[FFmpeg-devel-irc] IRC log for 2010-06-08

irc at mansr.com irc at mansr.com
Wed Jun 9 02:00:49 CEST 2010


[00:08:42] <ramiro_> does libfaac's license permit redistributing libfaac in binary form by itself? (as in not linked to ffmpeg)
[00:10:13] <Dark_Shikari> ramiro_: hard to say
[00:10:20] <Dark_Shikari> it's a mix of LGPL and LGPL-incompatible code
[00:16:16] <CIA-92> ffmpeg: maxim * r23518 /trunk/MAINTAINERS: Declare myself (Maxim Poliakovski) maintainer for the OMA Demuxer.
[00:35:38] <hyc> some robot has found my ffserver http listener
[00:35:44] <hyc> how annoying
[01:35:44] <peloverde> For a paper that is just algebra, there are a sure lot of algebra errors in it
[01:57:59] <CIA-92> ffmpeg: alexc * r23519 /trunk/libavcodec/aacsbr.c:
[01:58:00] <CIA-92> ffmpeg: Base downsampled SBR synthesis filterbank on a single IMDCT.
[01:58:00] <CIA-92> ffmpeg: Based on Han-Wen Hsu, et al. "Fast Complex Quadrature Mirror Filterbanks for
[01:58:00] <CIA-92> ffmpeg: MPEG-4 HE-AAC"
[02:03:05] <CIA-92> ffmpeg: alexc * r23520 /trunk/libavcodec/aacsbr.c: Cosmetics: Fold constants and re-indent after last commit.
[02:03:15] <peloverde> mru, The transforms should now be optimal from a complexity standpoint
[02:09:25] <kierank> how hard would it be to modify the mdct to add a selectable phase-term?
[02:21:30] <peloverde> easy for the C version, I don't know about the assembly versions
[02:22:11] <peloverde> imdct_half is DCT-IV backwards and usually I need to do some post scramble anyway
[02:24:45] <lu_zero> yawn
[02:24:48] <lu_zero> good morning
[02:24:54] * lu_zero should move
[02:25:04] <peloverde> Though I suppose with the right phase term I wouldn't need any scramble
[02:25:33] * lu_zero packs the baggage and heads to the airport
[02:59:20] <BBB> I think I fixed it :)
[03:09:20] <peloverde> good work
[03:20:52] * peloverde just subscribed to celt-dev.. ffceltdec to follow?
[03:22:31] <Dark_Shikari> bitstream isn't even final yet
[03:22:33] <Dark_Shikari> or even close
[03:23:45] <peloverde> yeah but it will be closer by the time I actually have a decoder written
[03:24:02] <peloverde> Which won't even have started until psdec lands
[03:25:58] <peloverde> My timeline is more october-ish for FOMS anyway
[03:26:16] <Dark_Shikari> then again, we're already using celt, so...
[03:26:27] <peloverde> debian is already packaging celt
[03:26:33] <Dark_Shikari> well they're on crack
[03:26:37] <Dark_Shikari> we at least control all ends of our use-chain
[03:26:55] <peloverde> OTOH, they just broke compatibility with 0.7.1 \o/
[03:27:33] <peloverde> Either way CELT is the main reason I'm going to FOMS, so I want to have a decoder written by then
[03:27:49] <bcoudurier> is celt any good ?
[03:27:53] <Dark_Shikari> yes
[03:28:22] <bcoudurier> xiph is definitely good at audio
[03:28:22] <Dark_Shikari> nothing else can get 5ms latency at mp3 bitrates
[03:29:49] <Dark_Shikari> we use celt in production, it works.
[03:31:26] <bcoudurier> using lavc ?
[03:31:33] <Dark_Shikari> well obviously not, there's no lavc wrapper
[03:31:49] <peloverde> yet
[03:31:53] <Dark_Shikari> even then we wouldn't
[03:31:59] <Dark_Shikari> why use a wrapper when we can use the original?
[03:32:06] <Dark_Shikari> using a wrapper only makes sense if you want to easily swap in other codecs
[03:32:10] <Dark_Shikari> but there's nothing else that can possibly suit our needs
[03:32:11] <bcoudurier> switch codec in the future ?
[03:32:16] <bcoudurier> yet
[03:32:21] <Dark_Shikari> new codecs don't appear in a month
[03:32:56] <bcoudurier> testing purpose
[03:33:01] <peloverde> USAC-LD?
[03:33:02] <bcoudurier> flexibility
[03:33:13] <Dark_Shikari> peloverde: 5ms latency?
[03:33:17] <bcoudurier> there are some good reasons to
[03:33:23] <bcoudurier> not like firefox using libogg
[03:33:37] <Dark_Shikari> bcoudurier: equally, we use libx264 directly
[03:33:38] <Dark_Shikari> for the same reason
[03:33:47] <Dark_Shikari> lavc is never going to provide all the api hooks we want
[03:33:51] <Dark_Shikari> and when we do, it'll be years late.
[03:34:15] <Dark_Shikari> we add tools to x264, we can use them the same day.
[03:35:04] <bcoudurier> hehehe
[03:35:07] <bcoudurier> same lame excuse
[03:35:36] <Dark_Shikari> Do you think you can maintain an lavc that wraps every single x264 capability?
[03:35:42] <Dark_Shikari> including external function calls?
[03:35:45] <bcoudurier> I know I can
[03:35:50] <Dark_Shikari> and the ability to pass arbitrary per-block quantizer offsets
[03:35:54] <Dark_Shikari> with a callback to free?
[03:36:00] <bcoudurier> again, I know I can
[03:36:09] <Dark_Shikari> Can you get it committed?
[03:36:12] <bcoudurier> no
[03:36:15] <peloverde> Dark_Shikari, I think it should be in that neighborhood, I don't have the latest spec
[03:36:26] <Dark_Shikari> peloverde: what bitrate is targeted for transparency?
[03:36:43] <bcoudurier> it's like all the people that were using only libx264 and that want now to use vp8
[03:36:55] <Dark_Shikari> bcoudurier: no, they all used lavc
[03:36:57] <Dark_Shikari> and they all suffered for it
[03:36:57] <bcoudurier> it they were using lavc, everything would be smooth
[03:37:04] <Dark_Shikari> and they all had broken x264 settings
[03:37:09] <bcoudurier> only newbies suffered
[03:37:15] <Dark_Shikari> no, everyone suffered
[03:37:19] <bcoudurier> no
[03:37:19] <Dark_Shikari> because there was NO WAY to set such settings in some apps
[03:37:22] <bcoudurier> only newbies
[03:37:27] <Dark_Shikari> there are dozens of bug reports around because lavc-based apps CANNOT ENCODE WITH x264
[03:37:34] <Dark_Shikari> because they don't expose the options in their interface necessary to un-break it
[03:37:54] <Dark_Shikari> unless you consider "newbies" people who don't want to modify the application's code and recompile it
[03:38:19] <bcoudurier> what applications are you talking about ?
[03:38:27] <bcoudurier> all the main ones work perfectly fine
[03:38:41] <peloverde> My guess would be that they *want* AAC or close bitrates for transparency, but I don't have good USAC software
[03:38:44] <Dark_Shikari> like, say blender
[03:38:50] <Dark_Shikari> there were a ton of linux-based lavc apps that broke
[03:38:53] <bcoudurier> blender ....
[03:39:01] <Dark_Shikari> might not have been blender
[03:39:07] <Dark_Shikari> I just recall seeing a lot of gentoo bug reports
[03:39:31] <bcoudurier> besides you made a point breaking lavc
[03:39:35] <bcoudurier> on purpose
[03:39:46] <peloverde> Dolby International taught mru sig/sys, I wouldn't write them off :)
[03:39:57] <peloverde> but I do like CELT, that's why I am going to FOMS
[03:39:59] <Dark_Shikari> bcoudurier: lavc was already broken
[03:40:00] <bcoudurier> if you cared about applications
[03:40:07] <bcoudurier> you would have been less reckless
[03:40:07] <Dark_Shikari> "utter shit" is broken
[03:40:11] <Dark_Shikari> just like ffvorbis is broken
[03:40:19] <bcoudurier> utter shit ?
[03:40:27] <bcoudurier> you mean lavc ?
[03:40:29] <Dark_Shikari> peloverde: they still don't understand the concept of energy maintenance
[03:40:32] <Dark_Shikari> bcoudurier: lavc's x264 defaults
[03:40:45] <Dark_Shikari> they generated shit video.  notifying application developers of this problem is better than ignoring it
[03:40:45] <bcoudurier> well too bad you maintain shit
[03:40:49] <Dark_Shikari> No I don't
[03:40:50] <bcoudurier> IMHO it's your fault
[03:40:53] <Dark_Shikari> I do not maintain lavc's defaults
[03:40:55] <bcoudurier> my wrapper works great
[03:41:04] <Dark_Shikari> options.c is not my file
[03:41:15] <bcoudurier> wtf are you talking about
[03:41:20] <Dark_Shikari> I only maintain libx264.c
[03:41:21] <bcoudurier> poor jason
[03:41:23] <Dark_Shikari> libx264.c does not have any defaults.
[03:41:33] <Dark_Shikari> therefore, I do not maintain lavc's libx264 defaults.
[03:41:39] <Dark_Shikari> how hard is this to understand?
[03:41:55] <bcoudurier> the wrapper you maintain is crap, easy to understand
[03:42:05] <Dark_Shikari> huh?
[03:42:12] <bcoudurier> it's the worst of all the external libs of ffmpeg
[03:42:16] <bcoudurier> x264 is not the only one
[03:42:18] <Dark_Shikari> I don't maintain the fucking options
[03:42:23] <bcoudurier> even vp8 is better
[03:42:30] <bcoudurier> so stop complaining
[03:42:30] <Dark_Shikari> I DONT MAINTAIN IT, I CANNOT FIX IT
[03:42:41] <bcoudurier> you maintain the wrapper
[03:42:44] <peloverde> then submit patches and make a stink
[03:43:17] <Dark_Shikari> I did that
[03:43:19] <Dark_Shikari> They didn't get in.
[03:43:42] <peloverde> Also it's a well understood fact that everyone gets one oops a year where you can commit something you shouldn't then exclaim oops if people object!
[03:43:52] <peloverde> why not?
[03:45:38] <Dark_Shikari> bcoudurier: you want me to make it like vp8 and swap out the preset system for an insane cpu_used system?
[03:45:45] <Dark_Shikari> and make -level 378923678264 map to the settings you want?
[03:46:53] <peloverde> anyway when we switch to git, you can have better defaults in your branch
[03:46:59] <Dark_Shikari> that doesn't mean shit
[03:47:01] <Dark_Shikari> I don't use my branch
[03:47:03] <Dark_Shikari> I don't use ffmpeg
[03:47:14] <Dark_Shikari> bcoudurier: http://pastebin.org/317440 Will you shut up if I commit this?
[03:47:26] <peloverde> then it makes me wonder what you are doing in this channel?
[03:47:28] <Dark_Shikari> Then it'll be just like Theora and you can be happy.
[03:47:35] <Dark_Shikari> peloverde: because my users use ffmpeg, so I develop it for them.
[03:47:57] <Dark_Shikari> Also, I indirectly use ffmpeg, like every person on earth
[03:48:04] <Dark_Shikari> via lavc in some various apps.
[03:48:09] <Dark_Shikari> like, say, my media player.
[03:48:24] <peloverde> tell them to use your new fancy ffms(2?) x264 front end?
[03:48:40] <Dark_Shikari> I'll do that at the end of the summer when our audio support is done
[03:48:47] <bcoudurier> this is the most stupid patch I've ever seen
[03:49:00] <Dark_Shikari> bcoudurier: it gets rid of exactly what you complained about
[03:49:02] <bcoudurier> ffms2 is unusable with big files
[03:49:09] <bcoudurier> _unusable_
[03:49:09] <Dark_Shikari> bcoudurier: that's why we have lavf input as well
[03:49:30] <bcoudurier> people want flexibility
[03:49:36] <Dark_Shikari> bcoudurier: 99% of users don't give a shit
[03:49:38] <Dark_Shikari> they want it to just work
[03:49:39] <bcoudurier> the most flexible transcoding tool today is ffmpeg
[03:49:44] <peloverde> also why use ffmpeg for audio? FFmpeg's mainstream audio encoders blow
[03:49:52] <Dark_Shikari> peloverde: we don't plan to
[03:49:57] <Dark_Shikari> we're going to use libvorbis
[03:50:01] <bcoudurier> and vp8 give even more flexibility
[03:50:10] <Dark_Shikari> vp8 gives fucking no flexibility at all
[03:50:16] <Dark_Shikari> it gives you two numbers, and that's the entirety of its external options
[03:50:18] <bcoudurier> it gives more flexibility to ffmpeg
[03:50:36] <bcoudurier> there is nothing to argue here
[03:50:42] <peloverde> patches welcome
[03:50:43] <Dark_Shikari> ok, then if there's nothing to argue, I'll commit this
[03:50:49] <Dark_Shikari> peloverde: to vp8?  no, they aren't welcome
[03:50:54] <Dark_Shikari> google has explicitly said they're not welcome
[03:51:02] <peloverde> I've already had one patch accepted
[03:51:18] <Dark_Shikari> and patches that actually matter, like fixing the bug that allows remote DOS of any vp8 client
[03:51:21] <Dark_Shikari> haven't been committed
[03:51:44] <Dark_Shikari> (and probably remote system takeover, too, given a good enough hacker)
[03:51:44] <bcoudurier> using ffmpeg for audio the same reason you use ffmpeg for video
[03:52:00] <bcoudurier> you want to encode to aac/mp3/flac/vorbis
[03:52:12] <Dark_Shikari> bcoudurier: for decoding, ffmpeg works well
[03:52:18] <Dark_Shikari> this is because decoding is pretty "easy" from an API perspective
[03:52:19] <Dark_Shikari> it has few options
[03:52:24] <Dark_Shikari> encoding is much harder.
[03:52:29] <bcoudurier> it's not harder
[03:52:37] <bcoudurier> it demands more work
[03:52:38] <Dark_Shikari> I don't believe it is possible to create a unified API that doesn't suck for tons of different encoders
[03:52:39] <bcoudurier> that's all
[03:52:44] <Dark_Shikari> Well, obviously that work was never put in
[03:52:46] <peloverde> I encoded all my music to vorbis once, and no portable device I've acquired since then supports it, I've learned my lesson
[03:52:49] <bcoudurier> I certainly believe it is
[03:52:54] <Dark_Shikari> half of ffmpeg's codec options can't be understood without reading the code
[03:52:58] <bcoudurier> and in that regard, mencoder is more flexible
[03:53:06] <Dark_Shikari> mencoder doesn't do that though
[03:53:09] <Dark_Shikari> it _explicitly_ does not
[03:53:09] <ohsix> key/value pairs with localized help dumps per codec!111
[03:53:13] <Dark_Shikari> every single encoder has its own options
[03:53:16] <Dark_Shikari> ohsix: SO MUCH THIS
[03:53:18] <bcoudurier> that's enough
[03:53:25] <Dark_Shikari> bcoudurier: the point is, that's not "unified"
[03:53:28] <Dark_Shikari> and it's _fine_ that is isn't
[03:53:29] <bcoudurier> what a lot of dev want is per codec/per format options
[03:53:41] <Dark_Shikari> I just think that ffmpeg's approach, the unified one, is inherently flawed
[03:53:48] <Dark_Shikari> and the mencoder approach, the per-codec approach, is inherently correct.
[03:53:53] <bcoudurier> it's inehrently flexible
[03:53:59] <Dark_Shikari> And flexible too, yes
[03:54:02] <ohsix> if there were a lot of parameters that were relevant to everything you wouldn't want for choice of diversity to pick more novel encodings, and the options thereof
[03:54:12] <peloverde> ohsix, welcome back
[03:54:17] <peloverde> :)
[03:54:23] <ohsix> thanks :]
[03:54:23] <Dark_Shikari> and yes, welcome back ohsix
[03:54:30] <bcoudurier> from an application point of view, recoding the wrappers is a lot of time
[03:55:02] <ohsix> if you know you want to use h264 chances are you want to play with the parameters to the extent that you can too
[03:56:54] <bcoudurier> it's certainly valid to unify common parameters
[03:57:03] <bcoudurier> like vp8 encoding will do
[03:57:08] <Dark_Shikari> bcoudurier: but you don't need to do that via the api
[03:57:12] <Dark_Shikari> you can just have the encoders agree with each other
[03:57:18] <Dark_Shikari> i.e. do it as a rule, not enforced by code
[03:57:26] <Dark_Shikari> i.e. "every codec calls its keyframe interval 'g'"
[03:57:34] <Dark_Shikari> and just have that as a requirement for new patches
[03:57:47] <bcoudurier> that's what is is currently
[03:57:47] <Dark_Shikari> it's no more code
[03:57:49] <Dark_Shikari> no it isn't
[03:57:55] <bcoudurier> ?
[03:57:58] <Dark_Shikari> every codec MUST be 'g' because they all share the same option system
[03:58:09] <Dark_Shikari> and you cannot have per-codec options and defaults
[03:58:16] <bcoudurier> you can
[03:58:19] <bcoudurier> implement it
[03:58:21] <Dark_Shikari> IMO, avcodeccontext should mostly not exist
[03:58:29] <bcoudurier> here we go
[03:58:34] <Dark_Shikari> It was implemented
[03:58:34] <bcoudurier> you don't use ffmpeg
[03:58:35] <Dark_Shikari> it was rejected
[03:58:41] <bcoudurier> how can you know ?
[03:58:43] <Dark_Shikari> bcoudurier: I use libavcodec
[03:58:47] <bcoudurier> you don't for encoding
[03:58:50] <Dark_Shikari> Yes I do
[03:58:53] <Dark_Shikari> we use sorenson
[03:59:00] <Dark_Shikari> and have an lavc interface to it
[03:59:20] <Dark_Shikari> and we use it for decoding in the general case
[03:59:36] <Dark_Shikari> for h264, sorenson, and others
[03:59:39] <bcoudurier> <Dark_Shikari> I don't use ffmpeg
[03:59:45] <Dark_Shikari> I don't use ffmpeg.  I use libavcodec.
[03:59:49] <Dark_Shikari> there is a big difference
[04:00:00] <Dark_Shikari> libavcodec is set of really great codecs hobbled by an awkward API.
[04:00:03] <bcoudurier> I'm not sure about that
[04:00:05] <Dark_Shikari> ffmpeg is mostly just a pile of shit.
[04:00:11] <Dark_Shikari> Fortunately, ffmpeg isn't libavcodec.
[04:00:19] <bcoudurier> listen
[04:00:22] <Dark_Shikari> libavcodec and libavformat are imo what this project really is about
[04:00:24] <bcoudurier> if you continue like this
[04:00:27] <Dark_Shikari> it would be great to have a better interface
[04:00:29] <Dark_Shikari> but they are what really matter
[04:00:30] <bcoudurier> I'd like you to leave the channel
[04:00:55] <bcoudurier> I've said multiple times, that it was enough
[04:01:07] <Dark_Shikari> a developer that cannot admit when code sucks is a failure.
[04:01:22] <Dark_Shikari> many things about the x264 API have sucked and continue to suck, so I try to fix them
[04:01:23] <bcoudurier> lol
[04:01:26] <Dark_Shikari> but if you cannot admit that they suck
[04:01:27] <Dark_Shikari> YOU CAN'T FIX THEM!
[04:01:43] <bcoudurier> that's your opinion
[04:01:46] <Dark_Shikari> no, it's fact
[04:01:51] <bcoudurier> no it's your opinion
[04:02:04] <Dark_Shikari> how can you talk about improving something if you can't admit it needs improvement?
[04:02:08] <Dark_Shikari> how is that even possible?
[04:02:12] <Dark_Shikari> what kind of doublethink would you need to do such an insane thing?
[04:02:16] <bcoudurier> what are you talking about
[04:02:27] <Dark_Shikari> You are saying that it's possible to improve X without admitting that X needs to be improved.
[04:02:31] <bcoudurier> because a guy says x264 sucks you gonna listen to him ?
[04:02:39] <Dark_Shikari> If a guy says that Y feature of x264 sucks
[04:02:41] <Dark_Shikari> I look into it
[04:02:45] <bcoudurier> LOL
[04:02:48] <Dark_Shikari> and I try to fix it.
[04:02:53] <bcoudurier> let me record what you said
[04:02:58] <Dark_Shikari> tell me something that sucks, and I'll go look into it.
[04:03:02] <bcoudurier> because I'm gonna throw that back at you
[04:03:09] <Dark_Shikari> um... what?  I'm absolutely fine with that
[04:03:12] <Dark_Shikari> I want you to tell me what sucks
[04:03:14] <Dark_Shikari> so I can make it better
[04:03:17] <Dark_Shikari> PLEASE
[04:03:19] <Dark_Shikari> THROW IT BACK AT ME.
[04:03:20] <Dark_Shikari> seriously
[04:03:23] <Dark_Shikari> >_>
[04:03:29] <Dark_Shikari> you see, I'm not like you
[04:03:37] <Dark_Shikari> you want to attack other developers, you want to avoid improving the code
[04:03:43] <Dark_Shikari> you want to dodge messages about things "sucking"
[04:03:44] <Dark_Shikari> I don't.
[04:03:46] <Dark_Shikari> I want to improve things.
[04:03:49] <Dark_Shikari> I want you to tell me what things are shit
[04:03:52] <bcoudurier> no I want you to stop saying crap
[04:03:54] <Dark_Shikari> I want you to tell me that my code is shit.
[04:03:54] <bcoudurier> and be arrogant
[04:03:59] <Dark_Shikari> bcoudurier: EXACTLY MY POINT
[04:04:02] <Dark_Shikari> you want me to stop saying crap
[04:04:03] <peloverde> the open issue on roundup prove where FFmpeg is deficient
[04:04:04] <Dark_Shikari> I want you to start saying crap.
[04:04:12] <Dark_Shikari> we're exact opposites
[04:04:14] <Dark_Shikari> I _want_ you to say crap
[04:04:15] <peloverde> FFmpeg for the most part does not suck
[04:04:27] <Dark_Shikari> I want you to tell me where things are broken, so I can fix them.
[04:04:31] <bcoudurier> there is absolutely no reason to say crap
[04:04:32] <Dark_Shikari> By comparison, you want to ignore where things are broken
[04:04:34] <Dark_Shikari> Yes there is
[04:04:36] <Dark_Shikari> That's how things get done
[04:04:38] <bcoudurier> absolutely not
[04:04:39] <ohsix> if the patches were rejected is it even up for debate that it could be improved?
[04:04:41] <Dark_Shikari> Well, not with people like you
[04:04:59] <Dark_Shikari> The first step to making something better is admitting that it sucks.
[04:05:13] <bcoudurier> when you think it doesn't, there is nothing to admit
[04:05:30] <Dark_Shikari> And there's your problem.
[04:05:35] <Dark_Shikari> Because everything sucks, to some degree or another.
[04:05:44] <Dark_Shikari> If you cannot admit why something you work on -- even something you wrote -- sucks
[04:05:45] <bcoudurier> saying it's "utter shit", "crap" whatever is just childish
[04:05:47] <Dark_Shikari> then you are a failure as a developer
[04:06:01] <bcoudurier> there is nothing constructive
[04:06:05] <Dark_Shikari> sure it is
[04:06:05] <bcoudurier> you are not constructive
[04:06:06] <Dark_Shikari> as I said
[04:06:08] <Dark_Shikari> 12:04 <@Dark_Shikari> The first step to making something better is admitting that it sucks.
[04:06:17] <bcoudurier> saying that everything sucks goes nowhere
[04:06:24] <j0sh_> s/sucks/could be improved in places
[04:06:29] <bcoudurier> I don't see you being so self criticizing
[04:06:36] <Dark_Shikari> Um, I do it all the time
[04:06:43] <bcoudurier> are you kidding me ?
[04:06:45] <Dark_Shikari> x264's ratecontrol code is a pile of hack, and I'm at fault for making it worse
[04:06:56] <Dark_Shikari> x264's API lacks documentation and I'm at fault for being too lazy to fix it
[04:06:56] <bcoudurier> you spit all around that x264 is best encoder in the _world_
[04:07:00] <bcoudurier> and that everything else is crap
[04:07:06] <Dark_Shikari> No, Mainconcept 8.5 is quite good
[04:07:12] <bcoudurier> how can this be self-criticizing ?
[04:07:17] <bcoudurier> after your arguement
[04:07:18] <Dark_Shikari> All the commercial products using Mainconcept use shitty old versions though
[04:07:22] <Dark_Shikari> which is really annoying
[04:07:22] <bcoudurier> you should say that x264 is crap
[04:07:25] <Dark_Shikari> because their actual latest version is good
[04:07:32] <Dark_Shikari> bcoudurier: something can be "crap" and "the best" at the same time
[04:07:33] <peloverde> Even in your own statements you didn't use the word suck
[04:07:40] <bcoudurier> I never heard you saying x264 is crap
[04:07:47] <Dark_Shikari> bcoudurier: I've said that things in x264 are crap
[04:07:50] <Dark_Shikari> equally, things in libavcodec are crap
[04:07:53] <bcoudurier> it's easy to talk about other projects
[04:07:55] <Dark_Shikari> libavcodec as a whole is not all crap
[04:08:22] <Dark_Shikari> I can easily make a list of a dozen things in x264 that I think are crap, including things I've written.
[04:08:33] <ohsix> least worst is an amicable place to be in a lot of disciplines
[04:08:52] <peloverde> celt.git doesn't even build on my system and gives some arcane autoconf error :-/
[04:09:05] <bcoudurier> that's fine
[04:09:09] <bcoudurier> that doesn't make x264 crap
[04:09:09] <ohsix> peloverde: what error?
[04:09:13] <Dark_Shikari> also, there are plenty of things x264 does that I know for a fact it does worse than many other encoders
[04:09:18] <Dark_Shikari> or at least one other encoder.
[04:09:20] <bcoudurier> the same thing goes to ffmpeg
[04:09:25] <bcoudurier> there may be crappy stuff
[04:09:26] <peloverde> ohsix, syntax error near unexpected token `tools="tools",'
[04:09:28] <bcoudurier> that doesn't make it crap
[04:09:29] <Dark_Shikari> bcoudurier: when I mean "ffmpeg" I generally refer to ffmpeg.c
[04:09:35] <Dark_Shikari> i.e. the wrapper application around libav*
[04:09:36] <bcoudurier> and again
[04:09:39] <peloverde> ohsix, `  XIPH_PATH_OGG(tools="tools", tools="")'
[04:09:41] <bcoudurier> that's your opinion
[04:09:43] <Dark_Shikari> which is just a tiny portion of the ffmpeg project
[04:09:49] <Dark_Shikari> bcoudurier: I'm just talking about what I mean when I refer to X
[04:09:55] <Dark_Shikari> I'm defining my terms.
[04:10:16] <ohsix> peloverde: weird heh; before you pasted it i pegged it on a missed quote somewhere
[04:10:18] <bcoudurier> yeah
[04:10:22] <ohsix> clone url?
[04:10:23] <bcoudurier> that's a great way to improve the discussion
[04:10:31] <bcoudurier> "but that's fine I understand myself"
[04:10:48] <peloverde> ohsix, git://git.xiph.org/celt.git
[04:10:57] <Dark_Shikari> bcoudurier: um....
[04:11:00] <Dark_Shikari> I'm defining terms
[04:11:05] <Dark_Shikari> so that others can understand specifically what I'm saying
[04:11:07] <Dark_Shikari> .........
[04:11:08] <j0sh_> nothing is perfect, and everything can be improved on in some way. but i think for what ffmpeg/lav*/etc does, it's pretty remarkable. 2c
[04:11:31] <j0sh_> thats why we develop on it
[04:11:41] <bcoudurier> exactly
[04:11:46] <bcoudurier> and no, ffmpeg.c is not crap
[04:11:48] <j0sh_> if it were perfect right now (or in your words, non-crap), we wouldnt have a reason to keep working on it
[04:12:12] <astrange> i have yet to understand -vsync in ffmpeg.c
[04:12:16] <Dark_Shikari> j0sh_: so you understand now
[04:12:17] <ohsix> peloverde: derp, goes here fine; good luck :O
[04:12:18] <Dark_Shikari> _everything_ is crap
[04:12:26] <bcoudurier> no it is not
[04:12:26] <Dark_Shikari> some things more than others
[04:12:33] <astrange> i'm not sure how both mencoder and ffmpeg have a/v sync algorithms that make no sense to me
[04:12:36] <Dark_Shikari> if it wasn't crap at all, there would be no reason to work on it.
[04:12:47] <bcoudurier> astrange, you don't understand it
[04:12:55] <ohsix> bcoudurier: if being perfect is an impossible thing, then real measures of things will always be to the degree that they are unperfect
[04:13:00] <bcoudurier> a/v sync algos make perfect sense to me
[04:13:12] <peloverde> just because there is a potential for better doesn't mean that something is crap
[04:13:14] <astrange> i also know it's wrong in two different ways
[04:13:17] <bcoudurier> you can see an 1/2 empty bottle if you wish
[04:13:21] <bcoudurier> I see it as 1/2 full
[04:13:31] <astrange> the treatment of initial delay is wrong (trying to fix it but i keep forgetting how the details work)
[04:13:36] <bcoudurier> that's the difference between an optimistic and a pessimistic
[04:13:45] <Dark_Shikari> bcoudurier: no
[04:13:46] <ohsix> whats empty/half full on the spectrum from not existing to perfectly existing?
[04:13:55] <astrange> and it uses avpkt.dts as pts because too many of the demuxers set wrong values for pts
[04:14:12] <Dark_Shikari> the optimist believes that the glass is half empty, which means he can work on the other half to make it full
[04:14:24] <Dark_Shikari> the pessimist believes it's half full, so it's good enough and has plenty left.
[04:14:27] <bcoudurier> astrange, proof that you don't understand it
[04:14:37] <j0sh_> Dark_Shikari: i think what we object most to is the choice of vocabulary. crap/sucks/etc does a disservice to lav* and the devs working on it
[04:14:47] <astrange> i understood it enough to make most of the changes and then forgot it
[04:14:52] <ohsix> peloverdey: using autoconf 2.65?
[04:14:53] <Dark_Shikari> j0sh_: some things are quite admittedly crap and if you cannot accept it, you've done something wrong
[04:15:00] <Dark_Shikari> and some things aren't crap
[04:15:01] <bcoudurier> are you kidding me ?
[04:15:17] <Dark_Shikari> I mean, to begin with, there's a lot of old code that isn't really maintained and is probably crap
[04:15:20] <bcoudurier> optimistic is half full
[04:15:22] <Dark_Shikari> by comparison, there's code like h264.c
[04:15:28] <bcoudurier> the last word emphasize the goal
[04:15:29] <Dark_Shikari> which is heavily maintained and somewhat less crap.
[04:15:37] <bcoudurier> !!!
[04:16:04] <Dark_Shikari> bcoudurier: the optimist believes he can make the cup fuller.  the pessimist believes the cup is plenty full enough and that people should stop complaining about the fullness of the cup.
[04:16:04] <bcoudurier> using dts with delay is acceptable
[04:16:36] <peloverde> ohsix, autoconf 2.65
[04:16:59] <ohsix> optimist pretty much is ok with the glass being half full, since they've got something at least; but that thing sort of falls apart with software when you start from nothing, and especially with code that isn't constantly worked over it just rots, and rarely gets the attention it needs for the first pass, which is how it exists after
[04:17:09] <bcoudurier> Dark_Shikari, that's wrong
[04:17:31] <ohsix> peloverde: no clue on a fix, the macros are supposed to be in aclocal and they appear to be here, not sure what component might mess with that sort of thing
[04:17:41] <Dark_Shikari> ugh, autoconf.
[04:17:56] <peloverde> why does CELT use autoconf?
[04:18:15] <peloverde> and why do they insist I shout their name?
[04:18:21] <ohsix> why not? i like it when it works, i haven't had to do anything complicated with it yet
[04:18:35] <peloverde> it's not working now
[04:18:39] <astrange> i don't like developing it even when it works
[04:18:43] <ohsix> ya, good luck :D
[04:18:50] <peloverde> it likes to check for fortran compilers when building ure C code
[04:19:01] <astrange> m4 is difficult to read and i prefer script languages over macro expanders
[04:19:12] <astrange> plus most people write a ton of default checks that they never use
[04:19:29] <ohsix> it wouldn't have m4 at the bottom if that wasn't some old school common denominator
[04:19:36] <bcoudurier> Is the glass half empty or half full? is a common expression, used rhetorically to indicate that a particular situation could be a cause for optimism (half full) or pessimism (half empty);
[04:19:40] <peloverde> FFmpeg's configure seems to work fine everywhere
[04:19:42] <astrange> wonder how tromey's quagmire is going
[04:19:44] <bcoudurier> WIKIPEDIA !
[04:19:45] <ohsix> astrange: its really hard to get those out too
[04:19:48] <bcoudurier> seriously guys
[04:20:03] <ohsix> even an empty thing checks for fortran and stuff if you don't throw a bunch of implied macro params in there
[04:20:05] <Dark_Shikari> bcoudurier: I'm making a different metaphor
[04:20:08] <bcoudurier> yeah yeah
[04:20:15] <bcoudurier> change your words now that you are wrong
[04:20:17] <astrange> or FFconf if someone made a project generator for it
[04:20:18] <Dark_Shikari> um...
[04:20:22] <Dark_Shikari> I intentionally made a different metaphor
[04:20:24] <Dark_Shikari> ......
[04:20:26] <Dark_Shikari> what the fuck
[04:20:34] <bcoudurier> what I said
[04:20:37] <bcoudurier> was exactly that
[04:20:46] <Dark_Shikari> and I said that there's another dimension to the problem
[04:20:51] <Dark_Shikari> that not of what the glass is, but what the glass can be
[04:20:56] <bcoudurier> <Dark_Shikari> the pessimist believes it's half full, so it's good enough and has plenty left.
[04:21:00] <Dark_Shikari> it doesn't matter if you believe it's half full
[04:21:01] <bcoudurier> you said exactly _wrong_
[04:21:09] <Dark_Shikari> um, how can a metaphor be "wrong"?
[04:21:13] <Dark_Shikari> I was intentionally invoking the opposite
[04:21:15] <jai> lol
[04:21:18] <Dark_Shikari> ....
[04:21:18] <ohsix> stuff that expands in place is conceptually simpler since you're just replacing the tokens that you care about, rather than being imperitive about it
[04:21:20] <bcoudurier> that's an answer to what I said
[04:21:22] <Dark_Shikari> yes
[04:21:27] <bcoudurier> so you are wrong
[04:21:29] <bcoudurier> holly shit
[04:21:31] <Dark_Shikari> ...wtf?
[04:21:34] <bcoudurier> can't you admit it
[04:21:34] <Dark_Shikari> How can a metaphor be wrong?
[04:21:38] <ohsix> any replacement for autoconf would be largely a token replacement/file generation task
[04:21:39] <Dark_Shikari> it's like saying
[04:21:41] <jai> i'm lmao IRL
[04:21:42] <Dark_Shikari> "what's your favorite color"
[04:21:43] <Dark_Shikari> "green"
[04:21:45] <Dark_Shikari> "WRONG"
[04:21:48] <bcoudurier> all right you win
[04:21:50] <Dark_Shikari> jai: same
[04:21:52] <bcoudurier> you are right
[04:21:53] <Dark_Shikari> this is hialrious
[04:21:57] <_av500_> tangerine
[04:22:01] <astrange> expanders are difficult to debug unless you can directly evaluate the unexpanded part
[04:22:04] <Dark_Shikari> also, which is better?  apples or oranges?
[04:22:06] <bcoudurier> I'm very tired of discussing crap like that
[04:22:10] <j0sh_> http://xkcd.com/386/
[04:22:22] <bcoudurier> when the person you are talking to can't even recognize what he said
[04:22:38] * _av500_ prefers some light xiph bashing when waking up...
[04:22:39] <peloverde> apparently "syntax error near unexpected token `tools="tools",'" is auto* for install libogg-dev, blol
[04:22:42] <ohsix> yea, it doesn't solve the "find the problem" stuff very well, i dunno what you could do about that too; maybe some sort of diff format to show every replacement
[04:22:55] <bcoudurier> that's being stubborn to the extreme
[04:23:15] <Dark_Shikari> bcoudurier: good job completely missing the point of everything I said
[04:23:17] <peloverde> and blol at DS for liking a project that thinks such is a good idea
[04:23:17] <Dark_Shikari> and furthermore, proving my own point
[04:23:32] <bcoudurier> blah blah
[04:23:32] <Dark_Shikari> peloverde: it's possible for a good project to have a bad build system
[04:23:37] <Dark_Shikari> or, equally, a bad project to have good ideas
[04:23:41] <Dark_Shikari> or equally, a good project to have bad ideas
[04:24:00] * Dark_Shikari adds bcoudurier to ignore so that he can get actual work done, since bcoudurier appears to not believe in the concept of work
[04:24:01] <ohsix> peloverde: hah nice, i looked at the aclocal, things ripped from icecast; it just picks from in tree or on system stuff
[04:24:26] <bcoudurier> I should request you to be banned
[04:24:34] <peloverde> DS, I don't see you saying that the CELT build system sucks, I'll I hear you say is how great CELT is
[04:24:47] <bcoudurier> afterall your contribution is very minimal
[04:24:49] <ohsix> bcoudurier: in my experience the "ok you are right" response doesn't mollify anyone if they really want you to think about their viewpoint
[04:24:51] <Dark_Shikari> peloverde: I have never built it
[04:24:54] <Dark_Shikari> oh wait, I built it once
[04:24:57] <Dark_Shikari> on my own box
[04:25:03] <Dark_Shikari> and it worked perfectly, but I've never built it and had it not work
[04:25:05] <Dark_Shikari> of course, that means nothing
[04:25:10] <Dark_Shikari> you can have a totally shit build system that just happens to work
[04:25:12] <peloverde> It requires libogg!
[04:25:15] <Dark_Shikari> It's just that I never _noticed_ it
[04:25:22] <Dark_Shikari> so I couldn't possibly comment on it.
[04:25:22] <peloverde> gg is pretty much the worst container ever
[04:25:24] <bcoudurier> ohsix, who are you ?
[04:25:25] <peloverde> *ogg
[04:25:35] <ohsix> nobody
[04:25:43] <peloverde> ohsix is mru's nemesis :)
[04:25:51] <bcoudurier> good
[04:25:51] <ohsix> if that matters then take it how you will, heh
[04:26:07] <Dark_Shikari> peloverde: all I know is that CELT is transparent at about 128kbps for stereo, at 5ms latency
[04:26:07] <bcoudurier> we don't need more trolls here
[04:26:13] <Dark_Shikari> and its quantizer system is really really really freaking cool
[04:26:26] <Dark_Shikari> to the point where if I ever make my own custom video format
[04:26:29] <Dark_Shikari> I will rip it off wholesale
[04:26:47] <peloverde> well I intend on making a CELT decoder that doesn't suck
[04:26:52] <peloverde> libcelt clearly does
[04:26:57] <Dark_Shikari> what about it?
[04:26:58] <Dark_Shikari> slow?
[04:27:02] <peloverde> auto*
[04:27:07] <Dark_Shikari> that's not a reason a program sucks
[04:27:10] <Dark_Shikari> a build system can be replaced
[04:27:12] <Dark_Shikari> Easily
[04:27:16] <peloverde> then they should replace it
[04:27:21] <peloverde> also dependency on libogg
[04:27:25] <Dark_Shikari> submit a patch
[04:27:46] <astrange> ffcelt and ffspeex would be appreciated
[04:27:50] <astrange> or does celt obsolete speex?
[04:27:52] <peloverde> meh, It's not a standard if there is only one implementation
[04:27:54] <astrange> couldn't tell
[04:27:54] <Dark_Shikari> astrange: no
[04:27:57] <Dark_Shikari> celt sucks at low rates
[04:28:11] <Dark_Shikari> it is not intended to be good below 32kbps per channel
[04:28:19] <Dark_Shikari> speex is higher latency
[04:29:04] <ohsix> peloverde: elementary streams seem to go into ogg, at least from the command line tools
[04:29:53] <ohsix> looks like theres a configure param to not build those
[04:30:17] <peloverde> why can't they use mkv be default or even riff (they are CBR)? ogg is full of fail
[04:30:25] <Dark_Shikari> peloverde: xiph
[04:30:33] <Dark_Shikari> and celt isn't necessarily cbr
[04:30:37] <Dark_Shikari> it has a vbr (one frame delay) mode
[04:30:39] <Dark_Shikari> to improve transcients
[04:30:50] <Dark_Shikari> they intend to improve this in the future as well, it's not pure cbr
[04:30:55] <ohsix> heh, gotta pick one, might as well be ogg
[04:31:01] <peloverde> they presentation I watched said CBR, it may be old
[04:31:19] <Dark_Shikari> yeah, they've been adding fancier stuff lately
[04:31:22] <peloverde> ohsix, http://hardwarebug.org/2010/03/03/ogg-objections/
[04:31:24] <Dark_Shikari> it was originally pure CBR
[04:32:01] <ohsix> if there were technical reasons for _them_ not to put it in ogg, they might have not
[04:32:18] <peloverde> technical reasons: http://hardwarebug.org/2010/03/03/ogg-objections/
[04:32:33] <ohsix> for them.
[04:32:54] <ohsix> does the guy doing celt object or agree? you'll have to ask him :P
[04:33:06] <ohsix> (jmspeex if you catch him on freenode)
[04:33:14] <peloverde> a bunch of people are doing celt
[04:33:28] <ohsix> plus they do speex all up in ogg too
[04:35:12] <peloverde> ogg still sucks
[04:35:31] <peloverde> how I feel about ogg: http://i.imgur.com/T5Tl2.png
[04:37:17] <Dark_Shikari> ogg is shit, but it gets a bit too much hate -- there are a lot of things that are equally shit and could use flaming, imo
[04:37:33] <peloverde> ogg does net get enough hate
[04:37:40] <ohsix> heh
[04:37:49] <peloverde> ogg has a large army of know-nothing defenders
[04:37:53] <Dark_Shikari> this is true
[04:37:54] <ohsix> it is a bit much for peoples preference, technological merits or not
[04:38:09] <Dark_Shikari> ogg has a massive army of idiotic retarded fanboys
[04:38:11] <Dark_Shikari> Then again
[04:38:12] <peloverde> mkv is a much better choice
[04:38:13] <Dark_Shikari> so does apple
[04:38:22] <Dark_Shikari> then again, apple does get a lot of flame
[04:38:29] <Dark_Shikari> not that there's anything wrong with that ;)
[04:38:46] <ohsix> but people critically deciding things will do just that; popularity doesn't change a discerning decision
[04:39:11] <Dark_Shikari> discerning decisions are horrifically rare in this day and age
[04:39:33] <ohsix> all i can trust is my judgement, other peoples is rarely my business :D
[04:39:45] <peloverde> Even blizzard who is straight-gay for xiph admits ogg sucks
[04:40:12] <Dark_Shikari> blizzard? what?
[04:40:13] <Dark_Shikari> oh
[04:40:15] <Dark_Shikari> christopher blizzard
[04:40:26] <Dark_Shikari> when I heard blizzard I think of that OTHER blizzard
[04:40:47] <peloverde> bah, chris blizzard is far more annoying
[04:40:52] <siretart> like in WOW? ;-)
[04:40:58] <siretart> morning, btw
[04:41:09] <Dark_Shikari> yeah
[04:41:22] <peloverde> he said suggesting people use ffmpeg2theora would be contributory infringement
[04:41:35] <Dark_Shikari> peloverde: people say a lot of things
[04:41:37] <Dark_Shikari> idiots, particularly
[04:41:40] <peloverde> he's convinced there are patents on mov but can't name any
[04:41:41] <Dark_Shikari> also, trolls
[04:42:07] <peloverde> Dark_Shikari, he works for mozilla which needs to die for exactly two reasons
[04:42:12] <Dark_Shikari> 1) memory leak
[04:42:13] <peloverde> a) he works for them
[04:42:15] <Dark_Shikari> 2) memory leak
[04:42:35] <peloverde> b) gecko is fully of legacy shit including XPCOM and memory leaks
[04:42:36] <Dark_Shikari> maybe some day firefox will stop taking 2 seconds to switch tabs
[04:42:43] <Dark_Shikari> and will stop locking up completely when I click a button
[04:42:48] <saintdev> Dark_Shikari: you forgot to free that 2nd statement :p
[04:42:49] <ohsix> you can't discount that maybe he has actually thought about it, and concluded something with a decent amount of thought; doesn't mean he's a troll outright if he never qualifies his opinions
[04:42:51] <Dark_Shikari> and will stop eating 1.5gb and then crashing when it runs out of address space
[04:43:07] <ohsix> Dark_Shikari: sounds like you got some crazy stuff going on in ff again :O
[04:43:16] <Dark_Shikari> ohsix: it's always done this
[04:43:19] <Dark_Shikari> the longer you leave it on, the laggier it gets
[04:43:30] <Dark_Shikari> chrome, by comparison is freaking instant, and even when it lags it doesn't log up the whole thing
[04:43:35] <ohsix> yea thats not what it does to me, i said that last time, huhuhuhu
[04:43:38] <peloverde> Mozilla to us: What are you doing to make sure your windows friends use firefox
[04:44:03] <ohsix> i disable everything though so no tabs are driveby loading embedded objects that bring in totem or wmp, or quicktime
[04:44:08] <peloverde> Diego to Mozilla: What are you doing to make sure your firefox friends don't use Windows?
[04:44:14] <peloverde> Mozilla: no response
[04:44:24] <saintdev> Dark_Shikari: except my flash chrome process eats up 1GB of ram by itself
[04:44:26] <peloverde> fuck firefox, fuck mozilla
[04:44:28] <Dark_Shikari> peloverde: mozilla is tsundere for megacorporations
[04:44:45] <ohsix> i get people on ff before i get them on lunix; they don't even understand the transition but it doesn't matter, browser is the same
[04:44:50] <Dark_Shikari> "n-no, google, we just did this because we felt like it, not because w-we're friends or anything"
[04:45:49] <peloverde> Mozilla leadership: GOOG is the evil use bing; everyone else: do you remember the 90s?
[04:45:52] <ohsix> 6528 ohsix     20   0 2381M 1246M 29636 S  2.0 44.2 13:10.90     0  |           `- /usr/lib/firefox-3.6.3/firefox-bin
[04:46:00] <ohsix> huhuhuu
[04:46:02] <peloverde> I used to be a Mozilla fan: http://www.mozilla.org/credits/
[04:46:03] <Dark_Shikari> >2381M
[04:46:03] <Dark_Shikari> lol
[04:46:08] <Dark_Shikari> enjoy your memory leak
[04:46:11] <ohsix> vm is free-ish though
[04:46:18] <ohsix> thats' not a leak
[04:46:21] <Dark_Shikari> ohsix: on my computer that would have already crashed
[04:46:30] <ohsix> its rss is 1246
[04:46:33] <peloverde> Organizations I want to die: Mozilla > Microsoft > Apple
[04:46:37] <ohsix> ah, what computer do you have?
[04:46:42] <Dark_Shikari> ohsix: I use 32-bit firefox
[04:46:52] <Dark_Shikari> 64-bit doesn't have an accelerated VM yet, or support flash
[04:46:54] <ohsix> oddly enough it runs worse on my x86_64 machine cuz theres no jit until like 3.7
[04:47:01] <Dark_Shikari> er, jit, yeah
[04:47:03] <ohsix> i use flash; works good
[04:47:11] <Dark_Shikari> there's no 64-bit flash 10.1
[04:47:20] <Dark_Shikari> and for business reasons I need 10.1 installed
[04:47:29] <Dark_Shikari> or wait, was that 10?  I can never remember
[04:47:41] <peloverde> flash-x64 doesn't work with hulu
[04:47:41] <ohsix> the alpha astro or whatever, they may call it alpha but theres no way it is an "alpha", i suspect they call it alpha just to push out possible support requests
[04:47:50] <ohsix> peloverde: it does again
[04:47:56] <astrange> safari 5 just added bing, i'm not about to switch to it though
[04:48:00] <peloverde> ohsix, on rare occasions
[04:48:06] <ohsix> really?
[04:48:24] <Dark_Shikari> astrange: in before an apple search engine
[04:48:29] <ohsix> last time, i think hyc was the one that said it was broke; it was flat broke
[04:48:30] <Dark_Shikari> iSearch
[04:48:45] <ohsix> since i checked again like 2 weeks ago though its been fine any way i hit it
[04:48:54] <peloverde> the pre-ads work fine
[04:49:06] <peloverde> most of the actual streams are broken
[04:49:21] <Dark_Shikari> I wonder if that relates to the way that flash media server keys work
[04:49:23] <astrange> large free websites are not something apple knows how to do
[04:49:32] <ohsix> they changed their throbber and roll in and stuff
[04:49:34] <Dark_Shikari> every few versions, flash fucks with the server keys
[04:49:38] <Dark_Shikari> and breaks red5
[04:50:41] <ohsix> keeping track of files in the client is the most insane shit ever; even regular development would mean a bunch of versions to track
[04:50:44] <peloverde> The funny part is because flash doesn't work with hului I sue rmtpdump
[04:51:07] <peloverde> *I use rtmpdump
[04:51:11] * peloverde can't type
[04:51:29] <ohsix> its telling me to check my internet connection on this random video i picked
[04:51:52] <ohsix> i take it the part that doesn't work is the hash on the browser plugin? along with the app sig?
[04:52:30] <ohsix> anyways; that sucks you have to use the 32bit version; i still use it on my windows machine with a retarded huge session, it breaks windows before ff breaks
[04:52:59] <Dark_Shikari> my firefox just kinda grinds to a halt
[04:53:07] <ohsix> you can still navigate the tabs and stuff in ff even after windows ate itself and couldn't draw anything on the screen
[04:53:12] <Dark_Shikari> I find myself constantly, unconciously, using chrome instead when I need to open something quick
[04:53:21] <ohsix> word
[04:53:27] <peloverde> chrome is my default browser
[04:53:33] <ohsix> i use different profiles when my laptop is on wifi
[04:53:52] <ohsix> just lets you remove so much junk from the web
[04:54:10] <peloverde> The high profile devs from the days or yore all work on webkit browsers, and the new high profile devs all have Macs
[04:54:58] <saintdev> peloverde: rtmpdump works on hulu?
[04:55:00] <Dark_Shikari> webkit is a classic example of free beating Free
[04:55:21] <peloverde> First thing Blizzard did when he left Red Hat for Mozilla was switched to Mac, coincidence, I think not
[04:55:23] <Dark_Shikari> because it was BSD, it became so widely adopted that eventually it because Free anyways
[04:55:28] <peloverde> saintd3v, works for me :)
[04:55:29] <bcoudurier> free in webkit == owned by google
[04:56:01] <saintdev> peloverde: haven't ever tried it, just was curious if it did
[04:56:02] <ohsix> i think its a case of good enough being picked up by a steamroller to beat it into shape
[04:56:28] <peloverde> saintd3v, sometimes it takes several tries, but in the end it always seems to work
[04:56:45] <peloverde> I use the ai get_hulu.sh script
[04:56:58] <bcoudurier> valgrind doesn't work on snow leopard
[04:57:03] <bcoudurier> painful
[04:57:13] <peloverde> I even found someone on hulu using AAC-Main
[04:57:23] <astrange> valgrind svn branches/MACOSX106
[04:57:24] <astrange> works
[04:57:28] <peloverde> zzuf doesn't work on Lucid
[04:57:32] <peloverde> :(
[04:57:33] <bcoudurier> really ?
[04:57:48] <astrange> just got there today, before that there was a large patch on the bug tracker
[04:57:50] <peloverde> valgrind doesn't work on libvpx :)
[04:57:55] <saintdev> oh yeah i remember you talking about that when you discovered it
[04:58:08] <bcoudurier> LOL
[04:58:15] <bcoudurier> google is fucked on safari 5
[04:58:23] <bcoudurier> lol seriously
[04:58:37] <astrange> it's not great though unless it's improved, ffplay/opengl don't render on my gpu and some non-memcheck tools crash immediately
[04:58:54] <peloverde> valgrind+libvpx: https://bugs.kde.org/show_bug.cgi?id=239992
[04:59:36] <peloverde> astrange, ffplay/ogl exists?
[05:00:04] <astrange> sdl goes through ogl
[05:00:21] <astrange> i meant "ffplay or other opengl programs"
[05:00:39] <peloverde> Everywhere on only on St. Steve's second fav OS?
[05:00:41] <bcoudurier> nice html5 support safari 5, you fucked google up
[05:01:27] <peloverde> fuck Google, they recently decided who was with them and who was against them
[05:03:52] <bcoudurier> mac is good, but still behind a good old ubuntu x64
[05:04:11] <bcoudurier> for dev I mean
[05:05:18] <peloverde> I bent over backwards to deal with their Chrome bugs and Sorenson was a zero day partner but I got crumbs from other people I worked for? bullshit! The zeroday libvpx stuff was not (L)GPL compatible, the patches weren't pre-vetted despite the number of people here under NDA
[05:06:54] <ohsix> one thing that bugs me is that 32bit has much less "weird" going on vs x86_64; kernel iowait times and other stuff relating to the _port_ (generics from the original port are still in there for spinlocks and stuff)
[05:07:14] <ohsix> i have some weird writeback livelock on xfs too :\
[05:08:03] <astrange> the gpl compatibility was odd
[05:08:16] <astrange> chris dibona definitely knows about those issues
[05:08:36] <ohsix> did you see his mail to the OSI?
[05:09:16] <astrange> no
[05:09:25] <ohsix> hrm he's on the board at osi too
[05:13:54] <ohsix> http://www.crynwr.com/cgi-bin/ezmlm-cgi?17:mss:1019:dphgapeaeaahenjiboco
[05:14:14] <ohsix> kind of makes you wonder if he does know :O
[05:21:52] <Dark_Shikari> peloverde: the NDAs were to let people read it and support it
[05:21:55] <Dark_Shikari> NOT to improve it
[05:21:55] <Dark_Shikari> imo
[05:22:08] <peloverde> agreed
[05:22:45] <bcoudurier> nice support, indeed
[05:22:50] <peloverde> A codec that was developed in secret and can't be modified has the monopoly on open, blol
[05:23:13] <av500> +1
[05:23:47] <av500> peloverde: what i find hilarious is how they now call ppl to "work" on it, so what has on2 been doing all these years? writing whitepapers?
[05:24:04] * elenril wakes up and reads the backlog
[05:24:07] <peloverde> av500, more or less
[05:24:12] <elenril> w00t, a night full of trolls
[05:24:24] <peloverde> staging fake comparisons against x264
[05:24:25] <bcoudurier> anyway it's a bit of a shame that the patches were not in a good shape given some people were aware ...
[05:24:26] <av500> elenril: but not a single golem..
[05:27:50] <Dark_Shikari> hmm.  is there an easy way in x86 to complement the carry flag if a value is 1, otherwise do nothing?
[05:27:54] <Dark_Shikari> Or some similar operation
[05:28:00] <Dark_Shikari> (without a jump, or more than 1 or 2 ops)
[05:28:08] <Dark_Shikari> where the value is 1 or 0
[05:28:33] <Dark_Shikari> currently if( x == (y & 1) ) takes 3 instructions: mov (to make a copy of y, as it's used later), and, and cmp
[05:28:44] <Dark_Shikari> as this code is extraordinarily critical I'd really really really like to avoid the mov
[05:28:50] <Dark_Shikari> and "bt" is a totally awesome instruction
[05:31:06] <ohsix> heh hulu desktop isn't bad; and i guess its good enough for whatever stream signing there is, gonna have to check it out on the netbook
[05:31:22] <peloverde> Can FFmtech squeeze anything out of Google for distributing an illegal FFmpeg for a few weeks?
[05:31:37] <Dark_Shikari> peloverde: that would involve a test of the gpl in court
[05:31:45] <Dark_Shikari> which is inherently risky
[05:32:06] <peloverde> Not necessarily, most big companies would rather settle
[05:32:35] <peloverde> especially for small amounts
[05:32:42] <Dark_Shikari> true
[05:32:56] <Dark_Shikari> of course that would just give them an excuse not to use ffmpeg
[05:33:18] <peloverde> I'd be ok with that
[05:34:49] <astrange> if the value is only 1 or 0?
[05:35:27] <astrange> that's obviously just bt
[05:36:07] <peloverde> It seems to me that their patches were fairly trivial and could have been written from the inside in less time than it took to approve them
[05:36:17] <Dark_Shikari> astrange: it's (x&1) != y
[05:36:19] <Dark_Shikari> not x&1
[05:36:30] <Dark_Shikari> small difference
[05:37:29] <Dark_Shikari> I want to do that in 2 instructions or less
[05:39:12] <Dark_Shikari> y can be clobbered, x cannot
[05:46:51] <astrange> 01:28 <@Dark_Shikari> currently if( x == (y & 1) ) takes 3 instructions: mov (to make a copy of y, as it's used later), and, and cmp <- sounds like y can't be clobbered?
[05:48:59] <Dark_Shikari> whatever I swapped the variables in my second example
[05:49:00] <Dark_Shikari> yes
[05:49:03] <Dark_Shikari> y cannot be clobbered, x can
[05:49:07] <av500> ok, i want this one: http://www.theinquirer.net/inquirer/news/1652779/powercolor-hd5970-display-outputs
[05:49:49] <astrange> can't see how it can be faster than and, test
[05:50:05] <astrange> even if x86 had multiple condition flags like ppc it'd still be 2+ insns
[05:50:14] <Dark_Shikari> the problem is the mov
[05:50:19] <Dark_Shikari> because we need to back up y
[05:50:45] <astrange> can you move it up somewhere else to schedule it?
[05:50:57] <Dark_Shikari> we're already saturating the execution units
[05:51:06] <Dark_Shikari> every op costs
[05:51:51] <Dark_Shikari> it just annoys me because the code _looks_ like it can be done in 2
[05:51:53] <Dark_Shikari> but it takes 3
[05:53:09] <astrange> xor y, x; test $1, x
[05:53:19] <astrange> does that work? (dest on right)
[05:54:06] <Dark_Shikari> doesn't that clobber y?
[05:54:08] <Dark_Shikari> oh
[05:54:11] <Dark_Shikari> dest on right
[05:54:17] <Dark_Shikari> lemme try
[05:57:52] <astrange> that's (x&1) != (y&1), which might fit if you reverse the branch condition
[05:58:17] <astrange> ppc has a not+xor instruction (called eqv) but not x86
[05:58:36] <Dark_Shikari> ah, nice
[06:46:46] <merbzt> peloverde: how much faster is the dct-iv method ?
[06:48:01] <KotH> moin girls
[06:48:03] <superdump> probably quite a lot i expect
[06:52:55] <Tjoppen> bcoudurier: did you have/use any samples to handle PixelLayout?
[06:53:10] <bcoudurier> I got one from opencube
[06:53:14] <bcoudurier> uncompressed rgb
[06:55:00] <Tjoppen> ok. it's not up on samples.mplayerhq.hu is it?
[06:55:29] <Tjoppen> although, assuming regular rgb24 it should probably match what I have in my patch
[06:55:37] <Tjoppen> (R8G8B8)
[06:59:08] <Tjoppen> anyway, I've sent some questions to our client, who may have a contact that can shed some light on this
[07:10:56] <bcoudurier> nice
[07:48:52] <ohsix> Dark_Shikari: do you use compcache at all? (re: firefox discussion earlier)
[07:49:43] <Dark_Shikari> compcache?
[07:49:45] <Dark_Shikari> what's that
[07:50:19] <ohsix> its a compressed page cache thing, midlevel swap that'll compress pages before they hit real swap
[07:50:43] <kshishkov> sounds like a wonderful thing to have misterious RAm-related errors
[07:51:10] <ohsix> it makes ff act much better with just a tiny bit of it, since most of the pages are junk and compress to almost nothing, or never reach the backing device
[07:51:18] <Dark_Shikari> I have never had an extension speed up anything
[07:51:22] <Dark_Shikari> =p
[07:51:39] <ohsix> :O its not a firefox extension, its a linux kernel module
[07:51:44] <Dark_Shikari> oh, linux kernel module
[07:51:46] <Dark_Shikari> I'm on windows
[07:51:52] <ohsix> ahhh
[07:51:55] <kshishkov> there is even FF extension (windows-only) for garbage collecting
[07:52:05] <ohsix> yea but ff does that already
[07:52:36] <Dark_Shikari> In theory
[07:52:37] <ohsix> junk that just gives you a button to do it any time sure are empowering! :D
[07:52:37] <Dark_Shikari> ;)
[07:53:14] * kshishkov remembers that system call in Java which "recommends runtime environment to start GC"
[07:53:19] <ohsix> i'd really like to help you figure out whats fuxed in your ff stuff sometime
[07:53:49] <Dark_Shikari> I suspect adblock plus
[07:54:11] <ohsix> like i said earlier, on windows the OS dies before ff starts messing up (for me at least) and i have outrageous browser sessions (300 tabs is nothing, maybe 1200)
[07:54:31] <KotH> o_0
[07:54:36] <ohsix> there was a thing with abp ages ago, 2.0 or something with the object tabs; but nothing since
[07:54:42] <Tjoppen> kshishkov: you can spite the jvm by catching OutOfMemoryError and calling System.gc()
[07:54:55] <KotH> ohsix: ok, you've done it
[07:55:08] <ohsix> hm?
[07:55:12] <KotH> ohsix: someone who has more ff tabs/windows open that i have... increadible
[07:55:34] <Dark_Shikari> the main thing to note is that firefox cpu just randomly spikes
[07:55:34] <ohsix> oh, hah; yea i use them as temporary bookmarks of a sort, but i open more in a day than i close in a day
[07:55:38] <Dark_Shikari> and during those spikes, the browser freezes
[07:55:47] <kshishkov> Tjoppen: still, that call is only a hint to it that user wants to do GC. And usually I had to work with web servers which had PermGen memory leak (by JRE design).
[07:55:53] <ohsix> Dark_Shikari: ya by the time it does that some shit already went down
[07:55:57] <Tjoppen> got another matrox sample today. mpeg2video with b-frames in avi. ffplay thinks it's 6.25 fps instead of 25 (gop size is 4)
[07:56:16] <kshishkov> Dark_Shikari: combine those spikes with Flash
[07:56:17] <Tjoppen> kshishkov: nasty
[07:56:23] <ohsix> even if ff can have 1200 tabs open, if you regularly navigate in a handful of them it'll eventually start doing that (on windows, on lunix it "feels" different)
[07:56:24] <Dark_Shikari> kshishkov: yup
[07:56:29] <Dark_Shikari> Flash playing a video locks up hard
[07:56:34] <ohsix> KotH: http://lez.ath.cx/~ohsix/images/1207.png
[07:56:35] <Dark_Shikari> Of course, ANYTHING playing a video locks up
[07:56:38] <Dark_Shikari> BECAUSE ITS ALL IN ONE FUCKING THREAD
[07:57:02] <Tjoppen> when I worked with a Wicket based web thing a while back PermGen crashes happened about once per hour
[07:57:08] * KotH puts a [CRAZY] stamp on ohsix forehead
[07:57:08] <ohsix> Dark_Shikari: its actually a bunch; but the one host thread is quite contended :O
[07:58:36] <ohsix> the cycle collector touches a lot of stuff too; and ff will try and collect every 30 seconds or so even when it's not gaining anything after the session has gone off the deep end (again windows only, i haven't seen it do that in linux, it may have to do with some windows-y resource lifecycle like other bugs i've encountered, not so much VM stuff)
[07:59:21] <ohsix> one thing that does suck about FF is theres no way to know whats going on in some tab with respect to the resources or plugins its using, so cleanup is untenable after a certain amount of time
[07:59:55] <ohsix> gotta really dive in to the stuff you've opened in the last few minutes _just_ as the freezing thing you describe starts to happen, scrolling iframes and stuff can do it a lot
[08:00:48] <Dark_Shikari> ohsix: could it be related to that massive gigantic sql database?
[08:00:52] <ohsix> and at least on windows, you might need to restart the computer along with the browser after you've squashed the likely candidate
[08:01:03] <Dark_Shikari> and it doesn't "start to happen", it always happens
[08:01:06] <Dark_Shikari> 100% of the time
[08:01:07] <ohsix> Dark_Shikari: nah; its actually rarely touched, and its the most async bit of it
[08:01:24] <Dark_Shikari> "rarely touched"?  every time I type in something to the search bar firefox lags for 10 seconds
[08:01:25] <ohsix> if its 100% of the time then the thing doing it is already somewhere in your session
[08:01:28] <Dark_Shikari> as it crunches the hard drive
[08:01:35] <Dark_Shikari> looking through its sqlite
[08:01:38] <ohsix> thats weird
[08:01:49] <Dark_Shikari> I'm guessing I have a big sqlite.
[08:02:29] <ohsix> well if you suspect it; http://lifehacker.com/5344418/make-firefox-faster-by-vacuuming-your-database (no restart required)
[08:02:38] <Dark_Shikari> places.sqlite: 52736KB
[08:02:48] <ohsix> there isn't anything periodic going on with sqlite that'll cause a ui freeze
[08:03:51] <Dark_Shikari> oh wow.  that did help
[08:03:56] <ohsix> thats pretty big, my history is set to 2 years and mines only 23mb
[08:04:08] <ohsix> for the awesome bar or the whole thing?
[08:04:21] <Dark_Shikari> aweombar
[08:04:23] <Dark_Shikari> *awesomebar
[08:04:31] <ohsix> upgrading ff should vacuum the db a bit, and most users have recent enough
[08:04:49] <ohsix> well cool; the freezy bits are a diff problem though, and the more insidious one
[08:04:55] <Dark_Shikari> hmm, seems to not be freezing much now.  it always gets worse when I open tons of tabs.
[08:05:01] <Dark_Shikari> only got ~30 now
[08:05:04] <ohsix> do you use flashblock and stuff?
[08:05:21] <ohsix> without flashblock a browser with even 10 tabs starts to get unruly
[08:05:32] <Dark_Shikari> flashblock
[08:05:34] <Dark_Shikari> yes
[08:06:18] <ohsix> one time i was having this freezing problem and you'll never guess what it was, it was a tab wayyy on the left (by order of opening, and months old) that had snowflakes stuttering around the page for the holidays
[08:06:30] <Dark_Shikari> LOL
[08:06:38] <Dark_Shikari> this is why you need tree-style tabs
[08:06:40] <ohsix> there are some sites with very poor image transition scripts and menus that you need to keep on top of too
[08:06:42] <Dark_Shikari> otherwise you can't manage a gajillion tabs
[08:06:56] <ohsix> i use tab kit, so they're colored depending on how they're open
[08:07:03] <Dark_Shikari> you can combine them
[08:07:06] <Dark_Shikari> also, I end having to periodically garbage-collect my tabs
[08:07:12] <Dark_Shikari> I go back every week or so and kill all the pastebins
[08:07:19] <ohsix> so generally the leftmost with the same color is where i started, if it turns out i didn't want to read it i can kill the whole group
[08:08:38] <ohsix> ah wacky, i close that stuff immediately; the junk i have in my session is articles or forays into wikipedia (or any wiki), or junk on rolling stone or the new yorker that i haven't gone back to read yet
[08:08:56] <Dark_Shikari> yup
[08:08:58] <ohsix> wikis suck hard when you can open browser instances so easy
[08:09:04] <Dark_Shikari> well, pastebins I end up with because "I might need it later"
[08:09:08] <Dark_Shikari> and I don't close tabs that I "might need later'
[08:09:27] <Dark_Shikari> it's like in RPGs, when you end the game with five billion missiles and a gajillion potions
[08:09:30] <Dark_Shikari> because "you might need it later"
[08:09:40] <Dark_Shikari> shooting peas at the giant crab out of the worry that there's a bigger crab behind him.
[08:09:52] <Dark_Shikari> can't throw away the tab, I might need it later.
[08:09:55] <ohsix> it would be cool to have a bogomips type number that shows what the elements on the page are doing with respect to the session; so you could get bad tabs before problems start
[08:10:25] <ohsix> haha i had a bunch of stuff for a japanese game i was playing (and learning japanese) like a year ago that i only just closed, hadn't played the game in 8mo
[08:11:16] <ohsix> what i'd really like is a tab that would exist as a title and a siteicon and nothing else; or some sort of bag you can throw stuff in (a lot like Places tags, but with an actual button in the button bar so i'm reminded to get to it)
[08:11:17] * kshishkov usually has billions of stuff mostly because he likes to hexedit savegames
[08:11:51] <ohsix> i only keep the tabs around for the first class visibility; but bookmarks would do well too
[08:11:59] <av500> kshishkov: you can edit ff save status to have millions of tabs?
[08:12:05] <Dark_Shikari> yeah, I use tabs as reminders too
[08:12:11] <Dark_Shikari> I leave a pastebin of a patch I need to look over at the bottom
[08:12:14] <Dark_Shikari> and then I look at it later
[08:12:49] <ohsix> if you really have 30 tabs and the freezing problem that i'm thinking; something weird is definitely up
[08:13:01] <ohsix> do you use zonealarm or macaffee?
[08:13:40] <ohsix> mcafee, even; stupid name
[08:14:03] * av500 has neither
[08:15:13] <ohsix> there are some weird software interactions that you really wouldn't expect to affect ff
[08:16:09] <Compn> what i want is a way to stop rendering on tabs
[08:16:16] <kshishkov> av500: but of course! I don't think it's hard
[08:16:22] <ohsix> it shouldn't take any effort  to keep ff notsuck but it does :[
[08:16:39] <Compn> e.g. any non active (for x hours) tab just goes to blank page but remembers url, or caches the page to html and then reads that when i activate said tab again
[08:16:46] <ohsix> Compn: yes, that'd be awesome; or at the very least some way to quantify how intensive the page is going to be to keep open
[08:16:55] <Honoome> j0sh_: I seriously would like to avoid bison (and flex) as much as I can ;)
[08:17:00] <av500> kshishkov: make me a ff save game that has every dilbert open in a tab each
[08:17:12] <Compn> its how i'm currently setup, i open my browser pointed at an invalid proxy, so all my tabs are blank , until i f5 them (after turning off proxy settings)
[08:17:48] <ohsix> ff sort of does that already, but if you pick the tab it makes it a full class deal with a bfcache and everything, so if you've visited that tab at all in that run session of ff its footprint doesn't go back to what it does when you initially start ff and it opens your session
[08:18:14] <Dark_Shikari> ff needs to stop dropping compressed data out of cache before it drops the uncompressed versions
[08:18:18] <Dark_Shikari> it's really freaking annoying
[08:18:52] <ohsix> uncompressed versions are generally owned by the os tho, its the more expensive time wise version to create
[08:18:57] <j0sh_> Honoome: understood, i kinda figured that
[08:19:06] <Dark_Shikari> ohsix: huh?  "owned by the os"?
[08:19:16] <ohsix> s/os/display driver setup; they're made into DDB's on windows for speed
[08:19:16] <kshishkov> av500: it stores it in sessionstore.js in profile, so it's easy to edit
[08:19:19] <Dark_Shikari> if you aren't vieweing a tab, firefox is free to drop it
[08:19:21] <merbzt1> umm, is it possible to loose git commits when merging ?
[08:19:24] <Dark_Shikari> Or, at least, should be.
[08:19:38] <merbzt1> sort of lost last days of commits
[08:19:51] <Dark_Shikari> merbzt1: git fsck ?
[08:20:32] <ohsix> post-reflow there is a sizable cache of the page kept around in case you might visit them again; on windows it basically means you can have a 2000 tab session, but if you actually navigate in _one_ of the tabs the whole thing will go tits up
[08:21:33] <Honoome> j0sh_: I'll take a look at json-c… we already need that for statistics in the first place, so…
[08:21:48] <Dark_Shikari> why can't firefox free the ddbs?
[08:22:08] <ohsix> its  hard to characterize, but navigating in a big session can get quite hazardous as far as ff's interactivity goes, and it'll let you just keep opening tabs
[08:22:35] <ohsix> Dark_Shikari: it can and will, but the lifecycle is different because they're generally so expensive to create
[08:22:42] <janneg_> twnqx: ping
[08:22:42] <j0sh_> Honoome: json parsers are small enough that you should just be able to include it without requiring an additional download
[08:22:56] <twnqx> janneg_: pong
[08:23:17] <Compn> i dislike firefox tho
[08:23:21] <ohsix> ddb = GDI to "native" color conversion and generally relocation to nonlocal memory (pci)
[08:23:33] <Compn> any time i leave it open for 6+ hours it starts to become unresponsive
[08:23:40] <Honoome> j0sh_: okay now I need to point you to the problem of bundled libraries I actually spend lots of time to work on ;)
[08:23:42] <wbs> bcoudurier: I've got some experience with the apple http live streaming stuff, what do you want to know?
[08:23:44] <Compn> with 2 tabs, i mean really ;\
[08:23:44] <ohsix> Compn: on windows too?
[08:23:56] <Compn> ya windows
[08:24:06] <ohsix> whats in the 2 tabs?
[08:24:19] <wbs> j0sh_: I'll look through your rtsp stuff in a while, it takes a while to get up to speed even after half a day without connection ;P
[08:24:27] <Compn> one static page, i forgot the other page, must have had some heavy html elements
[08:24:30] <bcoudurier> what kind of events does the device report to the application ?
[08:24:33] <Compn> i had to force kill it...
[08:24:36] <ohsix> Dark_Shikari: a minimized ff window is "special" on windows; it'll minimize its footprint too, freeing caches and ddbs and stuff
[08:24:47] <ohsix> Compn: weird
[08:25:26] <wbs> bcoudurier: I haven't used it embedded in an app, just done streaming in their player (launched from safari)
[08:25:26] <Compn> ohsix : er, that was my dad's firefox actually
[08:25:27] <j0sh_> Honoome: yeah, but the json parser is two files, surely it can't be that much trouble :)
[08:25:39] <j0sh_> wbs: cool, no rush. ive been working on vlc (scary)
[08:25:40] <Compn> ohsix : i can leave my firefox open for days no problem, with yahoo and gmail open even
[08:25:47] <Honoome> j0sh_: it surely can! bundled code is always bad…
[08:25:50] <Compn> 6+ tabs
[08:25:50] <janneg_> twnqx: rendering finished?
[08:25:58] <twnqx> dunno
[08:26:00] <Compn> not sure what the problem is, both are updated
[08:26:02] <twnqx> there... might be issues
[08:26:21] * Honoome got his new laptop! time to fetch sysresccd, and install Gentoo on it
[08:26:26] <twnqx> apparently it causes problems running multiple instances of the same scene in parallel
[08:26:30] <Compn> yahoo + gmail open at all times really slow down some browsers tho
[08:26:34] <ohsix> ff on windows can get windows itself into a zombie state after a looong session; because it _will_ leak gdi resources if intermediate operations fail, and last i looked ff created a brush for every window it created, when theres a stock object (always already created, and shared), so that already is a bit of pressure on it for a long session, it didn't destroy those brushes with the window either, some 4000~ windows in and it'll shit up
[08:27:11] <kshishkov> Honoome: that's cheating - fetch sysrescfloppy instead
[08:27:28] <ohsix> Compn: ah ya gmail will, news sites like huffington post reaaaally will too (they update their frontpage in your browser every 15 seconds or so, even when nothing changed, still causes a page reflow)
[08:27:57] <Honoome> kshishkov: I actually use a sysresc-usbstick ;)
[08:28:33] <kshishkov> Honoome: and you should use tftp-boot with networked installer
[08:29:07] <Compn> ya huffpo is really cpu intensive
[08:29:40] <Compn> in the olden days i used to surf without javascript, it would speed up things so much. but now that just breaks most sites
[08:29:45] <Honoome> hrmm this stuff _could_ use UEFI for booting… maybe another time
[08:29:53] * j0sh_ wtfs at 300kb gzipped json implementations
[08:30:01] <lu_zero> uh?
[08:30:04] <lu_zero> what's that?
[08:30:10] <lu_zero> (btw hi)
[08:30:17] <j0sh_> yo
[08:30:21] <ohsix> Dark_Shikari: do you have your swap file disabled?
[08:30:24] <Compn> anyways, we should probably move this browser talk, ffmpeg talk is starting up :P
[08:30:33] <j0sh_> a couple, json-c and M's json parser (that ive looked at so far)
[08:30:35] <Honoome> lu_zero: hey
[08:30:45] <Honoome> lu_zero: I'm not much positive about the scanners thing btw
[08:30:57] <bcoudurier> wbs, ah ok thanks anyway :)
[08:31:00] <Honoome> but j0sh_'s idea of using json for config file is not too bad I guess ;)
[08:31:33] <jai> which config file?
[08:31:38] <j0sh_> for feng
[08:31:39] <wbs> bcoudurier: but given it's apple, I don't think you get many or interesting callbacks, probably mostly "starting playback", "playback finished" or "error" ;P
[08:31:46] <jai> right, k
[08:32:10] <janneg> KotH_: rendering 1 last frame on fogir
[08:32:11] <bcoudurier> I was more interested in the discontinuity callback :/
[08:32:46] <wbs> bcoudurier: ah.. don't know about that, but I'd guess you don't get any such
[08:34:29] <Dark_Shikari> ohsix: no, otherwise my computer would be rather dead
[08:36:02] <merbzt1> Dark_Shikari: ok I found a few commits, how do I get the data from it ?
[08:37:04] <lu_zero> Honoome: json might be good
[08:37:20] <Honoome> lu_zero: would you consider replacing .ds and .sd files with json files as well at that point? ;)
[08:37:22] <Dark_Shikari> merbzt1: cherrypick?
[08:37:34] <lu_zero> agreed
[08:37:42] <lu_zero> you want me to do that now?
[08:37:59] <Honoome> lu_zero: not yet… I'll be looking into that myself, I want to make sure json-c is decent before staring to use it…
[08:38:19] <Honoome> and before I can do that, I have a laptop here to install :D
[08:38:39] <lu_zero> pff
[08:39:00] <j0sh_> Honoome: or you can use ragel to build a json parser yourself :)
[08:39:06] <lu_zero> I'd rather keep the working config files here and figure out why we clog in threads
[08:39:09] <ohsix> sometimes you can close an app you think has a ton of gdi handles and get the error dialogue and kill some more stuff, but in general until the window station (basically the computer) is restarted everything is a mess, there is a tipping point that you cross unknowingly long before it happens too; so its a major bitch avoiding it
[08:39:09] <Honoome> j0sh_: I could also avoid that :P
[08:39:19] <ohsix> derp that was supposed to be a message
[08:39:37] <Honoome> lu_zero: that's one thing, but I think it might require some config file to do that ……
[08:39:44] <lu_zero> Honoome: or we could import cpython and use those data structures...
[08:39:48] <Honoome> …
[08:40:05] <Honoome> who's at LinuxTag later on today? I need somebody to smack lu_zero in the head
[08:40:07] * lu_zero notices that this terminal isn't fully working
[08:40:18] <av500> Honoome: I can do that tomorrow
[08:40:37] * lu_zero is afraid...
[08:40:44] <av500> together with kshishkov .....
[08:40:46] <Honoome> av500: thank you :) lots of people will be grateful of that :P
[08:41:01] <av500> youtube later....
[08:41:16] <lu_zero> Honoome: so the 2 biggest devs around should smack my head?
[08:41:19] <merbzt1> yey found my lost code
[08:41:21] <av500> or we put it in videowall
[08:43:24] <Honoome> lu_zero: maybe you'll stop proposing silly stuff like… you know… PYTHON =_=
[08:43:53] <lu_zero> cpython data structs are nice
[08:43:59] <KotH> janneg: feel free to render anything :)
[08:44:08] <KotH> janneg: i'll leave the machine as it is until end of the week
[08:44:25] <j0sh_> lu_zero: if you want to embed a language, make it something lightweight like lua
[08:44:41] <av500> KotH: new project: highres render of lu_zero being smacked in the head
[08:45:06] <lu_zero> j0sh_: it's just a loong time inside joke
[08:45:24] <Honoome> j0sh_: that was actually one of my options before… I have quite a few applications here using lua already… might actually not be too bad
[08:45:24] <lu_zero> Honoome: cannot stand python bracketless
[08:45:52] <Honoome> I cannot stand python. Full stop.
[08:45:52] <KotH> av500: lol
[08:45:57] <lu_zero> and I'm not so happy of rails and ruby interpreters
[08:46:04] <j0sh_> hehe
[08:46:23] <lu_zero> Honoome: add brackets to python and you get a subset of what ruby parses =P
[08:46:31] <Dark_Shikari> complaining about python and ruby in the same sentence is like complaining about car horns and the nazis in the same sentence
[08:46:32] <lu_zero> pardon
[08:47:04] <lu_zero> adding an "end" statement would suffice
[08:47:11] <lu_zero> =)
[08:47:16] <Honoome> half the Ruby extensions' developers needs to be kicked in the nuts… but that's a different problem
[08:47:33] <av500> Honoome: send them to ffmpeg booth tomorrow...
[08:47:39] <Dark_Shikari> monkeypatching monkeypatching monkeypatching
[08:49:02] <lu_zero> theheh
[08:49:20] <lu_zero> Honoome: still python is a bit nicer IMHO
[08:49:36] <Honoome> Dark_Shikari: monkeypatching is the least of the problems with Ruby
[08:49:36] <KotH> boys boy boys
[08:49:52] <KotH> why do you quarrel over such petty languages, when you fully know that you should be using perl?
[08:50:08] <lu_zero> KotH: I was thining parrot
[08:50:31] <lu_zero> why using an higher level abstraction when you can code in interpreted-asm?
[08:50:33] <Honoome> you've been thinking parrot since 2007 or so … and I still have to see it doin anything useful
[08:50:56] * j0sh_ uses a magnetic needle and a steady hand
[08:50:57] <lu_zero> Honoome: I consider parrot a fun idea, that's all
[08:51:17] <lu_zero> j0sh_: Honoome uses emacs...
[08:51:19] <lu_zero> !
[08:51:21] <lu_zero> lisp!
[08:51:29] <j0sh_> hehe
[08:51:35] <lu_zero> that's the solution (and the cause) or our problems!
[08:51:56] * Honoome is a happy emacs user indeed
[08:52:02] <Honoome> so M-x butterfly
[08:53:59] <lu_zero> j-b: when you'll arrive?
[08:54:53] <j-b> to?
[08:55:08] <lu_zero> linuxtag?
[08:55:18] <Honoome> AAAAH! I forgot one thing
[08:55:22] <j-b> I won't come. Too much work. I'll send some minions for me
[08:55:39] <lu_zero> ahhh
[08:55:47] <av500> mignons?
[08:55:59] <j-b> av500: no idea, I am not gay (yet)
[08:56:30] <Honoome> I haven't chosen an hostname for the laptop yet! Memory Alpha, to me!
[08:56:36] <lu_zero> ...
[08:56:57] <av500> Honoome: what model?
[08:57:07] <Honoome> av500: Latitude E6500
[08:59:58] <Honoome> Saladin or Centaur… hmm
[09:00:15] <ohsix> bebop or rocksteady
[09:00:56] <lu_zero> ${klingon_ship}
[09:01:28] <lu_zero> additional bonus point if you actually write the name in the original characters
[09:01:35] <lu_zero> linux supports them...
[09:01:47] <Honoome> lu_zero: I use klingon names only for non-x86/amd64 boxes
[09:02:17] <lu_zero> ah
[09:02:41] <Honoome> and romulan names for second-hand boxes…
[09:02:46] <j0sh_> nice. which one is kahless?
[09:03:11] <spaam> Honoome: can you speak klingon?
[09:03:12] <j0sh_> lu_zero: im sure theres a unicode spec for klingon script somewhere
[09:03:13] <Dark_Shikari> lol
[09:03:17] <Dark_Shikari> j0sh_: yeah, there is
[09:03:22] <Honoome> j0sh_: none, my one and only sparc box was Klothos :P
[09:03:37] <Honoome> not an official one though, Unicode rejects “fake” languages
[09:03:46] <KotH> Honoome: fake?
[09:03:55] <KotH> Honoome: hey! there are people who are speaking it!
[09:04:00] <j0sh_> marc okrand invented it. it's as real as anything else
[09:04:14] <j0sh_> iirc some hospital in oregon had to hire a fulltime klingon interpreter
[09:04:20] <Honoome> I don't doubt so (thus my quotes btw, arguably _any_ language is invented)
[09:04:22] <KotH> j0sh_: what?
[09:04:24] <j0sh_> because they had enough patients coming in who refused to speak anything else
[09:04:29] <KotH> j0sh_: er..
[09:04:44] <KotH> j0sh_: i doubt that thera are more than a couple of people who speak klingon fluently
[09:04:47] <Dark_Shikari> lol
[09:04:49] <j0sh_> http://www.foxnews.com/story/0,2933,86550,00.html
[09:04:56] <Dark_Shikari> that sounds like an urban legend
[09:04:58] <Honoome> KotH: you underestimate the trekkies!
[09:05:24] <KotH> Honoome: i'm not understimating trekkies, but i know how complex the klingon language is.. especially its pronounciation
[09:05:26] * Dark_Shikari uses touhou names 
[09:05:43] <j0sh_> KotH: you would be surprised
[09:05:46] * lu_zero uses... random stuff
[09:06:00] * Honoome needs to fix his symbols' harvesting script, damn
[09:06:21] <KotH> j0sh_: "mental health patients"... i doubt they have the capacity and deligence to learn klingon :)
[09:06:35] <av500> no, they go mental after they learn klingon
[09:07:15] <j0sh_> lol
[09:07:21] <j0sh_> trying to find this awesome article...
[09:07:28] <j0sh_> some guy raised his child speaking klingon
[09:08:01] <av500> http://blogs.citypages.com/blotter/2009/11/dinkytown_dad_s.php
[09:08:09] <Honoome> Saladin will be… Centaur is slightly more difficult to pronounce ^^;;
[09:09:04] <av500> saladiere?
[09:09:10] <j0sh_> http://www.washingtoncitypaper.com/articles/10873/klingon-as-a-second-language
[09:09:38] <av500> Honoome: they look similar you have to admit: http://www.koelwebshop.nl/images/Greenland_S903.jpg
[09:09:53] <Honoome> rotfl :)
[09:09:58] <Tjoppen> I keep forgetting.. does gop_size == 1 also mean intra-only? or is it the number of frames between I-frames?
[09:10:13] <Honoome> http://memory-alpha.org/wiki/Saladin_class ;)
[09:10:22] <Tjoppen> == 0 is intra-only according to docs, but it's a bit unclear of what it means by "size"
[09:13:02] <Dark_Shikari> 1 means intra only
[09:13:41] <Tjoppen> so both 0 and 1 means intra only then, unless the doc is incorrect
[09:17:53] <hyc> Anyone object to adding preset support to ffserver config? https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-June/089897.html
[09:18:02] <hyc> should I ping the list again?
[09:19:08] <wbs> hyc: yeah, add baptiste's name explicitly there so he notices that you want an explicit answer on that
[09:19:43] <lu_zero> uhmm
[09:21:24] <hyc> okeydoke
[09:45:23] <Dark_Shikari> hyc: +1 to preset change
[09:47:46] <hyc> cool. will just have to see what Baptiste says
[09:48:18] <hyc> it's just so nuts having to manually insert those preset settings into ffserver.conf
[09:49:55] <Dark_Shikari> he said ok
[09:52:34] <hyc> ah, great. (just checked my inbox)


More information about the FFmpeg-devel-irc mailing list