Timestamps are in GMT/BST.
[0:06] * zorton (zorton@circle.xyxx.com) has left #freenet
[0:58] * lostlogic starting freenet for the first time in ... yeah
[1:05] <lostlogic> wow, immediate integration and bandwidth usage.
[1:05] <lostlogic> I'm impressed.
[1:05] <lostlogic> 60243 latest 60248
[1:05] <lostlogic> ugh
[1:05] <lostlogic> *builds CVS*
[1:16] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[1:17] * plixed (~plixed@p50823BB4.dip.t-dialin.net) Quit (Connection timed out)
[1:19] <Ash-Fox> http://www.2-spyware.com/images/2SPYRR1C.gif <- but I'm browsing this page using linux...
[1:19] <lostlogic> hahahahaha
[1:19] <lostlogic> I'm amused
[2:01] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit (Read error: 110 (Connection timed out))
[2:06] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) has joined #freenet
[2:58] * plixed (~plixed@pD9551FF5.dip.t-dialin.net) has joined #freenet
[3:07] * azerty (www-data@80.67.174.43) has joined #freenet
[3:08] * azerty (www-data@80.67.174.43) Quit (Client Quit)
[3:11] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[3:16] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[3:43] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit (Read error: 110 (Connection timed out))
[3:55] * spaetz (~spaetz@217-162-197-136.dclient.hispeed.ch) has joined #freenet
[4:01] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) has joined #freenet
[4:40] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[4:43] * TLF (~francisco@31.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[4:58] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[5:11] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit (Read error: 110 (Connection timed out))
[5:17] * TLF (~francisco@31.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[6:01] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) has joined #freenet
[6:12] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[6:17] * Don_PhrostByte (~phrosty@cpe-24-30-180-198.socal.rr.com) has joined #freenet
[6:44] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit (Read error: 110 (Connection timed out))
[7:00] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) has joined #freenet
[7:01] <d-ArkAngel> Morning all
[7:01] * pittaman (~mistery_b@81.164.16.202) has joined #freenet
[7:03] * pittaman (~mistery_b@81.164.16.202) Quit (Client Quit)
[7:08] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[7:14] * Eyecorners (~eyecorner@fia41-111.dsl.hccnet.nl) has joined #freenet
[7:15] * kers (~kers@55.ppp144.rsd.worldonline.se) has joined #freenet
[7:15] <Eyecorners> 'morning
[7:15] <d-ArkAngel> morning
[7:16] * Eyecorners goes to get some muesli
[7:20] <d-ArkAngel> yeah maybe I should go aquire some food as well.
[7:32] * Don_PhrostByte (~phrosty@cpe-24-30-180-198.socal.rr.com) Quit ()
[7:32] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Read error: 104 (Connection reset by peer))
[7:40] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[7:47] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[7:49] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[7:53] <Eyecorners> Woohoo! I think I have found a way to write uber-secure encrypted CD-R's with linux, wanna hear about it?
[7:59] <Eyecorners> Gosh it's quiet in here...
[8:01] <d-ArkAngel> lol, it's your fault, you mentioned food
[8:03] <Eyecorners> Want to hear about my encrypted cd's?
[8:03] <d-ArkAngel> are they lots of fun?
[8:03] <Eyecorners> I think they are, but that's a matter of taste.
[8:03] <Eyecorners> Ok... the story begins.
[8:03] * d-ArkAngel has no taste ;-)
[8:03] <Eyecorners> You need the following: A Linux installation, and dm-crypt
[8:04] <Eyecorners> Optional: Removable media like iPod, stick, memcard etc.
[8:04] <Eyecorners> Say we have a bunch of files at /home/user/badpr0n we want to burn to a cd but it must be really secure
[8:05] <Eyecorners> We create a key at /mnt/usbstick/keys, like this:
[8:07] <Eyecorners> # hashalot ripemd5 > /mnt/usbstick/keys/`makepasswd --chars=10`
[8:07] <Eyecorners> Then enter a password.
[8:07] <Eyecorners> Oops, s/ripemd5/ripemd160
[8:08] <Eyecorners> The makepasswd --chars=10 is to assign a random ID to a key, I have a bunch of then on my iPod
[8:08] <Eyecorners> Now the creating of the ISO9660 image:
[8:09] <Eyecorners> First we calculate the size of the filesystem, with this in the home: # mkisofs -v --print-size badpr0n
[8:10] <Eyecorners> There will be a number in the output. We then create a file in a temp dir with # touch badpr0nimage and then a # dd if=/dev/zero of=badpr0nimage bs=2048 count=[thenumberintheoutput]
[8:11] <Eyecorners> Then a # losetup /dev/loop7 badpr0nimage and then # cryptsetup -v -c aes -h ripemd160 -d /mnt/usbstick/keys/XXA4NjURRE create badpr0nimage /dev/loop7
[8:12] <Eyecorners> Now the iso9660 filesys, from the home: # mkisofs -v -o /dev/mapper/badpr0nimage badpr0n
[8:13] <Eyecorners> We now remove the cryptoloop: # cryptsetup remove badpr0nimage and then # losetup -d /dev/loop7
[8:13] <Eyecorners> The file in the work directory, badpr0nimage can now be burnt to a CD as an iso image.
[8:14] <Eyecorners> To access the CD, do the following:
[8:14] <Eyecorners> - Insert o'course
[8:14] <Eyecorners> - losetup /dev/loop7 /dev/cdroms/thecdrom
[8:15] <Eyecorners> - # cryptsetup -v -c aes -h ripemd160 -d /mnt/usbstick/keys/XXA4NjURRE create badpr0ncdrom /dev/loop7
[8:15] <Eyecorners> - mount -t iso9660 /dev/mapper/badpr0ncdrom /mnt/whereveryouwanttomount
[8:16] * kers (~kers@55.ppp144.rsd.worldonline.se) Quit (Read error: 238 (Connection timed out))
[8:16] <Eyecorners> Et voila! CD-R's, securely encrypted. When a person not aware of the encryption tries to access the CD it will just appear damaged.
[8:16] <Eyecorners> What do you think d-ArkAngel?
[8:17] <d-ArkAngel> could both of these be done at the same time to a R/W drive? ala encrypted HDD partitions?
[8:17] <plixed> the cryptoapi is not very new, it was included in redhat 8
[8:17] <Eyecorners> DM-crypt is relatively new...
[8:17] <plixed> just a new name
[8:18] <plixed> before cryptoapi there was loopaes
[8:18] <Eyecorners> d-ArkAngel, dm-crypt is originall meant to be used with r/w partitions ;-)
[8:18] <Eyecorners> plixed: I see
[8:18] <Eyecorners> I think it rules
[8:18] <plixed> loopaes was included in suse distris
[8:19] <plixed> years ago
[8:19] <Eyecorners> d-ArkAngel: Just do a google search for 'encrypted filesystem dm-crypt'
[8:19] <Eyecorners> dm-crypt is the only cryptoloop that works with the device mapper, right?
[8:20] <plixed> you didn't need a device mapper years ago to encrypt a cd or a file or whatever
[8:20] <Eyecorners> I know, but I think the device mapper is very convenient
[8:21] <d-ArkAngel> some of the new via chips have hardware implementations of aes which makes them damn fast for stuff like that.
[8:21] <Eyecorners> Wow that's amazing
[8:21] * cbreak (~cbreak@dhcp-32-094.via-eth.ch) has joined #freenet
[8:21] <plixed> looking at your description it's not more convenient than the old stuff
[8:21] <Eyecorners> I'm not a leet scripting hacker... so...
[8:23] <plixed> Eyecorners: the argument for dm-crypt / devicemapper is that it's more secure in multiuser environments
[8:25] <Eyecorners> Huh? I still need to be root to use it, soo?
[8:26] <d-ArkAngel> http://www.mini-itx.com/reviews/mii12000/?page=13
[8:26] <Eyecorners> Nice!
[8:27] * Eyecorners wants such a mobo!
[8:28] <Eyecorners> Encrypted root filesystem all the way!
[8:28] <d-ArkAngel> they're low power baords
[8:28] <plixed> encrypting / is dumb
[8:28] <d-ArkAngel> take a look at the site front page
[8:28] <cbreak> plixed: why?
[8:28] <d-ArkAngel> what encryption algols does a freenet node use?
[8:29] <Eyecorners> Good question
[8:29] <d-ArkAngel> is freenet using AES? or something else?
[8:29] <plixed> and hw accel for just one algorithm is dumb too
[8:29] <d-ArkAngel> not really
[8:29] <cbreak> I think it uses twofish.
[8:29] <Eyecorners> I don't think Rijndael will be obsolete soon
[8:29] <cbreak> (It used to use it for the DS)
[8:29] <d-ArkAngel> not when AES is used in SSL, SSH, and IPSEC
[8:30] <Eyecorners> Indeed
[8:30] <Eyecorners> What is twofish like?
[8:30] <d-ArkAngel> it's another standard one.
[8:30] <cbreak> a symetric crypto algo similar to aes.
[8:30] <d-ArkAngel> SSH uses it I know that much
[8:30] <plixed> if an algorithm is used everywhere it is a good argument to use a different one
[8:31] <d-ArkAngel> no it's not
[8:31] <d-ArkAngel> security through obscurity is just nieve
[8:31] <cbreak> If an algorithm is used everywhere it is a good argument to asume its security
[8:32] <d-ArkAngel> because it's had more testing for one thing.
[8:32] <plixed> like DES?
[8:32] <Eyecorners> Yes, but what do you mean by security by obscurity?
[8:32] <Eyecorners> (I know what it means though)
[8:32] <d-ArkAngel> using something that no-one else has heard of because you think that means no-one will be able to crack it
[8:33] <Eyecorners> All VNC variants use triple DES for the encryption of the server password. It can be cracked in no time
[8:33] <plixed> obscurity is not using a different strong algorithm
[8:33] <Eyecorners> I did it at my school
[8:33] <Eyecorners> Switching the monitor on/off with ultravnc on some innocent victim... ;-)
[8:33] <d-ArkAngel> but your argument is that if other people are using it then it's best to use soemthing else.
[8:33] <plixed> most vnc variants do not use encryption at all but assume you are using a ssh tunnel or similar
[8:34] <Eyecorners> plixed: No encryption for the stream, but for the storage of the server password
[8:34] <Eyecorners> I personally only use VNC over SSH tunnels.
[8:34] <d-ArkAngel> why?
[8:34] <cbreak> passwords should be stored as one way hash...
[8:34] <plixed> indeed
[8:34] <d-ArkAngel> you like all your data being none encrypted?
[8:35] <Eyecorners> ???
[8:35] <d-ArkAngel> you just said the stream isn't encrypted.
[8:35] <d-ArkAngel> so that means anyone can read what you're sending yes?
[8:36] <Eyecorners> But it is if I run it over an SSH tunnel
[8:36] <Eyecorners> You familiar with SSH, d-ArkAngel
[8:36] <d-ArkAngel> yes
[8:36] <Eyecorners> Well then?
[8:36] <d-ArkAngel> oh right
[8:36] <plixed> d-ArkAngel: it like adding SSL to HTTP
[8:36] <Eyecorners> Indeed
[8:36] <plixed> s/it/it's/
[8:36] <d-ArkAngel> VCN vis a SSH tunnle
[8:36] <d-ArkAngel> not that you use VCN instead of SSH
[8:36] <Eyecorners> No offence, but you should check your keyboard
[8:36] * sanity (~ian@81-178-76-180.dsl.pipex.com) Quit (Read error: 104 (Connection reset by peer))
[8:37] <d-ArkAngel> huh?
[8:37] <Eyecorners> Forget it ;-)
[8:38] * sanity (~ian@81-178-76-180.dsl.pipex.com) has joined #freenet
[8:38] <Eyecorners> Welcome back, sanity
[8:41] <cbreak> http://en.wikipedia.org/wiki/Advanced_Encryption_Standard
[8:41] <cbreak> It looks like AES has it's weaknesses...
[8:42] <Eyecorners> Mmm
[8:42] <Eyecorners> Mmm
[8:42] <Eyecorners> Who hasn't?
[8:43] <plixed> Eyecorners: encryting / is dumb because you are forced to type your secret in even for starting the system and you can't leave it running alone
[8:44] <Eyecorners> What about the key on a usb stick, plixed?
[8:44] * sanity (~ian@81-178-76-180.dsl.pipex.com) Quit (Read error: 104 (Connection reset by peer))
[8:44] <plixed> it changes nothing
[8:44] <Eyecorners> As long as you leave the key in, it will boot
[8:44] <cbreak> Use different Keys for different drives.
[8:44] * sanity (~ian@81-178-76-180.dsl.pipex.com) has joined #freenet
[8:47] <d-ArkAngel> I don't see any weakness in that information... just discussion about the security of the algorythm.... almost all of them are dependant on umproven facts, like the upper bound on prime factorisation
[8:48] <cbreak> It looks like you can break 9 of the 14 rounds of a 256bit encryption.
[8:48] <Eyecorners> Read on
[8:48] <d-ArkAngel> yeah I've read that.
[8:48] <d-ArkAngel> being able to partialy break something doesn't imply that you can break the rest....
[8:49] <cbreak> It's certainly not giving me confidence in the algorithm...
[8:49] <d-ArkAngel> there have been partial attacks against use algols for a long time, but it doesn't automaticaly follow that the remainder is breakable
[8:50] * TLF (~francisco@228.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[8:53] <d-ArkAngel> even the developers of fwofish think aes is better
[8:54] <d-ArkAngel> well one of the developers
[8:55] * kers (~kers@14.ppp144.rsd.worldonline.se) has joined #freenet
[9:01] * kers (~kers@14.ppp144.rsd.worldonline.se) Quit ("Leaving")
[9:08] * hobx_ (~chatzilla@ankh.math.chalmers.se) Quit (Remote closed the connection)
[9:17] * hobx_ (~chatzilla@ankh.math.chalmers.se) has joined #freenet
[9:17] * TLF (~francisco@228.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[9:19] * kers (~kers@14.ppp144.rsd.worldonline.se) has joined #freenet
[9:29] <spaetz> hobx_: do you live in the math department of your university? Gee, it's week end
[9:29] <spaetz> :)
[9:30] <spaetz> d-ArkAngel: Do you know if the Via M10000 has the encryption build in as well? (Nehemiah)
[9:30] <spaetz> Because if yes, I should put it to some use...
[9:30] * spaetz googles
[9:32] <d-ArkAngel> are you running linux on your M10000?
[9:33] <spaetz> yep
[9:33] <d-ArkAngel> do a cat /proc/cpuinfo
[9:33] <spaetz> It's a nice passive cooled router (and web surfing machine)
[9:33] <d-ArkAngel> and see what revision it is
[9:34] <spaetz> Model 9 Stepping 1
[9:34] <spaetz> VIA Nehemiah
[9:35] <spaetz> Would that show up in the flags section?
[9:35] <d-ArkAngel> (I've got a PD6000E as my router, and a ME 6000
[9:35] <spaetz> mmh, more googling necessary, seems it "just" has a random number generator
[9:35] <d-ArkAngel> not sure ;-)
[9:38] <d-ArkAngel> yeah looks like it's only the MII 12000 that has the new C5P core with the aes acceleration
[9:39] <cbreak> you can't emulate a good rng in software, but AES you can. :)
[9:40] <d-ArkAngel> did you see the benchamarks? the aes performance was blowing even a P4 away
[9:40] <d-ArkAngel> not bad for such a small chi[
[9:40] <d-ArkAngel> chip even
[9:41] <cbreak> It's not like all I do is de and encrypting aes :) (And P4 Sucks anyway...)
[9:41] <d-ArkAngel> well the P4 was beating the AMD they had in the same benchmarks ;-)
[9:42] <cbreak> Yes, AMD is more or less in the same category of heat generating units.
[9:43] <d-ArkAngel> the new low power opterons look quite good
[9:44] <spaetz> yep, they look nice, but are not especially cheap.
[9:44] <spaetz> So not yet for home users...
[9:44] <cbreak> I prefer G5. should be quite useable with simd for 128bit vectors.
[9:45] * d-ArkAngel has never looked at anything about a G5
[9:48] <spaetz> isn't that the one with the defective cooling fans? :)
[9:48] <cbreak> hmm?
[9:48] <spaetz> I think I remember some articles saying so...
[9:48] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Read error: 104 (Connection reset by peer))
[9:48] <spaetz> But I might easily be wrong
[9:48] <cbreak> the cpu has no fans.
[9:49] <cbreak> it has some metal heat spreaders or a water cooling on the fastest model.
[9:49] <d-ArkAngel> like the VIA's then
[9:49] <cbreak> well, the fans are in the enclosing :)
[9:49] <cbreak> 9 fans to be exact (in the dual cpu model). I read it would be as loud as a wisper...
[9:50] <spaetz> ahh, it was about strange fan noises o
[9:50] <cbreak> that was in the g4, I think.
[9:50] <spaetz> google serves well as a memory replacement
[9:51] * TLF (~francisco@10.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[9:51] * Eyecorners switches from Opera to X-Chat
[9:51] * Eyecorners (~eyecorner@fia41-111.dsl.hccnet.nl) has left #freenet
[9:52] * michaelkuijn (~michaelku@fia41-111.dsl.hccnet.nl) has joined #freenet
[9:54] * hobx_ (~chatzilla@ankh.math.chalmers.se) Quit (Remote closed the connection)
[10:06] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[10:06] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[10:06] * sanity (~ian@81-178-76-180.dsl.pipex.com) Quit (Read error: 232 (Connection reset by peer))
[10:07] * sanity (~ian@81-178-76-180.dsl.pipex.com) has joined #freenet
[10:11] * kers (~kers@14.ppp144.rsd.worldonline.se) Quit (Read error: 232 (Connection reset by peer))
[10:14] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[10:15] * kers (~kers@76.ppp141.rsd.worldonline.se) has joined #freenet
[10:18] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[10:18] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[10:19] * guido^pe (~unknown@dsl-213-023-234-251.arcor-ip.net) has joined #freenet
[10:22] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[10:26] * sanity (~ian@81-178-76-180.dsl.pipex.com) Quit ()
[10:30] * TLF (~francisco@10.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[10:33] * TLF (~francisco@253.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[10:42] * TLF (~francisco@253.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[10:43] * spaetz (~spaetz@217-162-197-136.dclient.hispeed.ch) Quit (Read error: 238 (Connection timed out))
[10:44] * TLF (~francisco@125.Red-81-40-116.pooles.rima-tde.net) has joined #freenet
[10:46] * TLF (~francisco@125.Red-81-40-116.pooles.rima-tde.net) Quit (Client Quit)
[11:02] * TLF (~francisco@58.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[11:04] <toad_> hi
[11:04] <d-ArkAngel> hi
[11:08] <toad_> freenet uses AES/Rijndael, and Twofish (both were finalists in the AES contest)
[11:08] <d-ArkAngel> why the combination?
[11:09] <toad_> <Eyecorners> All VNC variants use triple DES for the encryption of the server password. It can be cracked in no time - only if the password is weak, and you can do a dictionary attack - unless you have a cryptanalytic attack against 3DES, which is still used by a lot of banks??
[11:10] <toad_> <Eyecorners> What about the key on a usb stick, plixed? - only if the memory stick can be removed and utterly destroyed quickly and easily..
[11:11] <toad_> i'm completely ignoring physical access issues here - if you live with peoplw who might pull the memory stick, you have problems :)
[11:12] <d-ArkAngel> and if your ignoreing physical access then encrypted stuff on a removable disc is un-necessary ;-)
[11:12] <toad_> re hardware AES... I'm unreliably informed that next year's dual-core AMD chips will have a really heavy vector coprocessor a la AltiVec, which will speed up symmetric crypto greatly
[11:12] <d-ArkAngel> that would be neat
[11:13] <d-ArkAngel> I heard roumour that the AMD Dual cores might be ahead of schedule, and be sometime later this year
[11:14] <toad_> that'd be nice
[11:14] <toad_> they'll have to run at 1GHz under the current ones of course :|
[11:14] <toad_> 2GHz maybe when the top 1P ones are 3GHz
[11:15] <toad_> anyway, Freenet...
[11:17] <i2p_iip> <gott> *hugs toad*
[11:18] <d-ArkAngel> they're using the same tech as their current 35W Opterons
[11:18] <toad_> d-ArkAngel: 35W?!?
[11:18] <toad_> woah
[11:19] <toad_> how do they achieve that? (and at what clock speed) ?
[11:19] <d-ArkAngel> that's the ones made in their new 90nm process
[11:19] <toad_> ahhh
[11:19] <d-ArkAngel> same clocks as the current modles
[11:19] <toad_> even so, you don't usually get a factor of over 2
[11:19] <toad_> on a process shrink
[11:19] <d-ArkAngel> it's a SOI tech as well I think
[11:20] <toad_> you usually get something like 30% if you're lucky, and then you make it all back immediately by ramping the clock speed
[11:20] <toad_> d-ArkAngel: they always used SOI for the opterons, didn't they?
[11:20] <d-ArkAngel> not sure
[11:20] * TLF (~francisco@58.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[11:21] <d-ArkAngel> but they have 3 power usage versions of their opterone, 95W 35W and one somewhere in the middle...
[11:21] <d-ArkAngel> opteron (why do I keep adding that damn trailing e??
[11:21] * toad_ wonders if once people have figured out how to break google's DRM, there'll be a distributed project to download all the books, bypassing the per-IP portion restrictions by requesting from different IPs? :)
[11:22] <cbreak> how does that drm work?
[11:23] <d-ArkAngel> http://www.amd.com/gb-uk/Processors/ProductInformation/0,,30_118_8796_9240,00.html
[11:24] <d-ArkAngel> 55W and 30W
[11:24] <cbreak> It looks like the new G5 uses 24.5W
[11:24] <toad_> cbreak: don't know, couldn't get to the site describing it, it was /.ed
[11:25] <toad_> cbreak: which is a single core Power4 plus AltiVec, right?
[11:25] <cbreak> more or less, yes.
[11:26] <toad_> d-ArkAngel: nice
[11:27] <d-ArkAngel> So dual ops are probably going to be back up at 60-90 W tho there is talk of a number of the technologies from Athlon-M being transferred over to help keep Power consumtion down under normal usage.
[11:27] <toad_> of course the HE's and EE's aren't going to be available in europe for months :)
[11:27] <i2p_iip> <gott> the DRM is very simple
[11:27] <i2p_iip> <gott> it uses CSS
[11:27] <toad_> well a month or so
[11:28] <toad_> gott: to do what?
[11:28] <toad_> CSS has the ability to block save? I can see blocking print with CSS, but save?
[11:28] <toad_> d-ArkAngel: so EE means power < 40W ?
[11:29] <d-ArkAngel> AMD's site quotes it as 30W
[11:29] <d-ArkAngel> tho aparently only avaliable in the 1.4Ghz freq at the moment
[11:29] <d-ArkAngel> the 55W version is avaliable at the 2Ghz flavour
[11:30] <toad_> ah
[11:30] <toad_> d-ArkAngel: what's the mobile tech? it's peak power consumption that you have to keep down, not just average... average is easy(ish)...
[11:31] <i2p_iip> <gott> it blocks printing
[11:31] <i2p_iip> <gott> not sure about save
[11:31] <d-ArkAngel> I think it's the powersavings to keep the average down more than peak. but I'm not sure.
[11:31] <d-ArkAngel> IAMAPD
[11:31] <d-ArkAngel> IANAPD even
[11:32] * d-ArkAngel wishes he wasn't typing on a laptop keyboard.
[11:32] <toad_> what does that stand for?
[11:33] <toad_> so connect a real USB keyboard :)
[11:33] <d-ArkAngel> I Am Not A ... Processor Designer
[11:34] * toad_ wonders when the reversible (therefore virtually no power) electronics will start to come in
[11:34] <d-ArkAngel> no idea what you mean by that toad
[11:35] <d-ArkAngel> the "next big thing" I'm looking forward to is OLED/OLEP displays
[11:36] <toad_> d-ArkAngel: reversible computing - simplest model is billiard-ball cellular automata
[11:36] <toad_> uses no power
[11:36] <toad_> spintronics was the first idea
[11:36] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 54 (Connection reset by peer))
[11:36] <toad_> but that'll take 10-20 years
[11:37] <cbreak> a perpetoum mobile?
[11:37] <d-ArkAngel> ahhh I see.
[11:37] <toad_> somebody was working on an electronic version recently
[11:37] * pupok (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[11:41] <toad_> okay, lets have some graphs...
[11:43] <toad_> cbreak: spintronics in theory has zero power consumption except for the interfaces.. it's some wierd quantum CS thing that you only use power when you destroy information..
[11:45] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-200x25.png
[11:45] <toad_> well, HTL 14 plainly sucks
[11:45] <toad_> HTL 16 is clearly consistently above 95%
[11:46] <toad_> although it may yet nosedive; it's very noisy
[11:46] <toad_> HTL 15 appears to be falling
[11:46] <toad_> and certainly isn't reliably over 95%
[11:47] <d-ArkAngel> I love the fact that 94% "sucks" :-)
[11:48] <toad_> 94% doesn't suck
[11:48] <toad_> 94% and falling sucks
[11:48] <d-ArkAngel> :-)
[11:48] <toad_> but the target is 95% anyway
[11:49] <toad_> it may be that we always eventually fall significantly
[11:49] <toad_> but i don't think so...
[11:50] <toad_> in any case, that tendancy does mean that we can't just say "lets try for 80% instead, so we can use lower HTL"... well, maybe it falls and then slows down and goes to an asymptote?
[11:50] <toad_> HTL is probably a significant issue... if I'm right about the pattern then to get good performance on a 10K node network we'll need an HTL of 50 even under ideal conditions
[11:50] <toad_> and for a 1M node network, we'd need 90
[11:51] <toad_> and it might be worse than that even in ideal conditions
[11:51] <d-ArkAngel> maybe... but maybe our sims insert policy is still a little more aggressive than the real network, and we're probably not getting much small world going on in only 200 nodes with a HTL of 14
[11:52] <toad_> and it's certainly going to be worse than that IRL..
[11:52] <toad_> d-ArkAngel: then what is happening? it's a lot better than random routing...
[11:52] <toad_> (yes i have simulated that!)
[11:53] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-800x25.png
[11:53] <toad_> argh
[11:53] <d-ArkAngel> what we're probably seeing is specilsation, and caching
[11:53] <d-ArkAngel> tho I'm not sure how much specilisation
[11:53] <toad_> yes
[11:53] <toad_> and that's the same thing
[11:53] <toad_> no
[11:53] <toad_> random routing with caching (pcaching or regular caching) produces much lower numbers
[11:53] <toad_> which don't change much over time
[11:54] <d-ArkAngel> but random removes specilisation doesn't it?
[11:54] <toad_> the only real point is that the test bar is fairly high - we request the last N (nodes*storesize/40) inserts and only them, all have equal popularity.
[11:55] <toad_> d-ArkAngel: exactly. we must be having specialization
[11:55] <d-ArkAngel> indeed
[11:55] <d-ArkAngel> yes, but specilization doesn't imply small world development does it? if our HTL is large enough to still reach the entire network...
[11:56] <toad_> doesn't it?
[11:57] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-800x25-supersmooth.png
[11:57] <toad_> hmmm
[11:57] <toad_> i think perhaps that's just a little bit too much averaging!
[11:57] <d-ArkAngel> small world means that multiple nodes develop similar specilisations doesn't it? and if we've got access to the whole network within HTL then there's no reason for nodes to develop with the same specilisation is there? (given the requests will always be routed towards the best nodes.
[11:57] <toad_> no, small world means 1/d distribution of links
[11:57] <toad_> which is implied by LRU
[11:57] <d-ArkAngel> 1/d?
[11:57] <toad_> which further implies logarithmic scaling
[11:57] <toad_> yeah
[11:58] <d-ArkAngel> what's d?
[11:58] <toad_> d is keyspace distance
[11:58] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-400x25.png
[11:58] <toad_> do you think green and red are just oscillating?
[12:00] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-400x25-supersmooth.png
[12:00] <toad_> hrrrm
[12:02] <d-ArkAngel> offline rt is bound to be better in the long run, since node specilisations are more likely to be the same to all nodes so are less likely to change
[12:04] <toad_> the actual results are not convincing in that respect
[12:05] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-200x25-supersmooth.png
[12:05] <d-ArkAngel> interesting that the newbie hits line clearly learns faster then the competition
[12:05] <toad_> which one?
[12:05] <d-ArkAngel> =500
[12:06] <cbreak> the real network has to deal with nodes which are only online part time. offline rt may be helpfull in such an environment :)
[12:06] <toad_> 100 or 500?
[12:06] <toad_> cbreak: indeed
[12:06] <d-ArkAngel> =500
[12:06] * Overand_ (common@218.108.252.64.snet.net) has joined #freenet
[12:07] <d-ArkAngel> peaks lower, and then falls away, but rises noteably faster at the start
[12:07] <toad_> d-ArkAngel: does it?
[12:07] <toad_> doesn't look especially good to me
[12:07] <toad_> hmmm
[12:08] <d-ArkAngel> on supersmooth, the mim is bearly bellow .9 and the others are closer to .88
[12:08] <toad_> the problem with these things is they take so fscking long
[12:08] <toad_> and if you're running one you can't really run a node on the same machine
[12:08] <d-ArkAngel> by the point it hits .92 the other have caught up
[12:09] <toad_> which supersmooth?
[12:09] <d-ArkAngel> just thinking that the higher newbie might be useful to attempt to assist during slashdotting
[12:09] <d-ArkAngel> 400x25
[12:10] <d-ArkAngel> http://amphibian.dyndns.org/simulations2-epass-0.025-400x25-supersmooth.png
[12:10] <toad_> http://amphibian.dyndns.org/simulations2-epass-0.025-100x25-supersmooth.png
[12:10] <toad_> hmmm
[12:11] <toad_> is that going for an asymptote?
[12:11] * Overand__ (common@239.99.252.64.snet.net) has joined #freenet
[12:11] <toad_> at maybe 94%?
[12:12] <d-ArkAngel> possibly...
[12:13] <toad_> the problem is we need to run most of these for at least 100,000,000 requests to get convincing data :|
[12:13] <d-ArkAngel> the thread sim is coming, I promise!
[12:14] * Overand (common@69.0.81.177.adsl.snet.net) Quit (Read error: 104 (Connection reset by peer))
[12:14] <toad_> d-ArkAngel: not much use to us without a supercomputer
[12:14] <toad_> ideally an internet-connected one...
[12:15] <d-ArkAngel> :-)
[12:15] <toad_> although physical presence for say a week would be acceptable too :)
[12:15] <d-ArkAngel> have you thought about approaching a university?
[12:15] <toad_> what's in it for them?
[12:15] <d-ArkAngel> some of their undergraduates doing papers based on the stuff we're working on
[12:16] <d-ArkAngel> (that way we get free researchers and CPU time)
[12:16] <toad_> that would be fine by me..
[12:16] <cbreak> you could make something like seti. A freenet sim screensaver.
[12:17] <d-ArkAngel> :-) and then when the project is not needing it, sell time on the super computer to other people :-)
[12:17] <toad_> cbreak: d-ArkAngel's distribution code may work over the internet but i don't think we can reduce the inter-node bandwidth below about 60kB/sec
[12:17] <toad_> both ways
[12:17] * toad_ sends to support to ask for shells
[12:19] * Overand_ (common@218.108.252.64.snet.net) Quit (Read error: 232 (Connection reset by peer))
[12:21] * Ash-Fox (Hal-9000@aaw133.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[12:21] * Ash-Fox (Hal-9000@aau123.neoplus.adsl.tpnet.pl) has joined #FreeNET
[12:27] * d-ArkAngel attaches bluetooth keyboard via laptop's bluetooth connection and is surprised that it works first time
[12:39] * hobx_ (~chatzilla@ankh.math.chalmers.se) has joined #freenet
[12:41] * toad_ also sent mail to tech
[12:50] <michaelkuijn> I am worried.
[12:50] <d-ArkAngel> about?
[12:50] <toad_> michaelkuijn: about what?
[12:50] <michaelkuijn> toad_, you quoted me... I didn't say that
[12:51] <toad_> i did?
[12:51] <toad_> where, and what?
[12:52] <michaelkuijn> <toad_> <Eyecorners> All VNC variants use triple DES for the encryption of the server password. It can be cracked in no time - only if the password is weak, and you can do a dictionary attack - unless you have a cryptanalytic attack against 3DES, which is still used by a lot of banks??
[12:52] <michaelkuijn> [11:10] <toad_> <Eyecorners> What about the key on a usb stick, plixed? - only if the memory stick can be removed and utterly destroyed quickly and easily..
[12:52] <michaelkuijn> [11:11] <toad_> i'm completely ignoring physical access issues here - if you live with peoplw who might pull the memory stick, you have problems :)
[12:52] <toad_> the dash indicates the end of the quote :)
[12:52] <toad_> and the start of my comment
[12:53] <michaelkuijn> Aha
[12:53] <michaelkuijn> Well my school uses a vnc password like AhgkvMUk
[12:53] <toad_> sorry if that causes confusion
[12:53] <michaelkuijn> and it didn't take a second to crack it with a standard tool
[12:53] <toad_> maybe i should always use a newline :)
[12:53] <toad_> michaelkuijn: letters only?
[12:53] <michaelkuijn> Guess so
[12:54] <toad_> is it case sensitive?
[12:54] <michaelkuijn> Just paste the registry key into the dos box et voila
[12:54] <toad_> hmmm
[12:54] <d-ArkAngel> what letter does it start with? :-D
[12:54] <toad_> probably just encrypts it with a fixed key then
[12:54] <michaelkuijn> I suppose so
[12:54] <toad_> if you know the fixed master key, you can decrypt them
[12:54] <toad_> which is completely lame
[12:54] <toad_> everyone uses hashes nowadays
[12:55] <michaelkuijn> So says our sysop
[12:55] <toad_> wouldn't make any difference if it used AES
[12:55] <toad_> if the master key is known, and it's a bidirectional symmetric cipher...
[12:55] <michaelkuijn> toad_, fixed key again
[12:55] <michaelkuijn> RSA or DSA then?
[12:56] <toad_> michaelkuijn: hmm?
[12:56] <toad_> michaelkuijn: asymmetric crypto is much better, obviously
[12:56] <michaelkuijn> By the way, what did you think of my encrypted CD-R story?
[12:56] <toad_> I might have done the same
[12:56] <michaelkuijn> Heh
[12:57] <toad_> re encrypted CD-Rs
[12:58] <michaelkuijn> Oops, now I am exposed. My regular nick and my name are now linked, all people present know it now...
[12:58] <toad_> :)
[12:58] * michaelkuijn has a feeling that there was too much 'now' in that last sentence
[13:00] <toad_> it's the same IP address anyway isn't it?
[13:00] <toad_> bbiab
[13:01] <michaelkuijn> Mmm, if someone steals my iPod I'm screwed... with security, that is
[13:01] * toad_ thinks it's more likely they'd steal your iPod than your PC...
[13:02] <michaelkuijn> I have my keys on my ipod
[13:02] <michaelkuijn> SSH, GPG, dm-crypt keys etc
[13:02] <toad_> hehe
[13:02] <toad_> of course they're passworded, right?
[13:02] <michaelkuijn> Yup
[13:02] <toad_> what-you-know-and-what-you-have...
[13:03] <toad_> well i think it's quite likely somebody would steal your iPod while you were out and about with it
[13:03] <toad_> :)
[13:03] <michaelkuijn> Noone here I know
[13:04] <michaelkuijn> I think there are a lot of iPods in a yellow skin, I don't think I have to worry too much.
[13:04] <michaelkuijn> Oops
[13:04] <michaelkuijn> ;-)
[13:05] * michaelkuijn shouldn't play with security like this
[13:13] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[13:26] <toad_> michaelkuijn: i just mean for its resale value
[13:28] <michaelkuijn> Yeah, I understand
[13:28] <toad_> michaelkuijn: got any spare PCs I could have a shell on for simulations?
[13:28] <toad_> anyone else got any?
[13:28] <michaelkuijn> Simulations like...?
[13:28] <toad_> freenet simulations, of course
[13:28] <michaelkuijn> Sorry, no
[13:29] <michaelkuijn> Could try SDF, though
[13:29] <toad_> huh?
[13:29] <michaelkuijn> sdf-eu.org
[13:29] <toad_> a) why does it matter that they are freenet simulations?, b) what is SDF
[13:29] <michaelkuijn> Well, you asked for shell access
[13:30] <toad_> hmmm
[13:30] <toad_> suspect they wouldn't take kindly to running CPU hogs
[13:30] <toad_> it also uses a fairly high amount of memory
[13:31] <michaelkuijn> Mmm
[13:31] <michaelkuijn> Could make a deal with mr Jones
[13:31] <toad_> who's he?
[13:31] <michaelkuijn> One of the powerful guys
[13:32] <michaelkuijn> Nice guy. He congratulated me when I got the last white edition of their t-shirt
[13:32] <toad_> [01] WHAT IS SDF? (QUICK SUMMARY)
[13:32] <toad_> Welcome to the only all 64bit public access supercomputing center!
[13:32] <michaelkuijn> Hehe
[13:33] <Ribs> Can freenet sites have spaces in the filename?
[13:33] <michaelkuijn> The oldest one, too
[13:34] <toad_> Ribs: debatable :)
[13:34] <Ribs> What?
[13:35] <Zorix> doesnt the broswer put like %20 or something in it
[13:35] <Zorix> to make a space
[13:35] <Ribs> but will they upload?
[13:35] <Ribs> I just put a %20 in the links...
[13:35] <Zorix> ive seen spaces in freesites
[13:35] <toad_> it'll get encoded
[13:35] <toad_> i think it's valid encoded
[13:36] <toad_> but i'm not completely sure
[13:36] <toad_> suggest you try it at htl 0
[13:37] <michaelkuijn> Freenet is becoming more and more inseparable from my life
[13:37] <Zorix> lol
[13:38] <toad_> michaelkuijn: it has that effect on people...
[13:38] <michaelkuijn> I've become more paranoid, too
[13:42] <Ribs> I wish I had a seperate connection for Freenet :(
[13:42] <michaelkuijn> Yeah me too
[13:42] <Ribs> I have to shut down my node when I wanna play online games
[13:42] <Ribs> otherwise it's way too laggy
[13:42] <michaelkuijn> I wish I had a server AND a separate line for Freenet.
[13:42] <Zorix> toad_
[13:42] <Zorix> i heard java 1.5.0 is out of beta now
[13:43] <Zorix> should we move to this
[13:44] <Ribs> Does Freenet compile yet?
[13:44] <Ribs> like, with Jikes and similar?
[13:44] <michaelkuijn> To real binaries?
[13:45] * guido^pe (~unknown@dsl-213-023-234-251.arcor-ip.net) Quit ()
[13:45] <michaelkuijn> That'd be really great
[13:45] * guido^pe (~unknown@dsl-213-023-234-251.arcor-ip.net) has joined #freenet
[13:45] <toad_> Zorix: we cannot use 1.5 features
[13:45] <Zorix> why
[13:45] <toad_> you can use the 1.5 JVM or JDK if you like
[13:45] <Ribs> michaelkuijn: I remember someone was working on it
[13:46] <Zorix> toad_.. why are you sticking to old buggy java code
[13:46] <toad_> Zorix: because we don't want to make it even harder for Free Software JVMs etc (such as GCJ) to run freenet
[13:46] <toad_> not without a good reason
[13:46] <Zorix> hah
[13:46] <toad_> Zorix: using the new language features will make us much less buggy?
[13:46] <michaelkuijn> Mmm. Are blackdown and the like so bad?
[13:46] <Zorix> no the new jvm supposively fixes alot of features that freenet uses
[13:46] <toad_> not using the new language features will prevent us from achieving the performance gains of 1.5?
[13:46] <Zorix> something about multiple threads or something
[13:46] <toad_> michaelkuijn: blackdown is more or less the same as sun
[13:46] <toad_> it's definitely non-free
[13:47] <toad_> Zorix: so use the new JVM
[13:47] <Zorix> uhm
[13:47] <Zorix> does that even work
[13:47] <Zorix> if its compiled with the old version
[13:47] <toad_> yes, it's backwards compatible
[13:47] <toad_> because it's not compiled
[13:47] <Zorix> but will it take advantage of the fixes
[13:48] <toad_> it only gets compiled to bytecode
[13:48] <Zorix> hmm
[13:48] <Zorix> ok
[13:48] <toad_> one of the better reasons for wanting freenet to work with free software is that GCJ DOES compile to native code
[13:48] <Zorix> so you dont use sun java
[13:49] <toad_> no, we use sun java
[13:49] <toad_> but we'd like freenet to work on non-sun java
[13:49] <Zorix> i wonder how those other projects dont get sued
[13:49] <toad_> it would make distribution much easier, it would solve some political problems, it would even improve performance in some respects
[13:50] <Zorix> why hasnt it been used then
[13:52] <toad_> because right now it doesn't work
[13:52] <toad_> with freenet
[13:52] <toad_> due to some bugs
[13:52] <Zorix> does it make an exe or something
[13:52] <toad_> in their NIO implementation
[13:53] * michaelkuijn is sucking custard through a straw... hard work
[13:53] <toad_> michaelkuijn: eeek
[13:53] <toad_> why?
[13:54] <toad_> Zorix: GCJ does
[13:54] <toad_> Kaffe is just a JVM
[13:54] <toad_> but it's not Sun
[13:54] <toad_> Sun isn't included in most linux dists because of sun's refusal to open source it
[13:54] <Zorix> how big do they get to be.. i know the freenet.jar is like 20 mb... wouldnt it make it like 40 mb or something
[13:55] <toad_> no idea
[13:57] <toad_> freenet.jar is 2.5MB, not 20MB
[13:57] * TLF (~francisco@22.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[13:58] <Zorix> hmm ok
[13:58] <Zorix> only time i look at size is when it gets the snapshot update
[14:00] * toad_ decides to start both stable and unstable nodes up
[14:00] <toad_> despite the simulation, which no doubt will break things
[14:01] <michaelkuijn> Btw.... have you looked at KSK@TOAD? So cuuute
[14:01] <toad_> is that the one on the freenethelp page?
[14:02] <michaelkuijn> Don't know. It just displays the caption 'TOAD' and a picture of a toad most likely taken from the first google image search of 'toad'.
[14:02] <toad_> on stable?
[14:04] <michaelkuijn> Guess so
[14:04] <michaelkuijn> The KSK may be compromised though
[14:04] <toad_> what build are you accessing it from?
[14:05] <michaelkuijn> 5996
[14:05] <michaelkuijn> 5096
[14:05] <toad_> hehe, wait a year for 5996 :)
[14:06] <michaelkuijn> Heh
[14:08] * moskau23 (~Miranda@dsl-213-023-251-066.arcor-ip.net) Quit (Read error: 104 (Connection reset by peer))
[14:09] * unknown_ (~unknown@dsl-082-082-159-151.arcor-ip.net) has joined #freenet
[14:09] * guido^pe (~unknown@dsl-213-023-234-251.arcor-ip.net) Quit (Read error: 110 (Connection timed out))
[14:10] <Zorix> more like 10 years heh
[14:20] <KenMan> toad_ how well are the estimators working in the simulation ?
[14:21] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[14:21] <toad_> KenMan: hmm?
[14:21] <KenMan> i mean, is pDNF accurate ? are the others ?
[14:21] <toad_> I don't know
[14:21] <KenMan> you told ian you would investigate the accuracy of the various estimators...
[14:22] <KenMan> ok.
[14:22] <toad_> dumping the whole routing table on every node is something I haven't done yet
[14:22] <KenMan> I was just thinking of a running barcode graph. Suppose we created a barcode for each hour's worth of incoming requests (in real network) and watched whether they consolidated or not.
[14:23] <KenMan> We could break them out by requestor, or just plot the aggregate incoming requests.
[14:23] <KenMan> picture 24 (hours worth of) barcodes stacked vertically, and looking for a 'melding' effect
[14:24] <KenMan> what struck you most about that P2P hints page you mentioned yesterday (hosted at my old school - umbc.edu)
[14:25] <toad_> the which?
[14:26] <KenMan> I was hit by the "tit-for-tat you share a block, i share a block" description (re: bittorrent)
[14:26] <toad_> moving barcode graphs are something we used to do on dodo
[14:26] <KenMan> I mean, we couldn't do that so precisely with Freenet...
[14:26] <toad_> before we got rid of barcodes
[14:26] <toad_> KenMan: sure
[14:26] <KenMan> but the ratio of give:take is pretty extreme on stable network
[14:26] <toad_> KenMan: but we can and do approximate it
[14:27] <toad_> huh?
[14:27] <toad_> you think so? why?
[14:27] <KenMan> so far, I have only examined the ratio of requests (per peer-link) coming and going
[14:27] <toad_> leech nodes will suck heavily
[14:27] * unknown_ is now known as guido^pe
[14:28] <KenMan> request ratios are affected by mRIs in each direction
[14:28] <cbreak> suck in the sense of not working or such down data? :)
[14:28] <KenMan> which are currently independent values
[14:29] <toad_> cbreak: they will not work very well
[14:29] <toad_> and as a result they will not be able to download much
[14:29] <toad_> KenMan: MRIs in each direction are necessarily different
[14:29] <toad_> because different nodes have different load levels, capacities, etc
[14:29] * kers (~kers@76.ppp141.rsd.worldonline.se) Quit (Read error: 113 (No route to host))
[14:29] <KenMan> of course, but they can result in 100:1 imbalance in actual used query rates...
[14:30] <toad_> so what?
[14:30] <KenMan> mRIs can vary by 10000:1 , but the actual Q traffic is what I look at.
[14:30] <toad_> you're popular with a different set of nodes than the set of nodes you send most traffic to
[14:30] <toad_> no big deal
[14:30] <KenMan> I don't know, I'm not sure if it is something we want to control or not
[14:31] <toad_> KenMan: did you read my most recent tech post re simulations?
[14:31] <KenMan> I don't ever hope to see 1:1 requests and 1:1 transfer bytes with Freenet, but there may be an advantage if that is loosely approximated. I really don't know.
[14:32] <KenMan> I am not subscribed to tech, and Gmane only catches 50% of posts
[14:32] <toad_> we can't expect it on a per link basis
[14:32] <toad_> are you sure about that?
[14:32] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) Quit ("Going to the moon")
[14:32] <KenMan> i'll go look, but the last 5 times I did that, your references were not found.
[14:32] <toad_> gmane i mean
[14:33] <toad_> http://dodo.freenetproject.org/pipermail/tech/
[14:33] <toad_> yay
[14:33] <toad_> we have archives!
[14:33] <KenMan> the newest fresh post on gmane tech is sep 30
[14:33] <KenMan> yay for pipermail !!
[14:34] <KenMan> boo-hiss for gmane ;)
[14:34] <toad_> http://dodo.freenetproject.org/pipermail/tech/2004-October/001121.html
[14:35] <KenMan> how about this for a simulation : one fixed DS (simple as possible), one requesting node, how well do the estimators learn?
[14:36] <KenMan> you could control what percent of the requests are not present in the DS
[14:36] * toad (toad@82-32-16-91.cable.ubr03.azte.blueyonder.co.uk) has joined #freenet
[14:36] <toad> gah
[14:36] <KenMan> ack ?
[14:36] <toad> [14:35] <KenMan> how about this for a simulation : one fixed DS (simple as possible), one requesting node, how well do the estimators learn?
[14:36] <toad> [14:36] <KenMan> you could control what percent of the requests are not present in the DS
[14:36] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[14:36] <toad> well, we tried that on testnet, remember?
[14:37] <toad> or do you mean just the overall pDNF?
[14:37] <toad> or what?
[14:37] <KenMan> not exactly, but sort of, yes, i kind of remember
[14:39] <toad> anyway did you read my post?
[14:39] <toad> what's your opinion on the HTL issue?
[14:39] <KenMan> reading
[14:39] <toad> ok
[14:39] <toad> bbiab
[14:42] <KenMan> what can I say about HTL other than your findings are disappointing ? :(
[14:44] * TLF (~francisco@22.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[14:45] <KenMan> in a perfect, DHT-like world, HTL ~= log (base2) N would be optimal routing. We expect to be less than perfect, as we only approximate a DHT.
[14:46] <KenMan> I suppose the only path to follow would be to focus more heavily on connection management (choice of peers)
[14:46] <KenMan> sanity and hobx already did/are looking there, no ?
[14:47] <KenMan> perhaps, if we notice a *significant* part of our traffic involves nodes F & R, we could sew them together...
[14:48] * TLF (~francisco@82.Red-81-40-116.pooles.rima-tde.net) has joined #freenet
[14:53] * tav is now known as tav|offline
[14:55] * tav|offline is now known as tav
[15:00] <KenMan> the problem is that different nodes estimate a single node differently, and pull it in multiple directions. How can we ensure that they merge ?
[15:01] <KenMan> another part of the problem (or solution) is that each requestor makes a different number of requests when 'pulling' on the single node's spec
[15:04] <KenMan> HTL ~= log (base |RT|-1) N with PERFECT routing, but we are far from perfect
[15:07] <KenMan> it would be useful if we had a function that evaluated the "degree of specialization" over a datastore, or over a set of requests.
[15:09] <KenMan> i think we should aim for each node having a single point/area of specialization. The other model is just too complicated.
[15:17] <KenMan> What would happen if two peers agreed that their shared bidi link were to be specialized about a common point ?
[15:18] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[15:18] <KenMan> Compute this point from the two nodes' DSA keys. Then we should get very strong routing, and DS spec should naturally fall out from this.
[15:20] <KenMan> A node would have to work very hard to collect a certain set of peers, such that it could then censor key X with any effectiveness.
[15:22] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[15:30] * TLF (~francisco@82.Red-81-40-116.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[15:30] <tav> .google jxta p2psockets
[15:30] <xena> jxta p2psockets: http://p2psockets.jxta.org/
[15:33] * lostlogic (~lostlogic@node-4024215a.mdw.onnet.us.uu.net) has joined #freenet
[15:40] <toad> <KenMan> the problem is that different nodes estimate a single node
[15:40] <toad> differently, and pull it in multiple directions. How can we ensure
[15:40] <toad> that they merge ?
[15:40] <toad> hmmm
[15:41] <toad> there's that paper suggesting that some sort of fixed specialization will help; i suggest that perhaps something like bidi conns' current sort order would help, rather than LRU... the difficulty is you can prove that LRU gets 1/d
[15:41] <toad> well almost
[15:41] <toad> whereas MRI*estimateAverage you can't
[15:41] <toad> and you have to decide how many reqs to include in it and so on..
[15:43] <toad> KenMan: doing log at all is a triumph imho
[15:43] <toad> but if it doesn't give us a practicable system...
[15:47] <toad> bbl (probably on monday)
[15:47] * toad (toad@82-32-16-91.cable.ubr03.azte.blueyonder.co.uk) Quit ("BitchX-1.0c20cvs -- just do it.")
[15:47] * toad_ (toad@82-32-16-91.cable.ubr03.azte.blueyonder.co.uk) Quit ("using sirc version 2.211+KSIRC/1.3.11")
[15:49] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) has joined #freenet
[16:57] * kers (~kers@126.ppp141.rsd.worldonline.se) has joined #freenet
[17:15] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[17:30] * michaelkuijn (~michaelku@fia41-111.dsl.hccnet.nl) Quit ()
[17:44] * kers (~kers@126.ppp141.rsd.worldonline.se) Quit (Read error: 104 (Connection reset by peer))
[17:45] * kers (~kers@149.ppp132.rsd.worldonline.se) has joined #freenet
[17:49] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) has joined #freenet
[17:57] * Usurp (~Usurp@trudaine-8-82-230-34-86.fbx.proxad.net) has joined #freenet
[18:24] <Ribs> hrm, interesting
[18:25] <Ribs> a site which I loaded 10 minutes ago now takes ages to re-load
[18:29] <d-ArkAngel> that's not entierly unexpected behaviour....
[18:29] <d-ArkAngel> depends if your node has disconnected from the route that it used to get that page last time
[18:30] <cbreak> and it hasn't cached it.
[18:31] <Ribs> my poor node
[18:31] <Ribs> I think I killed it, even the main page doesn't load
[18:31] <cbreak> how many connections does your browser open to a single server?
[18:32] <Ribs> 50
[18:32] <d-ArkAngel> close all the other pages your trying to load from freenet
[18:33] <Ribs> I think it's coming back now
[18:33] * kers (~kers@149.ppp132.rsd.worldonline.se) Quit (Read error: 104 (Connection reset by peer))
[18:33] <Ribs> poor thing, heh
[18:34] <lostlogic> my node on unstable appears to be happy... I'm way behind the times... what indicates a healthy node these days?
[18:35] <d-ArkAngel> that it doesn't try and steal your wife.
[18:35] <lostlogic> I don't have a wife, so I'm clear on that one from the getgo
[18:36] <lostlogic> Connections open (Inbound/Outbound/Limit) 68 (42/26/200)
[18:36] <lostlogic> Transfers active (Transmitting/Receiving) 98 (50/48)
[18:36] <lostlogic> Data waiting to be transferred 4,495 Bytes
[18:36] <lostlogic> Total amount of data transferred 1,437 MiB
[18:36] <d-ArkAngel> or maybe it's already stolen her, hence you don't have one....
[18:36] <lostlogic> Uptime 7 hours 22 minutes
[18:36] <cbreak> looks similar to mine.
[18:36] <lostlogic> mayhaps
[18:45] * blume (~Franz@p5083B043.dip0.t-ipconnect.de) has joined #freenet
[18:46] <blume> hy
[18:46] <cbreak> hi.
[18:47] <blume> wie bekommen ich hier ne serverlist??
[18:47] <cbreak> von was? Eine Freenet seednodes.ref?
[18:48] <blume> so was z.b
[18:48] <cbreak> freenetproject.org/snapshots/
[18:48] <Zorix> hmmm germans
[18:49] <cbreak> sollte aber in jedem snapshot zip file schon enthalten sein.
[18:49] <blume> yes,german##
[18:49] <d-ArkAngel> I'm stunned that I can follow what's being said, even if I couldn't construct a reply :-)
[18:49] <blume> and you?
[18:49] <Zorix> yea i can too i took 3 years of german in high school
[18:49] <Zorix> plus my family is german
[18:50] <blume> where you come from?
[18:50] <Zorix> my family comes from straubing
[18:50] <Zorix> like 1.5 hours north east of munich
[18:50] <d-ArkAngel> I took it for 5 years, tho the first three we had by far the worst teacher, and my german was very poor
[18:51] <i2p_iip> <gott> Anyone from the Rheinland-Pfalz area ?
[18:51] <Zorix> its been a long time and i get confused with another languages too.. im picking up norwegian because a lot of my irc friends are from norway
[18:52] <blume> und wo bekomme ich chatkan??le??
[18:52] <Zorix> what is that
[18:53] <cbreak> tippe /list
[18:53] <lostlogic> ugh, german
[18:53] <lostlogic> :-P
[18:53] <blume> gracias
[18:53] <lostlogic> yay, a bit of spanish... I know some of that one!
[18:53] <lostlogic> :)
[18:54] <Zorix> yea if i went outside enough.. i could learn from the hispanics
[18:54] <lostlogic> hehe
[18:54] <Zorix> but im too much of a nerd
[18:54] <lostlogic> I took 3 years of spanish ih HS, and have variously worked with spanish speakers on and off...
[18:55] <cbreak> hmm... : http://irc.netsplit.de/channels/?net=freenode
[19:00] * blume (~Franz@p5083B043.dip0.t-ipconnect.de) has left #freenet
[19:42] * pittaman (~mistery_b@d51A410CA.kabel.telenet.be) Quit (Client Quit)
[19:56] * Zxcvb (~somebody@client-216-176-88-60.consolidated.net) has joined #freenet
[19:56] <Zxcvb> good news, it looks like indymedia is looking at using freenet or something similar
[19:57] * d-ArkAngel fears the FBI turning up and stealing his HDD
[20:00] <Zxcvb> isn't that kind of thing what freenet is for?
[20:00] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) Quit (Read error: 110 (Connection timed out))
[20:01] <d-ArkAngel> freenet is for what ever you want to use it for.
[20:07] * TLF (~francisco@206.Red-81-40-113.pooles.rima-tde.net) has joined #freenet
[20:19] * TLF (~francisco@206.Red-81-40-113.pooles.rima-tde.net) Quit ("Adi?s a todos y a todas, y que les vaya bien. || http://www.vitaliano.esp.cc")
[20:23] * Zxcvb (~somebody@client-216-176-88-60.consolidated.net) Quit ("Leaving")
[20:25] * digitalch (~askipper2@itsb198.itsnpt.com) has joined #freenet
[20:38] * eMBee (~mbaehr@sirius.iaeste.tuwien.ac.at) Quit (Read error: 110 (Connection timed out))
[20:39] * eMBee (~mbaehr@sirius.iaeste.tuwien.ac.at) has joined #freenet
[20:57] * guido^pe (~unknown@dsl-082-082-159-151.arcor-ip.net) Quit (Remote closed the connection)
[21:03] * Zorix (Brandon@fl-198-70-221-44.dyn.sprint-hsd.net) has joined #freenet
[21:27] * Usurp (~Usurp@trudaine-8-82-230-34-86.fbx.proxad.net) Quit ("Leaving")
[22:11] * cbreak (~cbreak@dhcp-32-094.via-eth.ch) Quit ("leaving")
[22:44] * d-ArkAngel (~Rob@cpc2-midd2-5-0-cust16.midd.cable.ntl.com) Quit (Read error: 104 (Connection reset by peer))
[23:11] * pupok_ (~r00t@81-178-76-180.dsl.pipex.com) has joined #freenet
[23:11] * pupok (~r00t@81-178-76-180.dsl.pipex.com) Quit (Read error: 54 (Connection reset by peer))
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.
