Timestamps are in GMT/BST.
[0:18] * Ash-Fox (i=UNKNOWN@ecv74.neoplus.adsl.tpnet.pl) Quit (Read error: 110 (Connection timed out))
[0:56] * Ash-Fox (i=UNKNOWN@ecg92.neoplus.adsl.tpnet.pl) has joined #FreeNET
[2:05] * MrNaughty (n=mrnaught@S01060013103e9ce9.ed.shawcable.net) has joined #freenet
[5:06] * hobx (n=hobx@c83-248-105-188.bredband.comhem.se) has joined #freenet
[5:07] * sleon|tuX (n=sleon|tu@85.180.50.190) has joined #freenet
[6:32] * rah_ (n=rah@cpc1-nott8-4-1-cust246.nott.cable.ntl.com) has joined #freenet
[6:32] * rah (n=rah@cpc1-nott8-4-1-cust246.nott.cable.ntl.com) Quit (Read error: 104 (Connection reset by peer))
[7:01] * NullAcht15 (n=NullAcht@dslb-082-083-232-195.pools.arcor-ip.net) has joined #freenet
[7:13] * sleon|tuX (n=sleon|tu@85.180.50.190) Quit (Remote closed the connection)
[7:30] * hobx_ (n=ag@ankh.math.chalmers.se) has joined #freenet
[7:33] * sleon|tuX (i=test@85.180.50.190) has joined #freenet
[7:51] * rah_ is now known as rah
[8:02] <toad_> hi rah
[8:02] <rah> hi
[8:02] <rah> I've been meaning to ask
[8:02] <rah> do you get paid to work on fred?
[8:03] <toad_> me yes :)
[8:03] <rah> where does the money come from?
[8:04] <linyos> we rob banks now
[8:04] <linyos> for the cause
[8:04] <toad_> ;)
[8:05] <toad_> http://freenetproject.org/index.php?page=donate :)
[8:05] <rah> does that pay the bills?
[8:05] <toad_> obviously we have to have some legal isolation between linyos's criminal syndicate and the freenet nonprofit that pays me
[8:05] <rah> obviously :)
[8:05] <toad_> hence the above sham
[8:06] <rah> I see
[8:06] <toad_> to make it look like it's funded out of donations ;)
[8:06] * rah hides his inland revenue badge
[8:06] <toad_> :)
[8:06] <toad_> and yes, it does pay the bills
[8:06] <rah> wow
[8:06] <toad_> my main problem is finding the time and motivation to do the work, not fundraising
[8:07] <toad_> having said that, i live with my parents, rather cheaply
[8:07] <rah> what else do you do?
[8:07] <rah> aaahhhh
[8:07] <rah> I seeeee :)
[8:07] <rah> ok
[8:07] <toad_> on what i theoretically earn, in theory i ought to be able to live on my own and pay the bills though
[8:08] <toad_> many people do, although i don't know _how_ exactly ;)
[8:08] <linyos> certainly in a low-cost country you could.
[8:08] <toad_> i could here
[8:08] <toad_> even in Bristol
[8:09] <rah> toad_: what kind of numbers do you get?
[8:09] <rah> toad_: how much per month?
[8:09] <rah> umm
[8:09] <rah> you live in Bristol?
[8:09] <hobx_> how ever much you make, it never seems like enough
[8:09] <rah> where I lived most of my life?
[8:09] <toad_> ?10/hr GBP... in theory I work 35 hours a week
[8:09] <toad_> bristol, england, yeah
[8:09] <rah> fucking small world
[8:09] <toad_> rah: indeed!
[8:09] <rah> do you go to LUG meets?
[8:10] <toad_> i didn't think there were any
[8:10] <toad_> there's just the mailing list afaik
[8:10] <rah> oh :/
[8:10] <toad_> however, i am involved in http://www.bristolwireless.net/
[8:10] <rah> there were meets when I was living there :/
[8:10] <rah> there's a channel as well
[8:10] <rah> cool
[8:11] <toad_> some years back there was a 2600
[8:11] <toad_> that seemed to double for the LUG mailing list
[8:11] <toad_> (which i'm not on anyway!)
[8:14] <toad_> bbiab
[8:14] <toad_> going for walk around oldbury
[8:14] <NullAcht15> toad_: Freenet 0.7 got me thinking a bit. Do you think it might be possible to take some of the ideas from 0.7 to create a completely decentralised, world-spanning network as an anlternative to IP?
[8:15] <linyos> NullAcht15: certainly that is in effect what we are doing with a darknet.
[8:15] <toad_> NullAcht15: I don't know if we can *replace* IP, but I do think we can run 1:1 streams, multicast, and store/fetch over a darknet
[8:16] <linyos> i mean, once you have the darknet links, it is no big deal to run any kind of service over them.
[8:16] <toad_> => de facto internet replacement
[8:16] <toad_> linyos: *as long as you can route*, yes
[8:16] <toad_> if you need it to scale, a routing algorithm is a big deal
[8:16] <toad_> but we have one
[8:16] <linyos> sure.
[8:16] <toad_> also, premix routing over darknet is hard
[8:16] <toad_> but from what i've seen, premix routing over anything is hard
[8:17] <toad_> and i do have some reasonable ideas how to do it over darknet
[8:17] <toad_> bbl
[8:17] <linyos> depends on your security objective.
[8:18] <toad_> also there may be capacity issues if you use streams a lot
[8:18] <toad_> linyos: indeed, you have any expertise on this? i'd like to talk about it, but i'm going out for a few hours
[8:19] <toad_> seey
[8:19] <toad_> a
[8:19] <linyos> clearly if the topology of the darknet is known to all, one can simply construct a route using global knowledge.
[8:20] <linyos> later.
[8:21] <NullAcht15> I have written down my first ideas about how this might be done. I'm not a big expert on that, though, so I thought I'd ask for comments from you folks. Care to see it?
[8:22] <linyos> sure, i'll take a look.
[8:23] <NullAcht15> http://www.stud.uni-hannover.de/~guido83/self-organizing-net
[8:24] <NullAcht15> I'm going to put up some wet clothes to dry now. bbl
[8:26] <linyos> i'm a fan of routing with more global (as opposed to local knowledge), if your security model does not demand that the topology of the darknet link-graph be kept secret.
[8:26] <linyos> (and i have little faith that it could be, anyway.)
[8:27] <linyos> then you can simply pick routes from A to B yourself.
[8:30] <linyos> i'm also a fan of trade in network services within the darknet, btw.
[8:36] <linyos> conceivably, a public topology and the secrecy of individual participants need not conflict...
[8:36] <linyos> since there is no need for them to say who they really are.
[8:45] <NullAcht15> uhm, an important point about that thing I've forgot to mention: Anonymity is not a goal here
[8:46] <NullAcht15> The goal is "just" to create a decentralized network, as in "if your neighbor is already connected, just hook up to him and you're connected, too"
[8:46] <linyos> how do you hook up?
[8:46] <sleon|tuX> NullAcht15, do you know how dark net works ?
[8:46] <linyos> i mean, physically what do you do to transmit data?
[8:47] <sleon|tuX> NullAcht15, borgnet ?
[8:47] <NullAcht15> sleon|tuX: never heard of borgnet
[8:48] <NullAcht15> linyos: Not decided yet. This may work on a number of different transport (link-layer) protocols, like ethernet, or even ipv[46]
[8:48] <sleon|tuX> NullAcht15, startrek ?
[8:48] <sleon|tuX> NullAcht15, :D
[8:48] <NullAcht15> ah
[8:48] <sleon|tuX> NullAcht15, you connect node then it get assimilated :D
[8:49] <NullAcht15> hm, somewhat like this...
[8:49] <sleon|tuX> heh
[8:49] <NullAcht15> Is there such a thing as borgnet?
[8:49] <sleon|tuX> NullAcht15, no it was a joke
[8:49] <sleon|tuX> NullAcht15, but look at how darknet works
[8:49] <sleon|tuX> NullAcht15, have you looked at it already?
[8:49] <NullAcht15> I'm not particularly fond of the name "civesnet"....
[8:49] <sleon|tuX> NullAcht15, it is on freenet -.7
[8:50] <linyos> i guess if everybody ran an ethernet cable to his next-door neighbors, it would be feasible.
[8:50] <linyos> nothing so hard about that.
[8:50] <NullAcht15> sleon|tuX: I've read some of the information I could get my hands on, but that wans'tvery much...
[8:50] <linyos> the problem is when you don't have anybody next door that wants to participate.
[8:51] <linyos> then you have to use (and pay for) the existing infrastructure.
[8:51] <NullAcht15> linyos: I'd say the hard part is that you sometimes need long-distance connections between peers
[8:51] <sleon|tuX> linyos, public assimilation entry points :D
[8:51] <NullAcht15> Even if eerybody were to participate, you'd still have to interconnect cities and continents somehow
[8:52] <linyos> yeah, eventually your traffic would have to reach a datacenter with long-distance, high-capacity fiber.
[8:53] <linyos> but it could easily travel across thousands of little links to get there, so long as you don't introduce latency at each hop.
[8:53] <NullAcht15> linyos: There are countries where cities are hundreds of miles apart...
[8:54] <NullAcht15> Personally, I think this will only work if bandwidth is paid for by the users
[8:54] <linyos> sure, but there will be plenty of paths through people's individual properties to get there.
[8:54] * TheBishop_ (n=bishop@port-212-202-175-197.dynamic.qsc.de) Quit (Read error: 104 (Connection reset by peer))
[8:54] <linyos> even if you have to pay farmers to run wires through the fields.
[8:55] <NullAcht15> That is, peer operators would pay each other for routing each others traffic
[8:55] <linyos> that, or just for the right to run an underground cable across somebody's property.
[8:56] <linyos> the people at each end would share the cost.
[8:56] <NullAcht15> You'd need to have a lot of idealists with a lot of money...
[8:57] <linyos> plenty of people would be happy to let you run a cable through their property for a hundred bucks.
[8:57] <NullAcht15> I mean, what does it cost to put fibre into the ground over a ditance of maybe a hundred miles
[8:58] <NullAcht15> More importantly, what would it cost to do the same thing inside a city?
[8:58] <linyos> the beauty of it is that you don't need to have long links like that.
[8:59] <linyos> it's enough to bury a cable from one guy'
[8:59] <linyos> s house to another's
[8:59] <linyos> do that enough, and you have tons of links to route across.
[9:00] <linyos> and running a cable a hundred feet costs nothing, you just need the cable and a few hours time.
[9:01] <NullAcht15> Mind you, if your traffic needs to take hundreds of hops, that means the routers in the middle of the city/country will be hopelessly overloaded, even if every individual ode produces only reasonable amounts of traffic
[9:01] <linyos> not if there is tons of parallelism.
[9:01] <NullAcht15> hm
[9:01] <linyos> which there would be in such a mesh of links.
[9:02] <NullAcht15> And how am I supposed to connect to the house across the road here? Digging up the road and putting the cable into the ground is definitly too expensive
[9:02] <linyos> might be hard to run your cables under streets.... that is more of a problem.
[9:02] <linyos> yeah.
[9:02] <NullAcht15> Letting the cable hang over the street from one window to another is probably not allowed here in germany
[9:03] <linyos> line of sight laser links, i suppose.
[9:03] <linyos> i dunno.
[9:03] <NullAcht15> like ronja
[9:03] <NullAcht15> http://ronja.twibright.com/
[9:25] <linyos> i like how the links are basically on private property. that makes them hard to regulate.
[9:26] <linyos> and such an infrastructure really could be used for all sorts of popular applications, like file-sharing and so on.
[9:26] <linyos> internet link sharing too.
[9:27] <NullAcht15> I just read on slashdot that some telcos who offer internet connections are starting to block voip traffic...
[9:29] <NullAcht15> (This might be just one more incentive for people to look into alternatives)
[9:44] <linyos> one might download movies in one minute, with a block-caching infrastructure
[9:44] <linyos> hell, you could stream them.
[10:01] <sandos> movie in 10 minute with 10Mbit? small movie =)
[10:01] <sandos> 1 minute! =)
[10:01] <sandos> building multiple ronjas would be very expensive, no?
[10:03] <NullAcht15> might cost a few hundred dollars. But I guess it would be still feasible
[10:03] <NullAcht15> maybe directed radio-antennas (for wlan) are actually cheaper
[10:04] <NullAcht15> But, then, this is why I think people would still have to pay for their bandwidth
[10:05] <NullAcht15> I mean, most of the filesharers out would max out even a Gigabit link if it didn't cost them anything
[10:05] <NullAcht15> out = out there
[10:08] <sandos> its too bad its not possible (easily?) to use 100Mbit for ronja.. I dont know how that would work practically with nosie and stuff either
[10:09] * jokern (n=jokern@tor/session/x-b4da833b8edaaf58) has joined #freenet
[10:10] * jokern (n=jokern@tor/session/x-b4da833b8edaaf58) Quit (Client Quit)
[10:10] <NullAcht15> you could use a number of ronjas in parallel. As it is optical, there would be no interference
[10:11] <sandos> expensive/labourintensive =)
[10:11] <NullAcht15> Also, I think they actually are working on getting it to 100 MBit
[10:11] <sandos> it should be possible to use the same optics and multiple wavelenghts, no?
[10:11] <NullAcht15> hm, dunno
[10:11] <sandos> like they do with fibreoptics
[10:12] <sandos> they can anyway, dont know if its common
[10:12] <NullAcht15> well, any comments about the "civesnet" thingie I posted?
[10:13] <sandos> havent read it yet, Ill skim through it
[10:17] <NullAcht15> It certainly has a few weaknesses: The protocol overhead will be rather large and every router node that handles a lot of traffic will need a lot of main memory
[10:18] <NullAcht15> (that's because the only way to detect loops in this net is to see whether a particular did already come through)
[10:22] <NullAcht15> +packet
[10:43] <linyos> well, i figured you'd use gigabit routers and lay a few bundled ethernet cables between each property.
[10:43] <linyos> so you'd have a huge amount of bandwidth to your neighbors
[11:10] <NullAcht15> this perspective sure makes such a network look tempting, doesn't it?
[11:25] <toad_> 1Gbps is possible too
[11:25] <toad_> you can only use them for backbone though
[11:26] <toad_> they'd be too expensive (when you factor in labour) for clients
[11:26] <linyos> you can buy really fast laser links right now.
[11:26] <linyos> they cost a lot.
[11:26] <NullAcht15> When you factor in labor, I don't think they're much more expensive than 100 MBit, right?
[11:27] <toad_> linyos: yes, it's possible to route if you know the topology, however it can be a very hard/expensive problem if the network is large and complex
[11:27] <toad_> complex = varying capacity links etc
[11:27] <toad_> linyos: what makes you think that you could detect the darknet topology?
[11:27] <NullAcht15> I was trying to get at something where you route without knowing the topology...
[11:28] <toad_> ahh, so not talking darknets, talking community networks... build your own ISP, stick it to the Man
[11:28] <NullAcht15> yes, something like that
[11:28] <toad_> there are many possible routing algorithms which work for reasonable sized IP networks...
[11:29] <NullAcht15> reasonable sized meaning?
[11:29] <toad_> http://www.bristolwireless.net/
[11:29] <linyos> toad_: finding an optimum route is probably some kind of horrible np-complete problem, yeah.
[11:29] <linyos> but i'm sure there are a dozen good heuristics.
[11:29] <toad_> anyway bottom line is *you do need infrastructure* and therefore *you need somebody to own it*. yes it can be a co-op, it can be indirectly user-owned, but it cannot be *totally* distributed with all nodes equal
[11:30] <NullAcht15> hmhm
[11:30] <toad_> long distance links - inter-city - i suggest 802.11a band C (in the UK), for the time being
[11:30] <toad_> [13:56] <linyos> but it could easily travel across thousands of little links to get there, so long as you don't introduce latency at each hop.
[11:30] <toad_> you ALWAYS introduce latency on each hop
[11:30] <linyos> i mean significant latency. an ethernet switch doesn't take much time, does it?
[11:31] <toad_> and NO you do not need to pay for forwarding any more than you do on freenet
[11:31] <toad_> you DO need to pay for internet
[11:31] <toad_> quid pro quo can work in practice. money usually brings out the worst in everyone. yes, it's immensely useful, BUT you need to only use it where it really is necessary.
[11:31] <toad_> look at mojonation for how this can go wrong
[11:31] <toad_> everyone joined mojonation because they thought it'd generate hard cash
[11:31] <toad_> of course it didn't
[11:32] <linyos> when you have more demand than supply, you have to ration somehow.
[11:32] <NullAcht15> toad_: Still, if there is no money involved, then how is anybody supposed to connect to, say, australia from europe?
[11:33] <toad_> internet link sharing is sometimes allowed...
[11:33] <toad_> you have to do it at the application level of course
[11:33] <toad_> and it's precarious - get a lot of people doing it and the ISPs will start saying you're not allowed
[11:33] <toad_> and holding you responsible for other people's porn that went through your outproxy
[11:34] <toad_> linyos: Dijjer!
[11:34] <toad_> [14:47] <linyos> one might download movies in one minute, with a block-caching infrastructure
[11:34] <toad_> [14:47] <linyos> hell, you could stream them.
[11:34] <toad_> dijjer
[11:34] <NullAcht15> dijjer?
[11:34] <toad_> ronja is pretty cheap on parts, but comparable to wifi. it's expensive on labour but if you know what you're doing you can do it in a couple hours.
[11:35] <NullAcht15> ah, this dijjer...
[11:36] <toad_> linyos: using ronja or wired links probably saves you TONS of latency
[11:37] <toad_> well
[11:37] <NullAcht15> If you can cross several kilometers in one hop instead of 50....
[11:37] <toad_> you have to build from the ground up
[11:37] <toad_> international links are far future at the moment
[11:37] <toad_> you can build a good network in bristol say, and have a sustainable business model
[11:37] <toad_> then you do the same in bath
[11:38] <NullAcht15> toad_: I was hoping for some comments from you on that link I posted
[11:38] <toad_> and connect them together with initially 802.11a(C) (2W power limit so you can do it... around 20Mbps)
[11:38] <toad_> NullAcht15: I don't think we need new routing algorithms
[11:38] <toad_> NullAcht15: I think the current stuff is fine, it's a matter of *building it*
[11:38] <NullAcht15> hm
[11:38] <toad_> then you put one in newport
[11:38] <toad_> etc
[11:38] <toad_> in the long term you use a 1Gbps laser link
[11:39] <toad_> between cities
[11:39] <toad_> and internally you use 100Mbps ronja as backbone
[11:39] <toad_> and 802.11g for clients
[11:39] <toad_> at ROOFTOP LEVEL
[11:39] <toad_> if you want street level it's a lot harder
[11:39] <linyos> i looked it up, ethernet switches have a latency of like 10 microseconds
[11:39] <NullAcht15> with IP, you always need someone to give you an IP address or a range, you can't just make one up...
[11:40] <toad_> we're in the process of expanding our network from 1 real node to 4 or 5
[11:40] <toad_> NullAcht15: sure, so you have to negotiate that
[11:40] <toad_> NullAcht15: big deal, you can just use 10.x.y.z for the time being
[11:40] <toad_> that's 24 million addresses
[11:40] <toad_> well maybe 22 in reality, but it's loads
[11:40] <toad_> or you use the equivalent on IPv6
[11:41] <toad_> and then you don't need to worry about it ever again
[11:41] <toad_> we got a grant to build a nice triangle...
[11:41] <toad_> this gives us around half of bristol at rooftop level
[11:41] <toad_> well maybe more like 40%
[11:41] <toad_> but a LOT
[11:41] <NullAcht15> still, you'll need to give everyone who connects a certain range of addresses, and they'll need to give away some of that range to people that connect to them, and so on
[11:42] <toad_> essentially yeah
[11:42] <toad_> but like i said, you have infrastructure
[11:42] <NullAcht15> At some point, the ranges will run out...
[11:43] <toad_> you don't need to give everyone who connects a /24
[11:43] <toad_> you need to give first level nodes a /24, then clients run off them
[11:43] <toad_> then you have second level nodes
[11:43] <NullAcht15> You'll get the same problem as with todays internet: Ranges aren't available where they're needed and a lot of ranges are wasted where there not needed
[11:44] <toad_> you don't necessarily need second level nodes to have /16's
[11:44] <toad_> you can juggle them since you control both the first and the second level nodes
[11:44] <toad_> you *do not give clients nodes*
[11:44] <NullAcht15> Also, this requires some central authority
[11:44] <toad_> the reason for this is that directly-user-owned infrastructure doesn't work technically unless the users are all prepared to lay out for 3 separate directional links
[11:45] <toad_> they aren't, so you build somewhat hierarchical infrastructure
[11:45] <toad_> and you form a coop to own it
[11:45] <toad_> in each city
[11:45] <toad_> yes it's centralized, but it's not owned by The Corporation
[11:45] <toad_> nor is it owned by the gov
[11:46] <linyos> still, central control
[11:46] <toad_> it can be sustainable, it can provide free access (given some installation), and high bandwidth
[11:46] <toad_> right
[11:46] <NullAcht15> It doesn't make a difference whether it's the gov. or a large corpration or a large "club". It's a central authority
[11:46] <toad_> the anarchist wet dream doesn't work unless everyone pays 3x as much as they normally would in capital outlay
[11:46] <toad_> i think that's a metaphor for something ;)
[11:47] <linyos> so you just have to make some cheap 100mbps ronja devices
[11:47] <linyos> and then the capital is insignificant
[11:47] <toad_> NullAcht15: well, do you think you can get people to pay ~ $250 for 3 separate directional links?
[11:47] <toad_> linyos: you still have to install them, point them, etc. and ronja is not cheap. it's $100 for parts plus 1-70 hours for labour.
[11:47] <NullAcht15> certainly not everybody
[11:47] <toad_> ronja's great for backbone
[11:48] <linyos> that $100 is retail though, manufacturers buy wholesale.
[11:48] <toad_> (well it will be when they get 100Mbps working ;) )
[11:48] <linyos> and it's all machines these days, from making the circuit boards to sticking in the parts, anyway.
[11:48] <toad_> linyos: yes, but then you have to put them together, and you have overheads, and you have profit...
[11:48] <linyos> of course.
[11:48] <toad_> for backbone links in a community network, you can get the geeks to build them themselves
[11:49] <linyos> bbiab...
[11:49] <toad_> linyos: everything on the market as far as i know, in terms of free space optical, is seriously expensive
[11:49] <toad_> anyway, i don't share your politics; if it was possible to cheaply do it totally decentralized that'd be great, but right now it isn't
[11:50] <toad_> it *is* possible to build a high bandwidth, non-corporate, non-govt, accountable network, for the purposes of promoting local content, providing tons of bandwidth, and bridging the digital divide (in combination with grant funding; refurb PCs and free training)
[11:50] <toad_> and we *are doing this*
[11:51] <toad_> http://www.bristolwireless.net/wiki/
[11:52] <NullAcht15> a few years ago, people in my city tried something similar. But somehow the project died away...
[11:52] <toad_> current plans are to use wifidog to close the network to the extent that you have to create an account, provide 128kbps to everyone for free, and charge small amounts (via bank standing order or paypal) for more bandwidth
[11:52] <toad_> NullAcht15: yeah, most of them have
[11:53] <toad_> there's one in new york that's going well
[11:53] <toad_> there's one in i think berlin which is on the totally distributed model
[11:53] <toad_> but has no routing
[11:53] <toad_> i think it's just a way of getting people to put in more hotspots
[11:53] <toad_> there's Bristol Wireless, which is the only one in the whole of britain that's really doing well
[11:53] <toad_> there's a big one in northern europe somewhere
[11:54] <toad_> sustainability is probably the biggest issue
[11:55] <toad_> most of BW's income at the moment is from grants; we are trying to create new revenue streams
[11:55] <toad_> we are reselling bandwidth (with a wifi failover) to a couple (more soon hopefully) of community centers at business rates
[11:55] <NullAcht15> bandwidth with internet access?
[11:55] <toad_> they're people we already knew, and they were already on the network (which is useful to them in itself)
[11:56] <toad_> yeah, internet bandwidth
[11:56] <toad_> we run a 20:1 DSL in each one, and put it on the network
[11:56] <toad_> if the local DSL fails, we instantly failover to the other one across the wifi
[11:56] <toad_> we use the spare bandwidth from each via traffic shaping to provide for the rest of the network
[11:56] <NullAcht15> But that gives new IP addresses every time, doesn't it?
[11:56] <toad_> hmmm?
[11:57] <toad_> we have static internal IPs
[11:57] <toad_> externally, I dunno, it's a biz class connection so it might have static IP
[11:57] <toad_> if it failover's then it'd be something else admittedly
[11:57] <NullAcht15> You need some official IPs to connect to _the_ Internet, don't you?
[11:57] <NullAcht15> Or are you using NAT for this?
[11:57] <toad_> getting a big block of real IPs is a big issue but i think when we're big enough to get 1:1 SDSL, it will be solvable
[11:58] <toad_> NullAcht15: if it failovers, it'd probably be NATted
[11:58] <toad_> if it's local, it uses the local IP
[11:58] <toad_> the non-paying clients don't get an IP afaik
[11:58] <NullAcht15> Even then, a switch of dsl-connections means TCP connections will be interrupted...
[11:58] <toad_> sure
[11:59] <toad_> but it doesn't interrupt web or email much
[11:59] <toad_> which is mostly what these places use
[11:59] <toad_> before they got us, they had major problems with DSL reliability
[11:59] <NullAcht15> the problem with Wavehan (same thing here in Hannover, Northern Germany) was, imo, that ordinarily, there would be no internet access through it at all...
[11:59] <toad_> we ran them for a week because their DSL was down for some reason
[12:00] <toad_> well, you need two things
[12:00] <toad_> you need routing
[12:00] <toad_> and you need internet
[12:00] <toad_> the former requires infrastructure - mostly capital outlay
[12:00] <toad_> the latter requires ongoing cash
[12:00] <toad_> either you sell internet (as well as possibly doing some free internet), or you sell services
[12:00] <toad_> you can get grant funding for capital outlay very often
[12:01] <toad_> including putting people onto the network (which has to be done rooftop level so is relatively expensive - as much as $300 sometimes)
[12:01] <toad_> you can sometimes get whole-package funding - put them on the network, give them a refurb PC, and train them how to use it
[12:02] <linyos> the hard (or expensive..) part is simply getting to somewhere where you can buy $10-15/mbps internet transit
[12:02] <toad_> linyos: once you have the scale to buy SDSL, you can get bandwidth very cheaply
[12:02] <toad_> because you can heavily contend it
[12:02] <toad_> lets say $700/month for 1:1 2048/2048 SDSL
[12:03] <toad_> now, how much does that make per half megabit at 50:1?
[12:03] <toad_> 700/(50*4) =
[12:03] <toad_> ~
[12:03] <toad_> $3/mo
[12:03] <toad_> the trick is finding somebody who's willing to give you resellable bandwidth
[12:03] <linyos> 50:1 is a typical oversubscription rate?
[12:04] <toad_> fortunately The Phone Coop is very friendly to BW, partly because it's also a coop
[12:04] <toad_> linyos: yeah
[12:04] <toad_> linyos: 50:1 for domestic, 20:1 for business
[12:04] <toad_> a LOT of people on broadband don't use the net much
[12:05] <toad_> unfortunately we also get a lot of people with laptops etc who manage to get a weak signal at street level or through a window
[12:06] <toad_> they drag the whole thing down because they get a low-bandwidth timeslice
[12:06] <toad_> this is why it can't be a totally open network; we're going to make them create an account so we can keep tabs on them
[12:06] <toad_> and if necessary send them a page saying "Fuck Off Or Get A Proper Aerial... call XXXX and we'll put one in for you"
[12:07] <toad_> so to balance out all the geeks etc who will use tons of bandwidth, we need to get lots of grants and put lots of Genuinely Disadvantaged People on the network... :)
[12:07] <toad_> which is exactly what we have been doing, so no problem
[12:08] <toad_> also obviously putting in 1TB squid servers is on the todo list
[12:08] <toad_> on the gateways
[12:08] <linyos> if you can afford a computer can't you afford dialup?
[12:08] <toad_> linyos: we give them a computer
[12:08] <linyos> oh, cool.
[12:08] <toad_> linyos: and a LOT of people don't have a landline phone
[12:09] <toad_> they have PAYT mobiles, but no landline
[12:09] <toad_> (and no credit!)
[12:10] <toad_> now, i propose to get on with some work
[12:10] <toad_> any reason i should be further distracted? :)
[12:10] <toad_> linyos: also, broadband is a lot more useful than dialup
[12:11] <linyos> quite so.
[12:11] <toad_> e.g. you can VoIP outdial to just about any landline in europe for 5 euros up front and no per call charges
[12:12] <toad_> http://www.voipbuster.com/
[12:12] <toad_> you can use KPhone or a hardware SIP phone with it
[12:13] <linyos> my interest here is that it may be possible, by laying down tons of ethernet and wireless links, to create a technically superior network, with new, exciting applications, that also happens to be highly decentralized and resistant to regulatory control.
[12:13] <toad_> okay not anywhere... the states, several european countries, and you can pay for other places... and it's standards compliant, unlike skype
[12:13] <linyos> in other words, something that people could actually be persuaded to participate in.
[12:13] <linyos> (ordinary people)
[12:14] <toad_> linyos: i agree, but i think you will always need some infrastructure... or *at least* when you don't have such a network, the first stage is to build infrastructure in order to get sufficient penetration to seed off
[12:14] <toad_> new technology, such as multi-antenna 802.11whatever, or really cheap ronja, may change the situation
[12:15] <toad_> i see the "resistant to regulatory control" as occurring primarily at the software/metanetworking level
[12:15] <NullAcht15> linyos: care to join me and see how this protocol proposal may turn out?
[12:15] <toad_> but it has to be a good thing to take it out of the hands of the megacorps - even if you end up setting up structures of your own
[12:16] <linyos> toad_: no, it's a lot easier to start up the "national trusted safe internet community initiative" than it is to regulate millions of ethernet cables buried between private properties.
[12:16] <linyos> people would just say "fuck off, we like our network just fine thank you"
[12:16] <toad_> linyos: there is no reason you need One Big Company controlling all the network
[12:17] <linyos> toad_: even if it's N smaller players, they can be regulated.
[12:17] <toad_> linyos: running ethernet down the street is cool, but it only works if there are people on the street you want to connect to
[12:17] <linyos> whereas a cable running through my back yard is harder to regulate.
[12:17] <toad_> linyos: if it's traceable, it will always be regulated
[12:17] <linyos> toad_: yeah, which is why the thing would have to be popular---ie, an obvious upgrade to the internet connections people already have.
[12:17] <linyos> to make them faster or better or cheaper or whatever.
[12:18] <toad_> linyos: which is what we can provide with a community wireless WAN
[12:18] <toad_> also it helps to restore the community, which can only be a good thing
[12:19] <linyos> i like wireless just fine. but when you have centralized infrastructure, you have a point of weakness.
[12:19] <toad_> linyos: lots of bandwidth within the network; fast filesharing etc with your peers
[12:19] <linyos> sure.
[12:19] <toad_> linyos: well, that's what we can do with current technology and current (more or less zero) penetration
[12:19] <linyos> i respect it as a first step.
[12:19] <toad_> if in future we have more penetration of community networks people will be more interested in their own networking
[12:19] <toad_> but right now, we need infrastructure; anything else is just too expensive
[12:20] <toad_> the future is bright though
[12:20] <toad_> but imho we will always need some infrastructure if only for intercontinental links :|
[12:21] <toad_> so anyway
[12:21] <toad_> now i really need to do some work
[12:21] <toad_> ttyl
[12:21] <linyos> well, that's a different thing. if governments closed down those links they would just fragment the network.
[12:22] <linyos> later.
[12:22] <toad_> linyos: you need something different if the government is hostile yes
[12:22] <toad_> that's where "darknet" comes in
[12:23] <toad_> i'm still here, i just can't talk about non-freenet things much :)
[12:23] <linyos> toad_: no, i mean that if everyone buys their networking through some set of ISPs, those ISPs can be bullied into imposing my evil proxies, or any kind of orwellian thing. and then it is just there.
[12:23] <linyos> whereas fragmenting a decentralized network is less of a victory.
[12:24] <sleon|tuX> linyos, it is always possible to avoid filters and proxies
[12:24] <toad_> linyos: IMHO it is better for it to be a set of citywide networks which are locally owned than a few international backbone providers
[12:24] <linyos> they might just say, it's not worth it, we'd like to have control but it's not realizable.
[12:24] <linyos> indeed
[12:24] <toad_> but yes i can see what you're saying, and it's a matter of user-owned darknet links
[12:25] <toad_> which might be PDAs with wifi, might be sneakernet, and might be wifi or ronja
[12:25] <linyos> sleon|tuX: not if they're designed correctly... we can argue this privately if you want.
[12:25] <sleon|tuX> linyos, yep
[12:26] <toad_> you may as well have it out publicly as far as i'm concerned
[12:27] <linyos> For example, imagine that residential Internet connections were locked
[12:27] <linyos> behind a web proxy. When someone views a web page, the proxy uploads
[12:27] <linyos> the page's data to a database and sends a (user, timestamp, uri,
[12:27] <linyos> page-hash) record to the police. The police then run algorithms over
[12:27] <linyos> these records to sniff out unusual or forbidden activity.
[12:27] <linyos> so, how would you work around such a system?
[12:27] <sleon|tuX> linyos, i could use normal words
[12:27] <sleon|tuX> linyos, or even centences and combinations of them
[12:27] <sleon|tuX> linyos, for data transfer
[12:28] * Hory (n=Miranda@82.78.27.85) has joined #FreeNet
[12:28] <sleon|tuX> linyos, so it is not noticeable at all
[12:28] <linyos> sleon|tuX: yes, but will your activity blend in? will the pages you're putting your stego-words in look normal? will the times and frequency at which you read them seem normal? will the servers serving the pages seem legitimate?
[12:28] <sleon|tuX> linyos, it is also possible to do it with html tags and words combinations :)
[12:28] <sleon|tuX> linyos, i hope
[12:29] <sleon|tuX> linyos, the information could be distributed along pages
[12:29] <linyos> remember, they are running global statistical analyses here. they're looking at a background of normal activity, and trying to pick out anything that deviates from it.
[12:29] <sleon|tuX> linyos, it is good to place such info by big servers hosting thousands pages
[12:30] <sleon|tuX> linyos, so you can get information by acceing pages in predefined order and only once
[12:30] <sleon|tuX> i think there are ways
[12:31] <toad_> sure
[12:31] <toad_> there are covert channels
[12:31] <toad_> but they are all LOW BANDWIDTH
[12:31] <sleon|tuX> all analyses and checks can be done only automated on a such huge information masses
[12:32] <toad_> and in general eventually they will be found out
[12:32] <sleon|tuX> so there will be always ways to fool them
[12:32] <toad_> yes, and the automation will be very expensive, which is why it hasn't been done yet, but it will be done, if the chinese get pissed off enough
[12:32] <sleon|tuX> when this problem will get solved
[12:32] <linyos> sleon|tuX: you pick out the most suspicious activity and have people review the web histories for signs of illicit use.
[12:32] <sleon|tuX> then touring thesis will be overthrown
[12:33] <toad_> sleon|tuX: turing's thesis?
[12:33] <sleon|tuX> linyos, how can you pick it up?
[12:33] <sleon|tuX> linyos, requests for random pages ?
[12:33] <sleon|tuX> linyos, always new pages , with normal content
[12:33] <sleon|tuX> without any special words in them
[12:33] <toad_> sleon|tuX: that in itself is detectable
[12:33] <sleon|tuX> how should it get suspicious?
[12:33] <sleon|tuX> toad_, interesting
[12:33] <linyos> (a) will other people ever download these pages? if so, that's a red flag.
[12:33] <sleon|tuX> and how ?
[12:34] <sleon|tuX> linyos, what do you mean ?
[12:34] <linyos> sleon|tuX: i mean, where would these pages be?
[12:34] <linyos> on what server?
[12:34] <sleon|tuX> linyos, distributed along lots of servers
[12:34] <sleon|tuX> cnn.com child games
[12:34] <sleon|tuX> and such things
[12:34] <sleon|tuX> common places
[12:35] <linyos> and how would the data be hidden?
[12:35] <sleon|tuX> you wirte an article for example and then the n'th word tells you information
[12:35] <toad_> <pre> Content-Type: message/external-body; access-type=URL;
[12:35] <toad_> URL*0="ftp://";
[12:35] <toad_> URL*1="cs.utk.edu/pub/moore/bulk-mailer/bulk-mailer.tar"
[12:35] <toad_> </pre>
[12:35] <sleon|tuX> on where to look next
[12:35] <toad_> woah
[12:35] <toad_> http://www.faqs.org/rfcs/rfc2231.html
[12:36] <linyos> sleon|tuX: try doing that on any large scale. it's impossible.
[12:36] <linyos> sleon|tuX: like toad says, there will always be ways to get a little information through.
[12:36] <sleon|tuX> linyos, it is very slow but when information is distributed that way it is imposible to detect it
[12:52] <toad_> sleon|tuX: well, personally i want a scalable system that can take freedom of speech to the masses, rather than something that's only usable for 0.1% of the population
[12:54] <linyos> my interest here is just that local, private connections are much, much harder to regulate than global, public infrastructure.
[12:55] <toad_> indeed
[12:55] <toad_> now, how many geeks live on your street?
[12:55] <linyos> so i am rather enthusiastic about anything that promises to create them
[12:55] <linyos> none that i know of.
[12:55] <toad_> exactly
[12:55] <linyos> but
[12:56] <linyos> "pay 100 bucks and get wired on the SUPER-LOCAL-NET like all the hip people!!!"
[12:56] <linyos> is not exactly something reserved for an elite few
[12:56] <toad_> right
[12:56] <toad_> so what we need is a $25 ronja device
[12:57] <toad_> also we need a network of insured enthusiasts who are willing to go up and install these on people's rooftops for free
[12:57] <linyos> and lots of gigabit ethernet cable too!
[12:57] <linyos> faster.
[12:57] <linyos> that would help, yeah.
[12:58] <toad_> also i note that anything over 10Mbps on ronja will have to be laser because ordinary diodes aren't fast enough
[12:58] <toad_> not necessarily more expensive, but obviously it is laser and has laser issues
[12:58] <toad_> which go away to a large degree because you're expanding the beam like crazy with optics
[13:02] * sleon|tuX (i=test@85.180.50.190) Quit (Read error: 104 (Connection reset by peer))
[13:03] <toad_> is a 4 byte integer enough for a DBR's period and offset?
[13:03] <toad_> yes
[13:03] <toad_> as long as it's in seconds
[13:03] <toad_> i don't think we should support milliseconds on DBRs !
[13:04] <toad_> oh
[13:04] <toad_> should we support calendar units?
[13:04] <toad_> as opposed to update-every-X-seconds, update-every-calendar-month ?
[13:04] <linyos> encode the DBRs into the keys!
[13:05] <toad_> linyos: this is metadata i'm talking here
[13:05] <toad_> and we will look into updatable keys... later
[13:05] <linyos> so SSK@BLHALBHABLAH automatically changes into SSK@BLAHBLAHBLAH/DATE based on the current date
[13:05] <toad_> not for 0.7.0
[13:05] <toad_> yes
[13:05] <toad_> that's how DBRs work
[13:05] <toad_> it'll be at the metadata level
[13:06] <linyos> i mean, have a flag in the key itself that instructs the client to append the date
[13:06] <linyos> then there is no need for metadata at all.
[13:06] <linyos> (for dbrs)
[13:06] <toad_> ugh
[13:06] <toad_> then it would be in the node itself
[13:06] <toad_> that sucks
[13:06] <linyos> it saves a redirect, right?
[13:06] <toad_> major layering violation
[13:06] <toad_> eh
[13:06] <toad_> how would it save a redirect?
[13:07] <toad_> ohhh
[13:07] <toad_> i see
[13:07] <toad_> hrrrrrm
[13:07] <toad_> well we can put a lot of stuff in keys if we want to
[13:07] <toad_> but it does tend to make them rather long
[13:08] <linyos> have them be a concatenation of optional binary segments, each one prefixed by its type identifier
[13:09] <linyos> (that allows for future additions)
[13:09] <toad_> that's a good idea
[13:09] <toad_> i've sent off to the list re that
[13:09] <toad_> maybe i'll get some feedback...
[13:12] <NullAcht15> toad_: It would be great if, in 0.7, we'd have have an easy way to browse through all past issues of a dbr site
[13:12] <toad_> NullAcht15: it would be great in 0.5 too
[13:12] <toad_> NullAcht15: you volunteering to code it ? :)
[13:13] <NullAcht15> hm...
[13:13] <toad_> 0.7's fproxy will be somewhat different, and we may separate it into a separate module
[13:13] <toad_> to make it easier for new coders to hack on it
[13:14] <NullAcht15> well, it would be a lot easier to code (or do manually, for that matter), if for every edition, the edition number would rise by one, starting at 0 (or 1)
[13:14] <NullAcht15> Right now, if I got that right, it increases by the update interval for that site
[13:15] <toad_> i don't see that it would be much easier to code
[13:15] <toad_> it would however be easier to do manually
[13:15] <NullAcht15> hm...
[13:15] <NullAcht15> I could imagine it like this:
[13:16] <toad_> we need to decide how the new DBRs are going to work, I suggest you reply to me on devl
[13:16] <toad_> i suppose if they are 20051021181900 then yes it will be harder to do the above in code
[13:17] <toad_> but still fairly easy (there are calendar classes in java)
[13:17] <NullAcht15> If you want to browse editions, you have a narrow frame on top of the site you're viewing (provided by fproxy) which offers you information about what editions are availiable, from what dates they come and links to those editions (which would open in the main frame)
[13:19] <NullAcht15> I'm not subscribed to devl
[13:19] <NullAcht15> only to announce and tech
[13:25] <toad_> well i sent it to devl and support :|
[13:25] <toad_> i'll bounce you it if you want
[13:25] <toad_> NullAcht15: you can do that with DBRs too
[13:25] <toad_> ?date=
[13:26] <toad_> should a non-redundant splitfile be an error normally (unless it's a whacky splitfile pieced together from different files)?
[13:29] <NullAcht15> toad_: yes, I know, but right now it's highly impractical
[13:29] <toad_> ?date=-1month is highly impractical?
[13:29] <linyos> NullAcht15: really, site publishers should provide those links themselves, if they want to.
[13:30] <NullAcht15> linyos: They should, but I don't know many who do
[13:30] <NullAcht15> And even then, that wouldn't help you if you wanted to browse a dbr site taht has been abandoned
[13:31] <NullAcht15> toad_: that's possible? I didn't know about that...
[13:31] <linyos> fair enough. no reason why fproxy shouldn't have the feature.
[13:32] <toad_> NullAcht15: :|
[13:32] <toad_> NullAcht15: like most free software projects freenet has a documentation problem
[13:32] <toad_> NullAcht15: oh, I offered you SVN access (you too linyos) earlier... does anyone on this channel want SVN access?
[13:32] <toad_> (SVN ~= CVS)
[13:34] * linyos tries to figure out what politeness dictates in this circumstance...
[13:34] <NullAcht15> toad_: You realize I am the guy who bugged you about splitting fred into different parts on tech?
[13:34] <toad_> NullAcht15: no, i didn't realize that
[13:35] <toad_> NullAcht15: do you want SVN access?
[13:35] <NullAcht15> well, yes
[13:35] <toad_> you're in a better position to do something about it if you have it
[13:35] <linyos> ok fine, make an account for me too.
[13:35] <toad_> okay, i need your email address, and a nym, and a password (gpg mail me at toad@amphibian.dyndns.org, i sign all my messages so you can get my key easily enough)
[13:35] <linyos> maybe i will find something to do with it.
[13:35] <NullAcht15> I was thinking about doing some documenting...
[13:35] <toad_> linyos: you need svn access to get to the website
[13:36] <toad_> i suggest you generate a password randomly
[13:36] <toad_> dd if=/dev/random bs=32 count=1 | md5sum :)
[13:37] <toad_> okay, 2 bytes or 4 bytes for "number of blocks"...
[13:37] <toad_> 2 bytes -> 65,535 blocks maximum => 32kB * 64k = 2GB maximum file size
[13:37] <NullAcht15> nice idea for generating random passws...
[13:37] <toad_> okay lets have 4 bytes then :)
[13:38] <toad_> NullAcht15: it is if either a) you have an exceptionally good memory, or are going to use this password REALLY often, or b) you keep a file of them somewhere on a supposedly secure system
[13:38] * toad_ is b)
[13:39] <NullAcht15> toad_: well, I am using kwallet, hoping it is secure enough...
[13:39] <toad_> so what's the limit...
[13:39] <toad_> 2 bytes -> 2^(15+16) = 2GB
[13:39] <toad_> 3 bytes -> 2^(15+24) = 2^39 ~= 512GB
[13:39] <toad_> 4 bytes -> 2^(15+32) = 2^47 = 128TB
[13:40] * toad_ thinks 4 bytes, nothing like planning ahead :)
[13:40] <NullAcht15> erhm, waht are you talking about right now?
[13:40] <toad_> maximum splitfile size
[13:40] <NullAcht15> ah
[13:40] <toad_> the number of bytes to use for the "number of blocks" variable
[13:41] <NullAcht15> toad_: Where can I get your PGP/GPG public key?
[13:41] <linyos> you should use the linus torvalds svn, it is pretty cool.
[13:41] <toad_> of course if i go for 4 bytes, then i'll have to use 5 bytes for the "length of subsection"...
[13:42] <toad_> NullAcht15: it's on the servers, my ID is on my messages...
[13:42] <toad_> just fetch it
[13:42] <toad_> and then try to verify one of my messages
[13:42] <toad_> if it verifies, you have the right key
[13:42] <toad_> gpg --keyserver wwwkeys.pgp.net --recv-keys 6554A22D
[13:42] <toad_> i think
[13:43] <toad_> okay, 512GB or 128TB file size limit? anybody?
[13:43] <linyos> toad_: who cares, whatever you like
[13:43] <toad_> actuall
[13:44] <toad_> actually 64TB as it's an int...
[13:44] * toad_ does that
[13:46] <toad_> and we'll define a soft limit...
[13:46] <toad_> based on memory usage
[13:46] <toad_> lets say 1 million blocks for now
[13:46] <toad_> nah
[13:47] <toad_> 100,000 blocks
[13:47] <NullAcht15> Is it likely that this whole thing will be redone again before 1.0 (or even 0.9)? If yes, and if the 4 bytes use some unnecessary resources, I'd say go with 3
[13:47] <toad_> not before 1.0 no
[13:47] <NullAcht15> Although, one byte - that won't really make much difference, will it?
[13:47] <toad_> no, it's one byte on the whole splitfile manifest
[13:48] <toad_> it really doesn't make any difference
[13:48] <NullAcht15> which is always transferred as a 32 KiB block anyway....
[13:48] <NullAcht15> (right?)
[13:48] <toad_> it's just that there's a practical limit which is below the theoretical one, if you have a splitfile as part of a manifest
[13:48] <toad_> which is fair enough really
[13:49] <toad_> NullAcht15: yes... well, maybe not
[13:49] <toad_> we might decide to have 1kB SSKs, for example
[13:49] <toad_> (these would gel rather well with doing pub/sub over passive requests)
[13:55] <nextgens> hi
[13:55] <toad_> hi nextgens
[13:56] * nextgens hasn't borked the website ;)
[13:56] <nextgens> not my fault this time :P :)
[13:57] <toad_> any idea what happened to the .mng then?
[13:57] <nextgens> toad_: I've got an Inet connection at home now :)
[13:57] <nextgens> no, I don't know
[13:57] <toad_> yay!
[13:58] <nextgens> we will share a 512k/128 and we are almost 38 :-D
[13:58] <nextgens> but it's better than nothing :)
[13:58] <nextgens> and I'm in charge of the QoS ^-^
[13:59] <nextgens> toad_: anything usefull I should do ?
[13:59] <nextgens> on my todo list, it remains :
[13:59] <nextgens> 1) move the noderef script to emu
[13:59] <toad_> nextgens: age or number? :|
[13:59] <nextgens> 2) convert old ml archive to mailman
[13:59] <nextgens> toad_: number :'(
[13:59] <toad_> nextgens: looks good to me
[14:00] <nextgens> anything else ?
[14:00] <toad_> nextgens: so you have to block filesharing? :(
[14:00] <toad_> nextgens: not that i can think of right now
[14:00] <nextgens> yes, and even more than that : there won't be any NAT
[14:00] <nextgens> only a proxy
[14:00] <toad_> nextgens: there's loads of other sources for old mailing list archives
[14:01] <nextgens> and b/w limits, quotas,...
[14:01] <toad_> :|
[14:01] <linyos> no you don't, just prioritize all the interactive traffic with iptables/tc
[14:01] <toad_> but you have ssh, right?
[14:01] <linyos> then you can have all the filesharing you want.
[14:01] <nextgens> yes
[14:01] <toad_> linyos: and hope nobody runs it over port 80 :|
[14:01] <nextgens> and as I said, I will be the admin ;)
[14:01] <toad_> oh well, you can use my node :)
[14:01] <linyos> heh, i guess.
[14:01] <toad_> and emu
[14:01] <toad_> and dodo
[14:02] <nextgens> cool
[14:02] <toad_> dodo will be dropped very soon
[14:02] <toad_> if it hasn't already been dropped
[14:02] <toad_> i have a backup of the seed* files if you need it
[14:02] <nextgens> 'k
[14:02] <toad_> i _think_ i have a backup of the scripts too
[14:02] <nextgens> dod is still there
[14:03] <nextgens> I'm logged on it
[14:03] <nextgens> I'll do a full backup of /home on emu
[14:03] <toad_> i gave up on pub/sub for now, i think the architecture is wrong (something with passive requests, 1kB SSKs and maybe TUKs, possibly with some minor extensions to the former, is probably better)
[14:03] <nextgens> just in case ;)
[14:03] <nextgens> I saw the ml
[14:03] <toad_> so i'm working on metadata and related stuff
[14:03] <toad_> nextgens: 3 bytes or 4 bytes for "number of splitfile chunks" ?
[14:04] <toad_> 512GB or 128TB maximum file size? ;)
[14:04] <NullAcht15> toad_: hm, for some reason my mail program doesn't want me to encrypt the mail with that public key...
[14:04] <toad_> NullAcht15: eek
[14:04] <toad_> NullAcht15: strange
[14:04] <nextgens> 3 will be enough IMHO
[14:05] <nextgens> toad_: .8 will be out before we need 128TB IMHO
[14:05] <toad_> nextgens: i think i'm doing 4 anyway
[14:05] <nextgens> ;)
[14:05] <nextgens> why ?
[14:05] <toad_> nextgens: well yeah but we don't want to have to change it (rather than extending it) unnecessarily
[14:05] <toad_> it IS possible to do backwards compatibility with a new format, but it's preferable not to need to
[14:05] <toad_> nextgens: good practice
[14:06] <toad_> i'm putting in a soft limit of 100,000 to prevent memory DoS though :)
[14:06] <nextgens> agreed, seen like that
[14:07] <toad_> hmmm
[14:08] <toad_> is a 256 byte filename component limit a problem?
[14:10] <toad_> NullAcht15: recommend you do a test commit
[14:10] <NullAcht15> toad_: Server was emu.freenetproject.org?
[14:11] <toad_> https://emu.freenetproject.org/svn/trunk
[14:11] <NullAcht15> Which protocol and which repository?
[14:11] <nextgens> NullAcht15: why was ?
[14:11] <NullAcht15> is...
[14:11] <nextgens> :)
[14:11] <toad_> https/dav
[14:12] <toad_> add a newline to a file somewhere or something
[14:14] <toad_> okay, we DO want names to be in UTF8 in the underlying metadata, right?
[14:15] <toad_> they can be converted on the way in from fproxy if necessary
[14:15] <toad_> filenames that is
[14:15] <toad_> we definitely want to support chinese/korean/greek/russian names in our filenames
[14:15] <toad_> at least in theory
[14:19] <NullAcht15> toad_: Did a test commit
[14:20] <nextgens> 20:22:59?nextgens?~ ?svn log -v --limit 1 https://emu.freenetproject.org/svn/trunk
[14:20] <nextgens> ------------------------------------------------------------------------
[14:20] <nextgens> r7446 | NullAcht15 | 2005-10-22 20:24:08 +0200 (sam, 22 oct 2005) | 2 lines
[14:20] <nextgens> Chemins modifi?s :
[14:20] <nextgens> A /trunk/test
[14:20] <nextgens> A /trunk/test/README.test
[14:20] <nextgens> Just a test commit
[14:20] <nextgens> ------------------------------------------------------------------------
[14:20] <nextgens> sorry
[14:20] <toad_> okay it just showed up on the cvs list
[14:21] <toad_> oh and your name is made lowercase by the mailout scripts
[14:21] <NullAcht15> can that create any problems?
[14:21] <toad_> i doubt it
[14:22] <toad_> your commit mails get sent out from nullacht15@freenetproject.org
[14:22] <toad_> that email address redirects to your real email address, so people can reply to your commits
[14:23] <NullAcht15> hm...
[14:23] <NullAcht15> does /trunk/freenet/src contain the current 0.7 code?
[14:23] <nextgens> yes
[14:24] <nextgens> has anyone tried to register a new user on mantis ?
[14:25] * nextgens is not convinced by it
[14:25] <toad_> okay
[14:25] <toad_> nullacht15 and conrads added
[14:25] <toad_> yay
[14:26] <toad_> we already had nextgens, toad, ian, sleon, edt
[14:28] <nextgens> maybe we should remove /var/svn/freenet/
[14:28] <nextgens> the old ssh+svn repository
[14:28] <toad_> probably a good idea
[14:28] <toad_> is it part of a package or anything?
[14:28] <toad_> is there an official way to remove it?
[14:28] <toad_> or can we just rm -R
[14:29] <toad_> yay, i've written the metadata parser
[14:31] <NullAcht15> I'm a complete newbie in Java. I noticed there are some javadoc comments throughout the code. How do I turn those into javadoc documentation. (I'm still of the opinion that that's no substitute for real documentation - but it looks like this would be the best I can get in terms of docs right now...)
[14:31] <nextgens> no, it could be removed with rm
[14:31] * nextgens will do
[14:32] <nextgens> 3) creating backups
[14:32] <nextgens> i forgot it on my todo
[14:32] <toad_> NullAcht15: not sure
[14:33] <toad_> ant javadoc for the 0.5/0.6 branches
[14:33] <toad_> not sure about the 0.7
[14:33] <nextgens> toad_: have you considered using svk to ensure that the main repository hasn't been compromised ?
[14:33] <NullAcht15> hm, there aren't that many of them anyway. I wa looking at the single file taken from i2p first, which has lots of javadoc...
[14:33] <toad_> nextgens: is svk compatible with SVN? i thought it changed everything except the very bottom layer?
[14:34] <toad_> depends on the file, a lot of the code does have javadocs
[14:34] <toad_> the dijjer stuff generally doesn'y
[14:34] <toad_> t
[14:34] <toad_> the stuff i write new often does
[14:34] <nextgens> toad_: no as you can use a svn repo. as a source ... building a local svk mirror
[14:35] <toad_> nextgens: basically all we need for data integrity is a way to establish that the mailouts reflect the reality
[14:35] <nextgens> and then, merge changes using svk between the online and the offline copy ...
[14:35] <nextgens> yes
[14:35] <nextgens> :|
[14:35] <toad_> i would have thought that would be pretty easy as a script
[14:35] <nextgens> but it's also a convenient way of backuping history
[14:35] <toad_> since the mailouts include *full* diffs
[14:36] <toad_> well
[14:36] <toad_> it'd certainly be nice to have some other options, yeah
[14:36] <toad_> but if the SVN is the master, how easy would it be to merge people's SVK stuff?
[14:36] <toad_> unless it was a 2-way gateway...
[14:37] <toad_> it'd be nice to have some mirrors anyway
[14:37] <toad_> actually i need a mirror
[14:37] <toad_> my backup scripts include the freenet repository
[14:37] <toad_> or rather they did while it was CVS
[14:37] <toad_> now the SVN repo is rather larger
[14:37] <toad_> i think
[14:38] <toad_> so ideally i'd want to keep something synced
[14:38] <toad_> otherwise i suppose i could rsync it...
[14:38] <nextgens> no
[14:38] <nextgens> as it won't be atomic
[14:38] <toad_> svn repo is 230MB last time i looked
[14:38] <toad_> nextgens: ahh, true
[14:38] <nextgens> toad_: svk is THE solution IMHO
[14:38] <toad_> so what's my best option?
[14:38] <nextgens> toad_: svk is THE solution IMHO
[14:38] <toad_> just for backups?
[14:39] <nextgens> yes, and integrity checking
[14:39] <toad_> nextgens: if you want to set it up, with the master still being SVN, that would be very cool
[14:39] <nextgens> each time you do merge, it computes a hash of the diff
[14:39] <nextgens> I can, but where ?
[14:39] <toad_> nextgens: if you want me to spend any project resources i.e. time on it, you'll have to convince ian
[14:39] <nextgens> :| I could do it
[14:40] <nextgens> but I think that doing it on emu is silly
[14:40] <toad_> nextgens: either a) run it on amphibian, b) run it on emu and ask ian afterwards, or c) ask people on the lists, they often have hosting capabilities
[14:40] <nextgens> (same disk, same computer, ... useless)
[14:40] <toad_> amphibian probably has enough bandwidth for that
[14:40] <toad_> i doubt it'd get much usage
[14:41] <nextgens> well, the first sync will be really painfull and long but afterwards it's only diffs
[14:41] <toad_> obviously i'd want a command to get an atomic copy of the repo
[14:41] <toad_> so i could back it up
[14:41] <toad_> well you're welcome to run it on amphibian if you want
[14:42] <toad_> otherwise i'm sure we can find some hosting somewhere
[14:42] <toad_> i've had a few offers in relation to the mailing lists
[14:42] <nextgens> 'k
[14:42] <toad_> certainly there should be a mirror on amphibian for backup purposes
[14:42] <nextgens> I finish what I'm doing and will launch it
[14:42] <nextgens> do you have enough space ?
[14:42] <toad_> if you're going to do a major transfer, it would be nice if you could warn me and ideally bwlimit it to say 100K down 10K up
[14:43] <nextgens> ok
[14:43] <toad_> hmmm
[14:43] <toad_> should I break trunk, or make a branch, for the metadata work, or should i just not commit it yet?
[14:43] <nextgens> break trunk
[14:44] <nextgens> we could still checkout and old compilling version ;)
[14:45] <NullAcht15> I think right now there isn't much reason to worry about breaking trunk. You might rethink that, though, when 0.7 reaches "late alpha" or "early beta"
[14:45] <toad_> well
[14:45] <toad_> anything that other folks commit will probably be in roughly the same area as what i'm doing right now, so i suppose...
[14:45] * toad_ commits
[14:46] <toad_> i like svn's instant diffs and fast commits :)
[14:46] <NullAcht15> or are there a number of people who regularly check out the code and compile it, expecting it to work?
[14:46] <toad_> NullAcht15: not that i know of
[14:47] <nextgens> not on .7 for sure
[14:49] <toad_> i do think it was right to do pub/sub on a branch though
[14:49] <toad_> especially as i think that was a bit of a blind alley now code-wise :)
[14:57] <NullAcht15> toad_: If I wanted to get an overview over how the current code works, where and how should I start to read?
[14:58] <toad_> NullAcht15: hmmm
[14:58] <toad_> that's a good question
[14:58] <toad_> probably freenet.node
[14:58] <toad_> the package, that is
[14:58] <toad_> Node.java is sort of the core
[14:58] <toad_> freenet.client is the new metadata handling stuff i'm doing
[14:58] <toad_> freenet.crypt is crypto
[14:59] <toad_> (from the old code)
[14:59] <toad_> freenet.io, freenet.io.comm, freenet.io.xfer are stuff we inherited from dijjer
[14:59] <toad_> stuff for shoving packets
[14:59] <toad_> freenet.keys and freenet.store are what they say
[14:59] <toad_> support is miscellaneous classes like LRUHashtable
[15:00] <toad_> net.i2p.util is NativeBigInteger, and org.spaceroots.mantissa is the Mersenne Twisters (fast PRNG) implementation
[15:00] <NullAcht15> hm...
[15:00] <toad_> test is old stuff, mostly testing the dijjer code
[15:00] <nextgens> we are not using Yarrow anymore ?
[15:01] <toad_> Yarrow is a hard RNG
[15:01] <toad_> Yarrow is a hard cryptographic PRNG
[15:01] <NullAcht15> What's PRNG? Pseudo Random number generator?
[15:01] <toad_> Mersenne Twisters are fast reasonable quality PRNGs for things like padding
[15:01] <toad_> yup
[15:02] <toad_> Yarrow is very difficult to reverse or predict even given an infinite stream of output
[15:02] <toad_> MTs aren't
[15:02] <toad_> but they are very fast
[15:02] <nextgens> and what about using both ? depending on how often we need them ?
[15:03] <toad_> we do
[15:03] <nextgens> I mean generating node's key with yarrow for example
[15:03] <toad_> we do
[15:03] <nextgens> ah, ok
[15:03] <nextgens> :)
[15:03] <NullAcht15> Oh well. The frustrating thing about trying to understand freenet, imo, is that you always get quite a lot of fairly complex topics thrown at your all at once....
[15:03] <NullAcht15> +head
[15:03] <toad_> we generate the node's key with yarrow (you might want to look into whether there's enough entropy in it on startup the first time... i wonder if we need to feed some files into it)
[15:04] <toad_> we generate CHK padding, for example, with MT's
[15:04] <toad_> NullAcht15: yeah :|
[15:04] * TheBishop_ (n=bishop@port-212-202-175-197.dynamic.qsc.de) has joined #freenet
[15:05] * toad_ thinks he will postpone FreenetURI.fromFullBinaryKey()...
[15:05] <toad_> hey bishop :)
[15:05] <NullAcht15> Looking at the stuff is a bit like looking at chinese characters... First thought is "I'll never be able to learn that all."
[15:05] <toad_> want SVN (CVS) access? :)
[15:05] <toad_> NullAcht15: it's reasonably modular once you get to know it
[15:05] <NullAcht15> toad_: Just like chinese characters...
[15:06] <toad_> NullAcht15: chinese characters are modular?
[15:06] <toad_> well i suppose with the modern phonetic stuff...
[15:07] <toad_> trad chinese is one character per word, roughly, isn't it?
[15:07] <NullAcht15> hm, more or less
[15:07] <NullAcht15> Many words are constructed with more than one of them
[15:07] <nextgens> toad_: don't forget to advertise that https://emu.freenetproject.org/ allows password changing when you create SVN accounts ;)
[15:07] <NullAcht15> And the structure of the individual characters is somewhat modular, too
[15:08] <NullAcht15> I'm only approaching the subject from the side of japanese Kanji, which is very similar but not quite the same
[15:08] <toad_> nextgens: is that set up now?
[15:08] * nextgens isn't a good webdesigner ... I know that it could be more userfriendly :$
[15:08] <nextgens> yes
[15:09] <nextgens> all in there is working
[15:09] <toad_> cool
[15:09] <nextgens> ^should be working ;)
[15:12] <toad_> okay
[15:12] <toad_> lets say the default document is the one which has an empty name
[15:12] <toad_> i think that's how we do it now
[16:36] * lucky644 (i=lucky@d207-216-197-24.bchsia.telus.net) has joined #freenet
[16:40] * Elly (n=elly@ool-457856bb.dyn.optonline.net) Quit (Connection timed out)
[16:41] * Elly (n=elly@ool-457856bb.dyn.optonline.net) has joined #freenet
[16:44] <lucky644> is it normal for after 3 days to only have 2 nodes o.O
[16:47] * unknown_ (n=NullAcht@dslb-082-083-223-136.pools.arcor-ip.net) has joined #freenet
[16:49] <toad_> lucky644: no
[16:49] <toad_> well
[16:50] <toad_> lucky644: are you behind some sort of NAT, router, or firewall/
[16:50] <toad_> ?
[16:50] <toad_> lucky644: you need to forward the listenPort port
[16:50] <toad_> check in your freenet.conf/freenet.ini
[16:50] <toad_> for its value
[16:52] <toad_> bbl
[16:52] <toad_> bbl, probably monday
[16:52] * toad_ (i=toad@pdpc/supporter/active/toad-with-underline) Quit (Remote closed the connection)
[16:53] * NullAcht15 (n=NullAcht@dslb-082-083-232-195.pools.arcor-ip.net) Quit (Nick collision from services.)
[16:53] * unknown_ is now known as NullAcht15
[17:02] <lucky644> no, its not
[17:02] <lucky644> its directly connected.
[17:14] * sp (n=adamf@82-70-138-244.dsl.in-addr.zen.co.uk) has joined #freenet
[17:27] * Sugadude (n=Sugadude@tor/session/x-a908a280048bf1b0) has joined #freenet
[17:38] * sandos (n=sandos@tor/session/x-f5bacd970050a152) Quit (Read error: 104 (Connection reset by peer))
[17:39] * MrNaughty (n=mrnaught@S01060013103e9ce9.ed.shawcable.net) Quit (Read error: 104 (Connection reset by peer))
[17:39] * sandos (n=sandos@tor/session/x-46c688fa51bb74fa) has joined #freenet
[18:40] * unknown_ (n=NullAcht@dslb-082-083-198-226.pools.arcor-ip.net) has joined #freenet
[18:41] * linyos (n=linyos@109-17.73-24.tampabay.res.rr.com) has left #freenet
[18:41] * toad_ (i=toad@pdpc/supporter/active/toad-with-underline) has joined #freenet
[18:42] * ChanServ sets mode +o toad_
[18:42] * toad_ (i=toad@pdpc/supporter/active/toad-with-underline) Quit (Remote closed the connection)
[18:55] * NullAcht15 (n=NullAcht@dslb-082-083-223-136.pools.arcor-ip.net) Quit (Success)
[18:58] * Ash-Fox (i=UNKNOWN@ecg92.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[18:59] * Ash-Fox (i=UNKNOWN@ecu49.neoplus.adsl.tpnet.pl) has joined #FreeNET
[19:13] * MrNaughty (i=MrNaught@S01060013103e9ce9.ed.shawcable.net) has joined #freenet
[19:43] * lucky644 (i=lucky@d207-216-197-24.bchsia.telus.net) Quit (Read error: 104 (Connection reset by peer))
[20:08] * unknown_ (n=NullAcht@dslb-082-083-198-226.pools.arcor-ip.net) Quit (No route to host)
[20:39] * lucky644 (i=lucky@d207-216-197-24.bchsia.telus.net) has joined #freenet
[20:44] * sp is now known as sp_
[21:08] * sp_ (n=adamf@82-70-138-244.dsl.in-addr.zen.co.uk) has left #freenet
[21:27] * Hory (n=Miranda@82.78.27.85) Quit ("http://CyberLore.net - Recommendations on the best games, freeware and websites.")
[22:14] * Sugadude (n=Sugadude@tor/session/x-a908a280048bf1b0) Quit ("Zzz")
[22:30] * Ash-Fox (i=UNKNOWN@ecu49.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[22:30] * Ash-Fox (i=UNKNOWN@edg12.neoplus.adsl.tpnet.pl) has joined #FreeNET
[22:31] * Ash-Fox (i=UNKNOWN@edg12.neoplus.adsl.tpnet.pl) Quit (Read error: 104 (Connection reset by peer))
[22:36] * Ash-Fox (i=UNKNOWN@ect15.neoplus.adsl.tpnet.pl) has joined #FreeNET
[23:12] * lucky644 (i=lucky@d207-216-197-24.bchsia.telus.net) Quit (Read error: 110 (Connection timed out))
[23:26] * Ash-Fox (i=UNKNOWN@ect15.neoplus.adsl.tpnet.pl) Quit ()
[23:39] * Ash-Fox (i=UNKNOWN@ece74.neoplus.adsl.tpnet.pl) has joined #FreeNET
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.