problemo
[23:18:51 CEST] <Compn> feel free to ask questions
[23:18:57 CEST] <Compn> i think people just arent on irc right now
[23:19:02 CEST] <jvanriper> Alrighty.  Take care, then, and thanks.
[23:28:26 CEST] <J_Darnley> yeah, people are out drinking too much at VDD
[23:29:28 CEST] <llogan> isn't that the norm?
[23:31:05 CEST] <llogan> does libfdk_aac support aac_eld profile? https://pastebin.com/raw/02JkC5xu
[23:33:40 CEST] <Compn> euhhhhhhhhhh
[23:33:42 CEST] <Compn> dont know
[23:33:49 CEST] <Compn> aacdec has that nice list of supported things
[23:33:56 CEST] <Compn> wonder if fdk has it too
[23:34:30 CEST] <llogan> it should...if the docs are correct
[23:40:01 CEST] <Compn> oh
[23:40:14 CEST] <Compn> is our profile correctly sending to libfdk 
[23:40:15 CEST] <Compn> ?
[23:41:03 CEST] <Compn> e.g. does -profile work with libfdk and ... is it correct profile name...
[00:00:00 CEST] --- Fri Sep 22 2017


More information about the Ffmpeg-devel-irc mailing list