[FFmpeg-devel-irc] IRC log for 2010-01-29
irc at mansr.com
irc at mansr.com
Sat Jan 30 01:00:02 CET 2010
[00:35:54] <astrange> http://forum.slysoft.com/showthread.php?t=37966 "FFmpeg maintainers receive SlySoft products free of charge"
[00:36:11] <astrange> wonder what maintainer means
[00:44:29] <Compn> lol no clue
[00:44:43] <mru> people in MAINTAINERS?
[01:07:49] <iive> what does slysoft produce?
[01:08:23] <iive> clonedvd/anydvd, cool
[01:09:36] <mru> hmm, there is a -Tbss ld flag
[01:09:40] <mru> let's try that out
[02:06:44] <beandog> wow, someone hook me up with a free copy
[02:06:52] * beandog wants to rip all his blurays
[02:30:35] <Dark_Shikari> mru: awesome patch series
[02:31:45] <mru> does it look correct?
[02:36:32] <astrange> didn't see any problem when reading it
[02:37:42] <astrange> iirc copy_block* in dsputil doesn't work with gcc-svn + strict aliasing (snowll regtest fails), i can check if this fixes it
[02:40:23] <mru> some of those should probably used aligned versions too
[02:40:39] <mru> no difference on x86 of course
[02:46:24] <mru> that blackfin is SLOW
[02:48:44] <mru> orders of magnitude slower than arm at same clock frequency
[02:49:14] <mru> the architecture isn't all that bad
[02:49:17] <mru> must be gcc
[05:55:02] <jai> morning all
[05:56:07] <pJok> mornings jai :)
[05:57:17] <jai> hey there pJok :)
[07:32:25] <peloverde> Does C have some sort of full switch directive like verilog?
[07:32:50] <Dark_Shikari> what do you mean?
[07:32:58] <Dark_Shikari> never used verilog
[07:33:09] <peloverde> let's say I read a 2 bit bitstream element
[07:33:19] <peloverde> and I switch on it
[07:33:23] <ohsix> where all cases have to be handled else failure?
[07:33:29] <kshishkov> ah, no
[07:33:36] <peloverde> where you tell the compiler that no other cases are possible
[07:33:42] <av500> default: exit(1);
[07:33:48] <kshishkov> there is no such thing in C
[07:33:54] <astrange> default: __builtin_unreachable()
[07:33:59] <astrange> ...in gcc 4.5
[07:34:05] <astrange> what do you want the compiler to do about it?
[07:34:19] <astrange> if it's 2 bit there's not a lot of cases anyway
[07:34:27] <ohsix> theres a gcc thing to make it pedantic, but it also checks the data type, concievably if it was a bitfield with 2 bits it might complain about anything >4
[07:34:37] <ohsix> s/4/3
[07:34:44] <peloverde> I don't want the compiler to consider the possibility of not hitting one of the cases
[07:35:06] <peloverde> And I certainly don't want to to warn that shit may be uninitialized
[07:36:34] <astrange> switch is equal probability iirc
[07:38:51] <peloverde> So Should I just replace case 3: with default then?
[07:39:01] <Dark_Shikari> you could place both there
[07:39:40] <ohsix> oic, i misunderstood what you meant
[07:42:23] <ohsix> i think the gcc thing i'm thinking about is with enums, you put the only possible items in the enum and switch on it; then gcc will tell you if theres a possible enumeration that isn't handled
[07:42:55] <peloverde> Hmm, that may work to my advantage
[07:44:01] <ohsix> i know it does it, but i'm gonna find it in the documentation
[07:46:23] <Dark_Shikari> ffs what's the option now to list pixfmts?
[07:46:27] <Dark_Shikari> -pixfmt list doesn't work
[07:46:48] <astrange> pix_fmt
[07:46:50] <Dark_Shikari> blah
[07:47:39] <ohsix> its -Wswitch-enum (and it complains even when theres a default case)
[07:48:21] <ohsix> -Wall only turns on -Wswitch which isn't raised when theres a default label even for enumerated types
[07:49:11] <ohsix> (without enumerated types, it doesn't matter, so -Wswitch-enum ftw)
[07:51:41] <ohsix> it'll complain if labels are in there that aren't in the enumeration too
[08:00:45] <superdump> mru: did you get fftrollbot working properly?
[08:00:51] <superdump> and where does the mail go?
[08:02:25] <kshishkov> ffmpeg-devel-irc
[08:02:45] <kshishkov> seems to work
[08:04:05] <thresh> oh noes i'm in ur mailing list bitching about theora
[08:05:14] <jai> heh
[08:05:27] <jai> anyone can subscribe to that ml?
[08:05:41] <kshishkov> you can do it yourself
[08:06:07] <ohsix> is that the place to send random irc logs and funny quotes or just from here!111
[08:36:08] <KotH> moin
[08:36:14] <KotH> .o0(no bbb)
[08:36:34] <KotH> when BBB comes online, he shall ping me
[10:33:52] * pJok wakes up Dark_Shikari
[10:35:05] <jai> already done by mru
[10:35:05] <superdump> CTCP requested by rdwkoafiv: VERSION GNAA > j00
[10:35:07] <superdump> heh
[10:35:28] <mru> who or what is j00?
[10:35:35] <superdump> you
[10:35:37] <superdump> :p
[10:35:37] <Dark_Shikari> he doesn't like jews.
[10:35:50] <mru> the gay nigger?
[10:36:01] <jai> yep
[10:36:19] <jai> ED has an obligatory page on them
[10:36:33] <pJok> and then all mods suddenly woke up ;)
[10:36:36] <Dark_Shikari> they are hilarious because their wikipedia page was nominated for deletion about 25 times
[10:36:39] <Dark_Shikari> and then finally deleted
[10:36:54] <mru> motivation?
[10:36:54] <superdump> hehe
[10:36:54] <jai> iirc they have a domain now
[10:37:27] <Dark_Shikari> mru: lulz
[10:38:06] <thresh> they did for years
[10:38:17] <thresh> since 2004
[10:38:27] <jai> ah
[11:10:58] * jai waits for kshishkov to go aussie
[11:12:07] <kshishkov> why? It's hot there even if Australia is girt by sea.
[11:12:22] <pross-au> Haha
[11:12:54] <pross-au> Whats Ukraine Beer taste like?
[11:13:22] <kshishkov> there was a good Australian word for it - 'piss'
[11:13:39] <jai> lulz
[11:13:50] <kshishkov> not that I've tastd it myself but I believe those who did
[11:16:34] <pross-au> The other name we have for crappy tasting beer is 'Fosters'
[11:17:22] <jez9999> the aussies dont actually frink Fosters, do they? :-)
[11:17:25] <jez9999> drink
[11:17:46] <pross-au> Correct: Real aussies dont drink it.
[11:19:30] <thresh> some ukrainian beers are really nice
[11:19:47] <thresh> with comparable price, they are really better than every russian one
[11:22:27] <pross-au> Thats good to know
[11:22:46] <pross-au> So when is the pub crawl?
[11:23:51] <jai> next ffmpeg dev meetup maybe ;)
[11:34:38] <mru> pross-au: pub crawl whenever you come to the uk ;-)
[11:39:00] <mru> people, I need suggestions for something witty to put on fosdem t-shirts
[11:39:19] <mru> I have to order them today or tomorrow
[11:39:39] <av500> "works with html5"
[11:40:18] <Honoome> "it's the other half that's hard" ? :P
[11:42:08] <kshishkov> "2010 is a year of Theora on desktop"
[11:42:16] <Honoome> HAHAHAHAHA
[11:43:55] <thresh> :)
[11:44:37] <jai> btw, maybe we should submit snow for http://www.acmmm10.org/authors/call-for-contributions/open-source-software-competition/
[11:46:37] <superdump> jai: maybe when it's cleaned up a bit more? :)
[11:46:45] <mru> what's that? the ioccc?
[11:46:50] <av500> "snow in a nut-shell"
[11:47:13] <jai> superdump: could you elaborate a bit?
[11:47:24] <superdump> av500: huhuhu, nice
[11:47:36] <jai> also i think it works pretty decently for a basic demonstration
[11:47:40] <superdump> jai: well, when the spec is completely finalised and the code is less buggy?
[11:47:43] <superdump> well, sure
[11:47:45] <jai> and might get some "publicity"
[11:47:52] <kshishkov> well, we don't have FFlossyaudiocodec yet :(
[11:47:59] <jai> superdump: indeed those are all desirable things
[11:48:01] <superdump> but it would be better to present when it's production or closer to production level
[11:48:05] <superdump> no?
[11:48:19] <kshishkov> no - look at Google
[11:48:26] <superdump> kshishkov: what was sonic?
[11:48:41] <superdump> kshishkov: haven't you heard of the phrase 'google beta'
[11:48:43] <superdump> ?
[11:48:54] <kshishkov> superdump: it was rip of Bonk lossless/lossy audio
[11:49:01] <superdump> ah
[11:49:17] <kshishkov> yes, same with Snow - permanent beta
[11:49:21] <superdump> so not a real lossy codec (MDCT-based or so)
[11:49:31] <kshishkov> no
[11:49:39] <superdump> i haven't actually tried snow recently
[11:49:40] <kshishkov> and no means of rate control either
[11:49:56] <superdump> but last time i did, it still had a chroma bug that rendered it not really usable
[11:50:21] <kshishkov> patchiswelcome
[11:50:47] <jai> superdump: is there a corresponding roundup entry
[11:51:09] <mru> if nobody has better ideas I might go with PH34R THE GR33N Z1GZ4G ...
[11:51:11] <superdump> jai: i think roundup was set up after i noticed the bug
[11:51:16] <mru> consider yourselves warned
[11:51:18] <jai> ah
[11:51:40] <superdump> mru: that'd be ok, but i think av500's html5 suggestion might be suitably humorous
[11:51:47] <superdump> albeit flame inducing
[11:52:02] <jai> superdump: isnt that the idea ;)
[11:52:19] <mru> just a zigzag and "works with html5" underneath?
[11:52:22] <superdump> because whatever codecs html5 <video> will use, ffmpeg should support them
[11:52:23] <superdump> :)
[11:52:30] <superdump> sounds brill
[11:52:39] <mru> I kinda like it
[11:52:46] <superdump> i thought you would
[11:52:47] <superdump> :)
[11:52:55] <av500> "works with html5 <video>"
[11:52:55] <superdump> well done av500 :)
[11:52:59] <av500> thx :)
[11:53:18] <mru> I don't think we need the <video> bit
[11:53:26] <av500> yes, makes it too long
[11:53:29] <kshishkov> "FFmpeg - worst multimedia patents offender"
[11:53:30] <av500> and ppl in the know
[11:53:34] <av500> know anyway
[11:53:53] <superdump> i read through all diego's comments and people's responses on the lwn chris blizzard thread
[11:54:07] <av500> superdump: url?
[11:54:08] <kshishkov> stupid last name
[11:54:20] <superdump> it was in yesterday's log
[11:54:29] <superdump> errrrm
[11:54:44] <superdump> http://lwn.net/Articles/370985/
[11:54:45] <jai> just hope people dont write <video src="foo.bik" />
[11:54:46] <superdump> that one i think
[11:55:16] <av500> superdump: yep, got it
[11:55:39] <superdump> a few points (sorry if this dredges up more discussion on the topic...)
[11:56:18] <superdump> we seem satisfied that distributing ffmpeg under lgpl 2.1 is orthogonal to downstream acquiring patent licenses from the mpeg-la
[11:56:20] <superdump> that's fine
[11:56:27] <superdump> and that that is no problem for us
[11:56:28] <av500> "..Your comparison with evolution vs creationists is insulting. .." lol
[11:56:46] <superdump> but, i do see that roc's points seem valid to me
[11:57:17] <mru> the truth is occasionally insulting
[11:57:30] <superdump> that is - maybe downstream can't afford, or doesn't want to afford the license fees
[11:57:55] <superdump> and more importantly in my mind, paying license fees to create and publish content is very undesirable
[11:58:04] <superdump> or rather, having to and being pursued to
[11:58:26] <superdump> though it's not clear with the gif example how strongly this might have been enforced, nor how enforceable it might be
[11:58:55] <superdump> i expect it might not be too enforceable for N million average joes publishing their own videos on their own personal websites
[11:59:33] <mru> mpeg-la will never try to enforce against random people with a website
[11:59:54] <superdump> nevertheless, ignoring these two points for real world usage seems... possibly ignorant
[11:59:57] <mru> they're not acting in desperation like the riaa/mpaa (yet)
[12:00:04] <superdump> no, indeed
[12:00:11] <superdump> and i think they appreciate ffmpeg actually
[12:00:16] <av500> mpeg-la cannot find out what encoder SW random user used for his lil video
[12:00:32] <mru> there are many ffmpeg users paying the fees
[12:00:35] <av500> but they can find out what bbc or gg uses
[12:01:03] <mru> you'll find ffmpeg-encoded stuff on directv
[12:01:17] <mru> not video though
[12:01:18] <superdump> but it was noted somewhere (in blizzard's article in fact iirc) that an mpeg-la representative thought it would be silly for them not to try to tap into internet distribution for revenue
[12:01:26] <mru> only still images in their silly interactive bits
[12:01:43] <superdump> i don't doubt people are paying fees, nor that it's even reasonable for some entities to do so
[12:02:02] <superdump> i don't even doubt that theora doesn't contain 'submarine patents'
[12:02:09] <superdump> erm
[12:02:12] <superdump> does*
[12:02:28] <superdump> and so is not the answer
[12:03:18] <superdump> but ideologically, having an unpatented and royalty-free codec that has at least reasonable quality might be worthwhile as a de facto baseline for web content
[12:03:35] <mru> these people are also forgetting that an encoder will use much more patented stuff than is required by the spec
[12:04:17] <av500> err, chrome uses system installed ffmpeg?
[12:04:22] <superdump> yeah
[12:04:26] <superdump> diego was wrong on that one
[12:04:45] <av500> on linux
[12:04:58] <ohsix> and his use of "sale" wrt the actual terms
[12:05:29] <superdump> yes
[12:05:35] <superdump> i think that was before he checked it out
[12:06:33] <ohsix> it was pure flame; pointless discussion :)
[12:07:20] <superdump> oh yeah, the stuff about fluendo proprietary codecs is slightly interesting too
[12:07:24] <superdump> yes, it's a solution
[12:07:35] <superdump> but it's not a spirit-of-foss-friendly one
[12:07:44] <superdump> not at all open
[12:07:56] <superdump> at least with using ffmpeg and paying a license fee you're using foss software
[12:08:08] <superdump> -software (no need to double up ;))
[12:08:27] <ohsix> but then you get into the head of the orgs who would buy such a thing and its not really a useful concern
[12:08:49] <ohsix> unless you plan to provide the same its out of your purview
[12:09:54] <ohsix> if anything the fact that you can break something like gstreamer into a bunch of parts and limit your exposure makes it pretty attractive
[12:10:42] <superdump> mmm
[12:46:11] <av500> superdump: so chrome does not use system ffmpeg
[12:46:26] * av500 is confused
[12:46:30] <superdump> i don't think it does, no
[12:46:35] <superdump> it uses ffmpeg-mt
[12:46:44] <superdump> maybe i can check
[12:46:45] <superdump> hang on
[12:47:10] <superdump> $ ls -l /opt/google/chrome/libffmpegsumo.so
[12:47:12] <superdump> -rw-r--r-- 1 root root 1180592 2010-01-20 22:05 /opt/google/chrome/libffmpegsumo.so
[12:47:15] <superdump> i'm guessing that is it
[12:47:30] <av500> right
[12:48:36] <thresh> the one which is made by google is not using
[12:48:41] <thresh> the one self-compiled does, though
[12:49:31] <av500> ok
[12:50:09] <av500> now, does it mean that end user that downloads chrome now has automatically a patent license for h264 decode?
[12:50:17] <av500> to be used with all h264 sw on his system?
[12:50:32] <av500> if so, why dont all distros ship chrome? :)
[12:51:24] <superdump> av500: cunning.... i like it
[12:51:32] <ohsix> av500: only the vendor was party to any agreements
[12:52:10] <superdump> mmm
[12:52:16] <superdump> i guess it's only a license for google software
[12:52:24] <av500> distros could also dl chrome as 1st post install step
[12:52:28] <superdump> and it doesn't proliferate throughout the system
[12:52:39] <av500> and use the ffmpeg that comes with it to hook to vlc, gst etc
[12:52:41] <ohsix> as they do with google earth :>
[12:55:19] <av500> mpeg-la: "every user needs to pay" -> gg "we pay for every copy" -> all users :"dl chrome", solved
[12:55:38] <av500> so ff needs a "get chrome" button :)
[12:56:07] <av500> instead of installing the flash plugin, users install the chrome pluing to watch web video
[12:56:10] <pross-au> s/firefox/chrome/g
[12:56:38] <pross-au> Seems to be the trend.
[12:56:49] <av500> problem solved as well
[12:58:02] <av500> mozilas phear of mpegla is so stupid, mpegla would love all the internet to be h264 for all the encoder and closed hw royalties
[12:59:02] <av500> not to mention the pr disaster of suing ff/mozilla
[12:59:39] <KotH> fear is offten irational
[12:59:39] <pross-au> Politics and Programmers Do Not Mix
[13:00:13] <KotH> and RMS fellows fears are often based on wild speculations of what might happen if one is to touch anything made by a comercial entity
[13:00:39] * KotH still wonders why too many FOSS people see companies as evil
[13:00:54] <pross-au> KotH: and vice versa
[13:00:56] <ohsix> like fluendo? :>
[13:01:05] * KotH wonders where these people work or whether they live from some form of pension
[13:01:13] <kshishkov> pross-au: Stallman managed to write some GNU progs
[13:01:28] <KotH> kshishkov: and we've seen the result
[13:01:40] <kshishkov> what, you don't like Emacs?
[13:01:49] <KotH> kshishkov: he build a church out of a political movement that started in his programming mind
[13:01:59] * KotH is from the vi fraction
[13:02:02] <pross-au> That's a four letter word in my books kshishkov
[13:02:12] <pross-au> ditto
[13:04:35] <pross-au> Ogg/Theora has set back open-source multimedia by <INSERT THIS MANY YEARS>.
[13:05:12] <kshishkov> pross-au: how? IIRC it's an abbreviation for 'Escape Meta Alt Control Shift' - the modifiers you need to push at once for basic function calls
[13:06:27] <pross-au> I share an office with an Emacs zealot, okay.
[13:06:57] <kshishkov> I'm well protected against Australian Zealots
[13:07:19] <pross-au> Patriot, Patriot.
[13:07:44] * kshishkov lives in ~60km from Russian border
[13:08:02] <pross-au> We're all on the same side, right?
[13:08:29] <pross-au> ciao
[13:14:19] * KotH still thinks that ffmpeg & mplayer should publish a statement to the ogg and theora afair and get it settled once for all
[13:14:43] <thresh> of an one-liner, "theora sucks"?
[13:16:10] <kshishkov> nah, it's better to do in scientific way
[13:16:42] <kshishkov> give pictures of frames, explain theory, whatever
[13:17:42] <kshishkov> preferably with all artifacts described, CPU meansurements for postprocessing vs. decoder itself
[13:18:36] <kshishkov> and show that mpeg-2 with really good postprocessing can do the same in less time and botrate
[13:21:51] <thresh> like someone will actually read it
[13:22:06] <thresh> those who are deep into codecs already know it i think
[13:22:13] <superdump> is theora included in the msu test?
[13:22:27] <superdump> or is that only for h.264 encoders?
[13:22:28] <kshishkov> thresh: do you count Xiph devs?
[13:22:38] <kshishkov> h.264 only
[13:27:59] <KotH> thresh: it's not about codec devs, it's about forming a public opinion
[13:28:18] <KotH> thresh: currently, people hear only the voice of uninformed freetards and xiph people
[13:29:13] <ohsix> i don't hear anything from xiph or freetards; just people that wont walk into a minefield because they have legal counsel :>
[13:32:31] <av500> "<vendor> is going to add a DEBLOCKing hardcore process inside CPU for next version...."
[13:32:41] <av500> no more blocky porn I guess :)
[13:34:11] <KotH> ohsix: these are freetards
[13:34:36] <KotH> ohsix: if you know even a little bit of video coding, you know that it is impossible to create a codec that does not violate any patents
[13:34:54] <ohsix> even if you dont
[13:35:17] <KotH> no, those who dont know believe taht theora is patent free
[13:35:41] <ohsix> but that kind of reasoning is a dead end, you cant decry something that isn't just a hand over eyes position
[13:36:17] <ohsix> you dont prove something isn't covered by patents; you wait for legal claims
[13:36:47] <ohsix> its like a crook robbing every house with an address ending in 6
[13:55:16] <KotH> ohsix: tell taht all those stupid people who think that the patent issue is the most important thing in h264 vs theora
[13:56:07] <ohsix> if you're picking one it is, regardless of the comparative merit
[13:56:28] <KotH> i dont understand what you mean
[13:56:52] <av500> just demand from all theora users to watch videos on a patent free PC and wear patentfree clothes
[13:57:19] <ohsix> h264 is "safer" because the people who would make claims are the people you license it from, with theora you are just waiting for something to pop up
[13:58:06] * superdump kicks CIA-27
[13:58:27] <ohsix> patents represent an unaccountable legal liability
[13:58:31] <mru> superdump: cia isn't identified
[13:58:38] <superdump> hahaha
[13:58:40] <mru> so he can't speak with +R
[13:58:47] <ohsix> superdump: chan is +R
[13:58:48] <mru> fftrollbot is identified
[13:58:52] <superdump> fail
[13:59:10] <KotH> someone give it voice, then!
[13:59:15] <KotH> ohsix: ack
[13:59:44] <mru> but there are so many cia
[14:00:00] <av500> actually, existence of patent pool does not save one from trolls
[14:00:45] <ohsix> but its a number; andthey have a stake in what you licensed and will defend it
[14:01:43] <ohsix> or not, depends on wether they think it can withstand a legal case :>
[14:01:49] <av500> ohsix: hmm, I doubt they give you protection from 3rd party claims
[14:02:06] <mru> probably not directly
[14:02:12] <ohsix> not protection, no; but a common interest
[14:02:46] <av500> well, I did not see that mpegla jumped at At&t when they tried to enforce some additional mpeg4 patents
[14:03:08] <ohsix> you CYA to a certain degree when you remove a large number of possible claims vs. completely open ended claims
[14:03:14] <mru> but if you were a troll, who would you prefer to battle, every broadcaster, film studio, and major internet company, or the freetards?
[14:03:35] <av500> the one that have money
[14:03:51] <av500> so "every broadcaster, film studio, and major internet company"
[14:03:51] <kshishkov> av500: no, battle the easy prey
[14:04:01] <mru> you want a target with enough money for you to profit but not enough to beat you
[14:04:02] <av500> kshishkov: what for, if they have no money
[14:04:17] <mru> you'll never win against everybody at once
[14:04:21] <av500> come on, mozilla foundation with 80 million, thats nothing
[14:04:44] <ohsix> ya you can set precedent and make an artifact more valuable by getting people to rule on them; sometimes that makes an improbable litigant the best target
[14:05:16] <av500> anyway, we will never find out
[14:05:33] <av500> as gg et al will not use theorafor technical reason
[14:05:41] <ohsix> not until someone steps into the minefield
[14:05:50] <av500> and trolls will not bother to sue mozilla over theora
[14:06:15] <av500> even if theora was 1000% patent free, it is still crap
[14:06:27] <av500> so it will not be adopted
[14:06:30] <ohsix> they would only lend name recognition to a possible competitor
[14:06:57] <jez9999> perhaps embracing 4264, and all the patent BS that it entails, will help to kill software patents? bring the issue more to the forefront? heh
[14:06:59] <jez9999> blind hope
[14:07:07] <av500> ohsix: trolls go after the lil ones 1st, that is true
[14:07:16] <av500> but not unless there is also big ones
[14:07:54] <ohsix> the whole thing is pretty perverse
[14:08:03] <av500> so whoever sits on a good theora patent will sit on it for a very long time
[14:08:16] <ohsix> not even in the spirit of patents
[14:08:19] <av500> its like having the patent on a steam engine these days
[14:08:39] <av500> you could use it to sue puffing billy
[14:08:57] <jez9999> there was a train called puffing billy? :-)
[14:09:20] <av500> was? is! http://www.puffingbilly.com.au/
[14:09:36] <ohsix> there are some judges that know its trollin' and not in the spirit of the law; but the patent trolls always file in that one little district in texas that will entertain them
[14:09:49] <av500> ohsix: sure, marshall texes
[14:09:50] <av500> ohsix: sure, marshall texas
[14:10:18] <av500> but again, trolls due not sue for fun, they sue for huge profits and starting with lil ones helps of course
[14:10:29] <av500> but only if there are large ones to follow up with
[14:10:29] <ohsix> its like how they picked a podunk town in florida to string up max hardcore
[14:17:48] <mru> http://unicorn.mansr.com/~mru/html5.png
[14:18:24] <kshishkov> maybe bracket zigzag ?
[14:18:27] <av500> you need to shout it out? :)
[14:18:56] <mru> kshishkov: <ZZ> ?
[14:19:01] <kshishkov> yes
[14:19:06] <kshishkov> it's funnier that way
[14:26:28] <mru> reload
[14:27:29] <av500> done
[14:29:15] <mru> kshishkov: like that?
[14:30:46] <KotH> mru: what is that logo for?
[14:30:59] <mru> KotH: t-shirts for fosdem
[14:31:10] <kshishkov> mru: yep, fine with me
[14:31:14] <KotH> \o/
[14:31:37] <mru> just a little subtle trolling...
[14:31:57] <KotH> but place some ffmpeg somewhere too
[14:32:10] <KotH> otherwise people will not get the slightest clue what it all is about
[14:32:19] <kshishkov> top on zigzag should be fine
[14:32:19] <superdump> mru: i'd say make the brackets black
[14:32:23] <av500> mru: brackets in black? so they dont look like part of the logo
[14:32:29] <superdump> :)
[14:32:32] <av500> :)
[14:32:46] <KotH> .o0(two idiots)
[14:32:50] <superdump> :p
[14:32:54] <KotH> ;->
[14:33:03] <mru> reload
[14:33:19] <superdump> yup
[14:33:21] <KotH> mru: what colour will the shirt be ?
[14:33:23] <mru> hold on, this is going on a black t-shirt
[14:33:27] <mru> so black will be white
[14:33:29] <av500> mru: dont need to, its like that in gimp here :)
[14:33:55] <av500> mru: white on black is even better
[14:33:55] <kshishkov> alternatively balck can have white outline
[14:34:22] <KotH> thin lines on t-shirts are bad
[14:34:26] <KotH> they easily break
[14:34:50] <av500> white on black is fine
[14:34:53] <mru> reload
[14:34:57] <mru> I don't quite like that
[14:35:27] <mru> the white brackets distract from the zigzag
[14:35:27] <av500> hmm
[14:35:38] <KotH> IMHO we should propose a new tag <ZZ> to w3c... for files that need to be played with ffmpeg based players ;-)
[14:35:46] <mru> lol
[14:36:07] <KotH> mru: the white on the pic is stronger than it will be on the t-shirt
[14:37:44] <mru> reload
[14:39:31] <KotH> doesnt look bad imho
[14:39:51] <superdump> pretty cool :)
[14:39:53] <KotH> but i'm missing the jab at theora in there
[14:40:03] <av500> you could also ask for light grey
[14:40:25] <KotH> av500: it wont be a pure white anyways
[14:40:34] <av500> right
[14:40:39] <superdump> i don't think it needs a direct jab at theora
[14:40:45] <av500> not after sleeping in it the 1st night
[14:40:53] <superdump> the point is, whatever happens, ffmpeg can handle the video codec
[14:41:14] <KotH> superdump: i'd like to have one, so that people at fosdem ask us about theora
[14:41:26] <KotH> superdump: then we can explain them why theora is a bad idea :)
[14:41:35] <av500> KotH: that is pointless
[14:41:40] <superdump> i think that t-shirt as it is should prompt discussion
[14:41:43] <KotH> superdump: you definitly have to think more subversive ;)
[14:41:59] <KotH> av500: that's why i'd like to make a point ;-)
[14:42:06] <av500> i dont think bashing another os project in public is helpfull
[14:42:13] * mru sharpens a stick
[14:42:36] <KotH> av500: it's not about theora or not theora
[14:42:44] <mru> colours aside, is the design ok?
[14:42:50] <kshishkov> av500: ever heard of cdrecord?
[14:42:53] <KotH> av500: it's about those zealots who believe that theora is the only viable solution because it's patent free
[14:43:00] <av500> mru: now we discuss the font?
[14:43:03] <KotH> mru: preliminary ack
[14:43:15] <mru> it's the same font I've always used
[14:43:18] <av500> ok
[14:43:34] <mru> I'd like to order them asap
[14:43:41] <superdump> KotH: what is your goal with countering that?
[14:43:44] <ohsix> is there something comparable to theora thats also open source?
[14:43:55] <KotH> superdump: global enlightment :)
[14:43:59] <mru> ohsix: gnash
[14:44:11] <mru> same mindset
[14:44:17] <av500> kshishkov: cdrecord, of course
[14:44:24] <ohsix> thats a video codec? :P
[14:44:27] <superdump> KotH: enlightenment to what point?
[14:44:28] <mru> ohsix: no
[14:44:36] <mru> ohsix: but they're equally stupid
[14:44:40] <KotH> superdump: there are very few things i hate more than being dumb and not knowing it. and promoting something with arguments that are simply wrong is one of them
[14:45:07] <ohsix> yea but pretend i dont care; i laid my original interest on the line in the question
[14:45:16] <mru> should I go with a light grey if they can do it?
[14:45:20] <mru> or stick with white?
[14:45:34] <av500> ohsix: that is the point with theora, normally open source tries to deliver state of the art stuff and does so with e-g- linux, apache, *cough* firefox...
[14:45:35] <KotH> mru: stick with white
[14:46:01] <kshishkov> av500: cdrecord
[14:46:13] <KotH> siretart: will you be at fosdem?
[14:46:15] <av500> kshishkov: yes, I know cdrecord
[14:46:17] <mru> white it'll be then
[14:46:23] <mru> how many t-shirts do we need?
[14:46:29] <KotH> mru: i need two
[14:46:34] <ohsix> my query is not unlike people looking for stuff, a viable answer to it is likely also a "viable alternative"
[14:46:38] <av500> I need 1 xxxl :)
[14:46:46] <mru> KotH: yeah, you'll spill beer on the first one ;-)
[14:46:52] <KotH> hehe
[14:47:05] <mru> av500: seriously, what size do you need?
[14:47:10] <av500> I was serious
[14:47:13] <KotH> mru: milk, my friend! milk!
[14:47:17] * av500 is fat
[14:47:20] <ohsix> to which nobody has spoken up so it might as well be true that there is no viable alternative; which is my point
[14:47:29] <KotH> av500: go to .us
[14:47:33] <KotH> av500: there you are thin ;)
[14:47:44] <mru> av500: really xxxl?
[14:47:45] <av500> KotH: I tried, could get thru the door
[14:47:47] <KotH> ohsix: what's your query?
[14:47:49] <av500> n't
[14:47:51] <kshishkov> av500: so what? for example, I wear that size.
[14:47:52] <superdump> KotH: i agree promoting theora on the basis that it is patent free, if it isn't, or as being as good as h.264 but patent free, which it isn't, is wrong
[14:47:54] <bilboed-pi> ohsix, I only know of two other patent-free open-source codecs : Dirac and the Sun OMC codec
[14:48:10] <av500> mru: 2x might do :)
[14:48:12] <bilboed-pi> ohsix, both use expired-patent technologies
[14:48:19] <siretart> KotH: yes!
[14:48:25] <KotH> siretart: good!
[14:48:32] <siretart> grr, why do you need to be authed to post in this channel?
[14:48:33] <KotH> siretart: dont forget to stop by at our booth
[14:48:37] <ohsix> people are of their own volition looking for bits that fit what they need, its not really anyones place to offer more than a suggestion to alternatives; not ravin' spittle :>
[14:48:51] <siretart> can someone please fix that?
[14:48:55] <KotH> siretart: and dont forget to reserve 2.6.2011-5.6.2011
[14:49:07] <siretart> KotH: I will arive with dondiego
[14:49:08] <av500> mru: if that tshirt place is sane, XXL is OK
[14:49:17] <ohsix> bilboed-pi: thanks, i haven't looked at dirac in ages; i didn't consider it comparable, didn't know of omc
[14:49:20] <siretart> KotH: reserve what?
[14:49:21] <KotH> siretart: which means two days late? ;->
[14:49:30] <siretart> I hope not
[14:49:43] <KotH> siretart: mark those dates in your calendar
[14:50:00] <ohsix> re: +R, those bots have chilled out for the night
[14:50:18] <siretart> ah, they are date. what's on that date? lugcamp?
[14:50:20] <KotH> ohsix: i'm lost
[14:50:26] <KotH> siretart: lugcamp.ch ;)
[14:50:38] <siretart> ah, I see :-)
[14:51:12] <mru> av500: the XL is 62cm wide when laid out flat
[14:51:24] * av500 strips
[14:51:35] <mru> ~75cm long
[14:51:39] * KotH closes his eyes
[14:52:07] <av500> ok, they are sane
[14:52:14] <siretart> would some op mind to -R this channel?
[14:52:24] <superdump> why?
[14:52:26] <superdump> we get spammers
[14:52:27] <mru> siretart: you want gay nigger spam?
[14:52:51] <siretart> no, i'm at work and would like to use my other client
[14:53:04] <KotH> ok.. gtg boys
[14:53:06] <KotH> to ALL: could those who come to fosdem send mru and me their names (so we can make name tags)
[14:53:21] <ohsix> it was very temporary in the first place; thet aren't andhaven't been on for a whilr
[14:53:37] <ohsix> (they were spamming other channels)
[14:54:04] <ohsix> siretart: nickserv can "link" your other client to the same reg
[14:54:20] <Compn> siretart : i'll voice you
[14:54:23] <Compn> or your other client
[14:54:24] <Compn> :P
[14:54:27] <KotH> bye boys...
[14:54:38] <mru> ohsix: the spammers were here this morning
[14:54:46] <mru> that's why I changed to +R
[14:54:49] <ohsix> Compn: that doesb't help that he can't join
[14:55:04] <ohsix> mru: yes, i was here
[14:55:18] <siretart> the other client is siretart`, is joined but cannot speak
[14:55:25] <superdump> siretart: as ohsix said, you can group your other nick with the nick you're using and have both registered and identified
[14:55:35] <siretart> I guess I need to read about that then
[14:55:45] <siretart`> thanks
[14:55:52] <Compn> ohsix : +R means they can join but unable to talk
[14:56:20] <mru> what good is a join if you are unable to speak?
[14:56:32] <J_Darnley> lol
[14:56:40] <ohsix> with these spammers in particular it doesn't help much to have +R set for more than a few minutes; they're the cgi irc client being driven in an iframe by people clicking on malicious links :>
[14:56:46] <av500> mru: you can lurk?
[14:57:23] <superdump> siretart`: it should just be /msg nickserv group siretart <password>
[14:57:25] <superdump> i think
[14:59:27] <siretart`> superdump: I think it worked
[14:59:32] <siretart`> thanks
[14:59:50] <ohsix> heh, last one that joined: 06:01 -!- kmbrkivptphq is "kmbrkivptphq" using irc.freenode.net on #ffmpeg-devel
[15:01:14] <mru> I don't see a problem with +R
[15:01:26] <mru> most people who have anything sane to say here are identified
[15:01:34] <ohsix> the terrorists win
[15:01:54] <mru> back to the t-shirts...
[15:02:09] <mru> av500: so what size do you need?
[15:02:09] <ohsix> i'm sure anyone that would want to talk but not register will consent to a cavity search ;)
[15:02:33] <mru> all you need to register is an email address
[15:02:46] <mru> what's the big deal?
[15:02:50] <Compn> ohsix : instead of bugging us, go bug freenode
[15:02:52] <ohsix> all you need to do is drop your drawers
[15:03:00] <Compn> i posted a blog detailing how the attack is being created
[15:03:04] <Compn> its easy as shit to block
[15:03:09] <av500> mru: the XL sounds OK, I have 3xl here that are smaller :)
[15:03:25] <ohsix> Compn: dont presume that i haven't :)
[15:03:29] <Compn> have you ?
[15:03:32] <Compn> what was their response ?
[15:03:33] <mru> av500: ok
[15:03:43] <mru> any other oversized people coming?
[15:04:05] <mru> M and L are easy to offload if there are spares
[15:04:09] <Compn> actually
[15:04:30] <Compn> ;\
[15:04:41] <Compn> lets just set this secret and spammers wont join ?
[15:04:43] <ohsix> i satisfied my own sense of wonder regarding their preparedness; and everyone i talked to pretty much said add freenode/staff/* to acls for emergencies
[15:05:14] <ohsix> Compn: if it is as you assume with the script thing, how will +s matter?
[15:05:32] <Compn> the script doesnt have any particular channel hardcoded into it
[15:05:37] <ohsix> the js isn't doing /list before it spams
[15:05:38] <Compn> i'm guessing it joins based on number of people ?
[15:05:41] <Compn> what is it doing ?
[15:05:44] * Compn forgot already
[15:06:16] <ohsix> its been modifed; people have come up with their own versions with internal lists
[15:06:19] <superdump> it sends ctcp version
[15:07:22] <ohsix> isn't there a cmode to block ctcps to the channel? it breaks /me but it'd stop the junk
[15:07:37] <ohsix> ignore gets them all, though
[15:08:00] <Compn> until they switched to non ctcp spam
[15:08:01] <Compn> :P
[15:08:40] <ohsix> they would have to come here personally and see what you are doing, and they cant know who is ignoring the ctcps already
[15:09:10] <ohsix> as it stands now, there are only 2 "strains" of this thing
[15:10:23] <ohsix> channel modes are no substitute for people handing out bans for something this sporaidic
[15:10:36] <Compn> you want ops ?
[15:10:39] <Compn> we'll let you do it then
[15:10:40] <Compn> lol
[15:11:22] <ohsix> heh maybe tommorow. kicking off in a bit, as it stands now i dont even see them, my client is doing my evil bidding
[15:12:03] <ohsix> someone setting up a script to ban for 5m on channel ctcps that match the spam will do great in place of a warm body
[15:13:13] <ohsix> afaik one of the staff that i kept messaging was already doing it for the old spam message, except it was klines instead of bans
[15:15:11] <ohsix> the cgi client should make you enter a cookie from an image and cut it off completely though
[15:15:45] <Compn> which client is it ?
[15:15:53] <ohsix> or check the document.location of a parent iframe if its in one
[15:16:54] <ohsix> chat.freenode.net, but it seems stuffed at the moment
[15:17:23] <Compn> huh? no its just connecting to http://irc.freenode.net:6667
[15:17:25] <Compn> directly
[15:17:36] <Compn> no cgi client, no html client, no js client
[15:17:40] <Compn> using html post
[15:17:56] <Compn> unless i'm reading something wrong ...
[15:18:19] <ohsix> ic, well then regular "restricted" class I can handle that, it makes them send back a cookie to finish connecting
[15:18:47] <ohsix> i find it utterly bizarre that freenode doesn't do this already
[15:19:20] <ohsix> (it sends PING :cookie, expects PONG :cookie before accepting a connection)
[15:20:25] <Compn> yes, freenode can block it very easily
[15:20:27] <Compn> go bug them
[15:20:29] <Compn> some more
[15:20:56] <Compn> http://unethicalblogger.com/posts/2010/01/using_browser_piss_irc_users_or_spamming_redditdowntime
[15:21:01] <Compn> give them that if they havent seen it :P
[15:21:05] <ohsix> not if their ircd doesn't do it, and not this close to the switchover to the other ircd
[15:21:40] <Compn> ugh
[15:21:53] <Compn> well come up with the ctcp ban script and someone will run it?
[15:22:05] <Compn> 'patch welcome'
[15:22:07] <mru> I'm off to the t-shirt shop then
[15:22:12] <ohsix> is this the "patches welcome" trtdtffgty
[15:25:31] <Compn> hehe
[15:55:08] <aclarke> http://multimedia.cx/eggs/ffmpeg-introspection/ -- someone should implement Mike's suggestion
[15:57:34] <mru> make it playable?
[15:58:18] <av500> use a tts and read the binary
[15:58:28] <av500> or the src if -g is on
[15:58:59] * Compn thinks he should troll mike on his blog
[15:59:04] <Compn> hmm
[15:59:06] <Compn> good idea or no
[15:59:19] <kshishkov> Compn: troll On2 into releasing TM2X
[15:59:52] <Compn> 'you keep calling it 'web video', but if flash player cant get more than 5 fps in fullscreen with h264, is that still video? i would call it 'adobe slideshow' ... i guess its better than 'BUFFERING'
[16:00:31] <Compn> help me punch this troll up
[16:00:40] <Compn> it needs to have more zingers
[16:08:03] <aclarke> @mru nah, just decode the ffmpeg binary to some raw audio bytes and/or raw video, and then let people write them out. So 'ffmpeg -i `which ffmpeg` out.mp4' produces something cool
[16:08:28] <aclarke> and sorry I used you as my suggested audio :)
[16:09:13] <ohsix> bad idea to say so on the channel; he can see it now :) 07:58 * Compn:#ffmpeg-devel thinks he should troll mike on his blog
[16:25:52] <Compn> ohsix : where are the logs ?
[16:27:13] <mru> where the other ML logs are
[16:27:22] * Compn goes to check his mail
[16:37:11] * Compn finds bunch of bugs in his mail client
[16:37:48] <Compn> cant edit account preferences while mail editor is open. cant read other mails while sending mail is timing out. takes 30 seconds to report dns error
[16:39:05] <BBB_> koth: ping
[17:27:45] <jez9999> BBB: g'day
[17:37:15] <BBB> hullo
[17:37:18] <BBB> how's it going
[17:39:38] <jez9999> not bad thanks
[17:39:57] <jez9999> i've created a proof-of-concept patch that i'd appreciate if you could have a look at sometime
[17:40:24] <BBB> I just noticed the bug email
[17:40:26] <BBB> will look
[17:41:10] <jez9999> k
[17:43:21] <BBB> si why are you so fixed on the source address specifically? can't you check by ssrc as was suggested in the bug report?
[17:43:42] <BBB> the specific situation that the sender trying to interfere with your cnx knows the ssrc, is highly unlikely, no?
[17:44:45] <jez9999> just remind me what the ssrc is?
[17:46:15] <BBB> ssrc is in the SDP and is also in each rtp packet
[17:47:00] <BBB> "synchronization source"
[17:47:10] <BBB> # SSRC : (32 bits) Synchronization source identifier uniquely identifies the source of a stream. The synchronization sources within the same RTP session will be unique.
[17:47:14] <BBB> (from wikipedia)
[17:47:36] <BBB> the rtp packet header contains a ssrc field, as does the sdp
[17:47:44] <BBB> that way you can validate the source of the stream
[17:47:59] <jez9999> right i see
[17:48:10] <jez9999> well this is still useful for DoS if the attacker knows the ssrc
[17:48:20] <BBB> how would he know?
[17:49:12] <BBB> are you trying to fix the case where you're receiving two sets of streams, and only one is valid? or is there something else that you're trying to do? I'm confused
[17:49:43] <BBB> the callback is a weird solution that reminds me of the udp-timeout-callback that we should really remove, it's is so ugly
[17:50:53] <jez9999> where in the SDP is the ssrc specified?
[17:52:09] <superdump> merbzt: do you know if mike is sorting out gsoc stuff this year?
[17:53:39] <BBB> I will probably do it with him
[17:53:48] <BBB> we need to get students
[17:53:49] <BBB> NOW
[17:53:51] <BBB> not next month
[17:54:00] <BBB> someone needs to send announcements to the devel & student-soc list
[17:54:07] <BBB> and we need to post on our blogs and in fora everywhere
[17:54:11] <BBB> we'll get in anyway
[17:54:18] <superdump> then we need to start sorting out ideas and qualification tasks and so on
[17:54:30] <jez9999> as an aside, who runs the ffmpeg bugtracker?
[17:54:40] <superdump> lu_zero i think
[17:54:44] <jez9999> i can't remove an attachment, i always get the error "Please describe the change.
[17:55:38] <BBB> don't remove attachments
[17:55:53] <BBB> and you need some comment in the text box to accompany the attachment
[17:55:57] <BBB> otherwise you can't add new attachments
[17:56:02] <BBB> superdump, let's start
[17:56:04] <BBB> do you have time now?
[17:56:24] <jez9999> why is there a 'remove' button if you dont remove them? :-)
[17:58:10] <BBB> I meant "please do not remove attachments"
[17:58:14] <BBB> it's an archive for us
[17:58:19] <jez9999> i want to anyway
[17:58:21] <BBB> including all mistakes and stuff that embarrasses everyone
[17:58:29] <jez9999> and when i enter text in there, it still gives me that error
[17:58:34] <jez9999> so is that button purely for decoration?
[17:58:37] <BBB> might be
[17:59:01] <BBB> lu_zero would know :-p
[17:59:06] <BBB> superdump, do you have time now?
[17:59:13] <BBB> we need to do this non-last-minute for once
[17:59:21] <BBB> I also want diego to provide early and often input
[17:59:34] <BBB> otherwise we'll get a repeat of last years' messes
[18:04:48] <superdump> BBB: hmm, not right now, no
[18:04:52] <superdump> nor probably this evening
[18:05:00] <superdump> i might be able to find a little time over the weekend though
[18:05:51] <superdump> BBB: i think unless we know the student will perform without it, it would be best if we really pushed mentors to communicate with their students
[18:06:28] <kshishkov> solution - use Mans for all students mentor
[18:06:29] <superdump> at the very least, weekly updates, but better, at least 10 minutes of chatting about what they're doing every day
[18:06:56] * av500 watches snow
[18:07:09] <BBB> daily
[18:07:10] <BBB> not weekly
[18:07:17] <BBB> diego was very specific about that
[18:07:17] <superdump> i think it will help mentors to keep track of what students are doing/finding difficult
[18:07:23] <superdump> and keep students motivated
[18:07:33] <superdump> fine by me
[18:07:37] <BBB> daily commits, daily chats, all the way
[18:07:39] <BBB> just like x264 does
[18:07:46] <BBB> the students get paid more than a burgerflipper
[18:07:52] <BBB> they should work at least as hard as one
[18:07:57] <BBB> so to say
[18:08:10] <superdump> mmm
[18:08:22] <av500> BBB: flipping the burger over twice does not help :)
[18:08:30] <jez9999> BBB: where in the SDP is the ssrc specified?
[18:08:40] <superdump> so, we need a list of qualification tasks that we think would be a reasonable representation of a student's aptitude
[18:08:43] <BBB> jez9999, in ms, it's in the transport line
[18:08:49] <BBB> jez9999, but I think it differs
[18:08:53] <superdump> or at least their persistence in getting code done and committed to trunk
[18:09:12] <jez9999> transport line?
[18:09:13] <superdump> and we need some good, well-contained, desirable and achievable ideas
[18:09:20] <jez9999> you mean m=?
[18:09:29] <BBB> transport maybe comes before the sdp
[18:09:36] <BBB> I forgot :(
[18:09:44] <BBB> grep for ssrc in a sdp file
[18:10:23] <jez9999> i'm looking in the sdp RFC
[18:10:26] <jez9999> i dont see anything about it
[18:10:40] <jez9999> and it contains a sample sdp file
[18:10:59] <BBB> hm...
[18:11:07] <BBB> I'd need to look and see my streams
[18:11:17] <BBB> it might be in another part of the connection, which isn't useful to you
[18:11:21] <BBB> (i.e. rtsp, not sdp)
[18:11:29] <jez9999> yeah i was thinking rtsp
[18:11:39] <BBB> but still, in your situation, can you look at whether the ssrcs are the same or different?
[18:11:43] <BBB> lu_zero also asked this
[18:11:49] <BBB> we really want to know that detail
[18:11:50] <BBB> all of us
[18:12:03] <jez9999> i can check
[18:12:06] <BBB> thnks
[18:12:13] <jez9999> but whether they happen to be or not, they could just all be set to 0 or something
[18:12:13] <BBB> superdump, here's a few: RALF, WMAlossless
[18:12:21] <BBB> superdump, AMRWB decoding
[18:12:26] <BBB> (might be too big)
[18:12:27] <jez9999> if a camera manufacturer couldnt be boethered to set it
[18:12:38] <BBB> I don't think that'd happen
[18:12:46] <BBB> superdump, some demuxers/muxers?
[18:12:56] <BBB> superdump, I still'd like more rtp work
[18:13:17] <jez9999> BBB: say the ssrc's are different. how does that help?
[18:13:19] <BBB> superdump, avfilter for audio would be nice, although getting video finished would be useful also
[18:13:26] <BBB> jez9999, at least you can distinguish
[18:13:35] <BBB> then the question simply becomes: which of the two is right?
[18:13:40] <BBB> and we'll figure out a way to answer that
[18:13:40] <jez9999> and that is unanswerable.
[18:13:44] <jez9999> ??
[18:13:52] <BBB> I'll need to look at std documents for that
[18:13:59] <BBB> but I can't imagine that there isn't some way
[18:14:03] <BBB> we're not the first seeing this problem
[18:14:08] <BBB> so I'm sure there's some solution
[18:14:23] <kshishkov> BBB: another funny project - VC-1 interlaced
[18:14:24] <BBB> but first, we need to know if this is a potential solution
[18:14:28] <BBB> kshishkov, I'm all for it
[18:14:31] <BBB> kshishkov, wanna do it?
[18:14:36] <BBB> I'll mentor you
[18:14:40] <BBB> :-p
[18:14:48] <BBB> ("what's vc-1?")
[18:14:55] <kshishkov> BBB: not so fast, I'm not a student anymore
[18:15:00] <BBB> ah shit
[18:15:08] <BBB> when did you graduate?
[18:15:18] <superdump> BBB: i think there's enough shared celp code/knowledge now to get a decent way through amr-wb decoding if not finish it
[18:15:21] <kshishkov> in 2007
[18:15:37] <BBB> in 2008/9 you participated
[18:15:39] <BBB> superdump, tru
[18:15:46] <BBB> superdump, so let's do that also
[18:15:47] <kshishkov> and dropped off postgraduate program November 2009
[18:15:48] <BBB> anything else?
[18:15:50] <superdump> and enough awareness of amr spec spaziness ;)
[18:15:57] <superdump> errrrm
[18:16:02] <BBB> l?
[18:16:10] <superdump> l?
[18:16:13] <BBB> laziness?
[18:16:19] <BBB> "spaziness" (?)
[18:16:24] <superdump> slang
[18:16:30] <av500> clang?
[18:16:35] <superdump> erlang!
[18:16:41] <BBB> let's make a list on the wiki
[18:16:43] <BBB> and start blogging
[18:16:45] <BBB> I want students
[18:16:47] <BBB> and I want a lot
[18:17:18] <superdump> i need to cook now
[18:17:25] <kshishkov> go to Indida and get a truckload or two then
[18:17:31] <av500> superdump: real-ly?
[18:17:55] <av500> kshishkov: at least they call you "sir"
[18:18:26] <kshishkov> av500: not a man from European variant of India
[18:19:17] <BBB> let me copy the 2009 page to 2010
[18:19:59] <kshishkov> BTW, why when I put a post specifically asked for job in Europe all real offers were from USA?
[18:20:10] <BBB> we have a flash screen video 2 encoder right?
[18:20:32] <kshishkov> yes
[18:20:39] <kshishkov> not committed though
[18:20:52] <kshishkov> because we don't have a working decoder for it
[18:20:59] <av500> kshishkov: they dont know where that europe is....
[18:21:14] <av500> mabye think "Europa, Texas"
[18:23:03] <Andrius> what's this "europe" you're talking about?
[18:23:12] <kshishkov> av500: maybe there are not so many places in Europe where people understand English?
[18:24:13] <mru> I have yet to find a place in america where they understand English
[18:24:23] <BBB> http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010
[18:24:25] <kshishkov> mru: learn Spanish
[18:24:29] <BBB> anyone want to help?
[18:24:49] <mru> kshishkov: my spanish-speaking friends keep telling me that...
[18:25:19] <kshishkov> mru: so it must be true
[18:25:41] <kshishkov> and Spanish is an official second (or first) language of USA
[18:25:58] <mru> the US doesn't have an official language
[18:26:25] <kshishkov> Sweden also did not have till July 2009
[18:26:26] <mru> simplified english is the most widely used though
[18:26:31] <BBB> what other soc tasks do we want
[18:26:33] <BBB> ?
[18:27:30] <kshishkov> let me see...
[18:27:39] <BBB> I added VC-1 interlaced as a primary
[18:27:45] <BBB> I volunteer kshishkov to do that
[18:27:52] <BBB> we'll invent some university for you that you attend somehow
[18:28:09] <BBB> also, I'm very interested in more decryption tasks
[18:28:35] <kshishkov> better keep decryption close to players
[18:28:50] <mru> you could start by decrypting some codec specs
[18:29:08] <kshishkov> mru: we do that
[18:29:21] <kshishkov> MN even managed to decrypt parts of H.264
[18:30:00] <kshishkov> huh? what is WMA Pro doing in that list?
[18:31:02] <BBB> ?
[18:31:17] <kshishkov> isn't it completed already?
[18:31:24] <BBB> I removed it a minute ago already?
[18:31:30] <BBB> I'm editing it as we speak :)
[18:31:41] <BBB> decryption belongs in ffmpeg
[18:31:44] <BBB> but key generation does not
[18:31:49] <BBB> decryption is completely ok imo
[18:32:36] <kshishkov> my suggestion: codec-indeendent ratecontrol
[18:33:59] <BBB> it's in the list already
[18:34:00] <BBB> I think
[18:34:06] <kshishkov> not in that form
[18:34:21] <BBB> http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010#Improve_Ratecontrol
[18:34:22] <BBB> true
[18:34:29] <BBB> add it, or improve the current entry
[18:34:52] * kshishkov adds "Improve 'Improve Ratecontrol' entry"
[18:35:08] <kshishkov> add "Reverse-engineer WVP2"
[18:35:15] <Dark_Shikari> that's a biggie.
[18:35:19] <BBB> yes
[18:35:26] <BBB> what's the fullname of that?
[18:35:27] <Dark_Shikari> btw, for reference
[18:35:29] <Dark_Shikari> http://wiki.videolan.org/SoC_x264_2010
[18:35:30] <BBB> Windows Video ... 2?
[18:35:34] <Dark_Shikari> Windows Media Screen Video 2
[18:35:36] <Dark_Shikari> I think.
[18:35:37] <BBB> ty
[18:35:48] <Dark_Shikari> check that out to see what we've set up for this year's SOC
[18:35:50] <Dark_Shikari> and qual tasks
[18:36:04] <kshishkov> no, that's MSS2
[18:36:14] <kshishkov> WVP2 is "presentation", I think
[18:36:34] <kshishkov> but yes, "MSS2" as "M$ stupid slideshow" would be better
[18:36:38] <BBB> added
[18:36:53] <BBB> http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010#Implement_a_Windows_Screen_Video_2_decoder
[18:37:44] <BBB> j-b, ping
[18:38:11] <BBB> j-b: are you interested in a combination task, where the student implements ffmpeg/rtsp into vlc?
[18:38:28] <j-b> BBB: sure.
[18:38:42] <j-b> BBB: but I believe the new avio: module of VLC makes that possible
[18:38:48] <kshishkov> BBB: better idea - rewrite any major player as FFplay2
[18:38:51] <BBB> it'd be a dual task, basically "implement missing formats into ffmpeg" (ffmpeg part) and "make it Just Fucking Work" (VLC part)
[18:39:10] <BBB> j-b: what's that?
[18:39:15] <j-b> it should work with all protocol of avformat, like RTMP://
[18:39:23] <BBB> hm... that sounds interesting
[18:39:28] <j-b> BBB: something new to leverage network parts of libavformat
[18:39:29] <BBB> rtsp is a demuxer though, not a protocol
[18:39:36] <BBB> :(
[18:39:41] <BBB> (yes it's a mess)
[18:39:42] <j-b> well, yes/no
[18:39:43] <BBB> another one: MMS?
[18:39:49] <BBB> anyone cares about mms nowadays?
[18:39:53] <j-b> nope :)
[18:39:53] <BBB> it's still used a lot
[18:39:59] <BBB> really?
[18:40:03] <BBB> that sucks
[18:40:05] <BBB> I'd mentor that
[18:40:18] <j-b> anyway, avio should makes that usable with vlc (mms, rtmp and other stuff)
[18:40:27] <BBB> so not rtsp...
[18:40:29] <BBB> that sucks
[18:40:33] <BBB> can you test it?
[18:40:33] <j-b> why not?
[18:40:38] <j-b> I can, of course
[18:40:38] <BBB> b/c rtsp is a demuxer
[18:40:40] <BBB> not a protocol
[18:40:44] <j-b> not really :)
[18:40:49] <j-b> anyway :)
[18:40:58] <j-b> let's not discuss terminology
[18:41:00] <BBB> yes
[18:41:18] <BBB> I mean it's implemented as a AVInputFormat, not a URLProtocol
[18:41:25] <BBB> whereas rtmp is a URLProtocol
[18:41:36] <j-b> http://git.videolan.org/?p=vlc.git;a=blob;f=modules/access/avio.c;h=2063f086e802f85147ff10f1bf6b6568aef8e66f;hb=HEAD
[18:41:56] <j-b> I could try that
[18:42:30] <kshishkov> BBB: please enlighten me, why RTMP is not a protocol implementation?
[18:42:55] <BBB> rtmp is
[18:42:59] <BBB> rtsp is not
[18:43:06] <j-b> av_url_read_pause is only for URLProtocol ?
[18:43:17] <BBB> not sure...
[18:44:21] <j-b> so I guess I need testing
[18:44:46] <j-b> but while it was true in the past that VLC couldn't leverage avformat access/demux, avio should makes that possible
[18:44:47] <BBB> kshishkov, sorry if I'm confusing, I think I'm doing too many things at once
[18:45:08] <kshishkov> indeed
[18:45:22] <BBB> j-b: test rtsp :) if it works, great, if not, let's make it a soc task
[18:45:30] <BBB> fenrir could mentor the vlcpart of it
[18:45:42] <j-b> BBB: kshishkov: about rtmp, we have our own access_demux. It is so bad, that I want it removed
[18:45:58] <j-b> and avio was the answer for that.
[18:46:33] <BBB> cool
[18:46:43] <j-b> -the +an
[18:46:44] <j-b> :)
[18:47:00] <Compn> ok i've caught up on my mail
[18:47:06] <Compn> where is the ml for chat logs ?
[18:47:48] <kshishkov> ffmpeg-devel-irc
[18:48:27] <kshishkov> with such attitude you're not going to RE any codecs
[18:50:47] <BBB> kshishkov, now seriously, can't we enroll you in some fake way so you can participate?
[18:50:56] * BBB wonders
[18:51:12] <BBB> maybe we should enroll you in some online university in the US
[18:52:20] <kshishkov> BBB: if you can help me moving to Europe, I'll gladly do it for free
[18:53:16] <j-b> European Univ? or European work?
[18:53:44] <kshishkov> work
[18:54:15] <kshishkov> universities = loss of both money and time
[18:54:24] <kshishkov> while work is only loss of time
[18:56:31] <j-b> research center in univ on multimedia matters is ?
[18:57:26] <kshishkov> dunno, we don't have research centers here
[19:03:48] * Compn must watch his tongue now that irc is logged
[19:03:52] <jez9999> BBB: the ssrc numbers are in there, i can see the data, but they're the same
[19:03:55] <jez9999> from multiple cameras
[19:04:41] <kshishkov> Compn: please add something about the fact we don't react on hello's
[19:04:53] <Compn> kshishkov : you are opped, you can add it
[19:04:54] <Dark_Shikari> http://forum.doom9.org/showthread.php?t=152419
[19:04:56] * Compn runs afk , bbl
[19:04:58] <Dark_Shikari> here is a patch to ffmpeg that might be useful
[19:05:00] <Dark_Shikari> it's a new bitstream filter
[19:05:06] <Dark_Shikari> it changes SPS settings in an h264 bitstream
[19:05:12] <Dark_Shikari> e.g. aspect ratio, fps, level, cropping
[19:18:55] <peloverde> I don't suppose there is any super fancy way to do multiple IMDCTs simultaneously?
[19:19:11] <mru> multiple CPUs ;-)
[19:19:20] <Dark_Shikari> imdct can already be done in simd
[19:19:24] <Dark_Shikari> not sure why you'd need to do multiple at once
[19:19:35] <peloverde> I mean like there are fast ways to take 2 FFTs at the same time
[19:20:13] <peloverde> BAsically I need an IMDCT and an IMDST which is just an IMDCT with ever other input negated and the output reversed
[19:20:31] <Dark_Shikari> oh, you mean two on the same input
[19:21:55] <mru> are the core calculations identical?
[19:22:54] <pengvado> imdct can already be done in simd, but it's not quite as fast as it could be if you imdcted 2 or 4 blocks interleaved. the latter wouldn't need any shuffles.
[19:23:15] <mru> depends on instruction set
[19:23:29] <pengvado> otoh, more cache pressure might negate any such savings
[19:34:19] <pengvado> peloverde: I don't see any arithmetic savings to be had from running imdct and imdst on the same input. but I'm just looking at the arithmetic as implemented in lavc, not searching over the space of trig identities.
[19:34:41] <peloverde> ok
[19:35:11] <peloverde> If somebody thinks it can be done better, he can complain during review
[19:36:15] <pengvado> toggling input signs to a fft could be exploited, but imdct's prepass destroys that regularity.
[20:00:41] <jez9999> BBB: so the only way to even distinguish is by source address... let alone determine which is the 'valid' incoming data.
[20:26:35] <BBB> jez9999, hmk... so could you describe the situation in more detail? i.e., not just "two streams one ssrc", but where do the streams come from (what IPs? cameras? which brand?), who sets them up to stream to this computer (ffmpeg? something else?) and why are they both sending to the same udp port and so on
[20:26:41] <BBB> (in the bug report)
[20:44:22] <jez9999> BBB: Sercomm brand cameras
[20:44:38] <BBB> in the issue tracker entry please, or I'll forget and will ask again next time :)
[20:44:41] <jez9999> they're running a custom firmware developed for our company, and they poll in to our server
[20:44:49] <jez9999> this is rather irrelevant
[20:45:02] <jez9999> the fact is that the DoS *can* happen, so it's reasonable to have a mechanism to stop it
[20:45:30] <BBB> right, but explaining your situation might help us pick a good default behaviour in case the DoS happens
[20:46:01] <jez9999> allow the calling code to inspect the packets and source address ;-)
[20:46:19] <BBB> that's one option, and we think it's too extensive/messy to implement in ffmpeg
[20:46:26] <BBB> we're hoping to find something "cleaner"
[20:46:34] <mru> I still think you're solving the wrong problem
[20:46:44] <BBB> preferrably something that's "on by default" and requires little to no configuration client-side
[20:47:00] <jez9999> i see no way to do that.
[20:47:05] <BBB> mru: I agree with that too, the custom firmware, same ssrc and same udp port sounds supciious to me
[20:47:16] <jez9999> no (standard) way to specify source IP, or SSRC, in an SDP file.
[20:47:31] <mru> that's like trying to play two video streams with the same pid in mpegts
[20:47:49] <BBB> basically
[20:48:28] <mru> I think an external prefilter is the better solution
[20:48:30] <mru> aka firewall
[20:48:35] <mru> or a proxy
[20:48:41] <BBB> "The RTP SSRC is meant to label streams from different sources, that is, each sender in a conference has its own SSRC."
[20:48:45] <jez9999> it would have to deep-inspect the packets
[20:48:48] <BBB> so why diff cameras have the same ssrc is beyond me
[20:49:06] <mru> probably hardcoded ;-)
[20:49:58] <BBB> yes
[20:50:28] <jez9999> well, ffmpeg does stuff to deal with dodgy implementations all over the place
[20:50:41] <jez9999> it's generally aimed at getting stuff to work, and it has a solution to this here
[20:51:34] <BBB> I'm going to read the RFC and see if it suggests anything at all regarding thsi
[20:51:42] <jez9999> which RFC?
[20:51:45] <BBB> I wonder if you can't change the ssrc manually from your different cameras
[20:51:47] <BBB> the rtp rfc
[20:52:22] <jez9999> even if that were possible, how would one distinguish which was the correct ssrc? it'd have to be reported to our servers during the poll-in or something
[20:52:27] <jez9999> the firmware just doesnt have that functionality
[20:52:30] <mru> it's always possible to construct a mutilated stream that can be carefully pieced together again
[20:52:39] <mru> that doesn't mean ffmpeg should have the ability to do so
[20:53:33] <jez9999> i'd hardly call this mutilated
[20:53:50] <BBB> jez9999, rtcp?
[20:54:00] <BBB> rtcp can change ssrc, or negotiate it, according to the rtp spec
[20:54:01] <jez9999> rtcp?
[20:54:04] <BBB> need to read more
[20:54:09] <jez9999> that relies on us connecting to the camera
[20:54:09] <mru> if it's not following spec, it's mutilated
[20:54:18] <jez9999> as it's at the client's end, 99% chance it's behind NAT
[20:57:51] <jez9999> the situation we have is tough as it's the reverse of usual
[20:58:23] <jez9999> the device less likely to have a static IP would need to be connected to, and the one with a static IP would need to connect
[20:58:27] <jez9999> that rules out standard RTSP
[21:18:13] <KotH> BBB: pong
[21:42:02] <BBB> dns?
[21:54:31] <peloverde> BBB, I saw your GSoC e-mail and I was wondering about the proposed changes to the FFmpeg way FFmpeg handles GSoC
[21:57:50] <peloverde> And by proposed changes I mean: "random thoughts about SoC (was: Re: random thoughts about refactoring)"
[21:58:25] <peloverde> Are we moving forward the way it was done before despite concerns and the high failure rate?
[21:59:53] <mru> I've been meaning to write up a short summary of good and bad projects in the past but haven't got round to it
[22:00:57] <peloverde> Also I was thinking about offering to mentor MPEG Surround
[22:01:26] <peloverde> or BSAC
[22:19:27] <BBB> peloverde, rather not
[22:19:33] <BBB> peloverde, I want to discuss that, and do it practically
[22:19:38] <BBB> i.e. no discussion "we should"
[22:19:40] <BBB> but "I will"
[22:19:48] <BBB> so step up, do things
[22:20:07] <BBB> I think the first thing is to select students NOW and work with them and see if we think they're mentorable
[22:20:40] <peloverde> I agree getting them involved early is a good thing
[22:21:48] <mru> and throw them out early if they're not up to scratch
[22:23:14] <peloverde> Ok, so would you guys rather see BSAC or MPEG Surround?
[22:24:52] <peloverde> BSAC is an alternate entropy coding scheme for AAC/HE-AAC, it's used bu T-DMB among other things
[22:25:13] <peloverde> MPEG Surround is a better multichannel scheme for MPEG-4 audio
[22:25:38] <peloverde> DivX likes to talka bout MPEG surround but I don't think they've actually released anything
[22:25:44] <astrange> is there a good encoder/an actual system using either?
[22:26:23] <astrange> arith coded aac sounds good to me but apparently the maximum bitrate is low (according to this wikipedia paragraph)
[22:26:28] <peloverde> T-DMB (korean digital TV) definitely uses BSAC in the wild
[22:26:55] <mru> they use some dreadful muxes too
[22:27:20] <peloverde> yeah, we wont be able to demux it
[22:27:32] <peloverde> but there are plenty of bsac conformance streams
[22:29:19] <peloverde> Why is DivX EOLing DivX 7 without DivX 8 being out?
[22:51:07] <kierank> [22:27] <@peloverde> yeah, we wont be able to demux it --> how come?
[22:51:45] <peloverde> because IIRC T-DMB uses LATM, and we still don't support demuxing from LATM
[22:53:09] <kierank> ah ok
[23:02:03] <peloverde> We'll still be able to demux BSAC from MP4
[23:58:55] <KotH> night boys
More information about the FFmpeg-devel-irc