[FFmpeg-devel-irc] IRC log for 2011-02-05

[irc at mansr.com]

[00:00:09] <michaelni> ohsix, key is if its not fun one doesnt do it and rather leaves or whetever
[00:00:39] <Sean_McG> y'know, something as simple as formatting could be enforced with a commit-hook
[00:00:44] <ohsix> fun doesn't have any part with it
[00:01:06] <ohsix> you can't get people to do things for fun; or if you can it all gets done, then the rest that needs to be done is decidedly not fun
[00:01:11] <michaelni> ohsix, i wrote FOSS for fun or money
[00:01:29] <michaelni> 99% fun
[00:01:36] <michaelni> 1% money
[00:02:00] <michaelni> if you take the fun away i loose interrest and iam not the only one
[00:02:07] <ohsix> maybe your usage of the word fun doesn't match mine; i find it entertaining, but it's different from fun, whether i'm being paid or not
[00:02:22] <ohsix> it's stimulating
[00:02:29] <michaelni> ohsix, maybe yes fun is the wrong word
[00:02:43] <kierank> michaelni: worth keeping in mind that in spite of issues in the community, ffmpeg is still a world class toolkit
[00:02:49] <kierank> nothing even comes close to it
[00:03:07] <Sean_McG> agreed
[00:03:12] * michaelni thinks rm -rf beats ffmpeg
[00:03:25] <Sean_McG> well, that trumps *anything* really :P
[00:03:29] <ohsix> feature branches ftw
[00:03:36] <ohsix> fighting over the tree isn't worth it
[00:03:49] <The_Tick> ohsix: I find working on oss fun
[00:03:52] <kierank> and as trite as it might sound it makes hundreds of millions of people's lives easier (they don't have to fumble with codecs they don't understand with vlc etc)
[00:04:08] <{V}> michaelni, must be multiple rm -rf 'cause of all the mirrors and repos :)
[00:04:09] <The_Tick> ohsix: I think trying to make someone simply be definitive on why they work on something
[00:04:12] <The_Tick> is not helpful
[00:04:25] <The_Tick> I think michaelni has a point that if he doesn't find it fun anymore
[00:04:30] <ohsix> The_Tick: i was just making allusions to the fact that there are lots of things that are not fun that need to be done to progress, can't let one block
[00:04:31] <The_Tick> that he doesn't want to do it
[00:04:41] <michaelni> The_Tick, yes
[00:04:43] <The_Tick> ohsix: and people like me
[00:04:46] <The_Tick> like doing those things
[00:04:47] <kshishkov> The_Tick: that's obvious, he hardly does any hard work anymore
[00:05:03] <michaelni> and people kicking me out like this also is really not making me want to stay for longer than i have to
[00:05:16] <ohsix> but it's a personal decision to do or not do something; i don't see whats so dramatic about running out of steam on something
[00:05:29] <ohsix> once you step back for a while you find new things to do with freshened eyeballs
[00:05:38] <The_Tick> ohsix: or you stop working on it
[00:05:40] <kierank> michaelni: the "kicking out" was not a reflection of the quality of your work but of your leadership style
[00:05:45] <The_Tick> ohsix: and then projects die
[00:05:50] <ohsix> well, you do stop working on it
[00:06:15] <ohsix> and projects that die are ones that aren't already self sustaining or a large enough chunk that it can't be written off and replaced wholly
[00:06:23] <The_Tick> michaelni: so essentially you think people are being too hard on themselves and you, making the project feel more like work?
[00:06:43] <The_Tick> sorry I lagged out so I did miss some conversation
[00:06:45] <ohsix> theres another type of death thats transformative in the OSS world, too; it's cathartic
[00:07:09] <iive> kierank: there was absolutely no reason to not give michael committer/puller/reviewer status. the leader position would still have been eliminated.
[00:07:23] <michaelni> iive, exactly
[00:07:43] <michaelni> the move was nothing but a very very strong insult
[00:07:45] <The_Tick> kshishkov: regardless, he obviously has some value to the project otherwise you guys would have picked him out as a poisonous person and ejected him
[00:08:12] <kierank> iive: i agree mostly but I guess the reason was that they didn't want michael to continue (in their view) flouting the rules
[00:08:13] <ohsix> you know a committer/reviewer/puller isn't a position to obstruct everything, or even have technical arguments against things
[00:08:24] <ohsix> they're a functionary for moving things to a blessed tree
[00:08:28] <kshishkov> The_Tick: like he feels we did?
[00:08:35] <The_Tick> kshishkov: most likely, yes :)
[00:08:38] <iive> kierank: what rules?
[00:09:04] <kierank> iive: commiting without review
[00:09:05] <michaelni> ohsix, like all the cosmetics&restructuring without asking file maintgainers
[00:09:21] <The_Tick> kshishkov: how do you view what you guys did?
[00:09:27] <ohsix> submission dread; or thinking that effort you spend will never be returned is a bad thing
[00:09:29] <iive> kierank: can you give example of michael doing that in the last 2 months?
[00:09:40] <The_Tick> guys, hold on
[00:09:42] <ohsix> the idea of a file maintainer is foreign to me
[00:09:47] <kierank> iive: i'm not saying that's my opinion. i'm saying that's their opinion
[00:09:57] <kshishkov> The_Tick: I think we did the right thing, that's why I supported it
[00:10:08] <The_Tick> kshishkov: what do you view that you guys did?
[00:10:09] <michaelni> ohsix, someone who know the code like the author and who works on it
[00:10:20] <The_Tick> *sigh*
[00:10:23] <ohsix> and if they don't "work" on it?
[00:10:40] <The_Tick> ohsix: what you're asking isn't helping
[00:10:43] <ohsix> the maintainers i'm familiar with do integration work; not development or review
[00:11:24] <ohsix> that's why it's a foreign concept; cosmetic and structual changes aren't something a maintainer cares about
[00:11:26] <The_Tick> specifically the point of this is for each person on either side to see the exact opinion of the other person on the same subject
[00:11:41] <The_Tick> which essentially helps 99% of the time
[00:11:46] <kierank> I think the most important thing that needs to happen is that the shit-slinging that's happened over the past few weeks needs to be put behind us
[00:11:58] <The_Tick> kierank: which is what I'm attempting to help with
[00:12:07] <The_Tick> so, back to my question
[00:12:10] <ohsix> i'm just larting my pet goat, attracting and keeping contributors is prime; they're the one thats going to foster productive happenings
[00:12:15] <The_Tick> kshishkov: what do you view that you guys did?
[00:12:42] <kshishkov> ohsix: which project is that?
[00:13:23] <ohsix> kshishkov: the ones with actual maintainers :D debian/redhat/ubuntu
[00:13:44] <kshishkov> ohsix: that's slightly different project IMO
[00:14:02] <ohsix> it's a completely different project, but there's not a large and arbitrary definition of a maintainer
[00:14:58] <ohsix> in linux, there's "ATA maintainers" and other subsystem maintainers, but they're also nto concerned with cosmetics and restructuring unless they're doing it, or guiding it themselves
[00:15:29] <ohsix> they just make it fit for purpose, act as a contact for broad issues with the subsystem; and triage
[00:15:41] <kshishkov> The_Tick: I think I've mentioned what we did
[00:15:43] <ohsix> they have pet trees but they still roll patches like anyone else
[00:16:31] <The_Tick> kshishkov: I want michaelni to see it
[00:16:45] <The_Tick> kshishkov: essentially he has a different opinion of the same actions
[00:16:46] <ohsix> if something is literally going to explode, or didn't account for something, or is plain wrong they'll say something to that effect, but otherwise they're not a position to judge the fitness or novelty of the change
[00:17:00] <The_Tick> and you guys need to reconcile that, or you'll continue to have problems
[00:17:05] <ohsix> ie. aren't an automatic blocker or impediment
[00:17:17] <The_Tick> I've had people drop out of projects for less percieved sleights
[00:18:30] <ohsix> you can't account for taste
[00:18:45] * Sean_McG sighs
[00:19:14] <iive> kshishkov: would you repeat what you think you all did? I don't see it in the back log, and "the right thing" is too vague.
[00:20:00] <kshishkov> iive: grew tired of the way FFmpeg was developed and restarted it with new set of rules
[00:20:22] <iive> that's all?
[00:20:26] <michaelni> kshishkov, i can agree with this
[00:20:46] <ohsix> why fight over the carcass; this is what forks are for
[00:20:54] <The_Tick> ohsix: forks always die
[00:21:05] <ohsix> eh, no
[00:21:23] <ohsix> you might be thinking high school or little kids playing around projects, and the indigant fork
[00:21:24] <kshishkov> iive: those are facts, the rest is opinions
[00:21:48] <The_Tick> ohsix: I don't know what projects you've been a part of, but if they have less than 10 people on them, then you aren't privy to this sort of problem most likely
[00:21:49] <iive> kshishkov: the problem is that more things were done, than changing the rules.
[00:22:02] <ohsix> projects that don't fork for spite; or fork to actually get anything done succeed in some measure, even if that is being adsorbed back into the primary project some time down the road
[00:22:05] <{V}> The_Tick, stop talking about forks man. Spoon! :p
[00:22:09] <kshishkov> iive: please list them
[00:22:10] <The_Tick> nor how "omg contributors! patch writers!" isn't going to help
[00:22:15] <kshishkov> {V}: sporks
[00:22:23] <iive> kshishkov: kicking michael out of the project.
[00:22:44] <ohsix> The_Tick: if a project is flagging and dying it's all about what you do about that
[00:23:08] <kshishkov> iive: that's your and/or his opinion. He has repo with admin rights and he's still considered developer
[00:23:29] <iive> kshishkov: not in the "reboot"
[00:23:38] <iive> it is fact.
[00:23:41] <michaelni> kshishkov, i have the main tree you ahev a dieing fork ;)
[00:24:03] <ohsix> The_Tick: there are already a handful of de-facto forks; patches that wont get in and separate trees that wont get merged
[00:24:06] <kshishkov> iive: nope. He just don't have privileges there
[00:24:17] <kshishkov> michaelni: as you like
[00:24:38] <The_Tick> ohsix: it's unfortunate that because of differing opinions and the capability to simply make a copy of the source code and host it somewhere
[00:24:40] <ohsix> now that's the spite i'm talkin' about :]
[00:24:46] <The_Tick> that people simply can't work together that have in the past
[00:24:49] <michaelni> ohsix, i could merge them if people hadnt pissed me off that much and would have asked me to merge them
[00:25:08] <The_Tick> we're likely going to have to freeze where we are in the ffmpeg version we have
[00:25:08] <ohsix> The_Tick: of course it is, but those are i'm sure the forks you're alluding to when you say they always fail
[00:25:12] <The_Tick> and look for alternatives
[00:25:15] <The_Tick> this is disappointing
[00:25:22] <ohsix> those aren't a serious effort in the first place
[00:25:22] <The_Tick> you guys are all adults, you should act like it
[00:25:55] <The_Tick> I don't want to spend my time trying to figure out your drama, but I have to because the project I love depends on it
[00:26:14] <ohsix> welcome to the jungle
[00:26:36] <The_Tick> this is all petty nonsense, and all of you know it
[00:26:57] <The_Tick> it all distracts from actual work
[00:27:02] <ohsix> that's not going to be very convincing
[00:27:09] <ohsix> they're already distracted from actual work
[00:28:30] <The_Tick> you guys need to talk about how to resolve this and move on
[00:28:38] <The_Tick> I've tried to facilitate that
[00:28:50] <The_Tick> but it's obvious that I'm wasting my time
[00:29:01] <ohsix> check your messages first
[00:29:09] <The_Tick> ohsix: I am, you're obnoxious
[00:29:09] <iive> The_Tick: not you don't
[00:29:26] <iive> The_Tick: making people talk is the first step.
[00:29:32] <ohsix> i may be obnoxious but i've been here
[00:29:40] <michaelni> The_Tick, i agree with iive what you do is not useless
[00:29:55] <kshishkov> iive: are people in Bulgarian government not talkative?
[00:29:57] <The_Tick> then what can we do to resolve this?
[00:30:06] <michaelni> i dont know
[00:30:12] <The_Tick> if this were an svn repo
[00:30:17] <The_Tick> the actions you guys took
[00:30:23] <The_Tick> would not have been an option as easily
[00:30:33] <The_Tick> this cannot be a constant threat for any member of your team
[00:30:38] <Compn> spoon!
[00:30:40] <The_Tick> a year from now if someone disagrees with you
[00:30:45] <The_Tick> they cannot think in the back of their head
[00:30:45] <iive> kshishkov: there was one party that made that their policy. they didn't got enough votes to enter parlament on these elections.
[00:30:50] <The_Tick> that if they are so disliked
[00:30:54] <kshishkov> Compn: have you seen The Matrix?
[00:30:56] <The_Tick> that they would be removed from the project in this fashion
[00:31:00] <iive> and they were rulling like 8 years ago.
[00:31:00] <ohsix> svn doesn't rule out a git-svn clone :]
[00:31:06] <The_Tick> this is not acceptable behavior for anyone who works on a project
[00:31:17] <The_Tick> ohsix: this is not helping
[00:31:20] <The_Tick> ohsix: please stop
[00:31:48] <michaelni> ohsix, yes please fuck with us later nit now
[00:31:50] <The_Tick> you guys need to grow up, and resolve this like grown ups
[00:32:24] <Compn> michaelni : i thought you were, maybe i am wrong...
[00:32:36] <ohsix> it's not my fight :]
[00:32:37] <Compn> kshishkov : yes, matrix, why ?
[00:32:39] <iive> The_Tick: the problem so far seems that this kind of effort is one sided. The other side took what they wanted and don't see any gain from participating.
[00:32:50] <The_Tick> then it's simple
[00:32:54] <The_Tick> remove them from the project
[00:32:59] <The_Tick> until they decide to play nice
[00:33:06] <ohsix> heh
[00:33:14] <The_Tick> someone please +q ohsix
[00:33:15] <Sean_McG> there's an idea
[00:33:31] <The_Tick> Sean_McG: unfortunately one group already did that
[00:33:36] <kshishkov> Compn: for one spoon quote of course!
[00:33:41] <Sean_McG> FOSS only works when everyone is playing nice in the sandbox.
[00:33:42] <Compn> oh
[00:33:50] <Compn> kshishkov : no, the spoon quote is from the tv show 'the tick' ...
[00:33:52] <The_Tick> Sean_McG: agreed
[00:34:26] <kshishkov> Compn: and cake quote is not from Portal either?
[00:34:47] <Sean_McG> yes, "the cake is a lie" is from Portal
[00:34:50] <Compn> which cake quote? but probably portal yes :P
[00:35:20] <The_Tick> I have an idea for how to resolve this
[00:35:23] <ohsix> q is not for me; i stopped as asked
[00:35:25] <Sean_McG> (on a completely unrelated note, why does openssl hate me so much?!)
[00:35:26] <kshishkov> Sean_McG: and "there's no spoon" is from what show?
[00:35:31] * Sean_McG sighs, bangs head from table
[00:35:37] <Sean_McG> kshishkov: that's from the first Matrix movie
[00:35:45] <The_Tick> the main problems as I see it are management
[00:35:51] <kshishkov> Sean_McG: tell that to Compn
[00:36:02] <The_Tick> someone to handle infrastructure who isn't a person who actually manages the project
[00:36:03] <Sean_McG> Compn: consider yourself told! ;)
[00:36:24] <The_Tick> and a reorganization of the repositories
[00:36:27] <kshishkov> Sean_McG: also XKCD claims Matrix was only one movie
[00:36:32] <The_Tick> so that you have a single main tree
[00:36:37] <ohsix> xkcd is right
[00:36:42] <Sean_McG> kshishkov: in 3 parts? I could agree with that.
[00:36:45] <Compn> i just said 'spoon'  not 'there is no spoon'
[00:37:03] <Sean_McG> Mister Anderson.
[00:37:12] <ohsix> once you figured out what the matrix was and they started coming up with new sequel lore it was over
[00:37:19] <Compn> i watched lord of the rings and never stopped saying mister anderson after every line
[00:37:35] <Sean_McG> Compn: hahahah
[00:37:42] <ohsix> theres a drum & bass album that has the movies edited together Compn
[00:37:43] <michaelni> The_Tick, someone to handle infrastructure is diego & mans replacing them has not been successfull
[00:38:17] <iive> that's the roots we were talking in the beginning.
[00:38:46] <michaelni> They are good friends of the new team ...
[00:38:55] <Sean_McG> why is mans not a part of this conversation?
[00:39:05] <The_Tick> michaelni: replacing them is not the point
[00:39:11] <The_Tick> michaelni: are these people developers?
[00:39:14] <kshishkov> probably he's enjoing life at FOSDEM now
[00:39:15] <iive> he may be sleeping, or traveling.
[00:39:18] <michaelni> mans yes
[00:39:21] <Sean_McG> oh, fair point
[00:39:22] <ohsix> Sean_McG: 10 hours idle probably has something to do with it
[00:39:25] <michaelni> diego develops only whitespaces
[00:39:30] * The_Tick sighs
[00:39:32] <kshishkov> Sean_McG: also he's tired of these conversations
[00:39:33] <The_Tick> michaelni: it's comments like that
[00:39:38] <The_Tick> passive aggresive comments
[00:39:42] <Sean_McG> kshishkov: oh, crap.
[00:39:50] <The_Tick> which do not help you
[00:39:58] <{V}> The_Tick,  remerging the two primary repos has already been suggested. In fact, it was my primary objective when I had a go at trying to find a compromise to take us out of this mess
[00:40:05] <iive> The_Tick: Diego entered mplayer as documentation maintainer.
[00:40:06] <michaelni> The_Tick, diego doesnt write C he never did, i had not intended to be sgressive but i see i was
[00:40:18] <ohsix> why not let them breathe a while and see what actually happens
[00:41:05] <kshishkov> he worked on build system at least
[00:41:29] <ohsix> he can be reasoned with too, that's a huge plus
[00:41:52] <iive> ohsix: not really.
[00:41:57] <The_Tick> essentially you guys need someone to manage the project
[00:42:02] <The_Tick> that isn't part of the whole politics
[00:42:06] <The_Tick> so you guys can focus on code
[00:42:17] <ohsix> iive: i'm prejudiced by a small success in doing so
[00:42:22] <iive> The_Tick: yep, and somebody who everybody would respect.
[00:42:45] <michaelni> And someone who would do that volunteerly
[00:42:45] <The_Tick> iive: someone they wouldn't have to question the motives of
[00:43:05] <The_Tick> michaelni: it's kind of what I do for perian and growl, and what I did for adium
[00:43:31] <ohsix> it's worth noting, software can be stable, and not dead
[00:43:46] <ohsix> flagrant comma abuse; my bad
[00:43:54] <ruggles> The_Tick: what kinds of project management things are you referring to
[00:44:06] <Sean_McG> I'm not sure what you're trying to add to the conversation, ohsix.
[00:44:35] <The_Tick> ruggles: in general, keeping the infrastructure running, resolving disputes, setting the guidelines for going forward usually
[00:44:51] <The_Tick> so this whitespace thing would have been a 10 minute conversation on my projects
[00:45:04] <The_Tick> I would have had the person having issues make sure the coding document is up to date
[00:45:10] <The_Tick> and that all devs agreed to that
[00:45:11] <Sean_McG> it has been stretched to rather absurd proportions
[00:45:13] <The_Tick> and that'd solve the problem
[00:45:34] <The_Tick> generally if you use tabs for whitespace, everyone can set their editor to whatever they want for whitespace, for instance
[00:45:43] <relaxed> So you're here to promote yourself as FFmpeg's project manager?
[00:45:44] <ohsix> Sean_McG: o noes
[00:45:46] <The_Tick> ruggles: also making sure developers get up to date hardware
[00:45:48] <The_Tick> relaxed: no
[00:45:53] <The_Tick> relaxed: I'm too busy
[00:46:08] <The_Tick> relaxed: I have a 10 month old son, a full time job and I manage 2 projects already
[00:46:24] <The_Tick> relaxed: he asked what I do
[00:46:24] <michaelni> The_Tick, tabs vs space is easy we have space after specific keyword rules :)
[00:46:32] <ohsix> people don't need to agree either, as long as it's not punitive and needlessly fluid
[00:46:54] <Sean_McG> michaelni: and again as I pointed out, tab/space can be enforced with commit hooks :P
[00:47:12] <The_Tick> michaelni: in general if everyone else agrees with that
[00:47:14] <The_Tick> you're wrong
[00:47:15] <michaelni> Sean_McG, yes i have no problem with tab vs space that one makes sense
[00:47:31] <The_Tick> and you need to get over yourself and not make things big issues
[00:47:59] <ohsix> Sean_McG: you can also put headers in the file that tell the editor what the formatting is supposed to be
[00:48:00] <michaelni> The_Tick, the rules are documented by the output of indent IIRC :)
[00:48:08] <Sean_McG> ohsix: that too.
[00:48:19] <The_Tick> ruggles: I'd be happy to help resolve disputes if that's needed on a say, 2 hour basis per week
[00:48:21] <Sean_McG> ohsix: provided everybody uses an editor that respects that.
[00:48:27] <The_Tick> but that's the extent of what I could spend on ffmpeg itself
[00:48:39] <The_Tick> or whatever
[00:48:39] <michaelni> The_Tick, we need 2weeks per hour i fear ;)
[00:48:49] <uau> The_Tick: "resolving disputes" may not be so easy, and anyone familiar enough with the code to do that well probably isn't an "neutral outsider"
[00:48:54] <ohsix> Sean_McG: well, worrying about that is for the person who would edit the file, the file is neutral and does its best effort to lay down the law ;]
[00:49:01] <The_Tick> uau: indeed
[00:49:07] <uau> and same for "setting guidelines for going forward"
[00:49:29] <The_Tick> uau: which is helpful since I'm not a coder and I force coders to make things easier for me to understand
[00:49:46] <The_Tick> which makes the problem easier to agree upon
[00:49:50] <The_Tick> typically :)
[00:50:10] <The_Tick> anyhow, that's essentially what I do
[00:50:25] <The_Tick> luckily with 2-3 people on perian, we don't have the problems of larger projects
[00:50:43] <The_Tick> but I had a large number of committers on Growl, and I don't miss the 3 day arguments about a single checkbox
[00:51:10] <ohsix> defer to a designer :]
[00:51:24] * Sean_McG sighs, pulls out /usr/bin/truss
[00:51:38] <The_Tick> ohsix: I'm not sure what you're adding to this conversation
[00:51:42] <The_Tick> other people have said the same thing
[00:51:50] <ohsix> letters and carriage reteurns
[00:52:09] <The_Tick> ruggles, uau essentially I'm not interested in your code any longer though if you guys aren't going to get along
[00:52:36] <The_Tick> so I need you to get along :)
[00:52:43] <iive> The_Tick: I agree that in the end, there should be somebody to pick one solution over the other, even if it is not the perfect one.
[00:52:58] <Sean_McG> thats what I think you call a mediator
[00:53:11] <ohsix> i'm not party to your growl conversation; but theres someone whos' job it is to say if its appropriate, unless they're all designers, then theres TooManyCooks
[00:53:14] <The_Tick> iive: I'm happy to be that person if you so need it
[00:53:34] <kierank> [00:42] The_Tick: essentially you guys need someone to manage the project --> mpc-hc has this except a lot of the decisions revolve around code
[00:53:34] <The_Tick> I'm not really really impartial since I'm involved in perian
[00:53:48] <uau> The_Tick: what do you mean by "not interested in your code"? you'll stop using ffmpeg?
[00:53:50] <The_Tick> kierank: but they can be talked about in a non-code way :)
[00:53:55] <kierank> not all of them
[00:54:05] <kierank> a lot of the things require a good level of understanding
[00:54:07] <Sean_McG> but I think it'd be hard to find someone completely impartial
[00:54:09] <The_Tick> uau: I mean that if you guys have a powerplay one month
[00:54:17] <The_Tick> where someone commits code, then someone removes the commit
[00:54:19] <The_Tick> back and forth
[00:54:23] <ohsix> uau: he already alluded to freezing the version they use earlier
[00:54:24] <The_Tick> I'm afraid of that
[00:54:40] <The_Tick> and of not accepting beneficial patches
[00:54:44] <The_Tick> we have a few
[00:54:59] <The_Tick> hell, one of the listed trees is from astrange
[00:55:01] <The_Tick> he's on perian
[00:55:08] <michaelni> The_Tick, i accept all  beneficial patches
[00:55:21] <The_Tick> michaelni: but does everyone else
[00:55:29] <michaelni> libmpcodecs ...
[00:55:59] <The_Tick> do you guys have a list of who has what hardware, and a wya to use donations to help provide better hardware?
[00:56:15] <The_Tick> like we bought stereo equipment, monitors, etc etc
[00:56:20] <The_Tick> for people who needed them
[00:56:23] <uau> The_Tick: OTOH that kind of "commit war" you're referring to is best avoided by NOT working on a common tree...
[00:56:32] <ruggles> The_Tick: no we don't have such a list
[00:56:34] <ohsix> uau++
[00:56:37] <The_Tick> uau: but in the end you don't get a released product
[00:56:42] <Sean_McG> uau: which git helps to facilitate
[00:57:01] <The_Tick> uau: it only fractures the team doing that
[00:57:03] <ohsix> The_Tick: releases are still basically, pick a revision; roll with it
[00:57:12] <The_Tick> ohsix: yes, then test it, then release it
[00:57:15] <The_Tick> I know what a release is
[00:57:21] <ohsix> but that's been improved as of late
[00:57:36] <uau> The_Tick: yes it can "fracture" the team, but that doesn't mean it'd be a reason to freeze your use to the current version
[00:57:37] <ohsix> The_Tick: no you misunderstand, i don't mean what ffmpeg does, i mean the people who consume ffmpeg
[00:57:53] <The_Tick> uau: why not? I could avoid all of this mess entirely
[00:58:01] <The_Tick> and figure out an alternative
[00:58:20] <iive> maybe we should stay on topic.
[00:58:24] <michaelni> The_Tick, if you find an alternative to ffmpeg tell me and ill join
[00:58:34] <bryno> i've frozen my use of ffmpeg before the coup happened :(
[00:58:59] <uau> why would freezing be an improvement? if there are multiple repos you could just pick one - if no opposed people share a repo there won't be commit wars, and the development is likely to be at least somewhat positive (better than using a frozen version)
[00:59:18] <The_Tick> uau: which one do I pick that I don't need to spend time figuring out the politics?
[00:59:27] <The_Tick> one repo has some changes
[00:59:29] <The_Tick> others have others
[00:59:33] <ohsix> michaelni: you probably wouldn't like any answers (or wouldn't accept them as an alternative :)
[00:59:35] <The_Tick> I have to spend time figuring out what to have pulled
[00:59:42] <The_Tick> why?
[00:59:46] <iive> The_Tick: and you forgot conflicting changes :)
[00:59:50] <The_Tick> you're wasting my developers time
[00:59:52] <wooster> can't you just make branches?
[00:59:54] <The_Tick> if you do that
[01:00:18] <ohsix> uau has his own tree for his own stuff; i don't think he thinks he's wasting his time
[01:00:38] <michaelni> The_Tick, i pull all features and bugfixes and everything else thats not worsening user experience from ffmoeg.org to ffmpeg at videolan
[01:00:41] <The_Tick> and at some point that makes it into the main tree right?
[01:01:03] <The_Tick> which is how branches are *supposed* to work, right?
[01:01:14] <uau> The_Tick: i don't think it would be that hard to find at least one tree that's better than the "frozen" version
[01:01:27] <ohsix> well theres a technical branch, which is something your source control does; and independent development
[01:01:36] <The_Tick> uau: I don't disagree about better, from a technical perspective
[01:01:41] <The_Tick> but say that's your tree
[01:01:51] <The_Tick> and then a year from now you stop working on ffmpeg-uau
[01:01:53] <ohsix> with git a branch is cheap and can either end up rebased on upstream or turned into a patch series; but it lives on its own
[01:02:00] <The_Tick> now my project has to make a decision
[01:02:15] <uau> the only ffmpeg tree i have has some minor tweaks (in case ohsix's comment confused someone)
[01:02:40] <ohsix> i didn't mean to say you forked it or anything uau, it just wasn't useful to use myself as an example
[01:02:42] <iive> i'm sorry but I'll be leaving the chat.
[01:02:51] <{V}> bye iive
[01:02:51] <michaelni> uau if you have patches or a tree to cherry pick from iam interrested
[01:03:00] <The_Tick> uau: the point I'm making
[01:03:13] <iive> kshishkov: don't forget to ask michael, if he will work with mans&diego on ffmpeg, if they are not roots anymore.
[01:03:16] <The_Tick> is I want us to use the main tree+our patches
[01:03:31] <The_Tick> I don't wnat to have to worry about the politics of the project
[01:03:34] <The_Tick> for the main tree
[01:03:46] <The_Tick> there shouldn't be politics for an oss project
[01:03:47] <Dark_Shikari> BBB: which patch
[01:04:05] <The_Tick> since you all likely want somewhat the same end goal
[01:04:08] <kierank> there's politics in all walks of life
[01:04:13] <The_Tick> so please get along so I don't have to think this way
[01:04:14] <uau> michaelni: i changed optimization level from -O3 to -O2 (smaller binary, faster, and quicker to compile in my tests) and fixed a couple of errors in generated .pc files
[01:04:22] <ohsix> iive: root means what it says on the tin right?
[01:04:32] <Sean_McG> oh boy, libtool
[01:04:35] * Sean_McG facefaults
[01:04:55] <uau> Sean_McG: libtool? where?
[01:05:07] <iive> n8 ppl
[01:05:15] <The_Tick> anyhow, I've said my peace
[01:05:17] <{V}> goodnight iive
[01:05:21] <The_Tick> please figure out getting along
[01:05:31] <The_Tick> I'll be happy to mediate if you guys need that
[01:05:40] <Sean_McG> wait sorry I mean pkg-config
[01:05:43] <Sean_McG> damn, I'm tired
[01:06:12] <ruggles> The_Tick: thank you
[01:06:16] <The_Tick> my email is chris at growl.info if anyone needs that
[01:06:18] <Sean_McG> haven't had dinner yet either...maybe head out for pizza in a bit
[01:06:42] <The_Tick> and I'm on freenode almost 24/7 idling :)
[01:07:01] <michaelni> The_Tick, thanks for your attempts at helping :)
[01:07:08] <uau> Sean_McG: you mean you'd have that reaction to pkg-config too, or was that due to the libtool confusion? (libtool would be far more deserving i think)
[01:07:20] <Sean_McG> uau: the former.
[01:07:35] <uau> why? pkg-config is by far the best solution there is
[01:07:38] <Sean_McG> but pkg-config doesn't bring the pain as much as libtool does.
[01:07:39] <Kovensky> it's lovely how the latest libtool release actually doesn't work
[01:07:46] <Sean_McG> Kovensky++
[01:07:50] <Kovensky> I have to use trunk libtool
[01:07:57] <Kovensky> for it to be slightly less broken
[01:08:29] <Kovensky> I mean, it refused .a libraries because the objdump string wasn't exactly the one it grepped for
[01:08:32] <Sean_McG> uau: but admittedly I cannot suggest an alternative.
[01:08:47] <Kovensky> the difference was in one character IIRC
[01:08:57] <Kovensky> either I fixed that regexp everywhere or I just used a new libtool <_<
[01:09:07] <Sean_McG> you mean you aren't building all your libs *dynamically*? what's wrong with you?!! </sarcasm>
[01:09:13] <Kovensky> one has to wonder wtf is libtool doing grepping .a though
[01:09:20] <Kovensky> Sean_McG: no, that's the problem
[01:09:24] <ohsix> uau: there's a default set of hates and gripes :]
[01:09:33] <Kovensky> Sean_McG: given how distros are all obsessive compulsive about dynamic libs, they actually test that part of libtool
[01:09:51] <Kovensky> static libs get shoved in the plan9 limbo :(
[01:10:20] <Sean_McG> s/plan9/Duke Nukem Forever/? ;)
[01:12:41] <ruggles> pross-au: nice patches :)
[01:13:40] <{V}> Sean_McG, around the beginning of May, check to see if you need to find a new famous example of vapourware :)
[01:14:14] <Sean_McG> {V}: so it's been suggested, but that trust has been violated so many times already :P~
[01:14:46] <Kovensky> Sean_McG: http://harmful.cat-v.org/software/dynamic-linking/
[01:15:15] <{V}> Sean_McG, yeah I certainly wouldn't start looking before May, if I were you :D
[01:17:24] <ohsix> Kovensky: that carmack quote was talking about windows, heh
[01:17:30] <ohsix> and getprocaddress
[01:17:34] <Sean_McG> yeah
[01:19:18] <ohsix> i like me some cat-v tho
[01:27:16] <ohsix> all the footnotes and the article seem to completely sidestep licensing and stuff (probably rightly so, but it's worth a mention; even if it's not a technical reality but a reality reality)
[01:27:26] <Sean_McG> "Of course it's kind
[01:27:26] <Sean_McG> of interesting: Posix threads are used by 'date'. I had no idea that
[01:27:28] <Sean_McG> printing a date could be so complex. Maybe that's why it's 40k"
[01:27:33] * Sean_McG had to snicker at that one
[01:35:24] <astrange> my /bin/date doesn't use pthreads
[01:36:53] <drv> mine does and is actually 59K ;)
[01:39:00] <astrange> cat-v is just some guy who reposts rob pike quotes that only apply if you think computers should be 3 xterms and no wallpaper
[01:39:19] <ohsix> could be overlinking; date is in a large pile of utilities built with mostly the same cflags
[01:39:31] <ohsix> astrange: that's a good way to put it
[01:40:21] <Sean_McG> sean at mahoro:~$ ls -l `which date`
[01:40:22] <Sean_McG> -r-xr-xr-x  1 root  wheel  80848 18 May  2009 /bin/date
[01:40:27] <Sean_McG> that's on OS X Snow Leopard
[01:40:50] <ohsix> it's like reading cave meb wall paintings
[01:41:19] <Sean_McG> -r-xr-xr-x   1 root     bin        10840 Jan 22  2005 /usr/xpg4/bin/date
[01:41:24] <Sean_McG> that's Solaris 10u8 x86
[01:41:29] <astrange> the 80kb date is tri-arch
[01:41:33] <Sean_McG> aye
[01:41:48] <beastd> And personal computers are a bunch of capable hardware with bloated operating systems and shitty applications that make possible the fanciest useless things while at the same time making you think your hardware is too slow.
[01:42:22] <astrange> so 26kb or so x86-64, i don't feel like investigating further
[01:42:49] <astrange> gnu utilities would be larger due to including locales and --version and stuff
[01:43:06] <ohsix> ahh date links to stuff thats implicitly thread aware but isn't neccisarily used in such a manner in the linked application
[01:43:24] <Sean_McG> that would make more sense
[01:43:53] <astrange> BBB: http://pastebin.com/BFiJxArx rebased, i still see the failures
[01:44:08] <ohsix> beastd: for all its largesse i don't recall waiting for date to do its thing
[01:45:12] <beastd> ohsix: My statement wasn't necessarily aimed at date. Much more global.
[01:45:58] <ohsix> mine wasn't either; thats just what everyone was looking at
[01:46:07] <astrange> your hardware probably is too slow
[01:46:13] <ohsix> all i wait for in like the last 6 years is firefox
[01:46:18] <astrange> HDDs are very slow things and everything is blocked on them
[01:46:30] <Sean_McG> yes, damn those spinning platters
[01:47:27] <beastd> astrange: probably my hw is too slow for many things. but even for the things it would be good enough it so often can't be used.
[01:47:55] <ohsix> i got by on a netbook for about 2 weeks; like as a proper daily driver
[01:48:01] <ohsix> it just needed an extra gig of ram
[01:55:45] <beastd> anyway, there is a lot more to the issue. google is trying to address it with chrome os but that is not the thing i am looking for (though for the bulk of users it will maybe is, there are some similarities in objectives like e.g. speeding up boot process but i visionary wise i am looking in the opposite direction). But it is OT here anyway (meaning i willl STFU now).
[01:56:13] <ohsix> wat
[01:57:11] <ohsix> everyone needs fast SSD's with a lot more pressure to swap stuff out; and XIP restore images for startup and deep suspends
[01:59:48] <ohsix> if your disk is fast enough you can do some nutty things, ubiquitous flash storage is gonna be pretty trick in 15 years
[02:06:43] <ruggles> what is below @subsection in texinfo?
[02:07:08] <ruggles> chapter, section, subsection, ?
[02:09:11] <ruggles> nevermind. found it. @subsubsection
[02:10:01] * ruggles is writing encoders.texi for ac3enc AVOptions before sending a new patch
[03:14:09] <BBB> astrange: I will compare to my tree
[03:14:13] <BBB> astrange: it passes here
[03:14:26] <BBB> astrange: thanks for rebasing
[03:19:16] <ruggles> any suggestions for passing dB (decibels) in commandline arguments? currently dB means 0.1 bits...
[03:25:12] <astrange> which arguments overload dB and bits?
[03:25:52] <ruggles> none, but av_strtod() doesn't distinguish on a per-argument basis
[03:26:23] <ruggles> currently B in the postfix means multiply by 8.
[04:13:02] <astrange> in that case, what arguments overload dB and anything else?
[04:13:14] <astrange> since you could pass "10" for 10 dB
[04:13:38] <astrange> i guess dB and linear multipliers?
[04:35:12] <BBB> astrange: I have the relevant changes under 1 page, I'll get you the "problem" tomorrow, do you have time to review and submit the -mt tomorrow? :)
[04:36:33] <BBB> astrange: or wait, I've got it already
[04:37:31] <BBB> astrange: http://ffmpeg.pastebin.com/aYmmY6V9 fixes make fate-vsynth2-wmv2 (and probably all others - untested) for me
[04:38:11] <BBB> astrange: enjoy submitting the patch
[04:39:33] <astrange> that'll make it draw edges twice for all decoding frames. but i'll do some gdbing and look at what state i can use
[04:39:36] <astrange> s->encoding didn't work
[04:42:08] <Sean_McG> gdb is a verb now? :P
[04:47:24] <The_Tick> Sean_McG: it's attained google status
[06:25:16] <pross-au> kshishkov: awake?
[07:41:59] * elenril yawns
[10:51:52] <_av500_> http://www.youtube.com/watch?v=fz_FmQ_v7Uw
[11:49:30] <pross-au> _av500_: looks awesome
[12:53:50] <_av500_> thx
[13:02:34] <thresh> yeah looks cool
[13:05:14] <_av500_> thresh: your man was not here yet
[13:05:23] <BBB> astrange: ah, you sent patches, thanks :)
[13:06:19] <thresh> _av500_: I can try texting him with your location, where is your booth?
[13:10:31] <_av500_> thresh: in AW building
[13:10:47] <_av500_> at the booth with the huge video wall
[13:11:44] <thresh> ok sent
[13:11:54] <kshishkov> wearing t-shirt with his nickname
[13:12:41] <thresh> how's fosdem hangover this year?
[13:16:51] <_av500_> thresh: not too bad, I was not there till 6am
[13:17:00] <_av500_> mru gave up quickly
[13:17:05] <_av500_> and so did i
[13:28:01] <Compn> Dark_Shikari : i replied to your mail in the equality thread, hopefully explaining what michaelni was trying to get at in his mail
[13:28:40] <Compn> michaelni : hopefully i've interpreted your mail correctly :)
[14:20:48] <Flameeyes> wbs: lscube bugzy should come up again soonish... at the end I called Luca at fosdem since nobody else was around =_=
[14:21:11] <kierank> is mru attending the autotools workshop?
[14:22:15] <wbs> Flameeyes: ok :-)
[14:31:54] <Flameeyes> kierank: oh god an autotools workshop? who the heck is supposed to handle that?
[14:32:14] <ohsix> it's become self aware
[14:32:57] <Flameeyes> ohsix: I just hope it's not one of those who pretended to know autotools when designing the kde3 build system
[14:45:20] <lu_zero> Flameeyes: pg up again
[14:45:39] <lu_zero> mru: where are you?
[14:45:50] <Flameeyes> lu_zero: thanks — am I supposed to have access to that box, you sais?
[14:46:01] <Dark_Shikari> http://i.imgur.com/XvEz6.png
[14:48:39] <Dark_Shikari> Is it just me, or is there no AV_RL32A?
[14:48:45] <Dark_Shikari> What's with that?
[14:53:01] <Dark_Shikari> mru: any reason for this?
[14:53:23] <Dark_Shikari> BBB: ping
[15:02:36] <Dark_Shikari> ok, I had a batshit crazy idea.  this is just insane, but might just work.
[15:02:53] <Dark_Shikari> in x264, we use trailing zero count to loop over nonzero coefficients in a dct block, while skipping the 0s
[15:03:06] <Dark_Shikari> thus making performance O(num nonzero coeffs) instead of O(last nonzero coefficient).
[15:03:20] <Dark_Shikari> ... how about, in ffmpeg, doing this to loop over coded 4x4 dct blocks?
[15:03:43] <Dark_Shikari> so if there's 3 coded dct blocks out of 16, it does 3 trailing zero counts on a mask of nnz values, skipping through them rapidly.
[15:03:55] <Dark_Shikari> (vp8, h264)
[15:33:33] <Dark_Shikari> Well, here's the incredibly evil/ridiculous code:
[15:33:36] <Dark_Shikari> x = 0;
[15:33:36] <Dark_Shikari>                     while( nnz4 ) {
[15:33:36] <Dark_Shikari>                         int shift = __builtin_ctz( nnz4 ) >> 3;
[15:33:36] <Dark_Shikari>                         x += shift; nnz4 >>= (shift << 3);
[15:33:36] <Dark_Shikari>                         if ((nnz4&0xFF) == 1) s->vp8dsp.vp8_idct_dc_add(y_dst+4*x, s->block[y][x], s->linesize);
[15:33:39] <Dark_Shikari>                         else                  s->vp8dsp.vp8_idct_add   (y_dst+4*x, s->block[y][x], s->linesize);
[15:33:43] <Dark_Shikari>                         nnz4 >>= 8; x++;
[15:33:45] <Dark_Shikari>                     }
[15:34:07] <Dark_Shikari> using a trailing zeroes loop for idct... makes me want to puke.
[15:35:36] <mru> beware, trailing zeros isn't always a single insn
[15:35:47] <mru> depends on cpu
[15:35:47] <Dark_Shikari> I know
[15:38:09] <Dark_Shikari> This would probably be more useful in h264
[15:38:23] <Dark_Shikari> because vp8 tends to have dc coefficients in every block due to the higher precision of the dc quant
[15:38:28] <Dark_Shikari> and the hierarchical transform
[15:38:52] <iive> Dark_Shikari: doesn't skipped blocks have dc component on their own?
[15:38:55] <Dark_Shikari> ?
[15:40:44] <iive> I'm not familiar with h264 working. But in the code above when you handle nnz4 you do either idct_add or idct_dc_add
[15:41:03] <Dark_Shikari> Yes.
[15:41:03] <iive> so I would assume that the skipped blocks should be handled with idct_dc_add too.
[15:41:10] <Dark_Shikari> No... the skipped blocks have no DC at all.
[15:41:15] <Dark_Shikari> Note the while loop SKIPS over them
[15:41:17] <Dark_Shikari> because of the ctz
[15:41:33] <Dark_Shikari> the original code does
[15:41:45] <Dark_Shikari> if( nnz) { if(nnz==1) {idct_dc} else {idct} }
[15:41:52] <iive> i was asking if this is the intension.
[15:42:01] <Dark_Shikari> Yes, the intention is to skip empty blocks.
[15:42:39] <Dark_Shikari> By virtue of being empty, empty blocks can be skipped.
[15:43:13] <iive> aha, so nnz4 is 4 different variables combined into 1?
[15:43:25] <Dark_Shikari> yes
[15:43:30] <Dark_Shikari> uint32_t nnz4 = AV_RN32A(s->non_zero_count_cache[y]);
[15:43:35] <Dark_Shikari> `-`
[15:44:03] <Dark_Shikari> nb: my code obviously breaks on bigendian atm
[15:48:54] <Dark_Shikari> hmm, interesting thought...
[15:48:59] <Dark_Shikari> we can fit 16 nnz values in a uint32_t
[15:49:12] <Dark_Shikari> because nnz is either 0, 1, or >1
[15:55:41] <kierank> av500: troll hunter now has english subtitles
[15:55:48] <kierank> _av500_ : i mean
[16:07:49] * Compn wonders why Dark_Shikari didnt read whole email :(
[16:35:55] * Sean_McG returns
[16:36:24] <Sean_McG> for a short bit anyways...gonna go out for breakfast methinks
[16:36:45] <Flameeyes> wbs: ah sorry there is a branch in which we reworked http tunnels
[16:36:54] <Flameeyes> they _should_ work now, if you can try them we'd be delighted :)
[17:08:25] <wbs> Flameeyes: oh, which one do you want me to test? http-tunnel-threads or -v2?
[17:08:52] <Flameeyes> -v2 please
[17:11:14] <wbs> ok, I'll test it in a while. since the current master seems to work for me at the moment, I guess it's hard to say if it really works better or not, but I can at least find out if it's more broken or not :-)
[17:33:33] <mmu_man> plop
[17:34:04] <_av500_> mmu_man: hows fosdem?
[17:38:37] <mmu_man> nice
[17:38:42] <mmu_man> crowded as always
[17:38:50] <mmu_man> but you know :p
[17:45:14] <_av500_> yeah
[17:48:59] <kierank> _av500_: did you go to the autotools talk?
[17:53:02] * elenril wonders if he should bother reading the drama
[18:14:09] <Sean_McG> ahahahah oh snap... "the Verizon iPhone is really good at holding on to calls, if you're into that kinda thing"
[18:42:27] <Compn> are ffmpeg devs getting lots of questions at fosdem ?
[18:54:57] * elenril lols at DonDiego's mail
[18:55:04] <elenril> "Vive la revolution!" << awesome
[18:57:05] <Compn> alienating a bunch of devels is awesome ?
[18:58:33] <wbs> Flameeyes: agh, my libev (the latest in debian unstable) doesn't have EVRUN_ONCE defined
[19:08:26] <lu_zero> Compn: that could be read in two different ways
[19:09:06] <lu_zero> wbs: which version it is?
[19:09:36] <wbs> lu_zero: 3.9
[19:10:28] <lu_zero> uhm I wasn't planning to use already the 4.0 stuff...
[19:11:54] <Compn> lu_zero : thats why i asked elenril to clarify
[19:15:12] <lu_zero> Compn: elf followed by vlr, that's pretty in line
[19:15:46] <lu_zero> the question is, which of the two expression would alienate who?
[19:20:34] <Compn> erm, overall attitude i would say instead
[19:20:56] <Compn> e.g. responding to sigs , ignoring content of main message, etc
[19:21:06] <Compn> if you are asking who left, you can look at ramiro
[19:22:09] <Compn> i dont know who else left, i am not prepared with a list.
[19:39:30] <BBB> Compn: to the best of my knowledge, ramiro is at fosdem discussing (hopefully) with the rest of the team?
[19:40:23] <Flameeyes> wbs: damn.. okay need to see if I can avoid using that :/
[19:40:39] <BBB> iive = ivan kalvachev?
[19:41:28] <Compn> BBB : good (if true)
[19:41:40] * Compn is not in any loops and knows nothing
[19:42:33] <Compn> yes iive is ivan
[19:42:38] <BBB> and yes I know various people are annoyed, I'm not blind or deaf, I'm listening and hoping to do whatever's best for FFmpeg also. doing a vote between "kill me now" or "let's hang michael" is not in my list of "best for FFmpeg"
[19:42:57] <BBB> cut the ridiculous crap with these votes please
[19:43:06] <Compn> who are you talking to btw ?
[19:43:17] <Compn> well, i guess michaelni is here
[19:43:35] <iive> BBB: doing something with best intentions is not always doing the best thing.
[19:44:27] <BBB> iive: could I please kindly request that you refrain from "contributing" to all these trollwars-to-be and leave that to the people that the discussion is actually about, i.e. the actual developers? thank you
[19:44:46] <Compn> iive has code in ffmpeg iirc ?
[19:44:50] <iive> BBB: I am actual developer. or at least used to be.
[19:45:08] <iive> and I am maintainer. one or 2 files but still. I am.
[19:45:23] <Compn> xvmc stuff ?
[19:45:29] <iive> yep.
[19:45:52] <BBB> that's great, but your comments aren't helpful, they're flamy, trolly and generally not helpful. it's BS rethoric. please don't
[19:46:42] <iive> BBB: I try to keep them to the minimum.
[19:47:10] <Compn> iive speaks his mind :)
[19:48:02] <iive> BBB: have you actually lived under dictatorship?
[19:50:10] <BBB> iive: please stop the rhetoric, again
[19:50:52] <BBB> I'm said the same to michael in private, either say nothing or work towards a solution. don't troll. I won't listen, and neither will anyone else. you're just making it more likely that I'll stop listening to you altogether
[19:51:09] <BBB> I don't care that someone in some country far, far away lived under a dictatorship because THAT IS NOT THE ISSUE HERE
[19:51:14] <BBB> please see that
[19:51:18] <BBB> then stop responding uselessly
[19:51:28] <BBB> and contribute, if at all, in a useful and constructive manner
[19:51:33] <BBB> so again, stop trolling, thank you
[19:53:19] <iive> in order to work towards a solution, we should know what is the problem first.
[19:53:37] <BBB> Dark_Shikari: that an interesting idea, if there's indeed correlation between neighbouring blocks being more likely than purely random to be zero... not sure if that's the case for vp8... can that be measured?
[19:54:17] * BBB decides to stop listening to the rhetoric BS from iive
[19:58:02] <elenril> isascii() is locale-independent, right?
[20:00:41] <BBB> seems so from the manpage
[20:01:24] <elenril> maybe i shouldn't bother with it anyway and just write my own :/
[20:11:03] <BBB> michaelni: opinions on "[PATCH 1/6] Adopt pkt_dts/pkt_pts in lavc clients"? it seems simple enough to me
[20:11:26] <BBB> elenril: should be ok really, why don't you want to use it?
[20:11:37] <uau> BBB: threading is a real issue
[20:12:01] <uau> people haven't used it much in mplayer svn because it doesn't work for most h264 videos anyway
[20:12:30] <uau> i disabled thread callbacks in mplayer2 (which i think most people needing multithreading have already been using, though not under that name)
[20:12:38] <uau> and that was necessary
[20:12:45] <elenril> BBB: dunno, i have an irrational distrust for them ;)
[20:16:06] * elenril kicks freenode
[20:16:23] <elenril> arrrgh, who told this system to accept RAs
[20:17:14] <BBB> uau: uh, that sucks :)
[20:17:27] <BBB> so you're telling me slice-based MT isn't really tested much
[20:18:27] <uau> BBB: yes - slice-based is pretty much useless (as it only works for stuff like mpeg2 which doesn't need performance)
[20:18:37] <uau> no form of threading has been tested much with svn
[20:18:51] <Compn> roo had svn mt builds for a while
[20:18:59] <BBB> I thought slice-based h264 was used for low-latency stuff
[20:19:00] <Compn> also i think sherpya did svn-mt too
[20:19:03] <uau> there have been a reasonable amount of people (at least hundreds) using mplayer2 with ffmpeg-mt
[20:19:16] <uau> and that has worked without many complaints
[20:19:39] <uau> (with callbacks disabled with threading)
[20:20:28] <uau> BBB: there are some sliced h264 videos
[20:20:30] <Compn> http://oss.netfarm.it/mplayer-win32.php
[20:20:31] <BBB> I guess I had always assumed that it worked since it was out there and mplayer people tend to compain a lot about this kind of breakage
[20:20:48] <BBB> uau: not just videos, dark_shikari's company uses it too and I thought it worked for them
[20:20:56] <uau> but i think few people enable threading with those
[20:21:14] <uau> because 1) testing slice-based threading with most videos shows no benefit, only breakage
[20:21:17] <BBB> sucks :-p
[20:21:29] <Compn> anyways, dont worry about mplayer
[20:21:31] <BBB> please define breakage?
[20:21:36] <Compn> if ffmpeg-mt breaks something, it will get fixed
[20:21:41] <Compn> dont wait for mplayer to merge it
[20:21:47] <uau> and 2) people who really need multithreading and know enough to search for manually-set options can find mplayer2 and use that
[20:22:16] <uau> BBB: i think -vo gl in svn at least was totally broken with slice threading, as in no working picture at all
[20:22:50] <elenril> anybody with WMP here?
[20:22:53] <BBB> hmk, right, so that's what reimar is talking about
[20:23:26] <Compn> elenril : yes ?
[20:24:09] <elenril> Compn: can you please test ftp://ftp.khirnov.net/out.mp3 ?
[20:24:23] <elenril> i want to know if the tags are working properly
[20:24:55] * Compn wonders if wmp can ahndle ftp
[20:25:01] <Compn> ok where do i get tags from ?
[20:25:23] <elenril> i mean id3v2 in this mp3
[20:25:34] <elenril> wmp should display them
[20:25:35] <Compn> i mean, how do i display the tags in it
[20:25:40] <Compn> its scrolling them
[20:25:45] <elenril> rightclick->properties iirc
[20:26:04] <Compn> it lists title artist album composer and genre
[20:26:18] <Compn> in the properties window
[20:26:22] <elenril> it should say 梶浦由記 - storm is coming
[20:26:35] <Compn> title storm is coming
[20:26:42] <Compn> artist is ????
[20:26:50] <elenril> meh
[20:27:08] <Compn> by ??? its boxes, probably japanese text
[20:27:15] <Compn> just that no fonts installed properly
[20:27:20] <elenril> ah
[20:27:39] <elenril> how can you live without japanese fonts ;)
[20:27:53] <elenril> i'll just assume it works fine
[20:27:55] <Compn> well i cant read japanese , so boxes or japanese doesnt make much difference to me
[20:28:10] <Compn> kanji i should say
[20:28:12] <elenril> it looks nicer
[20:44:38] <Flameeyes> elenril: Kanno? which ost is that? not .hack//sign I guess :)
[20:45:11] <elenril> Flameeyes: kajiura
[20:45:26] <Flameeyes> gha sorry always confuse the two =_=
[20:45:27] <elenril> Tsubasa Chronicles OST III
[20:45:39] <Flameeyes> but yeah the same one as .hack//sign :)
[20:45:50] <elenril> an incredibly awesome ost for an incredibly crappy anime
[20:46:00] <elenril> right
[20:46:25] <Flameeyes> well if it is anything to go by .hack's OST... I need to track that down :)
[20:46:35] * Flameeyes bought the two .hack OSTs from amazon.jp years ago
[20:47:20] <elenril> wait....you mean you don't have all the music she's ever written??!!
[20:47:24] <elenril> *shock*
[20:47:34] <Flameeyes> ehm not yet sorry
[20:47:45] <Flameeyes> I needed to complete the YUI discography first
[20:48:31] <elenril> you really should get the tsubasa chronicle ost
[20:48:37] <elenril> and the noir ost
[20:48:50] <elenril> and the xenosaga 2 & 3 osts
[20:49:36] <Flameeyes> oh speaking about which .. three new lives from YUI ¬_¬
[20:49:45] <elenril> ...who's that?
[20:49:47] * elenril runs
[20:49:52] <Compn> yoko kanno is great
[20:49:53] * Flameeyes loves jpop, if elenril knows what he's referring to
[20:50:01] <Compn> but my favorite is probably yasushi ishii
[20:50:39] <Flameeyes> damn, that site has nothing from kajura :(
[20:50:52] <Flameeyes> and I need coffe
[20:50:56] <Flameeyes> very very much coffe
[20:50:57] <elenril> yoko kanno is great, but her music isn't so good for standalone listening
[20:51:16] <elenril> Flameeyes: recommend me something from yui?
[20:51:20] <Flameeyes> I skipped an i here, and a u there =_=
[20:51:34] <Flameeyes> elenril: holidays in the sun
[20:52:18] * elenril raises the jolly roger
[20:52:55] <elenril> arrr!
[20:53:18] <Flameeyes> tell sony to publish their stuff in europe! >_<
[20:53:28] <elenril> >sony
[20:53:33] <elenril> EVIL!
[20:53:33] * Flameeyes actually has all of Utada's work paid for... since _that_ is released here...
[20:53:49] <elenril> also downloading music is perfectly legal, so i'm not doing anything bad ;)
[20:55:46] <Compn> did you try cdjapan.co.jp ?
[20:55:51] <Compn> they ship internationally
[20:55:58] <Compn> but like $50 per cd isnt a very good price
[20:55:59] <Compn> lol
[21:07:00] <Flameeyes> Compn: plus 60% over that for import duties in italy
[21:07:54] <elenril> still no usable music downloads services?
[21:09:33] <Flameeyes> elenril: covering wide j-pop range? not that I know
[21:09:47] <Flameeyes> itunes has some stuff (Utada, L'Arc~en~Ciel, ...)
[21:10:03] <elenril> bleh, itunes
[21:11:50] <elenril> wait, i've heard this one somewhere
[21:12:00] <elenril> ....oh, FMA opening
[21:12:32] <Flameeyes> L'Arc? second FMA opening, first GTO opening
[21:12:42] <Flameeyes> and DNA² opening as well
[21:12:57] * elenril np again by YUI on HOLIDAYS IN THE SUN
[21:21:47] * BBB sadly notices that yet again, michaelni appears completely not interested in finding solutions to problems :(
[21:39:43] <michaelni> BBB, i do try to find a solution, i just think you dont want  to accept where the problem is
[21:40:04] <michaelni> i dont care all that much, once my payed work is done iam out of here
[21:40:22] <BBB> stop trolling, stop fueling the trollwar, I don't even read your emails anymore if this continues
[21:40:40] <BBB> I've told you what I'd like you to do: figure shit out with mans - go do it
[21:40:54] <michaelni> i tried, i failed
[21:40:59] <BBB> I noticed
[21:41:35] <uau> IMO those votes discussed on the mailing list are a bad idea, whether done with open or secret ballots
[21:41:36] <CIA-38> ffmpeg: Alexander Strange <astrange at ithinksw.com> master * r6b47495397 ffmpeg/ (ffmpeg.c ffplay.c):
[21:41:36] <CIA-38> ffmpeg: Adopt pkt_dts/pkt_pts in lavc clients
[21:41:36] <CIA-38> ffmpeg: No behavior change; this makes DTS reliable with the next patch.
[21:41:36] <CIA-38> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[21:43:02] <BBB> uau: agreed
[21:43:26] <uau> democracy isn't such a useful system for opensource projects anyway - its main benefit is resistance to corruption towards obviously bad government that cannot be changed because the existing bad government decides changes
[21:44:02] <uau> but in opensource projects people can fork anyway if that becomes really obvious, so there's not so much benefit from democracy
[21:44:56] <uau> also who gets to vote in an opensource project if you do try to decide things "democratically"?
[21:45:09] <uau> who'd get to vote here? the people who had write access to svn? IMO that would be completely arbitrary
[21:47:37] <uau> if you'd want to make it really democratic you'd need to allow *anyone* to vote, not just those who got write access through some fairly arbitrary criteria and circumstances
[21:48:29] <uau> but i think most people can see that allowing all the idiots on the mailing list (and elsewhere) would not be particularly likely to produce a *good* solution, even if that'd be the most democratic way
[21:48:51] <uau> allowing [to vote]
[21:51:53] <BBB> yeah voting gets sort of complex now
[21:53:38] <kierank> is all the valgrind complaining in v_block_filter in error_resilience a false positive?
[21:55:30] <astrange> i haven't read the thread in detail, but i think AVCodecContext.active_thread_type is the field the mt thread wants
[21:55:46] <astrange> kierank: try --track-origins=yes
[21:56:09] <BBB> astrange: I don't think that works if pthreads are disabled, does it?
[21:56:15] <uau> astrange: is it set before any callbacks can be called? if so then it's probably an ok solution
[21:57:25] <uau> BBB: why wouldn't it?
[21:57:43] <uau> would it not be set to 0 in that case for some reason?
[21:57:57] <astrange> it's set inside avcodec_open, callbacks happen after that
[21:58:19] <astrange> by pthreads are disabled, do you mean thread_count=0, or --disable-pthreads?
[21:58:27] <BBB> both
[21:58:44] <BBB> I guess I primarily meant --disable-pthreads ./configure
[21:58:51] <astrange> first - should be 0 (i didn't check), second - FF_THREAD_SLICE because of pthreads emulation
[21:59:16] <astrange> which is used so the threading tests still work without threads
[21:59:18] <BBB> because avctx->thread_count appears to be set to 1 regardless
[22:06:38] <Flameeyes> elenril: did a good thing to get those two lives.. "Unsupported codec (id=102400) for input stream 2" =_=
[22:06:41] <Flameeyes> new test files!
[22:07:25] <astrange> 1
[22:42:59] <Flameeyes> somebody fancying some debugging? :D http://paste.pocoo.org/show/332816/ file is a "meager" 400M :P
[22:44:36] <Flameeyes> sorry wrong paste, but the same happens with current git
[22:46:48] <kierank> Flameeyes: what's the problem?
[22:47:17] <Flameeyes> kierank: mostly the ong amount of bad stuff printed out ;) my script seems to be aborting, trying to understand that myself
[22:47:44] <kierank> well the stuff printed out happens on virtually all h.264 ts files because they're not cut cleanly
[22:48:12] <Flameeyes> uhm it seems like it's the multi-pid that throws my script out, damn :|
[23:05:14] <Dark_Shikari> BBB: The speed seems to be similar, at least on this clip
[23:05:22] <Dark_Shikari> it would definitely be more useful in h264 though
[23:05:54] <Dark_Shikari> BBB: http://pastebin.com/2TzTDgFp
[23:05:57] <Dark_Shikari> hunks 1/2 are my evil trick
[23:06:10] <Dark_Shikari> hunk 3 helps, but it breaks bigendian because we have no AV_RL32A
[23:06:59] <Sean_McG> htonl()?
[23:11:54] <BBB> what does RL32A do?
[23:12:08] <BBB> 32-bit aligned read?
[23:12:27] <Dark_Shikari> yes
[23:12:36] <Dark_Shikari> the & 0xFF and >>8 bits assume little-endian
[23:12:54] <Dark_Shikari> but it helps a few clocks, at least.  the same thing in the top part would also help.
[23:22:13] <BBB> for a "few clcks", this is a little ugly :-p
[23:22:21] <Dark_Shikari> The bottom part is ugly?
[23:22:23] <Dark_Shikari> It's like two lines
[23:22:35] <Dark_Shikari> It doesn't seem very ugly to me
[23:23:07] <BBB> the top part is ugly
[23:23:14] <BBB> the bottom part is fine
[23:23:21] <Dark_Shikari> The top part was my idea.
[23:23:25] <Dark_Shikari> The bottom part was a simple optimization.
[23:23:34] <Dark_Shikari> 10:05 <@Dark_Shikari> hunks 1/2 are my evil trick
[23:23:34] <Dark_Shikari> 10:06 <@Dark_Shikari> hunk 3 helps, but it breaks bigendian because we have no AV_RL32A
[23:23:49] <Dark_Shikari> the "few clocks" was about the last hunk
[23:24:01] <BBB> ah ok
[23:24:05] <BBB> sorry, misunderstood
[23:24:11] <BBB> AV_RL32() makes it significantly slower?
[23:24:22] <Dark_Shikari> It's definitely less efficient than a hypothetical AV_RL32A
[23:24:24] <BBB> (shouldn't make a difference on x86 anyway)
[23:24:29] <Dark_Shikari> On non-x86, at least.
[23:25:02] <BBB> AV_RL/B8/16/32A() is fine with me
[23:25:33] <BBB> (but let's not worry about that)
[23:25:46] <BBB> does the top make any difference?
[23:25:49] <BBB> it's kind of ugly
[23:26:19] <Dark_Shikari> I haven't done much testing
[23:26:21] <Dark_Shikari> I just wanted to show it could be done
[23:26:26] <Dark_Shikari> it would be more useful in h264 than in vp8 as I said
[23:27:03] <Dark_Shikari> By the way, if you wanted to be a gigantic dick, you could do something crazy like iterate over all the DC-only ones first, then the non-DC-only ones.
[23:27:07] <Dark_Shikari> The whole approach is pretty generic.
[23:34:11] <astrange> any reduction in branches in h264 is good
[23:36:24] <Sean_McG> do the branch prediction hint macros help with that?
[23:36:34] <BBB> yeah h264 is pretty branchy and has some incredible spaghetti code in places :-p
[23:36:56] <BBB> but then again I don't think it'd speed it up much to get rid of that, better just get ffmpeg-mt in place