Timestamps are in GMT/BST.
[1:18] * Zorix- (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[1:23] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[1:39] * Connelly (Default@c-24-21-147-63.client.comcast.net) has joined #freenet
[1:40] <Connelly> what license is the freenet source under?
[1:50] <Connelly> ok GPL
[1:51] * Connelly (Default@c-24-21-147-63.client.comcast.net) has left #freenet
[1:53] * spaetz (~spaetz@217-162-197-136.dclient.hispeed.ch) has joined #freenet
[1:59] * Zorix- (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[2:38] * MrNaughty (MrNaughty@d198-166-246-103.abhsia.telus.net) has joined #freenet
[3:21] * Schmeez (~nobody@DA-207164198.113.trytel.com) has joined #freenet
[3:23] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[4:30] * tessier (~knoppix@wsip-68-224-172-77.sd.sd.cox.net) has joined #freenet
[4:30] <tessier> Hello all!
[4:30] <tessier> Where is most of the freenet discussion happening these days? On the freenet-tech list?
[4:30] <tessier> Somehow I never got around to subscribing to it and -devel has been on the quiet side lately.
[4:31] <ShaunMacPherson> yes
[4:31] <tessier> I just looked at freenet-tech on sourceforge and it's all spam!
[4:31] <ShaunMacPherson> really
[4:31] <tessier> http://sourceforge.net/mailarchive/forum.php?forum_id=6244
[4:32] <ShaunMacPherson> hmmm people have not been taking care of freenet :)
[4:37] <tessier> So perhaps that isn't where the mailing list chatter is these days.
[4:37] <ShaunMacPherson> devl isnt that bad these days
[4:43] <plixed_> try http://dodo.freenetproject.org/pipermail/tech/2004-October/thread.html
[4:48] <ShaunMacPherson> I think you should all try I2P, not that I am being a advertising fanboy but it seems to work well :)
[4:53] <i2p_iip> <gott> cunt
[5:04] <tessier> When did the freenet-tech list get created?
[5:04] * tessier wonders how much he has missed
[5:06] <tessier> What is it with toad and never trimming his replies?
[5:07] * tessier wonders how much time he has lost scrolling through toad's replies over the years
[5:07] <hobx_> tessier: Summer of 2000 I think.
[5:07] <tessier> holy shit!
[5:07] <tessier> freenet-tech was created in summer of 2000? So what is -devel for?
[5:07] <hobx_> But it didn't used to contain anything
[5:10] <plixed_> tessier: i don't think you missed much
[5:11] * kers (~kers@25.ppp135.rsd.worldonline.se) has joined #freenet
[5:14] <hobx_> It didn't used to be interesting because I only posted and read devel.
[5:14] <hobx_> Now I'm uninterested in devel, but I got back on tech to argue with Ian a little, so it is more interesting.
[5:14] <hobx_> :-)
[5:24] <ShaunMacPherson> thetower is back
[5:37] * Sugadude (~Sugadude@geezy.cs.purdue.edu) Quit ()
[5:38] * kers (~kers@25.ppp135.rsd.worldonline.se) Quit ("Leaving")
[5:39] * kers (~kers@25.ppp135.rsd.worldonline.se) has joined #freenet
[5:43] * Sugadude (fuckyou@dsl027-178-156.sfo1.dsl.speakeasy.net) has joined #freenet
[6:02] * d-ArkAngel (~Rob@213-131-104-86.onyx.net) has joined #freenet
[6:03] <d-ArkAngel> Morning all
[6:14] <hobx_> morning Ark
[6:14] <d-ArkAngel> How's things?
[6:14] <hobx_> tired, broken.
[6:14] <i2p_iip> <NewsByte> very good, thnk you
[6:15] <hobx_> hopeless
[6:15] <hobx_> typical Wednesday morning.
[6:15] <i2p_iip> <NewsByte> that's the spirit!
[6:15] <d-ArkAngel> :-)
[6:15] <i2p_iip> <NewsByte> toad around, somewhere?
[6:16] <d-ArkAngel> so what's going on?
[6:16] <hobx_> toad is a very lazy boy
[6:16] <i2p_iip> <NewsByte> I was thinking about sending in a /. article, as we've discussed before, about the simulator. But I wanted to check first
[6:16] <d-ArkAngel> what makes you say that?
[6:16] <hobx_> sleeps till noon, eats with a spoon, never gets out bed, wants to be fed
[6:17] <hobx_> what exactly are you submitting to slashdot?
[6:17] <i2p_iip> <NewsByte> and working on fred
[6:17] <i2p_iip> <NewsByte> ;-)
[6:17] <d-ArkAngel> huh?
[6:18] <d-ArkAngel> slashdot?
[6:18] <i2p_iip> <NewsByte> well, about the fact freenet has a simulator which shows some interesting graphs, and such
[6:18] <d-ArkAngel> there's nothing lazy about sleeping till noon.
[6:18] <hobx_> for gods sake don't
[6:18] <i2p_iip> <NewsByte> I'm an atheist ;-)
[6:18] <hobx_> All it will do is make the project look amateurish and silly
[6:19] <hobx_> "look at our pretty numbers"
[6:19] <hobx_> And the more braindead a submission, the more likely /. will run it...
[6:19] <i2p_iip> <NewsByte> well, perhaps, but that's inherent to it :-)
[6:20] <i2p_iip> <NewsByte> let's not be to pessimistic. People could try out the simulator itself, after all
[6:20] <i2p_iip> <NewsByte> it's an oportunity to get more folks and money; we shouldn't let it pass
[6:21] <d-ArkAngel> or more worryingly they could all start nodes... which it seems the generaly opinion is that it's not going to do the network any favours...
[6:21] <hobx_> It is an opportunity to look stupid and needy of attention.
[6:21] <i2p_iip> <Sonax> Right now, is the simulator ready for "use" by "normal" users? Is there a guide, or howto? I thought you had to have coded it your self to use it?
[6:21] <i2p_iip> <NewsByte> toad agreed, some time ago. I just don't understand why he didn't submit one himself, now that we have an improved simulator. That's why I wanted to check, first.
[6:21] <hobx_> Is there much point in spending a lot of time making a simulator user ready?
[6:22] <hobx_> A simulator is very fit for being "developerware"
[6:22] <hobx_> to the extent I make sims configurable and scriptable it is only to save myself the bother of a recompile.
[6:22] <i2p_iip> <Sonax> My point is that if it is not user ready, posting on slashdot that we have one wont make as much sence. Not much to "see" if you cant run one your self.
[6:22] <i2p_iip> <NewsByte> it may come as a surprise, but you have nerds and geeky devls, even on /. ;-)
[6:22] <d-ArkAngel> and the sim is not a end in it's self, surely it should be /. when we've got some results we understand enough.
[6:23] <i2p_iip> <NewsByte> which we have. Toad has posted enough examples + comments on the devl list
[6:23] <hobx_> Toad has posted a bunch of pngs with hundreds of lines on them.
[6:24] <i2p_iip> <NewsByte> though it could be that the simulator isn't user-ready yet, which is a valid point, as sonax says
[6:24] <i2p_iip> <gott> it takes away from the import of 0.6, as well
[6:24] <i2p_iip> <gott> the more freenet gets slashdotted, the less interested people are in it
[6:25] <i2p_iip> <NewsByte> that's an unproven statement, and thus I'll not respond further on it
[6:25] <hobx_> I agree. I think a lot of people already go "not freenet again"
[6:26] <i2p_iip> <gott> there aren't enough ian clarke fansites on freenet
[6:26] <hobx_> I realize that slashdot is a revenue source, but I think submitting every little step is short sighted.
[6:26] <i2p_iip> <NewsByte> that's because freenet was a dissapointment for them, thusfar. Now it's about a simulator, if you make that clear, then even /.ers will understand it's not about the network itself
[6:26] <i2p_iip> <NewsByte> hobx, I agree with that, however:
[6:27] <i2p_iip> <NewsByte> we have not made a /. for some time, and we certainly did not for every little step
[6:27] <hobx_> But that is like when you have stories about people writing a document about what they are going to develop.
[6:27] <hobx_> Those always piss me off.
[6:27] <d-ArkAngel> and they'll all click on the link to the site, they'll all download a node, they'll all run it, and they'll all be dissapointed when the million new nodes on the network can't route bugger all.
[6:27] <i2p_iip> <NewsByte> in fact, we're long overdue. Most other P2P have a FAR more normal version implementation
[6:29] <i2p_iip> <NewsByte> yes, well, that's a freenet-fault, then. If we're going to wait untill freenet works fine before /.ing, then we won't put another article in for ten years.
[6:29] <d-ArkAngel> till the work with the sim is finished and the lessons learned are in the freenet code I don't see anything worth /. ing
[6:30] <hobx_> Perhaps if we worried more about getting it working and less about /., then it wouldn't be ten years.
[6:30] <d-ArkAngel> the fact that there is a simulator that's been written and is being used to produce results is harldly news.
[6:30] <i2p_iip> <NewsByte> I can partly agree to that. But it shouldn't go the way of freenet itself, then.
[6:31] <i2p_iip> <NewsByte> I mean, if we're going to wait untill the simulator is 'perfect', it will just be like with the network: we can wait 10 years on it
[6:31] <d-ArkAngel> it's not about getting the simulator perfect
[6:31] <d-ArkAngel> it was never the point
[6:31] <d-ArkAngel> the sim is a tool.
[6:31] <i2p_iip> <NewsByte> sometimes, you just have to go for it. It's a devl-process, after all.
[6:32] <d-ArkAngel> what was the last news about freenet?
[6:32] <i2p_iip> <NewsByte> well, I sincerely hope so. But note the argument that freenet should work good before making another article. On itself it sounds great, but in that way, you could wait an eternity
[6:32] <i2p_iip> <NewsByte> on /.? A huge time ago (as a head-article)
[6:33] <d-ArkAngel> it shouldn't work good, but it should be at a point where the developers want to try adding a huge number of nodes
[6:34] <i2p_iip> <NewsByte> which will be when it works good ;-)
[6:35] <i2p_iip> <NewsByte> muy opinion is, that it's still a beta after all, and when a major release comes out, it should get a slashdot, 'good' or not.
[6:35] <d-ArkAngel> yeah, and there is no revision here
[6:35] <i2p_iip> <NewsByte> as for that matter, I think we're long overdue for a major release number
[6:35] <i2p_iip> <NewsByte> indeed, and since we don't have that, we have to focus on something else, like the simulator
[6:36] <d-ArkAngel> I give up.
[6:36] <i2p_iip> <NewsByte> good! ;-)
[6:36] <i2p_iip> <NewsByte> our views on the matter just differ, I guess. Though the argument of sonax was a good one, I thought
[6:37] <d-ArkAngel> what that it's not user ready?
[6:37] <i2p_iip> <NewsByte> indeed
[6:37] <hobx_> pain
[6:37] <d-ArkAngel> have you ever run the sim?
[6:38] <hobx_> if ever a body suffered so...
[6:38] <i2p_iip> <NewsByte> I must agree that, if one is going to make a :. article about the simulator, and offer it for d/l, at least ppl should be able to try it out (easily)
[6:38] <i2p_iip> <NewsByte> no, I have TRIED running the sim ;-)
[6:38] <d-ArkAngel> you realise that the sim doesn't produce a graph don't you?
[6:39] <d-ArkAngel> it's just a large amount of text.
[6:39] <d-ArkAngel> it's hearly something that's easy to pick up and play with.
[6:39] <i2p_iip> <NewsByte> does that have any merrits on the point sonax made? :-)
[6:40] <i2p_iip> <NewsByte> I would say, it's proof user-friendliness could improve
[6:40] <d-ArkAngel> which it won'y
[6:40] <d-ArkAngel> why would anyone spend the time writeing user-freindliness into a simulator?
[6:41] <i2p_iip> <NewsByte> well, that depends on who's working on it
[6:41] <d-ArkAngel> if you're not going to spend the time learning what the sim is doing then the results are meaningless to you anyway.
[6:41] <i2p_iip> <NewsByte> toad agreed that it should be made more user-friendly, so I'm guessing he'll work towards that as well
[6:41] <d-ArkAngel> *sigh*
[6:43] <i2p_iip> <NewsByte> angel, your objections are not fundamental in nature. You just say: "it's too difficult", why sonax and me agree, but our view is, that it should be made more user-friendly. Granted that not every joe doe will try it, but there are puter-geeks enough on :.
[6:44] <d-ArkAngel> and the point of them trying it is?
[6:44] <i2p_iip> <NewsByte> the pure thrill of trying it out. You know how nerds and geeky devls are; always in for a new try on something
[6:45] <d-ArkAngel> you know that it takes more than a day to draw a single line don't you?
[6:45] <i2p_iip> <NewsByte> that depends on the box you are running
[6:45] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[6:46] <d-ArkAngel> that's on a 2ghz opteron.
[6:46] <d-ArkAngel> so that accounts for at least 99.9% of the /. crowd being at least that slow.
[6:46] <i2p_iip> <NewsByte> what? they didn't even use the Earth Simulator?!! :-o
[6:47] <i2p_iip> <NewsByte> well, have to work a bit, see you later
[6:56] * bllx (~rog@host217-44-61-160.range217-44.btcentralplus.com) has joined #freenet
[6:56] * bllx (~rog@host217-44-61-160.range217-44.btcentralplus.com) Quit (Client Quit)
[6:59] * bllx (~rog@host217-44-61-160.range217-44.btcentralplus.com) has joined #freenet
[7:17] * kers (~kers@25.ppp135.rsd.worldonline.se) Quit ("Leaving")
[7:37] <hobx_> oohhh fate, why do you mock me!
[7:51] <i2p_iip> <fate> because you are set on this earth to be mocked, my son!
[7:58] * Hory (~Miranda@82.76.81.56) Quit (Read error: 113 (No route to host))
[8:04] * michaelkuijn (~michaelku@fia41-111.dsl.hccnet.nl) has joined #freenet
[8:05] <hobx_> that is true if anything is
[8:08] * kers (~kers@14.ppp132.rsd.worldonline.se) has joined #freenet
[8:11] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[8:11] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[8:21] * Zv2S (~rfc1413@squirrel.owl.de) Quit (Read error: 110 (Connection timed out))
[8:28] * nextgens (~nextgens@d80-170-239-27.cust.tele2.fr) has joined #freenet
[8:29] <nextgens> hi
[8:33] <toad_> hobx: you're an inconsistent idiot. first you criticise freenet for being impossible, for having no rigorous basis, and so on. then when we try to give it one you say it doesn't matter.
[8:34] <toad_> NewsByte: we haven't finished with the simulations yet
[8:34] <hobx_> ouch
[8:34] <hobx_> "inconsistent idiot". The gloves must be off :-)
[8:34] <hobx_> What exactly are you replying to?
[8:34] <toad_> <hobx_> Perhaps if we worried more about getting it working and less about /., then it wouldn't be ten years. - that is also bull****
[8:35] <toad_> hobx: the entire conversation above
[8:35] <hobx_> yeah
[8:35] <hobx_> that was demagogery. Sorry :-)
[8:35] * Ash-Fox (Hal-9000@aaw12.neoplus.adsl.tpnet.pl) Quit (Read error: 104 (Connection reset by peer))
[8:36] <toad_> <d-ArkAngel> it shouldn't work good, but it should be at a point where the developers want to try adding a huge number of nodes - or when we urgently need funds? :)
[8:36] <hobx_> toad_: The conversation was about trying to get freenet on slashdot's webpage the whole time.
[8:36] <toad_> <i2p_iip> <NewsByte> muy opinion is, that it's still a beta after all, and when a major release comes out, it should get a slashdot, 'good' or not. - it is quite likely that 0.6 will
[8:36] * hobx_ explains usage: I reconsidered. He changed his mind. You went back on your word. John Kerry flip-flopped.
[8:36] <toad_> <i2p_iip> <NewsByte> as for that matter, I think we're long overdue for a major release number - yes, because if we put out 0.6 and it works worse than 0.5, everyone will love it
[8:37] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit ("Going to the moon")
[8:37] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 54 (Connection reset by peer))
[8:37] * pupok (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[8:37] <toad_> <i2p_iip> <NewsByte> toad agreed that it should be made more user-friendly, so I'm guessing he'll work towards that as well - no, I am not going to spend serious effort on making the simulator idiot friendly
[8:39] <hobx_> so we agree
[8:39] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 104 (Connection reset by peer))
[8:39] <hobx_> what are you angry about?
[8:39] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[8:39] <toad_> n/m :)
[8:40] <hobx_> so fucking tired...
[8:40] <toad_> I still think we need a /. soon, with some solid sim results and possibly some smaller improvements to the network
[8:40] <hobx_> I'm considering going back on caffiene.
[8:40] <toad_> but NOT YET
[8:40] <toad_> if NB /.'s us _now_, I'll have him hung, drawn and quartered :)
[8:41] <toad_> okay, ttyl
[8:41] <hobx_> teh he
[8:41] <hobx_> can come back and reply in a couple of hours to what has been said since then
[8:42] <toad_> w.r.t. my alleged laziness, i only charge for the hours i work, so STFU.
[8:42] * michaelkuijn remembers when he first found freenet...
[8:42] <toad_> michaelkuijn: hmm?
[8:42] <toad_> michaelkuijn: wasn't that a few days ago?
[8:43] <michaelkuijn> I don't think so
[8:43] <toad_> did it work better then?
[8:43] * hobx_ remembers when he first found freenet. No wait a minute, I don't.
[8:43] <michaelkuijn> Nothing romantic *scratch* I was browsing the ebuild index at packages.gentoo.org ;-)
[8:43] <toad_> ah cool
[8:44] * toad_ read the original mini-article in new sci... years ago
[8:44] * jokern (~virus@213.184.217.12) has joined #freenet
[8:44] <toad_> but don't remember first time i lurked on devl
[8:44] * Hory (~Miranda@82.76.81.56) Quit (tolkien.freenode.net irc.freenode.net)
[8:45] * Hory (~Miranda@82.76.81.56) has joined #freenet
[8:45] * Hory (~Miranda@82.76.81.56) Quit (Connection reset by peer)
[8:46] <toad_> updated: http://amphibian.dyndns.org/simulations2-epass-0.025-800x25.png http://amphibian.dyndns.org/simulations2-epass-0.025-800x25-smooth.png
[8:47] <hobx_> I think the old freenet archives are lost to history
[8:47] <hobx_> I wonder if the wayback machine has them
[8:48] <toad_> also http://amphibian.dyndns.org/simulations2-epass-0.025-800x25-smooth.png and http://amphibian.dyndns.org/simulations2-epass-0.025-400x50-smooth.png
[8:48] * pupok (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[8:48] <toad_> both of which look moderately promising (the concern on 800x25 is how far it will fall)
[8:48] <toad_> ah doh
[8:49] <toad_> i mean http://amphibian.dyndns.org/simulations2-epass-0.025-400x50-supersmooth.png and http://amphibian.dyndns.org/simulations2-epass-0.025-400x50-smooth.png
[8:49] * michaelkuijn is making his own CD sleeves (cut, fold, glue...)
[8:49] <toad_> and the above
[8:49] <hobx_> michael: Glitter!
[8:49] <michaelkuijn> Glitter?
[8:50] <i2p_iip> <newsbyte> I'm glad all those graphs are being shown with such enthousiasm! Maybe we should finally make a /. article about it?
[8:50] <toad_> newsbyte: not yet
[8:50] <michaelkuijn> Publicity! Yay!
[8:50] <toad_> we don't have anything like enough to prove the point yet :)
[8:50] <toad_> we need more horsepower to generate more graphs, we need some technical advances i.e. serialization, ability to browse the RTs of each node
[8:50] <toad_> and finally we need the network to work a little bit better
[8:51] <michaelkuijn> The network will be spammed with on-off nodes and people running it long enough to integrate and then to leave :-(
[8:51] <toad_> some supercomputer time would be helpful, ideally, after d-ArkAngel's parallel code is ready (it's not entirely implausible ;| )
[8:51] * nextgens nextgens is happy: his node ran for 14hours without a crash, having transfered around 130Mo ... without request and without strange things in logfile :-)
[8:52] * nextgens starting Fuqid again
[8:52] <i2p_iip> <newsbyte> toad, are you saying, thus, that we won't send a /. article in ten years? ;-)
[8:53] <toad_> newsbyte: when we are ready i'll get ian to do ti
[8:53] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[8:53] <toad_> it
[8:53] <toad_> via an announce mail
[8:53] <toad_> if you preempt us, i'll terminate your cvs access
[8:54] <michaelkuijn> Heh
[8:54] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-200x25.png and http://amphibian.dyndns.org/simulations2-epass-0.025-200x25-smooth.png are also updated
[8:55] <toad_> Newsbyte: if with the threaded sim we can get some credible results on 1600x50 (or 1600x100), that might be enough on its own. ideally we want some on 3200 as well to show scalability, but that'll be sloow.
[8:56] <toad_> technically, we want: a) threaded sim, b) serialization, c) dumping of RT to text so we can see if nodes are specializing
[8:56] <toad_> none of that will take very long
[8:57] <toad_> when i've done b), i'll start a sim again on erica, and on two volunteers' nodes hopefully, and probably stop some existing ones
[8:57] <toad_> i'm not actually working on a), although i should check the code out
[8:58] * greycat (~wooledg@192.35.79.70) has joined #freenet
[8:59] <toad_> hi greycat
[8:59] <toad_> Newsbyte: we may need to reset stable in order to fix the insert issues also
[8:59] <toad_> I suspect there are still some problems
[8:59] <toad_> with timeouts
[9:01] <greycat> hello
[9:02] <michaelkuijn> Anyone using FMB by the way?
[9:03] <michaelkuijn> *silence*
[9:05] <michaelkuijn> Anyone here?
[9:05] <michaelkuijn> ECHO ))))))
[9:05] <michaelkuijn> (((((( ECHO
[9:06] <greycat> SSK@wlUk5niRuNDUL4lzvdHu6DamhW8PAgM/Greg_Wooledge// is back in stable now.
[9:06] <michaelkuijn> Mmm?
[9:08] <i2p_iip> <Sonax> greycat: got that this morning when i woke up. Good to see you back online :)
[9:08] <i2p_iip> <Sonax> online --> on freenet
[9:09] <michaelkuijn> Anyone using FMB?
[9:10] <i2p_iip> <Sonax> michaelkuijn: very few as far as i know. Frost is "the thing".
[9:10] <toad_> what was it about the tower being back?
[9:11] <michaelkuijn> I thought so... I can't get it to work, always searching for the announce stuff
[9:11] <michaelkuijn> However, FMB seems nicer than Frost. Real threads and stuff.
[9:11] <michaelkuijn> What's the tower?
[9:11] <greycat> It was a Freesite.
[9:11] <toad_> author of a major freesite
[9:11] <michaelkuijn> Ah
[9:11] <toad_> left ages ago
[9:12] <toad_> [5:24] <ShaunMacPherson> thetower is back
[9:12] <i2p_iip> <Sonax> He is running around over at i2p. Has an eeepisite (or what they call flogs over there).
[9:12] <greycat> without the space, he probably meant the person
[9:12] <michaelkuijn> Am I the only one who hates the way frost organizes its messages? (Like NOT)
[9:12] <toad_> ah, so he's just on I2P, okay
[9:13] <toad_> michaelkuijn: it's pretty primitive...
[9:13] <michaelkuijn> Mmm. So we have Entropy, Freenet and now I2P?
[9:13] <toad_> well, Entropy is dead, and good riddens
[9:13] <toad_> 4 bit fixed specialization, and home grown crypto algorithms that could be cracked in an afternoon by a bored NSA cryptologist
[9:13] <michaelkuijn> The Entropy folks seem to think otherwise
[9:14] <toad_> there was a /. article ages ago saying entropy is dead, so it must be true ;)
[9:14] <michaelkuijn> I don't trust slashdot
[9:14] <toad_> well the main guy left, i do know that
[9:14] <d-ArkAngel> I don't trust anyone ;-)
[9:14] <i2p_iip> <Sonax> Toad: How do you feel about a slashdot story about the simulations. Newsbyt seems to have a itch in his fingers, so if you want it (or don't) you might want to be very clear.
[9:15] <toad_> Sonax: I am clear. "NO".
[9:15] * lolo-laptop (~lostlogic@68.251.84.226) has joined #freenet
[9:15] <michaelkuijn> Yup yup
[9:15] <toad_> as in not now, not yet, we will do an announce when we need one
[9:15] <d-ArkAngel> Sorry I exited in the midsts of our conversation yesterday toad_ the internet connection here fail (got some engineers coming out to fix it.
[9:15] <i2p_iip> <Sonax> Toad: Nice and clear. Thank you.
[9:15] <toad_> d-ArkAngel: what were we talking about?
[9:16] <michaelkuijn> I'd ask for the checksum of the "NO" :-) sorry couldn't resist
[9:16] <toad_> 19541a2746e08a6b8f5145bdbaa23e45
[9:17] <greycat> haha
[9:17] <michaelkuijn> michaelkuijn@michael1 michaelkuijn $ echo NO | md5sum
[9:17] <michaelkuijn> 19541a2746e08a6b8f5145bdbaa23e45
[9:17] * greycat did the same
[9:17] <michaelkuijn> Verified!
[9:17] <toad_> hmmm
[9:17] <d-ArkAngel> toad_ I was being "impractical and defeatist and unhelpful" :-)
[9:17] <toad_> my node is definitely not coexisting with my simulation...
[9:18] <toad_> d-ArkAngel: by suggesting we do 5x more simulations? :)
[9:18] <d-ArkAngel> I was actually surgesting that we come up with a voodoo based formular to link RT with Network size.
[9:18] <toad_> essentially what I want is to show that we can MAKE it scale logarithmically, more or less, on network size, even if we have to slowly (and i mean log^2 at worst) increase the RT size
[9:18] <toad_> does that make sense?
[9:19] <d-ArkAngel> yes that makes total sense, and I agree with you.
[9:19] <toad_> btw how's the parallel sim going?
[9:19] <toad_> it's our only realistic hope for 1600... and it's still going to take weeks...
[9:19] * michaelkuijn is still cutting and folding
[9:20] <toad_> during which time i intend to work on the network
[9:20] <d-ArkAngel> what I'm thinking is that we should come up with some formular we can stick to with our experiments. I realise that it won't be a "good" relationship, and it won't "mean" anything. But it might be easier to analyse things if we can say that we've kept the RT size to be "25 log base RT of Network size" or something like that.
[9:20] <hobx_> What I don't understand is you said my sims weren't realistic because I only made it to 40,000 nodes or so.
[9:21] <toad_> hobx_: no
[9:21] <toad_> that wasn't why i said they were unrealistic
[9:21] <toad_> why i said they were unrealistic was another reason
[9:21] <toad_> the ratio of nodes added:inserts:requests
[9:21] <hobx_> But you said the sim was too slow
[9:22] <toad_> :|
[9:22] <hobx_> But the low ratio was to make it possible to get many nodes in a decent amount of time.
[9:22] <hobx_> Now you are running very few nodes in an indecent amount of time.
[9:22] <toad_> well, i can make it significantly faster by doing essentially what ian suggested - cut back to 1 estimator (from 3), and never DNF, just estimate based on the number of hops it takes to find the data
[9:23] <toad_> this seems grossly unrealistic to me but it could produce some interesting results...
[9:23] <d-ArkAngel> Toad_: WRT threaded sims - it's coming along, tho I'm at the point of adding the connection lists, and it seems like a very complex set of nested objects is used for selecting nodes, and I've got slightly side tracked looking to see if I can find an equivelent way of doing the same task, but in a single data structure....
[9:23] <toad_> the current sims use 3 estimators and an HTL - requests DNF
[9:24] <toad_> d-ArkAngel: sounds good, it is a bit of a mess...
[9:24] <d-ArkAngel> the thing that's takeing the time is trying to work out how the hell it's working, and what it's doing :-)
[9:24] <toad_> hobx_: the irony is that on the actual network, NGR is faster
[9:25] <toad_> d-ArkAngel: well, if you have any questions :)
[9:25] <d-ArkAngel> I'm sure I will :-)
[9:25] <d-ArkAngel> I'm just trying to let my brain settle on how to combine the bukets onto a 2D data structure that will allow better searching times.
[9:26] <jokern> Interresting :
[9:26] <jokern> 4:26:06 Consecutive same winner: freenet.node.QueueManager$NodeElement@17249fd: timeToSendWindow=0
[9:26] <toad_> d-ArkAngel: the buckets? eh?
[9:26] <d-ArkAngel> if it works out, it might mean that the sim doesn't suffer as badly when we increase the size of the RT
[9:26] <toad_> d-ArkAngel: please don't mess with the estimators directly unless you know what you are doing...
[9:26] <toad_> jokern: how many times?
[9:27] <d-ArkAngel> don't worry, I won't be making any changes to any for the code that's in use else where :-)
[9:27] <jokern> many :\
[9:27] <toad_> woah http://amphibian.dyndns.org/simulations2-epass-0.025-200x25-load.png
[9:28] <toad_> d-ArkAngel: rewriting the estimators is bound to lead to serious bugs and invalidate the whole experiment
[9:28] <toad_> please refrain from it :)
[9:28] <toad_> unless there is a GOOD reason and you've discussed it with me
[9:28] <toad_> by this i mean KeyspaceEstimator's, not the SNEs, which we don't use in the sim
[9:28] <d-ArkAngel> If I find a better way to implement lists of connections which needs any chances to the estimators object structure it'll get done in a new copy of the object in the sims directory, like the smaller LRUQueue object
[9:29] <toad_> i think when you say buckets you mean something else
[9:29] <toad_> by buckets i mean the buckets inside the keyspace estimators
[9:30] <d-ArkAngel> what I'm looking at is finding a way of self sorting the keys so that we can find the best node faster.
[9:30] <toad_> as far as connections go.. we need some way to find the LRU experienced connection, and if none, the LRU inexperienced connection... we need to be able to find out quickly if there are any of each set...
[9:30] <d-ArkAngel> but from what I can tell that'll require reading the insides of the estimators.
[9:30] <toad_> and we need to be able to check if we are connected to a node, and look up its peer
[9:30] <d-ArkAngel> yes
[9:30] <toad_> d-ArkAngel: why would that speed things up?
[9:31] <toad_> d-ArkAngel: when i have messed with rewriting the estimators for speed, it's been a disaster
[9:31] <d-ArkAngel> because with the current implementation how are we selecting which node to route to?
[9:31] <toad_> anyway we can't "self sort" the keys
[9:31] <d-ArkAngel> no I realise that
[9:31] <toad_> that would be classical routing
[9:32] <toad_> eventually I did rewrite the estimators successfully for speed, but that involved copying the file and only making trivial changes (bigint -> double)
[9:32] <d-ArkAngel> ok, just so I understand do you mind a quick chat about the estimators?
[9:32] <toad_> fire away
[9:33] <d-ArkAngel> I'm guessing they work by having a "bucket" which is a range of keys, (a start and an end of range) right?
[9:33] <toad_> afaics the ONLY way to do NGR is by actually calculating the estimates
[9:33] <toad_> especially if you're doing the 3-estimator variety
[9:33] <toad_> d-ArkAngel: yes, we have 8 buckets (for sim; 16 in reality)
[9:33] <toad_> each has a RunningAverage which is supposed to be in its center
[9:33] <toad_> we interpolate between the centers to get our guess
[9:34] <toad_> when we report a result, we report it on the RA of the nearest center, and move the sectors closer to the key slightly
[9:34] <toad_> so the sector boundaries are always movign
[9:35] <toad_> sector aka bucket
[9:35] <toad_> you cannot, for example, fix the bucket boundaries for all the nodes and speed things up that way
[9:35] <d-ArkAngel> yes I realise that.
[9:35] <toad_> because you lose half the algorithm and especially you lose the ability for nodes to specialize
[9:36] <d-ArkAngel> here's what I was thinking.
[9:36] <d-ArkAngel> we've got moving buckets that have fine granularty.
[9:37] <d-ArkAngel> so when a request for a key comes in then we've got a test to do on every estimator that invovles calculating the closes center for every estimator
[9:38] <d-ArkAngel> so what about if you were to split the key space into a larger number of smaller sections. and have the ones where both extreams are closest to the same center a anything in that range for that estimator is that center.
[9:39] <d-ArkAngel> if the two edges are for different centers then that means that one needs the calculations doing to find the nearest center for any key inside.
[9:39] <d-ArkAngel> so it becomes a space vs time tradeoff.
[9:40] <d-ArkAngel> based on the size of the sub estimators.
[9:40] <toad_> so you want to quantize the buckets?
[9:40] <d-ArkAngel> no...
[9:40] <d-ArkAngel> think of it as an index of the buckets...
[9:40] <d-ArkAngel> if you've got a bukect that covers the range 7-306
[9:41] <d-ArkAngel> and you've constructed these sub-markers at 50 unit intervals
[9:41] <d-ArkAngel> then 0-50 is marked that it needs a nearest center.
[9:41] <d-ArkAngel> but 50-100 100-150 etc...
[9:41] <toad_> so you have a linked list running off each element?
[9:41] <toad_> or what?
[9:41] <d-ArkAngel> become a lookup
[9:42] <toad_> you have to deal with two buckets being in one sub-bucket for one node
[9:42] <d-ArkAngel> yes
[9:42] <d-ArkAngel> that's my 0-50 case
[9:42] <d-ArkAngel> because there'll be another bucket from say -5000, 7
[9:42] <toad_> it sounds like it's going to use a lot of memory
[9:42] <d-ArkAngel> not really.
[9:42] <toad_> and you'll have to update the sub-buckets a lot
[9:43] <d-ArkAngel> yes, but it'll be a definabel update.
[9:43] <d-ArkAngel> because you know what you're moving
[9:43] <toad_> is it going to be a significant optimization?
[9:43] <toad_> d-ArkAngel: yes but you need EVERY sub-bucket to contain a pointer to the nearest real bucket at all times
[9:44] <d-ArkAngel> indeed
[9:44] <toad_> well not the nearest
[9:44] <d-ArkAngel> but there don't need to be huge number of sub bucketys
[9:44] <toad_> because that really complicates things
[9:44] <toad_> there do if there is any serious specialization going on
[9:44] <d-ArkAngel> an individual estimator won't benefit. in fact it'll probably get slower because of the additional updates needed
[9:44] <d-ArkAngel> but the benefit is going to be seen very quickly node that have more than one peer
[9:45] <toad_> will it actually be a speed gain, given that we're only using 8 buckets anyway?
[9:45] <d-ArkAngel> yes
[9:45] <toad_> what does it cost now to do a binsearch of doubles in 8 buckets?
[9:45] <d-ArkAngel> because we're using 50 peers
[9:45] <toad_> 4 comparisons?
[9:45] <toad_> we'd probably have to do 3 with your scheme
[9:45] <d-ArkAngel> it's the peer wise dimention that'll speed up, not the individual estimators...
[9:45] <toad_> or updating would be a total PITA
[9:46] <toad_> what individual operation would speed up?
[9:46] <d-ArkAngel> selecting the best peer
[9:46] <toad_> how?
[9:46] <toad_> what is the difference?
[9:46] <toad_> remember we are using 3 estimators
[9:46] <d-ArkAngel> because a key will fall into a fixed sub-bucket
[9:46] <toad_> so we need to calculate a value
[9:47] <d-ArkAngel> and all the sub bukets will be the same on all of the estimators...
[9:47] <toad_> yes, so we go to the sub-bucket
[9:47] <toad_> how do we determine exactly where the key is in terms of actual buckets?
[9:47] <d-ArkAngel> keys are comparable.
[9:47] <toad_> we have to compare to the buckets to the left and to the right to get the borders right and so on
[9:47] <toad_> that means 3 compares
[9:48] <toad_> as against 4 compares
[9:48] <toad_> no big gain eh?
[9:48] <d-ArkAngel> huge gain
[9:48] <toad_> why?
[9:48] <d-ArkAngel> because that's 3 compairs once, but the 4 compairs happens 50 times for 50 peers
[9:48] <toad_> well if you flag a bucket with a flag saying "don't need to check left and right", you cut it down to 1
[9:49] <toad_> d-ArkAngel: that makes no sense
[9:49] <toad_> you have to calculate a value for EVERY SINGLE FUCKING PEER
[9:49] <d-ArkAngel> no.
[9:49] <d-ArkAngel> that's the buty of it
[9:49] <toad_> certainly if you are doing it with 3 estimators
[9:49] <d-ArkAngel> the sub buckets don't move.
[9:49] <toad_> of course the sub buckets don't move, but the mapping of buckets to sub-buckets does change
[9:50] <d-ArkAngel> but that's only on an update action, and we're doing far more tests than updates
[9:50] <toad_> keeping those, and the flags, updated, will be a major source of bugs, and the results won't be valid anyway because they'll be using a different algo
[9:50] <d-ArkAngel> no, it'll be the same probabilites it'll return
[9:50] <d-ArkAngel> we're not loseing any granularity
[9:50] <toad_> d-ArkAngel: yes, and how precisely do we do the test i.e. the choice of whom to route to without calculating an estimate for each peer?
[9:51] <toad_> d-ArkAngel: yes, we can do it without losing granularity, but we then have to worry about comparing to left and right, and so on
[9:51] * FallingBuzzard (~srademach@207.152.112.129) has joined #freenet
[9:51] <toad_> and potentially further afield if several sectors fit inside one sub-bucket
[9:51] <d-ArkAngel> because the sub bukets can be pre sorted. as they have a fixed bundries. we only have to test those peers that have "misses"
[9:51] <toad_> i don't follow
[9:52] <i2p_iip> <newsbyte> "<toad_> if you preempt us, i'll terminate your cvs access" What? You mean you'll terminate your 'what's new' update?
[9:52] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 104 (Connection reset by peer))
[9:52] <i2p_iip> <newsbyte> ;-)
[9:52] <d-ArkAngel> if a sub bucket isn't totally inside a bucket then we have to do the normal comparisons for the estimator.
[9:53] <d-ArkAngel> but if the sub-buket is total inside then we don't need to.
[9:53] <i2p_iip> <newsbyte> besides, I will do it anonymously, ofcourse! You can't denie me cvs acces because some anonymous coward has send in an article!!!
[9:53] <toad_> d-ArkAngel: IF you are using one and only one estimator, then perhaps you can pre-sort the sub-buckets
[9:53] <toad_> newsbyte: seriously, don't do it, it will only hurt us
[9:53] * pupok (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[9:53] <d-ArkAngel> you can pre-sort them in sets, you never compare one nodes eDNF against it's eSucess.
[9:53] <i2p_iip> <newsbyte> well, I agreed with sonax that it must first be user-friendly
[9:54] <i2p_iip> <newsbyte> makes not much sense to make a /. about it, else
[9:54] <toad_> d-ArkAngel: even if a sub-bucket is totally inside an inter-center period, (not a bucket), the value still varies from minimum to maximum
[9:54] <i2p_iip> <newsbyte> at the minimum we need a HowTo
[9:54] <toad_> but i think i see what you're getting at
[9:55] <toad_> if you can pull it off, that'd be great, but please build a test framework for it and prove that it does exactly what the old stuff does experimentally
[9:55] <toad_> if it works, i'd consider it for the real routing table
[9:55] <d-ArkAngel> it's not a worst case improvement
[9:55] <d-ArkAngel> but the average performance of Routing should improve.
[9:56] <toad_> you'll have to keep quite a lot of data - worst and best estimate for each sub-bucket for each node
[9:56] <d-ArkAngel> and the worst case only increase by a constant not in n
[9:56] <toad_> but it shouldn't be enormous
[9:56] <d-ArkAngel> indeed
[9:56] <i2p_iip> <newsbyte> I2P looks quite interesting, btw
[9:56] <toad_> and some sort of sorted list of the buckets, which you can if you are clever re-sort every time you change something, just by moving one bucket up or down if necessary
[9:57] <d-ArkAngel> indeed.
[9:57] <toad_> how are you going to deal with the min/max issue? very often one sub-bucket won't be universally better than the next..
[9:57] <d-ArkAngel> they'll be the exceptions, and will need to be calculated.
[9:57] <toad_> i suppose you have to detect that, and do a calculation if needed?
[9:57] <toad_> cool
[9:58] <toad_> do you think that the exceptions will be rare?
[9:58] <toad_> okay, example
[9:58] <toad_> we have key 0x1234
[9:58] <toad_> this fits into sub-bucket 0x12
[9:58] <d-ArkAngel> they'll be made rare by the fact that the number of sub buckets will be greater than the main buckets
[9:58] <toad_> top node is C
[9:58] <toad_> this has a range of 150.6 to 180.3
[9:58] <toad_> second node is D, this has a range of 170.2 to 190.8
[9:59] <toad_> third node is X, this has a range of 216.5 to 278.5
[9:59] <toad_> so we don't need to consider X
[9:59] <toad_> but we do need to do a full calculation on C and D
[9:59] <d-ArkAngel> indeed.
[9:59] <toad_> we also need to keep the number of sub buckets high enough that getting more than one sector in a sub-bucket is rare
[10:00] <toad_> we need counters on how often this happens
[10:00] <d-ArkAngel> in the testbed yet.
[10:00] <toad_> HOWEVER
[10:00] <toad_> we don't always need to recalculate just because there are two sectors in a sub-bucket!
[10:01] <toad_> we just need to make sure there's an up to date min/max value
[10:01] <toad_> up to date and accurate for all key values covered by that sub-bucket
[10:01] <toad_> we only need to recalculate when there is significant overlap in the estimates for the top and the second (third? fourth?) nodes
[10:02] <toad_> okay, this is very cool.. please try it. but please make a test harness where stuff gets reported on both, and the results checked
[10:02] <d-ArkAngel> will do.
[10:03] <toad_> and thanks for reminding me that i'm just another cog in the machine... often the really cool stuff gets done by the volunteers, and for good reason
[10:03] <d-ArkAngel> lol
[10:04] <d-ArkAngel> it's just a fresh perspective that helps things like that.
[10:04] <toad_> yup
[10:04] <toad_> i look forward to some benchmarks
[10:04] <d-ArkAngel> things get biult up from small into large things, and it's not till someone who doesn't have the history looks at it that it's clear there's a different way of thinking about the same problem.
[10:05] <d-ArkAngel> it's where open source will always have the advantage over closed source.
[10:05] <d-ArkAngel> lots more fresh eyes
[10:05] <toad_> this should be staggeringly fast compared to what we have now - even if you have to recalculate the top 5 nodes on average
[10:05] <d-ArkAngel> indeed
[10:05] <i2p_iip> <newsbyte> toad, you're just an insignificant xian cog in the eternal machine of life! :-)
[10:06] <d-ArkAngel> but it all depends on how much the trade off is in memory usage.
[10:06] <toad_> yeah, and that depends on how sharp the specializations are
[10:06] <toad_> which in the long term could be a serious problem
[10:06] <toad_> but we'll see
[10:06] <d-ArkAngel> thining about that as nodes get to know one another the sim should speed up.. :-)
[10:06] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 104 (Connection reset by peer))
[10:06] <toad_> if you need N sub-buckets, then you have problems :)
[10:07] <d-ArkAngel> :-)
[10:07] <toad_> which is why we need to keep stats on how many buckets needed to be recalculated
[10:07] <toad_> on each routing
[10:07] <toad_> anyway, i've gotta go
[10:07] <toad_> ttyl
[10:08] * d-ArkAngel wonders what the update of freenet would be if the system required to hold a data structure in ram the size of the keyspace size * RT size
[10:08] <toad_> will be back in ~ 2-3 hours
[10:08] <toad_> keyspace size is >2^160
[10:08] <toad_> with doubles it's 2^53
[10:09] <d-ArkAngel> welll that's ok, we just put 5^162 bytes of system memory on the node's requirments spec.
[10:10] <d-ArkAngel> ;-)
[10:13] * kers (~kers@14.ppp132.rsd.worldonline.se) Quit (Read error: 110 (Connection timed out))
[10:13] <toad_> d-ArkAngel: you going to keep the current KeyspaceEstimators around as well, for reporting, and extend them to calculate the needed info?
[10:14] * guido^pe (~unknown@dsl-213-023-234-076.arcor-ip.net) has joined #freenet
[10:14] <toad_> d-ArkAngel: rewriting them is a PITA, partly due to wrapping issues, likely to run into serious bugs
[10:14] <toad_> and the structure you describe can't hold the actual buckets anyway
[10:15] <d-ArkAngel> I'll have to look at the code and see how it'll all fit together best anyway
[10:15] <d-ArkAngel> it'll probably keep the estimators.
[10:15] <toad_> we're talking Node (pointer), estimator max (double), estimator min (double), in each cell, and a grid of #nodes * #sub-buckets
[10:15] <toad_> okay good
[10:15] * pupok (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[10:16] <toad_> just trying to prevent you making mistakes that'll cost you many days of debugging
[10:16] <toad_> bbl
[10:16] <d-ArkAngel> and extend them, but I'll have to see what kind of calls we need to implement everything. it shouldn't be a problem
[10:16] <d-ArkAngel> :-)
[10:22] <toad_> d-ArkAngel: there's also a cpu tradeoff with cost of updating - the more sub-buckets, the more have to be updated when something changes
[10:23] <toad_> and calculating which ones to update will be somewhat tricky; for each estimator involved, you have to go out the width of the influence of the center value, and the edges moved, i.e. out a whole bucket in each direction... you get 3 of these ranges, and overlap them
[10:24] <toad_> oh sh*t
[10:24] <d-ArkAngel> what?
[10:24] * Hory (~Miranda@82.76.81.56) Quit (Read error: 104 (Connection reset by peer))
[10:24] <toad_> this won't work with queueing... hmm, well, it can probably be adapted to work with queueing
[10:25] <d-ArkAngel> why would it have a problem with queuing?
[10:25] <toad_> you know queueing? we make a big grid of requests against nodes, and send the request:node pair with the lowest estimate
[10:26] <toad_> we could just keep the min:max, most of the time there won't be much overlap among the queued requests.. when there is we recalculate
[10:26] <d-ArkAngel> ok, I'm clearly not following...
[10:26] * toad_ has to rewrite that code anyway
[10:26] <toad_> d-ArkAngel: we have 16 requests queued, and 10 nodes we could send them to
[10:27] <toad_> we need to choose which request to send to which node
[10:27] <toad_> so we make a big grid of all the estimates
[10:27] <toad_> and pick the lowest one
[10:27] <toad_> then either we send it, if the node is immediately available
[10:27] <toad_> or we wait
[10:27] <toad_> but either way we knock out the request and the node from the grid
[10:28] <toad_> and repeat until we run out of nodes or requests
[10:28] <toad_> that's basically the queueing algorithm
[10:28] <toad_> it means we need actual estimates for every node:request pair - but we COULD have min,max,actual, where actual is only calculated if we absolutely need it
[10:29] <toad_> min,max just came from the sub-bucket:node cell
[10:29] <toad_> it's not an issue for the simulation, but it is an issue IRL
[10:29] <toad_> anyway it looks like a solvable one, so no big deal
[10:29] <d-ArkAngel> ohhh, right I see, it's part of the load balanceing.
[10:29] <toad_> d-ArkAngel: kind of
[10:30] <toad_> d-ArkAngel: load balancing says we can send requests at a given rate
[10:30] <toad_> we need to take maximum advantage of this
[10:30] <toad_> hence the above algorithm
[10:30] <d-ArkAngel> I think I follow,
[10:30] <toad_> before queueing, load balancing had a heavy toll on routing accuracy
[10:31] <toad_> nowadays things are much better behaved
[10:31] <toad_> that's why requests have been working a lot better
[10:32] <toad_> ttyl
[10:32] <d-ArkAngel> so routing on live is based on load as well as routability... that's interesing, I wonder what implications that has on network development.
[10:32] * hobx_ (~chatzilla@ankh.math.chalmers.se) Quit (Remote closed the connection)
[10:32] <toad_> d-ArkAngel: yup
[10:32] <toad_> d-ArkAngel: the conflict between load balancing and routing has been a big problem for freenet for many years
[10:33] <toad_> if we don't take it into account on the sender, we have to reject queries on the receiver
[10:33] <toad_> which is even worse
[10:33] <toad_> the simulations suggest if routing works, it ought to eventually balance load
[10:33] <toad_> but that's in an ideal homogenous network where all nodes have equal abilities, and where there is no centralized introduction
[10:34] <d-ArkAngel> assuming we don't add any nodes and let the network have a 100 million requests per 400 nodes :-)
[10:34] <toad_> after we have proved that freenet scales, we should look into centralized introduction's impact
[10:34] <toad_> :)
[10:34] <toad_> :)
[10:35] <toad_> there's loads we need to look at, but it'll be much more feasible with your changes, assuming that the tradeoffs aren't a lot worse than they look (on memory, and on cpu usage updating the cells)
[10:35] <toad_> ttyl
[10:35] <d-ArkAngel> sure thing
[10:38] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[10:52] * kers (~kers@212.ppp132.rsd.worldonline.se) has joined #freenet
[11:02] <i2p_iip> <Sonax> "CANNOT RENAME NODE FILE node-temp TO node" is that somthing for me to vorry about (latest stable, sun java 1.5 lots of free space etc.)
[11:13] <michaelkuijn> Permissions
[11:20] * Gunnar (~gunnar@34.80-203-209.nextgentel.com) has joined #freenet
[11:21] * Hory (~Miranda@82.76.81.56) Quit ("Visit http://CyberLore.net for recommendations on the best games, websites and freeware. Don't settle for anything else!")
[11:22] * michaelkuijn (~michaelku@fia41-111.dsl.hccnet.nl) Quit ()
[11:52] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 60 (Operation timed out))
[12:15] * cedric_ (~cedric@d213-103-119-228.cust.tele2.fr) has joined #freenet
[12:16] <cedric_> does someone have a datastore on a fat32 partition on linux?
[12:16] <d-ArkAngel> do I even want to know why it's on a fat32 partition?
[12:18] <cedric_> to use the same datastore on linux and windows
[12:18] <d-ArkAngel> and that worsk does it?
[12:18] <cedric_> avoid duplicates
[12:23] * Ash-Fox (Hal-9000@abj127.neoplus.adsl.tpnet.pl) has joined #FreeNET
[12:24] <d-ArkAngel> so what problems is it giving you?
[12:24] * bllx (~rog@host217-44-61-160.range217-44.btcentralplus.com) Quit ("Leaving")
[12:32] * Ash-Fox (Hal-9000@abj127.neoplus.adsl.tpnet.pl) Quit (Read error: 54 (Connection reset by peer))
[12:33] <cedric_> apparently no problem yoyo seems to load corectly
[12:34] <cedric_> frost retreving message coretly
[12:34] <cedric_> fuqid dosent look good on wine
[12:34] <cedric_> hungover?
[12:36] * Ash-Fox (Hal-9000@abj127.neoplus.adsl.tpnet.pl) has joined #FreeNET
[12:39] * Ash-Fox (Hal-9000@abj127.neoplus.adsl.tpnet.pl) Quit (Client Quit)
[12:41] <plixed_> cedric_: i hope you system never crashes / has power outtages while freenet is running either in windows or linux, fat32 has no journaling
[12:43] <toad_> hi
[12:44] <toad_> i think it has a fsck now though doesn't it?
[12:44] <toad_> on linux?
[12:45] <toad_> d-ArkAngel: how are you going to calculate the range in each sub-bucket?
[12:45] <toad_> example:
[12:45] <toad_> we are simulating, so our formula is (1-pDNF)*tSuccess+tFailure/(1-pDNF)
[12:47] <toad_> if we assume that the sub-bucket is entirely contained within a single inter-center period, so each of the 3 KeyspaceEstimators is a straight line
[12:47] <cedric_> d-ArkAngel: the problem is about having a journaled filesystem with acess from both linux and windows
[12:47] <toad_> EVEN THEN, the resulting estimator curve is not necessarily a straight line, which means you have to find the minimum point via differentiation
[12:47] <toad_> granted that's relatively easy
[12:47] <toad_> but you have to evaluate 3 points to do it probably
[12:48] <toad_> and if there is more than one inter-center period within the sub-bucket - e.g. if there's a turning point in both the pDNF and tFailure estimators - then things get even messier
[12:48] <toad_> it's quite feasible but it's going to be tricky and probably quite expensive
[12:48] * Ash-Fox (Hal-9000@abj127.neoplus.adsl.tpnet.pl) has joined #FreeNET
[12:52] * jay (jcl@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[12:52] <toad_> hi jay
[12:52] <jay> hey toad
[12:52] <toad_> grrr
[12:52] <toad_> looks like running a node kills simulation speed, and running a simulation kills the node... even if I nice the simulation way below the node...
[12:53] <toad_> probably just a thread/process handling issue - if I ran with NPTL it would be better perhaps - but if I ran with NPTL, the node would crash
[12:58] <jay> if your pc was a supercomputer you wouldn't have this problem
[13:04] <toad_> jay: heh
[13:04] <toad_> perhaps
[13:04] <toad_> if it were i'd be running dozens of simulations
[13:05] <toad_> each of which would be parallel via d-arkangel's code
[13:05] <toad_> bbiab
[13:05] <jay> sounds nice
[13:05] <toad_> bbiab
[13:07] <toad_> back
[13:07] <jay> that was quick
[13:07] <i2p_iip> <gott> toad, you are beautiful
[13:11] * tav|offline is now known as tav
[13:11] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) has joined #freenet
[13:20] <d-ArkAngel> I'm back toad_ just about to head home from work tho
[13:21] <d-ArkAngel> I'll be back online in an hour or less.
[13:21] <d-ArkAngel> catch you shortly
[13:21] <d-ArkAngel> robert@scabserver.com
[13:21] * d-ArkAngel (~Rob@213-131-104-86.onyx.net) has left #freenet
[13:21] <jay> oops.. chan is logged
[13:22] <jay> spam proof email addy's
[13:23] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[13:27] <jay> thelema|away: lo?
[13:27] <jay> ygh metadata handling
[13:27] <jay> err URI metastring handling really
[13:32] <KenMan> toad_: sorry for the HTML mail. Mozilla seems to have changed - I can't seem to set nonHTML as default :(
[13:34] * cedric_ (~cedric@d213-103-119-228.cust.tele2.fr) Quit (Read error: 238 (Connection timed out))
[13:37] * Sugadude (fuckyou@dsl027-178-156.sfo1.dsl.speakeasy.net) Quit (Remote closed the connection)
[13:37] * Sugadude (~Sugadude@erika.csbnet.se) has joined #freenet
[13:40] <toad_> KenMan: so use thunderbird?
[14:09] * guido^pe (~unknown@dsl-213-023-234-076.arcor-ip.net) Quit (Read error: 113 (No route to host))
[14:09] * guido^pe (~unknown@dsl-082-083-229-126.arcor-ip.net) has joined #freenet
[14:16] <jay> this is a valid ksk, yes? -> freenet:KSK@testksk,htl=1/filename//
[14:17] <toad_> bbl, back many hours later (aikido, travel to and from aikido, dinner)
[14:17] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) has joined #freenet
[14:17] <toad_> jay: hmmm
[14:17] <d-ArkAngel> evening all
[14:17] <toad_> ask me later, or read the source...
[14:17] <toad_> d-ArkAngel: hi
[14:17] <d-ArkAngel> I return from lands afar...
[14:17] <toad_> d-ArkAngel: gotta go, sadly
[14:17] <d-ArkAngel> well work anyways
[14:17] <jay> toad_: no prob
[14:17] <toad_> d-ArkAngel: i will read the logs
[14:17] * moskawa23 (~Miranda@dsl-082-082-238-225.arcor-ip.net) Quit (Read error: 60 (Operation timed out))
[14:18] * spaetz (~spaetz@217-162-197-136.dclient.hispeed.ch) Quit (Read error: 60 (Operation timed out))
[14:18] <toad_> also email me if you have anything important
[14:18] <d-ArkAngel> toad: no fear, I'm off to get some food in 15 min or so anyway
[14:18] <toad_> what time is it for you?
[14:18] <d-ArkAngel> same time as it is for you no doubt
[14:18] <d-ArkAngel> 6:18pm
[14:18] <toad_> ah ok
[14:18] <toad_> well i might see you later or i might see you tomorrow
[14:18] <toad_> either way, seeya
[14:18] <d-ArkAngel> indeed. have a good evening if I don't see you
[14:19] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[14:20] <jay> toad_: lates bro
[14:27] * jay (jcl@ool-18bf6dac.dyn.optonline.net) Quit ("bbl")
[14:27] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 238 (Connection timed out))
[14:33] <i2p_iip> <newsbyte> well, toad, I'm most impressed with I2P
[14:33] <i2p_iip> <newsbyte> very
[14:34] <i2p_iip> <newsbyte> It works considerably better then freenet
[14:34] <i2p_iip> <newsbyte> I fear a bit that, unless you can get freenet to work better soon, ppl will start moving to i2p in droves :-/
[14:35] <d-ArkAngel> nice to see you're being as constructive as ever newsbyte
[14:35] <greycat> freenet also works incredibly well with only 20 nodes. how big is I2P these days?
[14:37] <i2p_iip> <newsbyte> good question
[14:38] <i2p_iip> <newsbyte> constructive? I'm saying like it is, angel, that should suffice
[14:39] <d-ArkAngel> so would you like toad to make it work before or after he's spent the two days you wanted him to spend making the simulator user friendly?
[14:40] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) Quit ()
[14:41] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[14:41] <Ribs> I personally think Freenet has come a long way recently
[14:42] <i2p_iip> <newsbyte> not from an end-user perspective
[14:42] <i2p_iip> <newsbyte> 71, they say
[14:43] <i2p_iip> <newsbyte> well, we could use a new influx, so I'd say before, angel
[14:43] <i2p_iip> <newsbyte> but then again, freenet that sucks isn't of the last month, is it? More like the last year(s).
[14:44] <greycat> freenet has sucked and rocked, alternately, for many years.
[14:44] <i2p_iip> <newsbyte> I2P seems to make substantial improvements (that ARE noticable)
[14:45] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) has joined #freenet
[14:47] <i2p_iip> <newsbyte> jrandomn doesn't seem to see the two projects in competition, and maybe he's right. But (human) nature just may disagree. I'm not sure there is enough to go around for two major anon projects
[14:47] <i2p_iip> <newsbyte> though for the moment, freenet is still the higher profiled
[14:49] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) Quit (Client Quit)
[14:49] <greycat> I wish jrandom all the success he can manage. Even if it kills freenet, we'd get something better. If it doesn't (which I consider far more likely) then the two projects will help each other.
[14:58] * jay (~jay@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[14:59] <i2p_iip> <newsbyte> maybe
[14:59] <i2p_iip> <newsbyte> or they'll kill eachother ;-)
[15:00] <i2p_iip> <newsbyte> but the real test is the scalability of it
[15:00] <i2p_iip> <newsbyte> everything falls or not, with that
[15:17] <i2p_iip> <newsbyte> toad, you there? I had a splendid idea, just now!
[15:20] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[15:23] <d-ArkAngel> toad is out for the evening he said, he said he might call in later on.
[15:25] * nextgens (~nextgens@d80-170-239-27.cust.tele2.fr) Quit (Read error: 60 (Operation timed out))
[15:27] * hirvox (~hirvox@cs181027153.pp.htv.fi) has joined #freenet
[15:34] * JustMe_ (JustMe_@cs68206255-46.houston.rr.com) Quit (Read error: 110 (Connection timed out))
[15:40] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[16:03] * Hory` (~Miranda@82.76.81.56) has joined #FreeNet
[16:11] * nextgens (~nextgens@d80-170-176-253.cust.tele2.fr) has joined #freenet
[16:12] * kers (~kers@212.ppp132.rsd.worldonline.se) Quit (Read error: 110 (Connection timed out))
[16:24] * Hory (~Miranda@82.76.81.56) Quit (Connection timed out)
[16:36] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) has joined #freenet
[16:38] * jay (~jay@ool-18bf6dac.dyn.optonline.net) Quit ("Leaving")
[17:24] <Redb3ard> wyeh
[17:28] * nextgens (~nextgens@d80-170-176-253.cust.tele2.fr) has left #freenet
[17:38] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) Quit ("Trillian (http://www.ceruleanstudios.com)")
[17:39] * greycat (~wooledg@192.35.79.70) Quit ("Lick Bush in '04")
[17:39] * nextgens (~nextgens@d80-170-176-253.cust.tele2.fr) has joined #freenet
[17:39] * nextgens (~nextgens@d80-170-176-253.cust.tele2.fr) has left #freenet
[17:44] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) has joined #freenet
[17:46] * kers (~kers@188.ppp141.rsd.worldonline.se) has joined #freenet
[17:47] * tav is now known as tav|offline
[17:48] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) Quit ()
[17:58] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) has joined #freenet
[18:06] * hirvox (~hirvox@cs181027153.pp.htv.fi) Quit (Connection timed out)
[18:16] * lolo-laptop (~lostlogic@68.251.84.226) Quit ("Client exiting")
[18:25] * Hory` (~Miranda@82.76.81.56) Quit ("Visit http://CyberLore.net for recommendations on the best games, websites and freeware. Don't settle for anything else!")
[18:29] * robilad (~topic@mpiat2313.ag2.mpi-sb.mpg.de) Quit (Remote closed the connection)
[18:34] * FallingBuzzard (~srademach@207.152.112.129) Quit (Read error: 110 (Connection timed out))
[18:40] * jokern (~virus@213.184.217.12) Quit ()
[19:43] <toad_> <d-ArkAngel> so would you like toad to make it work before or after he's spent the two days you wanted him to spend making the simulator user friendly? - ROFL
[19:43] <toad_> <Ribs> I personally think Freenet has come a long way recently
[19:43] <toad_> <i2p_iip> <newsbyte> not from an end-user perspective
[19:43] <toad_> i think so too as far as requests go
[19:43] <d-ArkAngel> all things come in time
[19:43] <toad_> <i2p_iip> <newsbyte> toad, you there? I had a splendid idea, just now! - uh oh...
[19:45] <d-ArkAngel> well I've spent the evening scribling on paper and I think I understand how the estimators work now :-) just need to start looking at how to code up what we discussed earlier :-)
[19:46] * plixed (~plixed@pD9E25484.dip.t-dialin.net) has joined #freenet
[19:49] * hirvox (~hirvox@cs181027153.pp.htv.fi) has joined #freenet
[19:50] <toad_> d-ArkAngel: we can talk tomorrow
[19:50] <toad_> bbl zzz (builders coming, must get up early)
[19:53] <d-ArkAngel> work tomorrow must get up early (damn day job)
[19:53] <d-ArkAngel> did you get my message about me wanting to turn the server off tomorrow night sometime?
[19:56] <d-ArkAngel> oh clearly you did since you replied to it :-) sorry my bad :-)
[19:56] <d-ArkAngel> night all
[20:01] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[20:03] * sanity (~ian@adsl-68-123-90-162.dsl.lsan03.pacbell.net) Quit ()
[20:04] * plixed_ (~plixed@p50822FDD.dip.t-dialin.net) Quit (Read error: 110 (Connection timed out))
[20:20] * greycat (rfc1413@wooledge.org) has joined #freenet
[20:29] * kers (~kers@188.ppp141.rsd.worldonline.se) Quit (Read error: 242 (No route to host))
[20:40] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[21:03] * sanity (~ian@66.153.58.52) has joined #freenet
[21:14] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[21:15] * vsalento_ (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[21:22] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 238 (Connection timed out))
[21:23] * guido^pe (~unknown@dsl-082-083-229-126.arcor-ip.net) Quit (Remote closed the connection)
[21:39] * Redb3ard (~oylerj@c-24-125-12-101.va.client2.attbi.com) Quit ("Hey! Where'd my controlling terminal go?")
[21:40] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[21:48] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[21:59] <i2p_iip> <Jonva> I have a juicy NPE that toad may like, i'll post it in a second. If anyone has a problem with this, speak now.
[22:01] * vsalento_ (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[22:05] <i2p_iip> <Jonva> (freenet.interfaces.LocalNIOInterface, YThread-289, ERROR): Unhandled throwable while handling connection
[22:05] <i2p_iip> <Jonva> java.lang.NullPointerException
[22:05] <i2p_iip> <Jonva> at freenet.client.inFlightRequestTrackingAutoRequester$EventSequence.consume(inFlightRequestTrackingAutoRequester.java:172)
[22:05] <i2p_iip> <Jonva> at freenet.client.inFlightRequestTrackingAutoRequester$EventSequence.access$400(inFlightRequestTrackingAutoRequester.java:151)
[22:05] <i2p_iip> <Jonva> at freenet.client.inFlightRequestTrackingAutoRequester.toHtml(inFlightRequestTrackingAutoRequester.java:127)
[22:05] <i2p_iip> <Jonva> at freenet.node.http.infolets.DownloadsInfolet.toHtml(DownloadsInfolet.java:58)
[22:05] <i2p_iip> <Jonva> at freenet.node.http.Infolet.toHtml(Infolet.java:60)
[22:05] <i2p_iip> <Jonva> at freenet.support.servlet.HtmlTemplate$VarFragment.toHtml(HtmlTemplate.java:295)
[22:05] <i2p_iip> <Jonva> at freenet.support.servlet.HtmlTemplate.toHtml(HtmlTemplate.java:224)
[22:05] <i2p_iip> <Jonva> at freenet.node.http.NodeInfoServlet.doGet(NodeInfoServlet.java:199)
[22:05] <i2p_iip> <Jonva> at javax.servlet.http.HttpServlet.service(Unknown Source)
[22:05] <i2p_iip> <Jonva> at javax.servlet.http.HttpServlet.service(Unknown Source)
[22:05] <i2p_iip> <Jonva> at javax.servlet.http.HttpServlet.service(Unknown Source)
[22:05] <i2p_iip> <Jonva> at freenet.interfaces.servlet.ServletContainer.handle(ServletContainer.java:82)
[22:05] <i2p_iip> <Jonva> at freenet.interfaces.LocalNIOInterface$ConnectionShell.run(LocalNIOInterface.java:268)
[22:05] <i2p_iip> <Jonva> at freenet.thread.YThreadFactory$YThread.run(YThreadFactory.java:285)
[22:06] * hobbs (~andrew@hobbs.user) has joined #freenet
[22:08] <i2p_iip> <Jonva> That NPE came when i tried to access the current downloads page, it wasn't displaying anything, just a big blue box. So i clicked on Switch to simple mode, and clicked it again and boom, i got about 15 of them, all the same.
[22:40] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) has joined #freenet
[23:04] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) Quit (Read error: 104 (Connection reset by peer))
[23:23] * Zorix (Brandon@fl-69-34-0-219.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[23:39] * greycat (rfc1413@wooledge.org) Quit ("Lick Bush in '04")
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.
