Timestamps are in GMT/BST.
[0:00] * Redb3ard (~chatzilla@host-64-83-0-201.cavtel.com) Quit ("ChatZilla 0.9.61 [Mozilla rv:1.7.3/20040910]")
[0:06] <newblistic> suprnova.org is down again so i cant download movies :(
[0:06] <Bonkers> looks like it's down forever
[0:07] <newblistic> yup
[0:07] <newblistic> The request couldn't even make it off of your node. Try again, perhaps with the GPL to help your node learn about others. The publicly available seed nodes have been very busy lately. If possible try to get a friend to give you a reference to their node instead.
[0:09] <salahx> suprnova is down permentantly
[0:09] <salahx> Word on the street is some of the mare movign to Frost
[0:40] <TheSeeker> yay, another 'ign'er!
[0:41] * TheSeeker types ign instead of ing more than half the time when typing that !@#$ign suffix :P
[0:49] * newblistic (~dragonmor@cs24174111-205.houston.rr.com) Quit (Read error: 60 (Operation timed out))
[0:49] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 60 (Operation timed out))
[0:58] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) Quit (Read error: 110 (Connection timed out))
[1:30] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[1:57] * TheSeeker looks pleased with his "do two node's spec ranges overlap?" function code
[2:10] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) has joined #freenet
[2:27] <TheSeeker> yay, and it seems to work even :D
[2:30] * mazzanet (~mazzanet@mazzanet.user) Quit (Read error: 104 (Connection reset by peer))
[2:30] <TheSeeker> using randomly generated specLeft and specRight until I get my 'setSpecFromCenter(nodenum, distance)' function written... I get output that looks like:
[2:30] <TheSeeker> node a:
[2:30] <TheSeeker> 7367485A1878025A70F9BA4C27C3CB400F68E88CA4A9E1
[2:30] <TheSeeker> F99F69F7E2019E0549FEC7CCE9DFC76022FE7B524549C1
[2:30] <TheSeeker> nodeb:
[2:30] <TheSeeker> 3A20F7B464B63341052EBBF629D723F244249383BC29C7
[2:30] <TheSeeker> 5950E0F19D7ED798C5AED2FC2E2C366C0D3E333A09295F
[2:30] <TheSeeker> The node specialization ranges do not overlap.
[2:33] <TheSeeker> code to account for range crossing the keyspace boundaries seems to work fine too
[2:33] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 104 (Connection reset by peer))
[2:33] <TheSeeker> node a:
[2:33] <TheSeeker> 24FD3FD2A689CDAD5BEC8CED2185F1F5ED594FEB99214F
[2:33] <TheSeeker> ACB04BB3ACAFE44A8231A338586F0395E6DB8199718B05
[2:33] <TheSeeker> node b:
[2:33] <TheSeeker> A091A2FEFAB86777E49F0434D04F1742353E133EEB68E5
[2:33] <TheSeeker> 10AE72680F087C6D3DA69353FEBB3ABB3348DBE0810476
[2:33] <TheSeeker> The node specialization ranges overlap.
[2:36] * TheSeeker starts working on the functions to set the spec range left and right based on the center, and test for if a key lies within a node's range...
[2:40] * mazzanet (~mazzanet@mazzanet.user) has joined #freenet
[2:45] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) Quit (Read error: 110 (Connection timed out))
[2:52] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[3:00] * spaetz (~spaetz@80-218-144-36.dclient.hispeed.ch) has joined #freenet
[3:05] * sdogi (~java@84-50-17-209-dsl.kjj.estpak.ee) has joined #freenet
[3:05] * jdot (~jdot@ip70-185-16-76.ma.dl.cox.net) has joined #freenet
[3:06] * vsalento_ (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[3:07] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 60 (Operation timed out))
[3:17] * sdogi_ (~java@84-50-17-13-dsl.kjj.estpak.ee) Quit (Read error: 110 (Connection timed out))
[3:27] <TheSeeker> what's a reasonable minimum number of freenet nodes to use in a sim? anyone know about how many nodes there are running stable right now?
[3:29] <sanity> hard to goess, probably 500-10,000
[3:29] <sanity> s/goess/guess
[3:29] <TheSeeker> as many as 10k you think?
[3:29] <sanity> its possible
[3:29] <Bonkers> heh, that's a rough simulation
[3:30] <sanity> of course, freenet needs to be able to handle far larger networks than that (over 1M nodes)
[3:32] <TheSeeker> I'm only looking to simulate a self-organization of peers for the moment, to see what the network would look like if nodes knew absolutely what eachother's specialization points were, as well as the highest and lowest area of the keyspace they'll cache keys in (specCenter, specLeft, specRight [though left can be > Right die to wraping])
[3:33] <sanity> well, then 100-500 is probably a good place to start
[3:33] <sanity> TheSeeker: are you artificially enforcing specialisation?
[3:35] <TheSeeker> yes. on node creation it's given a random key to specialize around, as well as a range of keys the node will cache based on upload bandwidth avaliable. I'm going to see if by knowing about only one other node on the network if it will be able to organize with simple rules for connections
[3:36] <sanity> hmmm, i am not terribly fond of forced specialisation. the lru algorithm should specialise without being forced to :-/
[3:36] <TheSeeker> I don't see how.
[3:36] <sanity> TheSeeker: its a positive feedback loop
[3:37] <sanity> as other nodes perceive you to have a specialisation, they send more requests in that area, which makes you specialise even more in that area, which becomes a self-fulfilling effect
[3:37] <TheSeeker> the lru key is as likely to be any given key in the keyspace. if everyone specialized based on the popularity of keys, everyone would have identical specializations, and probably be perfectly flat.
[3:38] <TheSeeker> wouldn't it be easier just to tell other nodes "I have a specialization' so they don't have to spend thousands of requests trying to find it? (if they even keep asking you for kewys after you've failed so many)
[3:39] <TheSeeker> -w
[3:39] <sanity> the network-wide popularity of keys will be flat, but the popularity of keys received by a given node* won't be
[3:40] <sanity> how should a node determine what its specialisation should be and how broad it should be such that it doesn't overlap with other nodes? LRU determines this in a very natural emergent way
[3:40] <TheSeeker> it will if no node is ever percieved to specialize, which it won't, given that all areas of the keyspace are equally popular,.
[3:40] <sanity> TheSeeker: well, if that were true then we wouldn't see specialisation, but we do
[3:40] <TheSeeker> rarely to the point of almost never, and usually due to a bug :P
[3:41] <sanity> TheSeeker: the reason is that all areas of the keyspace are equally popular for the network as a whole, but not for a particular node
[3:42] <sanity> its like, if you look at the postal system as a whole, the number of letters sent will be reasonably evenly distributed (lets pretend), yet the letters received by a particular sorting office will be skewed towards addresses close to that sorting office.
[3:42] <sanity> the same is true of freenet nodes
[3:42] <TheSeeker> in a perfect world. I've never seen evidence of this working.
[3:43] <sanity> i have, it happened all the time pre-NGR, reliably. it is increasingly happening with NGR now that we have fixed a number of disruptive bugs.
[3:45] <sanity> if you can play mng animations, take a look at this animation of a node's datastore specialising: http://freenetproject.org/image/histanim.mng
[3:46] <TheSeeker> how old is that animation? I haven't seen the barcode in a LONG time.
[3:48] <sanity> it is pre-ngr - but the exact same logic applies
[3:48] <TheSeeker> I say: routing doesn't work... I guess: routing doesn't work because we guess too damn much what we could just as easily know with no less anonymity. I also stipulate: force a specialization of the datastore and routing, and the whole network will run a hellofalot smoother
[3:48] <sanity> kenman has some surprisingly similar animations somewhere using the NGR simulator
[3:49] <sanity> TheSeeker: so how do you explain the extremely successful results of the simulations Toad and KenMan have been doing?
[3:50] <TheSeeker> they don't take bandwidth into account.
[3:50] <sanity> and why would taking bandwidth into account cause NGR to stop working?
[3:51] <TheSeeker> I'm saying I would bet anything that simulating bandwidth limits in the simulations would show rather dismal results in comparison.
[3:52] <TheSeeker> sure, routing MAY find the key you're looking for eventually, but what are the odds that the data will make it back before you time out?
[3:52] <sanity> TheSeeker: well, rather than assuming your hypothesis is true and developing code on that basis, why don't you test your hypothesis?
[3:53] <TheSeeker> that's why I'm currently tryign to write my own sim :P
[3:53] <sanity> TheSeeker: but from the sound of it it will implement your forced specialisation from scratch - will it also implement simple LRU to act as a baseline?
[3:53] <sanity> and will it simulate bandwidth?
[3:54] <sanity> and wouldn't it be better to work with Toad and KenMan on their sim rather than reinventing the wheel?
[3:54] <TheSeeker> no
[3:54] <TheSeeker> because I dont' know java
[3:54] <sanity> why not?
[3:54] <sanity> oh, ok
[3:54] <sanity> well, will you implement the current algorithm for comparative purposes?
[3:55] <TheSeeker> if I can figure it out, I suppose.
[3:55] <sanity> if you don't, your simulation won't prove anything
[3:55] <sanity> basic experimental method - you need a control
[3:56] <sanity> bbiab
[3:56] <TheSeeker> what I'd like to show, is that a much simpler set of rules can be given to achive superior results, and be much more friendly to new nodes.
[3:59] <TheSeeker> anyway... backing way up... I was wondering about the number of nodes on the network to determine a good formula for setting the specialization bredth.
[4:02] <TheSeeker> given that I'd like the specialization to be proportional in some way to upload bandwidth, as narrower specialization should lead naturally to fewer routing requests (something I'd like to test)
[4:02] <sanity> TheSeeker: simpler how? it is hard to get simpler than "route to the best node for this key as per my prior experience"
[4:03] <sanity> ah, and here comes the alchemy :-)
[4:03] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[4:03] <TheSeeker> right, because "my prior experience" is ALL alchemy :P
[4:03] <sanity> your algorithm won't be simpler because you need to choose lots of constants like "specialisation breadth"
[4:04] <TheSeeker> aren't constants a lot easier than rolling average samples over time?
[4:04] <sanity> not if they change
[4:05] <TheSeeker> huh? ... constants... as in, not changing?
[4:05] <sanity> but that is the problem, the reason we use rolling averages is that they *do* change
[4:06] <sanity> so, if your definition of easy is "quick to code but won't actually work in practice" then yes, constants will certainly be easier
[4:06] <sanity> in the same way that building comfortable seats is easier if you assume that people are spheres
[4:06] <TheSeeker> not if you have a predefined specialization point, and connect primarily to nodes near your specialization point... everyone always knows you're good at routing keys near your specialization. you always have been, you always will be.
[4:06] <sanity> define "near"
[4:07] <sanity> and define "good"
[4:07] <TheSeeker> nodes with keyspace ranges that overlap your own.
[4:07] <sanity> how big is a range, and where should it start and end?
[4:08] <TheSeeker> that's where I was backing up to. that depends a bit on mow much redundancy is desired and how many nodes there are to start...
[4:08] <sanity> yes, but you will find that as you try to answer these questions you will be forced to make inaccurate assumptions, or get into a world of complexity that will make the current algorithm look like childsplay
[4:08] <TheSeeker> I suppose that bredth at node start could be based upon known other nodes, then narrow (though always around the predefined spec point) as it learns about more nodes on the network...
[4:10] <sanity> trust me, you will spend a few weeks thinking about this and all of these various little things, and then you will have a sudden revelation that the way were are already doing it actually achieves the things you are trying to achieve.
[4:10] <TheSeeker> innacurate assumptions... I like that :D kinda like assuming infinate bandwidth in simulations, then calling the sims a success ;)
[4:10] * vsalento_ (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[4:10] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) has joined #freenet
[4:10] <sanity> some simplifying assumptions are acceptable, some are not. the kind you will need to make won't be
[4:11] <sanity> but be my guest, waste a few weeks finding out the hard way what we have already discovered :-)
[4:11] <TheSeeker> What I'm trying to achive, is a network of more than 1k nodes self organizing into something userful in less than an hour, from scratch with each node only knowing one other node at the start.
[4:12] <sanity> TheSeeker: you should look at dijjer's routing algorithm
[4:13] <sanity> TheSeeker: it is probably closer to what you have in mind than Freenet. of course, it isn't anonymous and it is vulnerable to attack in ways that wouldn't be tolerable for Freenet
[4:13] <TheSeeker> true or false: a node will specialize faster if several nodes perceive the node as having the same specialization vs several nodes thinking it has a different one.
[4:13] <sanity> TheSeeker: see http://dijjer.org/index.php?page=arch_basic
[4:15] <sanity> TheSeeker: true. but this doesn't mean that its specialisation must be forced
[4:16] <TheSeeker> explain to me any negative of a forced specialization (where all nodes can always agree on who is closer to what area of the keyspace)
[4:18] <sanity> makes an attack much easier (chose to specialise in an area of keyspace containing data you wish to censor), precludes multiple specialisations, ignores issues such as network latency, the list goes on
[4:19] <TheSeeker> umm
[4:20] <TheSeeker> you could MAYBE choose to specialize in a single key that you want to censor, but nodes will just route around you...
[4:20] <sanity> not if you have told them that you specialise in that area
[4:20] <TheSeeker> I percieve multiple specializations as a Bad Thing(tm) that's probably been debated to death already though.
[4:20] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[4:20] <sanity> with NGR, you can't just tell nodes what your specialisation is, you have to prove it
[4:21] <TheSeeker> just because you connect to someone and they tell you "I'm specialized in X" doesn't mean you don't make sure that they actually behave normally.
[4:21] <sanity> with your approach i can just tell everyone what my specialisation is and they will believe me
[4:21] <sanity> define "normal"
[4:22] <sanity> i mean, if they DNF on a specific key that they are trying to censor, how will you know that this is abnormal?
[4:24] <TheSeeker> because it's likely that the key will be re-requested multiple times, and that node isn't going to be the one chosen every time, because it's likely to be busy some of the time (already transferring data) it's also not goign to be able to keep an area isolated enough that requests only come to it.
[4:25] <sanity> so, you set up 5 nodes in the same area of specialisation and ensure they have plenty of bandwidth. this is certainly not beyond the means of even the least resourceful censor
[4:27] <TheSeeker> how would that be different from NGR?
[4:27] <sanity> because with NGR other nodes must decide what your specialisation is, you don't just tell them
[4:27] <sanity> it raises the bar significantly
[4:28] <TheSeeker> censor sets up an ubernode. it's the best node at routing all requested keys except one, and because it never RNFs, you always want to route to it.
[4:29] <sanity> and you are saying that setting up a node so good that it is better than the rest of the freenet network combined is easier than setting up a few ordinary nodes which just tell other nodes that they are specialised and are believed unquestioningly?
[4:30] * vsalento_ (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[4:30] <TheSeeker> I never said "unquestioningly" but any node that's not tampered with WILL behave propperly.
[4:32] <TheSeeker> assume everyone is the enemy and nothing can possibly get done in a timeframe that is acceptable to end users.
[4:36] <TheSeeker> back to your question of setting up a few nodes to try and suck up the routing to a particular key... since I'm assuming that everyone is going to be primarily connected to nodes of similar specialization, it doesn't really matter which node I route to on a given hop as long as it's in the general area of they key I'm looking for.
[4:37] <TheSeeker> This means that if I'm already downloading from a "best match" peer, I can safely route to someone else nearby. I'm thinking that per-node failure tables would eliminate the threat of the attack you suggested...
[4:38] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[4:38] <sanity> your simple algorithm is starting to sound more complicated
[4:39] <TheSeeker> my "simple algorithm" was to be for getting the network to self-organize, not to route all keys devoid of any potential attacks.
[4:40] <sanity> if that is all you care about then just use the same algorithm as Dijjer
[4:41] <TheSeeker> I can't read Python any more than I can java. :P
[4:41] <TheSeeker> ah, looked at the url wrong
[4:41] <TheSeeker> it is java :P
[4:42] <sanity> the description of the algorithm is in English
[4:43] <sanity> and its so simple that you could probably implement it in whatever language you like in a few hours
[4:52] <TheSeeker> I can see that you have an interest in me not trying to learn things for myself, perhaps because you've spent a lot of time wrking on the current implementation of freenet, or because you've done something similar yourself before and got frustrated... I don't know. What I do know, is that to your average end user, Freenet -doesn't work-.
[4:52] <TheSeeker> I see what appear to be glaring flaws in the design, but certainly don't expect anyone else to sit around and code in implemtation changes and push it live so I can easily be proven right/wrong. So I'm doing the only thing left open to me, which is try and write a sim for myself.
[4:53] <sanity> what are you talking about?
[4:53] <sanity> exactly when did i say that you shouldn't learn things for yourself?
[4:54] * pupok (~r00t@81-178-106-165.dsl.pipex.com) has joined #freenet
[4:54] <sanity> you are free to do whatever you want, just as i am free to point out what is wrong with your proposal
[4:55] <TheSeeker> Perhaps, but you come across (to me) as thinking that any proposal is doomed to failure, as the "right" way of doing things is already in production.
[4:56] <TheSeeker> [01:11.19] <sanity> trust me, you will spend a few weeks thinking about this and all of these various little things, and then you will have a sudden revelation that the way were are already doing it actually achieves the things you are trying to achieve.
[4:57] <sanity> no, i think there is merit in your ideas - clearly a problem with NGR is that nodes may not learn the specialisation of other nodes quickly enough. but we have considered forced specialisation many times and discounted it for the reasons i have outlined. if you want to ignore our experience and find out for yourself then please do-so, i certainly won't try to stop you
[5:00] <TheSeeker> it's not just that nodes won't learn the specialization fast enough, it's that there's no way of knowing what any other node thinks about the specialization, or even what the node itself thinks it's good at.
[5:03] <TheSeeker> with NGR: do you trust a node to have actually sent your requests and that it's giving you legitimate DNFs?
[5:03] <sanity> not really
[5:04] <sanity> nodes which return an above-average number of DNFs are penalised
[5:05] <TheSeeker> in determining the success probablity of a key, you go by the 'perceived specialization' range, and pick the best range that's not backed off, do you not?
[5:06] <TheSeeker> (best = narrowest)
[5:06] <sanity> we go for the node we estimate will result in us getting the data in the smallest amount of time (including the possibility that it will DNF and we will need to re-request, and also the possibility that we will need to wait for it to not be backed off)
[5:07] <sanity> TheSeeker: this page offers a good overview of the mechanism: http://freenetproject.org/index.php?page=ngrouting
[5:11] <TheSeeker> hmmm... So how is a utopian "everyone's been running long enough to have perfect estimators with no connection churn" netowrk different than one that was organized from predefined specialization points ?
[5:12] <sanity> isn't it obvious?
[5:12] <sanity> anyway, it doesn't assume that estimators are perfect, they only need to be good enough
[5:12] <TheSeeker> define "good enough"
[5:13] <sanity> that the request gets closer to the data on each hop such that it finds the data within the htl
[5:15] <TheSeeker> I'd really like to see how freenet would perform if everyoen were simply limited to their upload speed in KB/s * 2 connections (instead of a lot of nodes with 100-200 open connections)
[5:16] <TheSeeker> we RNF because we overload, we overload because we overconnect to overcome other people RNFing ... negative feedback loop.
[6:28] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) Quit (Read error: 110 (Connection timed out))
[6:34] * d-ArkAngel (~Robert@213-131-104-86.onyx.net) has joined #freenet
[6:47] * pupok (~r00t@81-178-106-165.dsl.pipex.com) Quit (Read error: 110 (Connection timed out))
[6:56] * Hory` (~Miranda@82.76.81.56) has joined #FreeNet
[7:01] * Hory (~Miranda@82.76.81.56) Quit (Connection timed out)
[7:03] * pupok (~r00t@81-178-106-165.dsl.pipex.com) has joined #freenet
[7:12] * toad_ (toad@toad-with-underline.active.supporter.pdpc) has joined #freenet
[7:12] * ChanServ sets mode +o toad_
[7:16] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) has joined #freenet
[7:25] * Hory` (~Miranda@82.76.81.56) Quit (Connection timed out)
[7:33] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[8:40] * Elly (~Elly@ool-182c3b26.dyn.optonline.net) Quit (Read error: 104 (Connection reset by peer))
[8:40] * toad_ (toad@toad-with-underline.active.supporter.pdpc) has left #freenet
[8:54] * FallingBuzzard (~srademach@207.152.112.129) has joined #freenet
[8:58] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) Quit (Read error: 110 (Connection timed out))
[8:58] * vsalento_ (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[9:21] * toad_ (toad@toad-with-underline.active.supporter.pdpc) has joined #freenet
[9:21] * ChanServ sets mode +o toad_
[9:21] * greycat (~wooledg@192.35.79.70) has joined #freenet
[9:24] * guido^pe (~unknown@dsl-213-023-244-227.arcor-ip.net) has joined #freenet
[9:27] * d-ArkAngel (~Robert@213-131-104-86.onyx.net) Quit (Read error: 113 (No route to host))
[9:34] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) has joined #freenet
[9:39] * salahx (salahx@cc3-24.217.203.238.charter-stl.com) has left #freenet
[9:47] * spaetz (~spaetz@80-218-144-36.dclient.hispeed.ch) Quit ("Verlassend")
[9:49] * sdogi (~java@84-50-17-209-dsl.kjj.estpak.ee) Quit (Read error: 104 (Connection reset by peer))
[9:54] * sdogi (~java@84-50-19-192-dsl.kjj.estpak.ee) has joined #freenet
[9:56] * wern (~wern@h102n1fls20o845.bredband.comhem.se) has joined #freenet
[10:02] * sdogi_ (~java@84-50-19-194-dsl.kjj.estpak.ee) has joined #freenet
[10:14] * goatee (~goatee@ip216-239-89-99.vif.net) Quit (Read error: 110 (Connection timed out))
[10:14] * sdogi (~java@84-50-19-192-dsl.kjj.estpak.ee) Quit (Read error: 110 (Connection timed out))
[10:16] * guido^pe (~unknown@dsl-213-023-244-227.arcor-ip.net) Quit (Remote closed the connection)
[10:28] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[10:29] * orange_ (~orange@pm2-40.bahnhof.se) has joined #freenet
[10:29] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[10:33] * robilad (~topic@mpiat2305.ag2.mpi-sb.mpg.de) has joined #freenet
[10:35] * Nebukadneza (~daddel9@i3ED6E310.versanet.de) has left #freenet
[10:56] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[11:21] * mazzanet (~mazzanet@mazzanet.user) Quit (Read error: 54 (Connection reset by peer))
[11:22] * mazzanet (~mazzanet@mazzanet.user) has joined #freenet
[11:29] * sanity (~ian@81-178-106-165.dsl.pipex.com) Quit ()
[11:29] <greycat> did everyone see the big ol' freenet/kiddieporn thread on the slashdot suprnova-goes-dark story?
[11:30] * FallingBuzzard (~srademach@207.152.112.129) Quit (Read error: 104 (Connection reset by peer))
[11:30] <greycat> it seems to start here: http://yro.slashdot.org/comments.pl?sid=133299&cid=11131489
[11:30] * FallingBuzzard (~srademach@207.152.112.129) has joined #freenet
[11:34] * d-ArkAngel (~Robert@213-131-104-86.onyx.net) has joined #freenet
[11:40] * vsalento (~vsalento@cs136227.pp.htv.fi) Quit (Read error: 110 (Connection timed out))
[11:40] * kevlAway (~kevloral@CZ1-RAS-8-u-0179.du.onolab.com) has joined #freenet
[11:42] * Nomadic_Prophet (UserID@24-168-39-156.si.rr.com) has joined #freenet
[11:42] <kevlAway> g'afternoon all
[11:42] * kevlAway is now known as kevloral
[11:43] * kmkezlr_ (~kmkezlr@195.25.40.253) has joined #freenet
[11:50] <kmkezlr_> hi hello gouteun tag
[11:50] <kmkezlr_> bonjour
[11:51] <i2p_iip> <Sonax> greycat: So, freenet is evil... What else is new? (besides, some of the replies to the thread seemed to be fair enough).
[11:51] <i2p_iip> <Sonax> Hi kmkezlr_
[11:51] <greycat> Just figured I'd point it out.
[11:51] <greycat> Of course, by the time *I* see something on slashdot, it's usually been there for 16-24 hours, so...
[11:53] <i2p_iip> <Sonax> greycat: What i find interesting is if freenet can pick up on the bt users once they start looking for new ways of p2p'ing. There is talk on frost of the potential, but so far noone seems to be doing anyting for it (preparing bt-tracker sites/boards etc.).
[11:54] <i2p_iip> <Sonax> Just talking about the potential users will do very little...
[11:54] <i2p_iip> <Sonax> said sonax, and went back to ideling.
[11:55] * greycat doesn't use Frost, so isn't in a good position to comment on how best to use Frost :)
[11:55] <d-ArkAngel> maybe someone should host a site full of bt links on a freesite? that way they're safe from "attack" but people get the speed of BT for their downloads....
[11:55] <d-ArkAngel> just a thought
[11:56] * greycat also needs to learn all the BT specific terminology like "seeds"
[12:03] <TheSeeker> how can you possibly hope for anonymity using BT? all any agency must do is connect to a tacker and start a torrent to get a huge list of everyone currently downloading the material... and if BT sites are shutting down due to pressure from the MPAA...
[12:03] <d-ArkAngel> the current attacks against BT are against the centralised portion of it
[12:04] <d-ArkAngel> so if the centralised portions were moved to freenet they'd have a harder time doing anything
[12:04] <d-ArkAngel> and they'd have to start going after individual users of BT clients.
[12:05] <TheSeeker> nah, they'd just start using RIAA tactics and sue everyone with an IP.
[12:05] <d-ArkAngel> which would bring us a lot of nodes accessing small amounts of content, and provide a stepping stone for people into using freenet properly
[12:06] <d-ArkAngel> just a thought ;-)
[12:06] <d-ArkAngel> anyway I've gotta go home!!
[12:06] <d-ArkAngel> catch you laters
[12:17] * yonkeltron (~yonkeltro@pcp04665066pcs.wilog501.pa.comcast.net) Quit (Read error: 60 (Operation timed out))
[12:22] * yonkeltron (~yonkeltro@pcp04665066pcs.wilog501.pa.comcast.net) has joined #freenet
[12:33] * Hory (~Miranda@82.76.81.56) Quit (Read error: 104 (Connection reset by peer))
[12:36] * d-ArkAngel (~Robert@213-131-104-86.onyx.net) Quit (Read error: 110 (Connection timed out))
[12:43] * vsalento (~vsalento@cs136227.pp.htv.fi) has joined #freenet
[12:51] * moskau23 (~Miranda@dsl-213-023-252-161.arcor-ip.net) has joined #freenet
[13:21] * moskau23 (~Miranda@dsl-213-023-252-161.arcor-ip.net) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[13:22] * jdot (~jdot@ip70-185-16-76.ma.dl.cox.net) Quit ("Leaving")
[14:03] * toad_ (toad@toad-with-underline.active.supporter.pdpc) Quit (Remote closed the connection)
[14:04] * toad_ (toad@toad-with-underline.active.supporter.pdpc) has joined #freenet
[14:04] * ChanServ sets mode +o toad_
[14:15] <TheSeeker> 'ello toad.
[14:16] <TheSeeker> what's your take on forced and shared specialization vs guessed specialization?
[14:17] <toad_> forced specialization has a whole set of problems of its own
[14:17] <toad_> specifically how to enforce it
[14:18] <toad_> if we wanted forced spec, we'd use a classical DHT
[14:18] <Bonkers> how about suggested specialization?
[14:19] <TheSeeker> what's wrong with a classical DHT connection sctructure with all teh anonymous goodies placed on top?
[14:20] <Bonkers> well you'd need lots of overlap and redundancy
[14:20] <TheSeeker> ok, and this is bad?
[14:20] <Bonkers> and in the web-like way that freenet is connected, I think it'd be really tough to work out in a less than random way
[14:21] <Bonkers> if every node knows every other, it's not so bad, but without any knowledge of how to get to anyone but your neighbors, it's tough to assign chunks of a DHT to each node
[14:21] <Bonkers> I Think some suggested specializtion from you neighbors would be the best, generate some probability distribution based on incoming suggestions and store/drop chunks depending
[14:22] <Bonkers> if you were going to go the route of cooperative specialization that is
[14:23] <TheSeeker> I was thinking more along the lines of "on conenct, get a list of conencted nodes and their spec points" then start connecting to a disproportionately larger number of "near" nodes (overlapping spec range with your own) than "far" ones.
[14:24] <Bonkers> wouldn't it be in your interest to connect to far nodes?
[14:25] <Bonkers> I assume near and far mean distances between specializatoin, it would be in the interest of everybody to minimize the average distance between each node and someone with each part of the hash table
[14:26] <TheSeeker> not for efficient routing ... if you are specialized in an area of the keyspace, and everyone knows it, then you primarily want to be connected to other nodes near that specialization, as that's what the majority of your requests are going to consist of
[14:27] <Bonkers> hmm
[14:27] <Bonkers> is that necessarily true thoguh?
[14:27] <TheSeeker> I see forced specialization points and preferred conenctions as a way of just skipping past the days of "learning" that would otherwise take place.
[14:27] <Bonkers> a greater number of requests will be served at the neighbor nodes, so maybe you'd receive more request for information not related to neighbors
[14:28] <Bonkers> it could do more than skip past the days of learning, I think so sort of planned thought out organization could speed the network in general
[14:30] [ClientsSurvey VERSION]
[14:31] <TheSeeker> I had two thoughts on organization before (three now) 1) 50% of nodes with specialization ranges overlappign your own, the rest spread evenly throughout the keyspace. 2) all nodes conencted overlap specialization ranges with ouyr own node 3) (new) dijjer's method of having an inversely proportionate chance of allowing a connection to a node base on the distance between the two node's specialization centers.
[14:33] <TheSeeker> I see 1 as being succeptable to attack by seeign which keys are gettig "short circuited" across the network you can tell it's that node requesting a key. having a chance of not short circuiting would be protection against this.
[14:34] <Bonkers> 3 seems interesting
[14:34] <TheSeeker> 2) routing of all keys requested is achievable, but request time will grow as you try and find keys further from your specialization. wider specialization ranges will help with this, but that is counter-intuative to my original goal of optimized transfer speed
[14:36] <toad_> TheSeeker: you get dijer's method anyway
[14:36] <toad_> LRU produces it
[14:36] <toad_> you don't have to force it
[14:36] <TheSeeker> for 3) I'd suggest doing a "if connection refused, relay the N closest connected nodes" to the client trying to connect.
[14:36] <toad_> i'm quite willing to consider ripping out the routing algo (although ian may not be), once we've proved that it doesn't work
[14:36] <toad_> the existing one, that is
[14:36] <toad_> and we haven't to my satisfaction
[14:37] <TheSeeker> toad: perhaps if everyone agreed on what your specialization point was, but I'm not seeing that :P
[14:37] <toad_> TheSeeker: there are bugs in implementation, and there are bugs in architecture
[14:37] <i2p_iip> <Michelle> Satan has taken music and he has counterfeited it, convoluted it, twisted it, and exploited it.
[14:37] <TheSeeker> I think that if you JUSt had a forced key range that you will cache, that would force others to realize your spec a lot easier, even with guessing... but at that point, why make them guess?
[14:38] <toad_> if we can prove that even in an ideal world simulation, the architecture simply won't work/scale, then I'd be willing to ditch the present routing algorithm
[14:38] <TheSeeker> let me know when you're accounting for bandwidth limitations in your sim, I'll be very interested in the results then.
[14:39] <toad_> [16:52] <i2p_iip> <Sonax> greycat: So, freenet is evil... What else is new? (besides, some of the replies to the thread seemed to be fair enough).
[14:39] <toad_> uhh, what thread?
[14:39] <TheSeeker> teh suprnova getting canned thread on slashdot.
[14:39] <TheSeeker> http://yro.slashdot.org/comments.pl?sid=133299&cid=11131489
[14:39] <i2p_iip> <Sonax> taod: http://yro.slashdot.org/comments.pl?sid=133299&cid=11131489
[14:39] <toad_> [16:57] <d-ArkAngel> maybe someone should host a site full of bt links on a freesite? ?that way they're safe from "attack" but people get the speed of BT for their downloads....
[14:40] <toad_> maybe we should fix freenet so they don't need to rely on external unanonymous systems :)
[14:40] * yoss is now known as deuxexmachina
[14:40] <greycat> sounds good to me!
[14:40] <i2p_iip> <Sonax> Toad: Then get on with it... :)
[14:40] * deuxexmachina is now known as deusexmachina
[14:40] <toad_> Sonax: is it the usual freenet-is-only-used-by-pedos-and-it'll-never-work-because-it's-not-kademlia crap?
[14:41] <TheSeeker> I am trying to think of bandwidth first isntead of routing first, since bandwidth is a real issue that can be easily quantified and measured while routing is very alchemical :P
[14:41] <greycat> toad_: pretty much, yes
[14:41] <toad_> the point is, the only reason freenet would only be used by pedos (and corporate whistleblowers, and chinese dissidents, and...) is that it's so fscking (allegedly) slow
[14:41] <greycat> TheSeeker: the metric for routing is "Given a file inserted by node A and requested by B, how many times does it successfully find the file, and how long does it take?"
[14:41] <TheSeeker> "freenet, harbors pedos so I'm not going to use freenet, ensuring that only pedos will ever use freenet."
[14:42] <toad_> yeah, like they did with the internet
[14:42] <toad_> some still do in fact
[14:42] <TheSeeker> shh! don't let them know! ;p
[14:42] <greycat> it's not that it's "slow" per se. It's that it can't find the data, so you have to try five hundred times.
[14:42] * toad_ has a friend who basically accused me of aiding and abetting paedos, and who doesn't have an internet connection
[14:43] <toad_> no, it's not slow... TFE has been loading absurdly fast lately
[14:43] * kevloral is now known as kevlAway
[14:43] <i2p_iip> <Michelle> toad, I accuse you of aiding and abetting p?dophiles, but I do it as a compliment
[14:43] <i2p_iip> <Michelle> I SAY IT OUT OF LOVE
[14:43] <Bonkers> hah
[14:44] <TheSeeker> Morality is someting people made up to make them feel superior to others based upon the actions they choose. In the end nothing matters, and one idea or action is no more harmful than any other, but on a microcosmic scale of individual egos it tends to be the focus of much debate.
[14:45] <greycat> uh... I disagree, but you're clearly trolling, so I won't bother giving any coherent response
[14:45] <TheSeeker> ultimate freedom of expression reveals true human nature, which is to say that all actions and ideas are natural, even if we'd rather not admit it.
[14:45] <toad_> TheSeeker: please take the philosophy elsewhere
[14:45] <toad_> e.g. #freenet-politics
[14:46] <Bonkers> start up a flog with your philosophy
[14:46] <toad_> I also disagree, vehemently, but I have work to do
[14:46] <TheSeeker> heh
[14:46] <toad_> and an op bit ;)
[14:46] <i2p_iip> <Michelle> TheSeeker is a post-modernist. He gives head to Paul Martin in his spare time, who recently declared Canada to be 'the most post-modern nation' in existance.
[14:47] <toad_> okay fetchfrac=5% causes major problems... 60% success ratio over long term
[14:47] <TheSeeker> Who's Paul Martin?
[14:47] <toad_> fetchfrac=2.5% causes major niceness
[14:47] <i2p_iip> <Michelle> Thus, I am now referring to the prime minister of canada as 'Post-Modern Paul Martin' most of the time.
[14:47] <i2p_iip> <Michelle> He is the prime minister of canada
[14:47] <i2p_iip> <Michelle> He has declared dozens of times that Canada is a 'post-modern nation'
[14:48] <toad_> fetchfrac=2.5% gets 98%+ on 100x25@10... and nearly that good on 200x25@10 !
[14:48] <TheSeeker> ah, all I know about canada is that it's up there *points up*, is cold, and that all publicly displayed signs have to be in english and french ...
[14:48] <toad_> heh
[14:48] <greycat> and it's metric
[14:49] * toad_ would mumble his real views but really i suspect most people here would rather i spent the time writing simulations
[14:49] <Bonkers> can't be too much colder than here right now
[14:50] <Bonkers> it's -8C in NYC right now with gusts up to like 20km/h
[14:50] <TheSeeker> any idea when the current simulation routing model will make it to unstable for testing?
[14:52] <toad_> hmm?
[14:52] <i2p_iip> <Michelle> no, toad
[14:53] <i2p_iip> <Michelle> I want to know more about your views
[14:53] <Bonkers> I gotta put that damn philsophy course to use some time
[14:54] <TheSeeker> hey, I appologize for my eariler tangent... when I start trying to comprehend freenet as a whole, my head gets hooked into large number and macro-scales... it's not that hard to see the entire struggle of humanity as insignificant from far enough away on a great enough timescale
[14:55] * Hory (~Miranda@82.76.81.56) has joined #FreeNet
[14:56] <TheSeeker> watching that web-cast of a lecture on entropy didn't help either :P
[14:57] <toad_> TheSeeker: over entropy?
[14:57] <toad_> hmmm
[14:57] <toad_> i wonder... does it work just as well on 400x25@10?
[14:59] <toad_> that would have interesting consequences... the network can learn, and work remarkably well, as long as the live set is 2.5% or less of the available datastore slots i.e. as long as your level of redundancy+old data is above a certain level
[14:59] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[14:59] <toad_> also there is quite clearly some learning involved...
[14:59] <toad_> and disturbingly, with higher fetchfrac it doesn't seem to recover no matter how long the timescale
[15:01] * sdogi_ (~java@84-50-19-194-dsl.kjj.estpak.ee) Quit (Remote closed the connection)
[15:01] * sdogi (~java@84-50-19-194-dsl.kjj.estpak.ee) has joined #freenet
[15:01] <toad_> even with higher HTL
[15:01] <TheSeeker> hmm... I'm not at all sure what any of that means, heh.
[15:01] <toad_> it seems that fetchfrac is the determining factor...
[15:01] <TheSeeker> what does fetchFrac represent?
[15:01] <toad_> both 10 and 14 HTL produce roughly the same result with 5% fetchfrac - 60% success over the long term
[15:02] <toad_> TheSeeker: the ratio of the working set, which we randomly request, to the total number of datastore slots (nodes * datastore size in number of files)
[15:02] <toad_> so 0.025 = working set of 250 files for a 100 node network with datastores of size 100
[15:04] <TheSeeker> so as long as you don't have more unique data than will fit in 2.5% of the total network's datastore, you'll have high success with a low number of connections and HTL?
[15:04] <toad_> so for a more realistic network, 10,000 nodes with an average 5G store, with 1MB files.. that would be a total of 10,000 * 5,000 = 50M files... which becomes 1.25M files.. which is still over a terabyte...
[15:04] * Elly (~Elly@ool-182c3b26.dyn.optonline.net) has joined #freenet
[15:05] <toad_> TheSeeker: that's what it looks like...
[15:05] <toad_> if the scaling holds out
[15:05] <toad_> i.e. if it still works at 400 nodes
[15:05] <toad_> without increasing the HTL too much
[15:06] <TheSeeker> that would make for... a redundancy factor of 40x the active set... ?
[15:06] <Bonkers> wasn't the problem with local info pages slowing when loading a freenet page supposed to have been fixed a really really long time ago?
[15:06] <toad_> Bonkers: huh?
[15:06] <toad_> local info pages?
[15:06] * yonkeltron (~yonkeltro@pcp04665066pcs.wilog501.pa.comcast.net) has left #freenet
[15:06] <Bonkers> like environment, open connections, that stuff
[15:06] <toad_> ah
[15:06] <TheSeeker> Bonkers: what browser are you using?
[15:07] <toad_> no, it's mostly a browser problem, but there is still a node side component of it
[15:07] <Bonkers> firefox, maybe I don't have my connections set high enough
[15:07] <TheSeeker> Have you altered the max conenctions per server from the default?
[15:07] <toad_> the fix is to fix the browser connection settings, and make the web interface use NIO directly instead of emulating it with one thread per conn
[15:08] <TheSeeker> any given freenet page is likely to max out your allowed connections to a single host (localhost) at default settings :/
[15:08] <toad_> well there is a limit of 24-36 on the node itself
[15:08] <toad_> for client http conns
[15:09] <Bonkers> that's a bit better now
[15:09] <greycat> TheSeeker: it's even more fun in dillo. dillo has NO browser-side limits on open connections, and will happily open one connection for every image on a page. This runs you out of file descriptors (on, say, HP-UX where you get like 60 open files by default), and then dillo locks up and nothing works until you close and restart it.
[15:10] <greycat> and dillo over an ssh tunnel seems to hit some sort of limit in the ssh tunnel, too...
[15:10] <greycat> (that's if you raise the open FD limit from the default 60)
[15:10] <Bonkers> score one for dillo
[15:11] <Bonkers> is that 60 per process I hope?
[15:11] <greycat> per user
[15:11] <toad_> lol
[15:11] <toad_> well they take up kernel memory...
[15:11] <greycat> no, wait, that would be per process. the "max user processes" is per user.
[15:11] <Bonkers> makes sense
[15:11] <greycat> and of course there will also be a system-wide kernel limit on open FDs
[15:12] <greycat> but that's typically in the thousands
[15:12] <Elly> I've had bittorrent error because it runs out of filehandles.
[15:13] <Bonkers> thousands? shouldn't it be at least like 2^32-1 ?
[15:13] <greycat> I probably would too, if I were crazy enough to run it on an HP-UX box at work :)
[15:14] <toad_> 400x25@10 is definitely turning...
[15:14] <TheSeeker> yey?
[15:14] <toad_> Bonkers: only if the kernel is written so that filehandle metadata can be paged
[15:15] <toad_> well i dunno about yay
[15:15] <toad_> but something interesting is happening even if it's only a bug
[15:16] <toad_> okay i think it's probably going up, but it'll take an hour or two to get to the nineties
[15:29] * sanity (~ian@81-178-106-165.dsl.pipex.com) has joined #freenet
[15:32] * moskau23 (~Miranda@dsl-213-023-252-161.arcor-ip.net) has joined #freenet
[15:37] * sanity (~ian@81-178-106-165.dsl.pipex.com) Quit ()
[15:38] <toad_> ugh
[15:38] * sanity (~ian@81-178-106-165.dsl.pipex.com) has joined #freenet
[15:38] <toad_> sanity: having problems?
[15:38] <toad_> did you get my pm's?
[15:39] <toad_> if i was a simulator bug, that produced better results than should happen, where would I live...
[15:39] <Elly> What's the difference between anonymous IRC and an open proxy with no session limit?
[15:41] <toad_> ease of tracing somebody
[15:41] <toad_> decentralization
[15:41] <toad_> etc
[15:41] * Elly shrugs
[15:42] <toad_> http://amphibian.dyndns.org/simulations3-400x25-avg25.png
[15:42] <toad_> hmmm
[15:42] <toad_> coul
[15:42] <toad_> could be stabilizing, maybe starting to fall...
[15:44] * toad_ should start a 400x25@14 (say) on a simulator node... and an 800x25@14
[15:44] <TheSeeker> perhaps it's just buildign a larget base and will continue back up ?
[15:45] <toad_> TheSeeker: fluctuations perhaps, yeah.. but if they increase in size they can cause big problems
[15:45] <sanity> toad: why do you think its a bug?
[15:45] <sanity> toad: 1) not decrementing HTLs
[15:46] <toad_> sanity: because of the results which verge on absurd
[15:46] <toad_> sorry brb
[15:48] <toad_> sanity: if you don't decrement the HTL you get 100% success
[15:48] <toad_> I fixed that bug yesterday :)
[15:49] <toad_> http://amphibian.dyndns.org/simulations3-400x25-avg25.png
[15:49] <toad_> ooh
[15:49] <toad_> looks a lot like an asymptote... well below 100% in this case... but then it IS at 10 HTL
[15:50] <TheSeeker> hmm, looks like the rise is more or less linear now ...
[15:53] <TheSeeker> and drawing the line out... yeah, it'll take quite a while to get to the 90s ...
[15:54] <toad_> you think it'll keep going?
[15:54] <toad_> btw reload it
[15:55] * moskau23 (~Miranda@dsl-213-023-252-161.arcor-ip.net) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[15:57] * toad_ runs htl 13 on hard-core
[15:58] <toad_> one interesting point here: htl doesn't make much different on 5%
[15:58] <toad_> that suggests a bug; htl ought to make a huge difference
[15:58] <toad_> anyway i'm trying higher HTL on 400x25... htl 13
[16:01] <toad_> makes a huge difference at least to start with on 400...
[16:01] <toad_> but it's early days... http://amphibian.dyndns.org/simulations3-400x25-avg25.png
[16:03] <TheSeeker> I estimate: 4.75e06 for 90% sucess
[16:04] <TheSeeker> hmm, htl 13 seemed to help a bit...
[16:04] <Elly> I wish I knew enough to care.
[16:07] <toad_> looks like it's going to level out
[16:07] <toad_> Elly: X axis is the number of requests
[16:07] <toad_> Y axis is the success probability i.e the fraction of requests during this simulation cycle (5000 requests) which succeeded
[16:07] <toad_> (it's smoothed a bit)
[16:07] <toad_> green is with 13 HTL, red is with 10 HTL.. HTL is the number of hops a request goes to
[16:08] <TheSeeker> oops, I think I meant 6.75
[16:09] <TheSeeker> because my line didn't change that much with the last update (with the green line) and it's just over 6 on that one...
[16:10] <TheSeeker> 6.25e+06 keys requested for 90% at 10 HTL... ?
[16:12] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[16:13] <toad_> if this works, then there'll be a threshold htl for each network size below which it goes to an asymptote that is way below 100%...
[16:13] <toad_> or perhaps, they always go high, it just takes a lot longer on lower htl?
[16:14] * toad_ hmmz, can i run two simulations at once on hard-core...
[16:14] <toad_> d-ArkAngel isn't here...
[16:15] <TheSeeker> is it terribly difficult to run a simulation? I've got a half-decent machine.
[16:16] <TheSeeker> 1.8 GHz A64 (running 32 bit windows :P) a gig of RAM, and more than enough free hard drive space...
[16:16] <toad_> TheSeeker: get current unstable, run something like this:
[16:16] <toad_> java -cp freenet.jar:freenet-ext.jar freenet.node.simulator.newsim.Main --fetchFrac 0.025 --fullyconnected false --nodes 400
[16:16] <toad_> there are lots of options
[16:17] <toad_> graphing the output is harder
[16:17] <toad_> but i can provide scripts
[16:17] <TheSeeker> I have perl installed, but probably nothing else that might be used...
[16:18] <toad_> hmm, on windows? it's feasible
[16:18] <toad_> but harder to process the results
[16:18] <TheSeeker> latest unstable was built on Dec 9?
[16:19] * toad_ fixes
[16:20] <TheSeeker> on the plus side, it was very fast to download, seems the server problem was fixed.
[16:21] <TheSeeker> oh, and if you didn't read about it in the logs, someone verified that the winstaller datastore size is being set correctly now.
[16:21] <TheSeeker> allocating 2.5% of the hdd space on install instead of 256 MB.
[16:22] <toad_> i thought it was supposed to be 10%
[16:22] <TheSeeker> so did I, but then he would have had a 4 GB store instead of a 1 GB sotre...
[16:22] <toad_> TheSeeker: hmmm
[16:22] <toad_> that's a bug then :)
[16:22] <toad_> anyway bbiab
[16:25] <TheSeeker> yeah, current freenet-unstable.jar has no newsim subdirectory in teh jar x.x...
[16:25] <TheSeeker> woo, update
[16:28] <TheSeeker> hrm, "can not find main class" ... :/
[16:30] <TheSeeker> aha
[16:30] <TheSeeker> had to change the : to ; between the classpaths
[16:36] * Ash-Fox (HAL-9000@acw201.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[16:36] * Ash-Fox (HAL-9000@acw201.neoplus.adsl.tpnet.pl) has joined #FreeNET
[16:38] * greycat (~wooledg@192.35.79.70) Quit ("This time the bullet cold rocked ya / A yellow ribbon instead of a swastika")
[16:48] * TheSeeker runs the sim at a thousand nodes with 10 htl and goes to find food
[16:50] <TheSeeker> 40% -> 60% -> ? (dropping)
[16:58] * kevlAway is now known as kevloral
[17:14] <TheSeeker> bottomed out at 48.6
[17:16] <TheSeeker> currently at 51% and climbing at a rate of about .01% per tick...
[17:17] * Asciiwhite (~scegs@ppp213-43.lns2.syd3.internode.on.net) has joined #freenet
[17:17] <Asciiwhite> hey fellas,
[17:17] <Asciiwhite> anyone here use frost?
[17:18] <kevloral> Asciiwhite: yep
[17:19] <Asciiwhite> kevloral: i booted up frost for the first time yesterday, added some boards from the list of known boards, left it to update overnight
[17:19] <Asciiwhite> all the fronst default boards loaded though.
[17:19] <Asciiwhite> but not one of the boards i added retrived any messages.
[17:20] <TheSeeker> that could be because thge boards you chose are dead, or because your freenet node is unable to route messages very well yet...
[17:20] <Asciiwhite> ok
[17:20] <kevloral> Asciiwhite: many of the boards in the known boards list have no traffic.
[17:21] <Asciiwhite> kevloral, yeah but forums like e-books ?
[17:21] <Asciiwhite> and 2600
[17:21] <Asciiwhite> etc.
[17:21] <TheSeeker> yeah, I noticed that only a few of the boards I added retrieved any messages over the course of a couple days... check the "fiels search" tab though, and you'll see that people are uploading more than talking about it.
[17:21] <Asciiwhite> sorry boards.
[17:22] <Asciiwhite> mmm maybe i also have retevial days too low.
[17:22] <TheSeeker> "movies" was one of the only added boards to get regular traffic... "successful" has a bit of traffic too, but it's 'just' people saying they were able to retrieve a file...
[17:22] <kevloral> Asciiwhite: I have subscribed to 2600 and e-books. Let's see if I get any messages there.
[17:24] <Asciiwhite> ok
[17:27] <Asciiwhite> TheSeeker: im using a fcp proxy from i2p network.
[17:27] <Asciiwhite> I also use to to retreive freenet sites, and its reasonibly integrated, aswell as been up for i'd say over a month.
[17:27] <TheSeeker> toad: one thing I notice with the sim... it counts 'total requests'/'total successes'... is there any way to see 'last tick requests'/'last tick successes'? I suppose I could do it manually by using calc and doing: '(current-previous) requests'/'(current-previous) successes' ...
[17:28] <TheSeeker> over a month uptime? what version is the node? 5100 supposedly has some pretty good improvements.
[17:28] <Asciiwhite> shrugz, im not running the node :)
[17:28] <Asciiwhite> im just using it
[17:28] <TheSeeker> if you can access freesites, can you access the fproxy nodestatus page?
[17:29] <Asciiwhite> good point one sec.
[17:29] <Asciiwhite> ill check
[17:30] <Asciiwhite> yeah its 5100
[17:31] <TheSeeker> oh, -.-
[17:31] <TheSeeker> now that I look at it while it's not scrolling, the last tick request/success ratio is listed :P
[17:34] * sanity (~ian@81-178-106-165.dsl.pipex.com) Quit ()
[17:35] * tav (~tavino@62.206.114.179) has joined #freenet
[17:36] <Asciiwhite> ?
[17:36] <kevloral> Asciiwhite: I am not getting anything on those boards, sorry.
[17:37] <TheSeeker> kevloral: what about the file search for those boards? any uploads, or are they just really dead boards?
[17:37] * Hory (~Miranda@82.76.81.56) Quit ("CyberLore.net - Recommendations on the best games, freeware and websites.")
[17:37] <Asciiwhite> kevloral: ah ok so its not me, thanks
[17:38] <kevloral> TheSeeker: if there are no messages, there are no files.
[17:38] <TheSeeker> not true
[17:38] <Asciiwhite> i was trying to get tinkerit files ,aswell as some subwarez files, both sites are accessible, but the files they host are not.
[17:38] <TheSeeker> I've seen serveral boards where I showed no messages, but lots of files.
[17:38] <Asciiwhite> Both sites though recommended, some boards where their files are posted, so i thought i'd give them a try.
[17:38] <kevloral> TheSeeker: but they had files in the past, and messages "expire", but files don't.
[17:38] <kevloral> TheSeeker: but they had messages in the past, and messages "expire", but files don't.
[17:39] <TheSeeker> you can upload content to a board without posting a message.
[17:39] <kevloral> TheSeeker: and people do that?
[17:39] <TheSeeker> I've done it myself a few tiems to test.
[17:39] <TheSeeker> yes, quite frequently.
[17:39] <TheSeeker> (it would seem)
[17:39] <kevloral> TheSeeker: doh, I thought that wasn't being used.
[17:40] <kevloral> TheSeeker: you users are weird people. ;-)
[17:40] <Asciiwhite> yeah there are some files in books, not many though :)
[17:41] <Asciiwhite> thanks fellas
[17:41] <TheSeeker> though, it seems a lot of people also use fuqid to upload content then just post a keylist instead of re-inserting with frost later ... ideally, the inserts should all collide [quickly] upon a frost re-insertion... unfortunately that doesn't seem to happen the way one would expect.
[17:42] <kevloral> TheSeeker: yes, that was I thought: that everybody was using fuqid to upload (because uploading with frost is a pain) and then sharing the keys via messages.
[17:42] <Asciiwhite> can any of you access any of the files here , SSK@ZoZZWiG3Wd0GLteyyIE3AZEdcmQPAgM/tinker/11// ?
[17:43] <kevloral> TheSeeker: aaagh! s/that was/that is what
[17:43] <TheSeeker> seems a few figured out that frost's settings can be adjusted to allow for more insert/download threads, and actually use frost to upload content sometimes.
[17:44] * goatee (~goatee@ip216-239-88-163.vif.net) has joined #freenet
[17:46] <TheSeeker> this doesn't appear to be turning very fast... I'm at 2.8 million requests and still at 53% overall success, with a 54%-57% 'last 5k requests' success
[17:47] <TheSeeker> s/2.8/3.8/
[17:47] <TheSeeker> hmm, that last / would probably mess thigns up huh? :P
[17:51] * FallingBuzzard (~srademach@207.152.112.129) has left #freenet
[17:52] * sanity (~ian@81-178-106-165.dsl.pipex.com) has joined #freenet
[18:11] * pupok (~r00t@81-178-106-165.dsl.pipex.com) Quit (Read error: 110 (Connection timed out))
[18:20] <toad_> hi
[18:21] <tav> hi
[18:21] <toad_> http://amphibian.dyndns.org/simulations3-400x25-avg25.png
[18:21] <toad_> hrrm
[18:21] <toad_> red is definitely falling
[18:22] <toad_> <TheSeeker> toad: one thing I notice with the sim... it counts 'total requests'/'total successes'... is there any way to see 'last tick requests'/'last tick successes'? ?I suppose I could do it manually by using calc and doing: '(current-previous) requests'/'(current-previous) successes' ...
[18:22] <toad_> it does both
[18:29] <toad_> http://amphibian.dyndns.org/simulations3-400x25-avg25.png
[18:29] <toad_> does anyone have any idea how the performance of a network can decline?
[18:33] <TheSeeker> sending keys to the fastest node for any range of keys instead of the closest range to yoru request? (*shrug*)
[18:40] <toad_> no
[18:40] <toad_> a simulated network
[18:40] <toad_> ah
[18:40] <toad_> hmm
[18:40] <toad_> maybe
[18:43] <TheSeeker> http://theseeker.bounceme.net/1000x25@10.png
[18:43] * Ash-Fox (HAL-9000@acw201.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[18:43] * Ash-Fox (HAL-9000@acw109.neoplus.adsl.tpnet.pl) has joined #FreeNET
[18:44] <TheSeeker> I'm going to try again @15.
[18:47] * Elly is now known as Elly|afk
[18:49] <TheSeeker> hmm, just a thought... should # of requests per cycle be proportional to the number of nodes on the network? perhaps you're seeing a slower/worse spec rate with a larger network because you're using a lower number of requests per node?
[18:52] * TheSeeker remembers his laptop is faster than his desktop, and fires it up to run simulations too.
[18:55] * guido^pe (~unknown@dsl-213-023-240-162.arcor-ip.net) has joined #freenet
[18:57] * Elly|afk is now known as Elly
[19:02] <TheSeeker> hmm, there's an update available for Java 1.5.0 already...
[19:03] * sanity (~ian@81-178-106-165.dsl.pipex.com) Quit ()
[19:05] * pupok (~r00t@81-178-106-165.dsl.pipex.com) has joined #freenet
[19:17] * guido^pe (~unknown@dsl-213-023-240-162.arcor-ip.net) Quit (Remote closed the connection)
[19:38] * kevloral is now known as kevlAway
[19:42] <Bonkers> how do I set where files like lsnodes go?
[19:44] <Bonkers> they seem to be going to storePath/.. or something like that
[19:46] <TheSeeker> I haven' tseen a place to set it, but that is correct, storepath/.. is where they go on my system as well
[19:46] <Bonkers> hmm, they didn't used to go there, and that's not a terribly convenient place
[19:46] <Bonkers> I guess I'll have to chagne storepath
[19:46] <TheSeeker> so I made a freenet\datastore directory and pointed the datastore there, that way the lsnode data stays in freenet
[19:47] <TheSeeker> (on a different drive than my freenet application dir)
[19:48] <Bonkers> yep, exactly what I'm doing
[19:49] * setre (~jussi@h136n1fls301o291.telia.com) has joined #freenet
[19:49] * setre (~jussi@h136n1fls301o291.telia.com) has left #freenet
[19:53] <TheSeeker> heh, rounding error
[19:53] <TheSeeker> 1282326/2160000=0.5936694444444445
[19:57] <TheSeeker> the concept of repeating decimals makes my head hurt a bit... 1/3 = .3333? 2/3 = .6666? so 3/3 = .9999? ... which is < 1 by an infinately small amount :P
[19:58] * Asciiwhite (~scegs@ppp213-43.lns2.syd3.internode.on.net) has left #freenet
[19:58] * mikeeusa (~r00t@ool-18be3a49.dyn.optonline.net) has joined #freenet
[19:58] <TheSeeker> "There are no such things as whole numbers, only crude forms of measurement." ;p
[20:04] * verl (verl@h152n2fls33o877.telia.com) Quit ()
[20:09] <TheSeeker> htl=25 didn't help 1000 nodes, it stayed around 57% through 2M requests...
[20:11] * verl (verl@h152n2fls33o877.telia.com) has joined #freenet
[20:11] <TheSeeker> which is about what htl=13 is for 4M keys requested
[20:20] * leexx (~bbtt@cpc3-warr1-5-0-cust181.bagu.cable.ntl.com) has joined #freenet
[20:20] * leexx is now known as leexgx
[20:20] * leexgx is now known as |UK-Monster|
[20:20] <|UK-Monster|> ?
[20:21] <TheSeeker> ?
[20:21] <|UK-Monster|> heh
[20:21] <|UK-Monster|> dam anti nick change thing :P
[20:21] * |UK-Monster| is now known as leexgx
[20:22] <leexgx> any one around?
[20:22] <TheSeeker> a few people
[20:22] <leexgx> i guess freenet and other secure networks are going to be started to be used
[20:22] <leexgx> more
[20:23] <leexgx> BT sites have been pulled
[20:23] <TheSeeker> that, or new traker sites will pop up.
[20:23] <TheSeeker> I've only heard of two so far, there's a lot of other torrent tracking sites out there...
[20:23] <leexgx> ed2k sites has allso been pulled
[20:23] <TheSeeker> yeah, but who cares about ed2k? :P
[20:24] <leexgx> works for me
[20:24] <leexgx> well
[20:24] <TheSeeker> you are q position #349123 have a nice day.
[20:24] <leexgx> and is that any dif from freenet
[20:24] <leexgx> but thats an dif matter :)
[20:25] <leexgx> i have no problems with emule
[20:25] <leexgx> just ignore the ques
[20:25] <leexgx> going to fire up me node any way
[20:25] * leexgx freenet starting
[20:26] <TheSeeker> I find shareaza to be easier to use with better featuers... I'd be happier if freenet worked though...
[20:26] <leexgx> shareaza is incompatble with emule untill 2.2 comes out
[20:27] <TheSeeker> I dont' care, I don't connect to that network...
[20:27] <leexgx> as it does not use the rules of asking for an slot
[20:28] <leexgx> i norm get files in an day
[20:28] <TheSeeker> I may be able to find 'more content' when conencted to edonkey, but I can't actually retrieve much of what I can find.
[20:28] <leexgx> it can
[20:28] <leexgx> i used it before
[20:29] <TheSeeker> with the Gnutella/G2 hosts, I know that if I see a hit, I'm probably going to be able to download it.
[20:29] <leexgx> but with 0.44 of emule thay put something in that inforece the rules
[20:29] <leexgx> i think emule is going to much with rules
[20:32] <leexgx> is there an change log for freenet
[20:32] <leexgx> wundering whats been fixed and broke from the last time i been here
[20:32] <TheSeeker> sortof, buried in the archves of the devl and cvs lists
[20:34] <leexgx> seed nodes is 5mb now bit lighter then before
[20:38] * kevlAway (~kevloral@CZ1-RAS-8-u-0179.du.onolab.com) Quit (Read error: 238 (Connection timed out))
[20:41] <TheSeeker> http://theseeker.bounceme.net/1000x25@15.png
[20:41] <leexgx> lol 0.6%
[20:42] * leexgx asumes thats an bad number
[20:42] <TheSeeker> that;s actually 60%, oops
[20:42] <TheSeeker> mad label
[20:42] <TheSeeker> bad label
[20:42] <leexgx> lol
[20:42] <leexgx> ok 60% ant so bad
[20:42] <TheSeeker> it's just a simulation though
[20:43] <leexgx> just warring up me freenet
[20:43] <TheSeeker> and it's a lot worse than the 2-400 nodes at 10 htl sim figures
[20:44] <leexgx> lol i getting 20k out of freenet
[20:44] <TheSeeker> 20k what?
[20:44] <leexgx> 20KB
[20:44] <leexgx> was
[20:45] <leexgx> for 10 secs :P
[20:45] <TheSeeker> data you are requesting, or just in use by your node?
[20:45] <leexgx> my node is cold
[20:45] <leexgx> 10 mins old
[20:45] <leexgx> so freenet is getitng better
[20:46] <TheSeeker> it doesn't take too long to eat up whatever avaliable resources you give it...
[20:48] <leexgx> one i haev see is ants
[20:48] <leexgx> p2p
[20:48] <leexgx> but what i can see it is an proxy based p2p
[20:58] * orange_ (~orange@pm2-40.bahnhof.se) Quit ()
[21:07] * pupok (~r00t@81-178-106-165.dsl.pipex.com) Quit (Read error: 110 (Connection timed out))
[21:23] <mikeeusa> hello
[21:24] <mikeeusa> freenet working?
[21:24] <mikeeusa> I remeber when we could pull divx movies off of it
[21:24] <mikeeusa> then something happened
[21:24] <mikeeusa> and it got slow again
[21:25] <mikeeusa> how's it like now?
[21:27] * root (~r00t@ool-18be3a49.dyn.optonline.net) has joined #freenet
[21:39] <leexgx> dono
[21:39] <Elly> wow, root is here.
[21:39] <leexgx> heh
[21:39] <leexgx> toad_ around ?
[21:40] * leexgx hats maxtor hdds
[21:40] <leexgx> hates
[21:40] <Elly> my maxtor is okay,.
[21:40] <Elly> *.
[21:41] <leexgx> mines clicking an little
[21:41] <leexgx> died on me last night
[21:41] <leexgx> then came back to life to day
[21:41] <leexgx> so i dumped it onto an WD
[21:41] <leexgx> i getting to the point of going to be mirrioing all me drives
[21:42] <leexgx> RAID 1
[21:43] <leexgx> i got an 2x samsungs that have been going for 2 years nerly (running all the times an am stipe aray(radi 0) its been asking to die just coes its striped
[21:44] <leexgx> 1 die lose data on both but thay just keep on running
[21:44] <leexgx> 5400rpm is king
[21:45] * mikeeusa (~r00t@ool-18be3a49.dyn.optonline.net) Quit (Read error: 110 (Connection timed out))
[21:45] <Elly> wow
[21:45] <Elly> I just have two disks
[21:45] <Elly> a 160gb maxtor and a 250gb seagate
[21:45] <leexgx> be scaered
[21:45] <leexgx> (HDDs) (showing free space), C:\ (No label) 3.4GB/9.54GB(36%), F:\ (40gb store f) 14.5GB/37.2GB(39%), J:\ (80gb store H) 2.36GB/74.5GB(3%), M:\ (240) 84.5GB/223GB(38%), O:\ (No label) 4.44GB/115GB(4%), P:\ (No label) 23.0GB/115GB(20%) (Local total) 132GB/575GB (23%)
[21:46] <leexgx> only got 125gb free now
[21:46] <leexgx> c drive not count
[21:46] <leexgx> as thats my format and reload drive
[21:46] <leexgx> 1hr and i be back up
[21:46] <leexgx> not that i have needed it
[21:46] <leexgx> (OS) Windows 2000 Enterprise Server, Service Pack 4 (5.0 - 2195), (installed for) 96w 1d 20h 5m, (uptime) 6w 1d 11h 38m, (record) 10w 1d 42m 38s (set on 06.02.2004 23:34 using Win2K)
[21:47] <Elly> Nice.
[21:47] <Elly> this box is not in a good way.
[21:47] <Elly> Drivers are missing, hardware is starting to fail, it doesn't boot 60% of the time
[21:47] <leexgx> heh
[21:48] <Elly> occasionally I get this error:
[21:48] <Elly> "The operation could not be completed."
[21:48] <leexgx> wundering where i have head that before
[21:48] <Elly> it happens seemingly at random
[21:49] <leexgx> probly mem problem
[21:49] <leexgx> or mobo
[21:49] <leexgx> what mobo you got
[21:49] <Elly> nvidia nforce2
[21:49] <leexgx> pc chips
[21:49] <leexgx> more likey ram then
[21:49] <Elly> heh
[21:49] <Elly> my RAM has been okay
[21:49] <leexgx> you usen 2 sticks of rim
[21:49] <Elly> I have a corsair 512MB chip
[21:49] <leexgx> ram
[21:49] <Elly> nope, just one
[21:49] <leexgx> have you got it in the middle slot
[21:50] <Elly> nope
[21:50] <leexgx> not the one on its own
[21:50] <Elly> it's been working fine for two years
[21:50] <Elly> heh
[21:50] <leexgx> hmm
[21:50] <leexgx> reload of window
[21:50] <leexgx> s
[21:50] <Elly> yeah...not worth it
[21:50] <Elly> if it dies totally I'll reload it
[21:50] <leexgx> should only take you 20 mins
[21:51] <leexgx> if that
[21:51] <leexgx> << nforce2
[21:51] <leexgx> in my pc not this server
[21:51] <leexgx> its king
[21:51] <leexgx> 3000+ 512x2 200gb
[21:52] <leexgx> you only get the perfomance when you use 1
[21:52] <leexgx> i cant use my pc with only 512
[21:52] <leexgx> what mobo have you got
[21:52] <leexgx> make
[21:52] <leexgx> FFS
[21:52] <Elly> Not sure.
[21:52] <leexgx> 2 not 1
[21:52] <Elly> Haven't looked.
[21:53] <leexgx> dule chan ddr mode
[21:53] <leexgx> you dono what your own bord is ?
[21:53] <Elly> *headshake*
[21:53] <leexgx> heh
[21:53] <leexgx> whats your cpu speed
[21:54] <leexgx> ru the ram at the cpu rated speed
[21:54] <leexgx> run
[21:54] <leexgx> i got an abit nf7-s v2.0 mobo
[21:55] <leexgx> i had it for 1 year and 3 months
[21:55] <root> <leexgx> you usen 2 sticks of rim
[21:55] <root> lol
[21:55] <leexgx> ram
[21:55] <leexgx> DDR
[21:55] <root> Random Interger Math?
[21:55] <leexgx> i not usen P4 crap
[21:55] * leexgx ops speeking hs mind
[21:55] <Elly> <--- Athlon 3200+ 64bit, 2000mhz
[21:56] <leexgx> when he usen an p4 server
[21:56] <leexgx> probly ausu then
[21:56] <root> <---------- Dual Athlon 2Ghz's, 1 GB Ram, ~500GB Hdd space
[21:56] <root> (463 GB when formatted with reiser
[21:56] <root> (2 250s)
[21:57] <root> 7200rpm
[21:57] <leexgx> Elly its cant be an nforce 2 with AMD64
[21:57] <Elly> Why not?
[21:57] <Elly> that's what the box specs say
[21:57] <leexgx> Nforec 3 is the only one that supposrts it
[21:57] <leexgx> and 4
[21:57] <root> man my setup used to turn heads
[21:57] <leexgx> 2 is for k7
[21:57] <root> no i just get ignored :(
[21:57] <leexgx> :P
[21:57] <leexgx> robilad
[21:58] <leexgx> root
[21:58] <root> ?
[21:58] <leexgx> some one els has an r in there name
[21:58] <root> over 200 shell users tho
[21:58] <leexgx> press r then TAB
[21:58] <root> beat that
[21:58] <root> ahahha
[21:58] <leexgx> linux is cool
[21:58] <leexgx> or should i say fast
[21:59] <leexgx> next upgrade will be an
[21:59] <TheSeeker> hmm, if I assume max hops of 25, with N nodes, and C connections... 2N/C has to be less than 50 for an "overlapping specialization" only connection scheme to work (ring shaped network)... meaning it breaks down after hitting about 1k nodes :/
[21:59] <leexgx> 4000+ (pos FX?) nforce 4(probly 5 by the time i upgrade)
[22:00] * TheSeeker scratches idea 2 off the list.
[22:01] <leexgx> Elly the nforce pin out for the amd 64 is compleaty dif it taken them 1 year to bring out an fast mobo
[22:01] <leexgx> ffs
[22:01] <Elly> then maybe it's something else
[22:01] <leexgx> i am going stuped
[22:01] <Elly> I don't really know heh
[22:01] <leexgx> there was an nf2 i think
[22:02] <leexgx> but it was bad me thinks
[22:02] <leexgx> narr
[22:02] <leexgx> its was the nf3 150 that was rubbish
[22:02] <leexgx> 1 year later the nf3 250gb came out to fix the speed problems
[22:02] <leexgx> now the nf4 is out
[22:03] <leexgx> with 4xHW SATA - 2xGB eth ans SLI support (plus 2x SLI)
[22:04] <leexgx> the bord will rock
[22:05] <leexgx> Elly me older chip is faster then yours even tho the pr rating is higher then mine
[22:05] <Elly> okay
[22:05] <Elly> when I care I'll leave you a memo
[22:05] <leexgx> only the 939 is faster then the k7 range
[22:05] <leexgx> :P
[22:05] <leexgx> want to sell it
[22:06] <leexgx> so i can get amd 64
[22:06] <leexgx> ?? low at this time Xmas and all
[22:06] <leexgx> perhaps in 6 months i will
[22:07] <leexgx> when 3000+ chips are considered slow
[22:08] * leexgx is now known as leexx-zzz
[22:09] <TheSeeker> they alread yare considered slow ;p
[22:09] <leexx-zzz> 4000+ at some silly price i think
[22:09] <leexx-zzz> at the mo
[22:10] <TheSeeker> that's what I'm running, value 64-bit, waiting for windows x64 because linux scares me...
[22:10] <leexx-zzz> heh
[22:10] <leexx-zzz> its fun in an geek way
[22:10] <leexx-zzz> spent all day setting up me router
[22:11] <leexx-zzz> then spent another 1 hr looking at preeconfered firewalls
[22:11] <leexx-zzz> www.ipcop.org
[22:11] * Elly is now known as Elly|afk
[22:11] <leexx-zzz> the 1hr was understanding what red and green is
[22:12] <leexx-zzz> any way i sleeping
[22:12] <leexx-zzz> and typeing as well
[22:12] * goatee (~goatee@ip216-239-88-163.vif.net) Quit ("Leaving")
[22:12] <leexx-zzz> does not mix
[22:13] <leexx-zzz> bbbl
[22:14] * goatee (~goatee@ip216-239-88-163.vif.net) has joined #freenet
[22:17] <TheSeeker> ok, if I can ignore the max HTL now as a limit to the number of seperate spec ranges I can have [# of ranges overlapping at a single point it will take to cover the keyspace] with the assumption that after short cicuiting the network in teh first 5-10 hops the data should be easily found in the next 15-20 ... then my minimum specialization rannge in % of the keyspace becomes 4 * total_nodes / connections_per_node
[22:17] <TheSeeker> (* 100)
[22:19] <TheSeeker> so, with a million nodes, and 40 connections per node (average of mid/upper range broadband) we get a % range of the keyspace needed as 0.001%
[22:20] <root> ?
[22:20] <TheSeeker> now how do I turn that into a range of keys? x.x
[22:20] <root> I have UPPER range broadband
[22:20] <root> hahahaha
[22:20] <root> noobs have dialup
[22:20] <TheSeeker> what's your upload speed capped to?
[22:21] <root> 1Mbit
[22:21] <TheSeeker> decent... with my sim you'd be connecting to ... 175 nodes.
[22:22] <TheSeeker> still working on what specialization range you'd get, heh :P
[22:26] <TheSeeker> oh, oops, I should be dividing by 30, not 40 ...
[22:28] <TheSeeker> and that should be connections_per_node / 4 * total_nodes :P
[22:28] <TheSeeker> bah, all mixed up... I had it right in my notes -.-
[22:29] * newblistic (~dragonmor@cs24174111-205.houston.rr.com) has joined #freenet
[22:29] <newblistic> im back
[22:29] <TheSeeker> wait, hmm
[22:30] <newblistic> man this freenet is speeding up now
[22:35] <TheSeeker> hmm, I suppose the number of sections the keyspace is divided up into for determining minimum spec range should be such that the average connection should be able to see most of the keyspace ... which means, with a 50% connection to overlapping nodes and the rest evenly distributed, you'd have to have a spec range on average of 1 / (C/2 + 2)
[22:37] <root> ok so what is it?
[22:37] <root> freking math noob
[22:38] <TheSeeker> I'm thinking out loud... and I'm not all that great at math either or I'd be working the statistics of everything I'm thinking about instead of trying to write code that'll try and show them to me :P
[22:39] <root> DAMN you're *SUCH* a math n00b!!!
[22:39] <root> wowzers
[22:40] <TheSeeker> yeah, at least I'm trying...
[22:40] * root pokes TheSeeker VIOLENTLY with the "dumb n00bler" stick
[22:40] <newblistic> can i be hit with the same stick?
[22:40] * root spraypaint's n()()b onto TheSeeker
[22:40] <newblistic> freenet not working out to well for me
[22:41] * root impales newblistic on the NOOOOOOOOOOOOOBLER! stik
[22:41] * root bites off newblistic's leg and beats TheSeeker with the NEW n00b stick!
[22:41] <root> (which is newblistic's leg)
[22:41] <TheSeeker> having fun?
[22:41] <root> yes :P
[22:42] <TheSeeker> carry on then
[22:42] <newblistic> you are asuming i have legs?
[22:42] * root rips out newblistic's intestines (which are the n00b rope) and hangs TheSeeker by the kneck with them?
[22:42] <newblistic> what is the probability of that mister seeker
[22:43] <root> newblistic, well if you didn't AND you were a girl... then you'd be like a snale
[22:43] <root> that's why females have legs
[22:43] <root> so they don't leave a trail like snails
[22:43] <TheSeeker> I don't know how to do statistical math ... the magical art of making numbers say whatever I want them to is beyond me...
[22:46] <root> ok what im trying to convey is that
[22:46] <root> if women didn't have legs
[22:46] <root> and they tried to move around
[22:46] <root> there would be a slimey like trail behind them
[22:46] <TheSeeker> we get it....
[22:47] <root> because they are lustious whores
[22:47] <root> or perhapse lustious virgins
[22:47] <root> or both
[22:50] <root> ...
[22:50] <root> noob
[22:53] * Elly|afk (~Elly@ool-182c3b26.dyn.optonline.net) Quit (Read error: 110 (Connection timed out))
[22:54] * sanity (~ian@81-178-106-165.dsl.pipex.com) has joined #freenet
[22:59] <TheSeeker> hmmm... a 1 Mbit upstream only network needs 1.1% keyspace range, and a modem user network needs a 12.5% range... but it's all one network... so if I assume the average user will be broadband, and the average uplaod speed between the 128 and 384 connections is ~30 KB/s....
[23:00] <TheSeeker> then that's 40 connections and ~0.15% per KB/s of upstream... making modem users get a 0.9% range and 1 mbit users get a 19.65% range...
[23:04] * newblistic (~dragonmor@cs24174111-205.houston.rr.com) Quit ()
[23:53] <TheSeeker> hmm, freenet has to download an entire trailer before it can start seding it along to the next node in the chain so that the data can be verified, correct?
[23:59] <Hirvox> entire trailer?
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.
