Timestamps are in GMT/BST.
[0:05] * salahx (salahx@sc1-24.217.174.147.charter-stl.com) has joined #freenet
[0:23] <TheSeeker> dnf @ htl 15
[0:25] <TheSeeker> rnf @ htl 25
[0:30] <mikeDOTd> ouch
[0:31] <salahx> i tohugh maxHTL=20 ?
[0:36] * jmercier (~jmercier@4.10.164.111) has joined #freenet
[1:02] <TheSeeker> It's been 25 in my config for over a year...
[1:15] <lscd> salahx: it's documented that way, i think
[1:17] <TheSeeker> argh, I f'ng hate phone companies
[1:17] <TheSeeker> they input my address wrong and that's probably going to make my DSL order get backlogged another 2 weeks while they fix the damn paperwork.
[1:18] <salahx> SBC? :)
[1:18] <TheSeeker> Verizon
[1:18] <salahx> ack
[1:18] <salahx> Verizon = Evil
[1:18] <salahx> Versign = Evil
[1:18] <salahx> Note how simialr the two sound ? :)
[1:18] <TheSeeker> Ver* = Evil
[1:19] <TheSeeker> Veritas?
[1:20] <salahx> isn';t that an STD?
[1:20] <TheSeeker> lol
[1:23] <TheSeeker> http://www.veritas.com/aboutus/AboutUsHome.jhtml
[1:49] * Pascal666 (Pascal@c-24-13-17-191.client.comcast.net) has joined #Freenet
[1:58] <TheSeeker> Hops To Live: 4
[1:58] <TheSeeker> Error: Route Not Found
[1:58] <TheSeeker> Attempts were made to contact 0 nodes.
[1:58] <TheSeeker> 0 were totally unreachable.
[1:58] <TheSeeker> 0 restarted.
[1:58] <TheSeeker> 0 cleanly rejected.
[1:58] <TheSeeker> 48 backed off.
[2:43] * m[a]zzanet (~mazzanet@203-173-44-61.dyn.iinet.net.au) has joined #freenet
[2:50] * mazzanet (~mazzanet@203-206-239-92.dyn.iinet.net.au) Quit (Read error: 110 (Connection timed out))
[2:50] * m[a]zzanet is now known as mazzanet
[3:01] * aituioau (~bla@195.25.40.253) has joined #freenet
[3:01] <aituioau> Hi,
[3:02] <aituioau> I was asking if it is possible to do intranet tests with freenet. I mean to deploy a freenet network under an intranet. The idea is to test this topo under really less large network and send you the results.
[3:03] <ShaunMacPherson> that is a good idea
[3:03] <TheSeeker> toad is alseep, dun know if any other devs are awake
[3:03] <TheSeeker> sounds a lot like the testnet they have going already though
[3:03] <aituioau> Ok
[3:04] <ShaunMacPherson> I think someone asked about that in the past, having Freenet on a local university campus to reduce bandwidth, and have a small outside link to regular freenet
[3:04] <aituioau> ok like to contribute to this one ... have some machines which can be used for that
[3:04] <ShaunMacPherson> local university network
[3:10] <lostlogic> aituioau: yes, it is possible... We've also run various test nets and whatnot in the past.
[3:10] <lostlogic> I believe toad has a local test net of several machines each running several nodes
[3:11] <aituioau> ok When he will get up I'll ask him how to do :)
[3:11] <TheSeeker> I thought all the nodes were on the same machine...
[3:12] <lostlogic> TheSeeker: you may be correct...
[3:12] * Hory (~Miranda@81.196.25.110) has joined #FreeNet
[3:12] <TheSeeker> I miss the keyspace histogram charts that used to be in the stats pages...
[3:12] <lostlogic> aituioau: It's simple really ... create one node with a fresh configuration, on a port which you haven't already used for freenet on that machine, get it's node reference, use that to seed another node, get both of their refs to seed your next node, etc.
[3:13] <aituioau> how could I seed a node with fresh conf ?
[3:13] <aituioau> by default the seed file is full of ips
[3:13] <TheSeeker> "get this node's reference"
[3:14] <aituioau> do I need to erase those ips
[3:14] <aituioau> ?
[3:14] <TheSeeker> delete the seednodes.ref before you start.
[3:14] <lostlogic> aituioau: you would shut down the node, change it's listen port, remove seednodes.ref and all the routing related files from your freenet dir, and then start it back up
[3:14] <aituioau> ok. And get this node's ref is references inside the web front end of freenet ?
[3:15] <aituioau> ok thank you a lot lostlogic :)
[3:15] <lostlogic> aituioau: I believe so, but I couldn't swear to it... I know there is a command line argument to the freenet.node.Main to get the reference out, but I haven't done it in a longish time
[3:16] <lostlogic> and on that note, I think I'm going back to bed ... I was just on break from sleeping (it's 2:30am here ;-))
[3:16] <aituioau> lol
[3:16] <aituioau> ok
[3:16] <TheSeeker> central time...
[3:16] <aituioau> good night
[3:16] <lostlogic> have fun
[3:16] <lostlogic> g'night
[3:17] <Iakin> toad: re keeping the binaries and source separate.. I _do_ keeps the binaries in './lib', isn't that separate enough?
[3:22] <TheSeeker> I thought that at least the native biginteger stuff was going into freenet-ext?
[3:25] <aituioau> Couldn't we use this lib : http://jade.dautelle.com/
[3:27] <Iakin> TheSeeker: It will
[3:27] <Iakin> I think toad talked about how it was stored in cvs..
[3:28] <TheSeeker> oh
[3:38] <vsalento> should the file 'jcpuid.so' be in freenet-ext.jar or somewhere in the cvs?
[3:41] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[3:42] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[3:44] * salahx (salahx@sc1-24.217.174.147.charter-stl.com) Quit ()
[3:57] * bismuth (~bismuth@222.228.39-62.rev.gaoland.net) Quit (Read error: 104 (Connection reset by peer))
[4:11] <Pascal666> any particular reason CPUID.java has to contain unmappable characters for encoding UTF8?
[4:12] * bismuth (~bismuth@158.231.39-62.rev.gaoland.net) has joined #freenet
[4:31] * ShaunMacPherson (~ShaunMac@Sudbury-HSE-ppp3978722.sympatico.ca) Quit (Connection timed out)
[4:45] * aituioau (~bla@195.25.40.253) Quit ("Leaving")
[4:49] * Gunnar (~gunnar@34.80-203-209.nextgentel.com) Quit (Remote closed the connection)
[4:58] * taltipoa (~bla@195.25.40.253) has joined #freenet
[5:20] * bismuth (~bismuth@158.231.39-62.rev.gaoland.net) Quit (Remote closed the connection)
[5:25] <Pascal666> I'll take that as a no. Fixed now.
[5:45] * Hory (~Miranda@81.196.25.110) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[6:06] <Iakin> pascal, right.. where where those characters?
[6:06] <Iakin> vsalento: I currently neither
[6:06] * taltipoa (~bla@195.25.40.253) Quit (Connection timed out)
[6:06] * taltipoa (~bla@195.25.40.253) has joined #freenet
[6:07] <Iakin> It will be when someone has compiled it though
[6:16] * robilad[afk] is now known as robilad
[6:25] <Pascal666> Iakin: the abbreviation for micron
[6:45] * Hory (~Miranda@81.196.25.110) has joined #FreeNet
[6:47] <Iakin> Aha..
[7:04] <sanity> Iakin: if you like i can give you an account on dodo to compile linux stuff
[7:04] <sanity> Iakin: just email me with your ssh public key
[7:05] <sanity> how come I can only seem to find yesterday's TFE (last 3 days)?
[7:05] <sanity> perhaps CofE has set his clock incorrectly...
[7:08] * edt (~knoppix@dsl-139-28.aei.ca) has joined #freenet
[7:10] <sanity> wow, i am seeing search died probabilities all around 0.5-0.6 and search died times of around 30 seconds - 1 minue
[7:10] <sanity> s/minue/minute
[7:18] <toad_> hi sanity
[7:20] <toad_> <aituioau> I was asking if it is possible to do intranet tests with freenet. I mean to deploy a freenet network under an intranet. The idea is to test this topo under really less large network and send you the results. - well, I've been doing tests with 9 nodes on localhost recently
[7:20] <toad_> what would be more useful for testing would probably be a real "testnet" i.e. nodes on the real inet for more realistic test conditions, but still having overall control
[7:21] <toad_> a huge local-ish test network distributed over a lan IS an interesting idea though
[7:22] * toad_ hrrm
[7:22] <toad_> is 700 keys per node enough for testing? hrrm
[7:23] <toad_> <TheSeeker> I miss the keyspace histogram charts that used to be in the stats pages... - they still are
[7:23] * _dude (lebowskii@r403A.mag.afb.lu.se) has joined #freenet
[7:24] <TheSeeker> toad: how does one go about making their node specialized in one area of the keyspace? could I just delete the folders 0-4 and 6-f in my datastore dir to be specialized in "5" ?
[7:24] <TheSeeker> toad: I think I meant to say the barcode type charts ...
[7:24] <toad_> <Iakin> toad: re keeping the binaries and source separate.. I do keeps the binaries in './lib', isn't that separate enough? - that's not what I was told
[7:24] <toad_> but if so, cool
[7:27] <sanity> toad: can you get TFE on unstable?
[7:27] <taltipoa> | |
[7:27] <taltipoa> / X \
[7:27] <toad_> <aituioau> Couldn't we use this lib : http://jade.dautelle.com/ - I don't know, maybe
[7:27] <TheSeeker> the barcode chart of keys in our local DS was neato :D though I realize that the barcode charts for the nodes in the RT went away wen routing changed from local datastore to routing efficiency...
[7:28] <toad_> TheSeeker: no
[7:28] <toad_> because the folders don't reflect the keyspace spec
[7:28] <taltipoa> this is implementation of hashtable array and so one but much more speed
[7:28] <toad_> the folders are intended to be equally distributed
[7:28] <toad_> sanity: no
[7:28] <toad_> sanity: why should I be able to?
[7:29] <toad_> taltipoa: eh?
[7:29] <sanity> toad: well, i can get the previous day's edition, and have been able to for the last few days
[7:29] <toad_> TheSeeker: you want to delete all files called 1-[0-4]*
[7:29] <sanity> <sanity> wow, i am seeing search died probabilities all around 0.5-0.6 and search died times of around 30 seconds - 1 minute
[7:29] <toad_> and 1-[6-f]*
[7:29] <toad_> sanity: strange
[7:30] <sanity> what causes searches to die?
[7:30] <toad_> ummm
[7:30] <toad_> RNF i think
[7:30] <TheSeeker> toad: ok, thanks. that'd be useful for certain tests. what about the index file though?
[7:30] <toad_> could be wrong
[7:30] <toad_> TheSeeker: delete it
[7:31] * toad_ forces spec on his testbed
[7:31] <toad_> in order to do some tests to try to induce the SBKE bugs
[7:32] <toad_> sanity: unstable?
[7:33] <TheSeeker> I've left opera running and re-loading the "bondage fairies" freesite every 45 minutes for about five hours now ... it's now retrieving 5.5 MB when "done"
[7:33] <toad_> umm, I really wanted to know that
[7:34] <TheSeeker> well, it's a good site to test how well routing is working
[7:34] <TheSeeker> it has some 30 sub-pages that pre-fetch in i-frames
[7:34] * toad_ starts fetches, hopefully it will create the Bug on node #09
[7:34] <toad_> TheSeeker: woah
[7:35] <TheSeeker> each page has 18-22 images
[7:36] <toad_> okay, why isn't 7888 responding?
[7:36] <TheSeeker> TFE used to be a good site for testing back when all the "known good sites" didn't just have "go" for the icons, and DFI dissapeared as well, so I'm using the next best thing :P
[7:36] <toad_> significant idle time and yet :7888 isn't responding
[7:37] <toad_> the burning man site is reasonable for such things too
[7:37] <TheSeeker> oh?
[7:37] <TheSeeker> don't think I've seen that one.
[7:38] <toad_> lock contention here, there, and everywhere... :|
[7:39] <toad_> uh oh..
[7:39] <toad_> hmmm
[7:39] * toad_ reminds himself that stack dumps are not atomic
[7:40] <toad_> and just because it looks like 1.5 has THAT BUG with NPTL on, doesn't mean it does...
[7:41] <Iakin> toad: I was talking about in cvs.. maybe you are talking about in the jars?
[7:41] * toad_ tests the theory...
[7:41] <toad_> uh oh
[7:41] <toad_> it looks like it...
[7:41] <toad_> LOTS of idle
[7:41] <toad_> nothing in top, but that's normal with NPTL
[7:41] <toad_> sh*t
[7:42] <toad_> SH*T
[7:42] <TheSeeker> S*IT?
[7:42] <toad_> so why is it running the gzipper?
[7:42] <toad_> strange
[7:42] <toad_> SHIT!!
[7:42] <toad_> well, it looks to me like NPTL under Sun STILL has the EvilBastardBug
[7:42] <TheSeeker> o.O
[7:43] <toad_> which consists of the node just dying as a result of it losing track of an object that was supposed to be being unlocked
[7:43] <toad_> all threads wait for it despite the fact that it HAS been unlocked
[7:43] <toad_> we discovered this when debugging NIO, aaages ago
[7:43] <TheSeeker> ouch
[7:44] <toad_> I was assured that freenet DID work with NPTL
[7:44] <toad_> apparently it doesn't
[7:44] <toad_> not on 1.5.0b2 anyway
[7:44] * toad_ is REALLY ANNOYED !
[7:44] <TheSeeker> shaking angry fists at SUN probably didn't get much along the lines of fixes eh?
[7:45] <toad_> neither does telling them to just run freenet under NPTL for a day
[7:47] <TheSeeker> stupid Q ... what's NTPL?
[7:47] <TheSeeker> NPTL too
[7:47] * lostlogic attempts to build freenet-ext.jar and a new freenet.jar from CVS
[7:47] <TheSeeker> soemthing something Thread Locking?
[7:48] <toad_> TheSeeker: linux 2.6's alternate threading implementation
[7:48] <lostlogic> how would I know if it is using the native bigint?
[7:48] <TheSeeker> ah, nothing I'd be concerned about since I'm too dumb to use Linux properly :P
[7:48] <toad_> lostlogic: there's a message...
[7:48] * lostlogic watches for it.
[7:49] <TheSeeker> it'll either go ERROR!!! you suck! or INFO: yay, you're good to go!
[7:49] <TheSeeker> I think I paraphrased a little...
[7:49] <lostlogic> TheSeeker: figured :-P
[7:51] <Pascal666> Well I just managed to waste a few hours. Word of warning, do not do "yum install ant". The version you get is old and very buggy. You have to manually download and install ant if you want it to work.
[7:52] <lostlogic> I don't see any mention of bigint in the logs as the node started up ...
[7:52] <lostlogic> nvmd
[7:53] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[7:53] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[7:53] <TheSeeker> heh
[7:54] <lostlogic> it's in stderr
[7:54] <lostlogic> thought it woudl be in logs
[7:54] <lostlogic> where should the .so files be? in freenet-ext.jar?
[7:54] <TheSeeker> it's in my log, but I'm in windows...
[7:54] <TheSeeker> the paths were in freenet.jar when I looked.
[7:55] <TheSeeker> and when I copied the dlls into the correct paths in freenet.jar I got "INFO" instead of "WARNING"
[7:55] <lostlogic> hmm... ok
[7:57] <TheSeeker> toad: you mentioned a burning man site... do you have a link to it? I'm not finding any reference to "burning" on TFE.
[7:57] <TheSeeker> I suppsoe the mameroms site would be another good one, heh.
[7:58] <toad_> http://slaphack.com/freenet.log
[7:58] <toad_> horrible
[7:59] * toad_ would like to meet the idiot at sun that decided that NumberFormatException should be a RuntimeException and therefore unchecked someday
[7:59] <lostlogic> yikes
[7:59] <toad_> and congratulate him for his good design.. with a crowbar
[7:59] <toad_> actually... hrrm
[8:05] <lostlogic> damn, how long does nativebiginteger take to compile? looks like it is compiling it's own C library or something
[8:06] <Pascal666> toad: you try NPTL on 1.5 with -server ?
[8:08] * robilad hands toad_ a crowbar for rectal ractification of sun's engineers
[8:09] <toad_> Pascal666: I haven't tried -server recently because when I did try it it produced NPEs without stacktraces
[8:09] <Pascal666> TheSeeker: Native Posix Thread Library
[8:09] <toad_> I suppose I should try it again sometime
[8:09] <Iakin> ll: It takes about forever :)
[8:09] <toad_> what makes you think it will work with -server with NPTL when it doesn't work with client with NPTL?
[8:09] <lostlogic> Iakin: I'll kill you for this ... yer going to make me late for work ... right, it's your fault ;-)
[8:10] <Iakin> HEhe.. when I compile it under mingw in my VMWare virtual machine it takes a few hours (on my AthlonXP 2200+)
[8:10] <lostlogic> a few HOURS?
[8:10] <Iakin> Yup.. but it _has_ lousy I/O
[8:10] <TheSeeker> maybe you should go to work, then come home and see if it's done?
[8:11] <lostlogic> hmm... maybe I should go to work and finish getting nativebigint when I get home
[8:11] <lostlogic> hehe
[8:11] <lostlogic> ok, I'm off to do that ... talk to yous when I'm there or home or whatever
[8:12] <Pascal666> toad: I read something about thread optimizations with the server VM in 1.5 a bit ago. Not sure if it would be applicable or not
[8:12] * _dude (lebowskii@r403A.mag.afb.lu.se) Quit ()
[8:17] <toad_> http://www.newsforge.com/article.pl?sid=04/07/19/2315200
[8:17] <toad_> hrrm
[8:18] <toad_> well maybe allegations aren't a problem.. but interim relief may well be
[8:18] <TheSeeker> I forced specialization...
[8:18] <TheSeeker> peaks (count/mean)
[8:18] <TheSeeker> 0 --> (0.22857143)
[8:18] <TheSeeker> 9 --> (0.45714286)
[8:18] <TheSeeker> c --> (13.257143)
[8:18] <TheSeeker> e --> (0.22857143)
[8:19] <TheSeeker> :D
[8:19] <TheSeeker> I chose c because it had the highest success rate before (0.03something)
[8:20] <Pascal666> sleep
[8:22] * guido^pe (~unknown@dsl-213-023-232-066.arcor-ip.net) has joined #freenet
[8:24] <lostlogic> *L0gical
[8:24] <lostlogic> crap
[8:24] <lostlogic> well there goes that password.
[8:24] * greycat (~wooledg@192.35.79.70) has joined #freenet
[8:26] <TheSeeker> heh
[8:34] <sanity> lostlogic: its ok, it just showed up as ******* ;-)
[8:35] * lostlogic believes you :-P
[8:36] * lostlogic having difficulties remembering where he's used that password to change it :-(
[8:45] <toad_> hehe
[8:45] <toad_> sanity: did you get my email?
[8:46] <sanity> toad: yes, will read and respond now
[8:46] <toad_> well 2 emails
[8:46] <toad_> 9 Jul 21, 2004 2:04:34 PM (freenet.node.rt.NGRoutingTable, YThread-7, NORMAL): New fastest rate seen: 227708.33333333334 bytes per second
[8:46] <toad_> hehe
[8:46] <toad_> localtestnets are nice :)
[8:48] <TheSeeker> mmm, that's quick
[8:49] <sanity> toad: what is wrong with simply reporting a fake sample of 0 in pDNF when it first starts?
[8:50] <toad_> sanity: huh?
[8:50] <toad_> I'm not talking about "when it first starts"
[8:50] <toad_> we already have that covered
[8:50] <toad_> in any case we don't want it to return 0, that's completely absurd
[8:51] <toad_> I was trying to get away from the "initial report" stuff, it's hideously inaccurate
[8:52] <sanity> ok#
[8:52] <sanity> well, what is the problem right now?
[8:52] <toad_> edge kludging
[8:53] <toad_> it makes the graphs look horrible, and it skews the typical estimates
[8:53] <sanity> yes, but is it having no successes and therefore returning pDNF of 0?
[8:53] <toad_> 1
[8:53] <toad_> DNF is a failure, so it retturns pDNF of 1.. it would, if we didn't have edge kludging
[8:53] <toad_> this would result in the estimate being infinite
[8:53] <sanity> what if the estimator estimates 1/pDNF rather than pDNF ?
[8:54] <TheSeeker> hmm... when your node is overloaded, and you tell people "back of for N seconds" ... do you give a lower N as you approach when you'll be not backed off, or do you give the same N all the way up to the point you aren't backed off anymore?
[8:54] <toad_> then it would have problems at the other end
[8:54] <toad_> and we'd still end up dividing by zero
[8:54] <toad_> in longEstimate()
[8:54] <toad_> and we'd have to report infinities every time we got a success :<
[8:55] <toad_> what do you think of the non-alchemical option I suggest at the end of the first mail?
[8:56] <toad_> Single hop probability of search timeout 0.48947368421052634
[8:56] <toad_> hrrm
[8:56] <toad_> i think that's just the RNFs...
[8:58] <sanity> toad: yeah, that looks ok to me at first glance
[8:58] <toad_> TheSeeker: neither, you tell them "send me requests every N seconds"
[9:00] <toad_> sanity: does the inverting average bit of it seem sane?
[9:01] <toad_> i think it emphasizes nodes that perform very well in a small area
[9:01] <toad_> which is probably what we want
[9:01] <toad_> also what about the issues raised in the following mail?
[9:04] <sanity> i am a little concerned about the notion of comparing nodes with infinite estimates - not sure it is the best solution
[9:04] <sanity> drafting an email to devl now
[9:04] <toad_> any alternatives?
[9:04] <sanity> not yet, still getting my head around the issue
[9:04] <toad_> any pDNF value other than 1.0 for a node that has not received any successes but has received DNFs is alchemy...
[9:04] <toad_> and the alchemy I implemented seems rather destructive
[9:05] * Ash-Fox (Hal-9000@aav39.neoplus.adsl.tpnet.pl) Quit (Read error: 110 (Connection timed out))
[9:05] <toad_> another possibility would be different alchemy - I outlines a few possibilities
[9:05] <toad_> a third option would be to reevaluate the estimation formula of tSuccess + tFailure / pSuccess
[9:06] <sanity> i am swaying towards the third option
[9:06] <toad_> wonderful :|
[9:07] <toad_> AFAICS anything else will be alchemy
[9:07] <toad_> brb
[9:07] <sanity> well, the core problem is the assumption: "if a node has not yet succeeded, we can infer that it never will"
[9:07] * Ash-Fox (Hal-9000@aak18.neoplus.adsl.tpnet.pl) has joined #FREENET
[9:10] <toad_> sanity: we can't infer anything else
[9:10] <toad_> that's the problem
[9:11] <toad_> sanity: will you be around today?
[9:11] <toad_> i.e. for the coming few hours, at least by email?
[9:12] <toad_> my epileptic and largely housebound brother in law wants to play starcraft
[9:16] <sanity> toad: i will be around
[9:17] <toad_> ok
[9:17] <sanity> one sec
[9:17] <sanity> what about initialising pDNF to the global average pDNF?
[9:17] <toad_> that would be de facto optimistic init
[9:18] <sanity> not very optomistic...
[9:18] <toad_> just because new nodes have better than average MRIs - if they have average estimates as well, we'll route to them a LOT
[9:18] <sanity> is that such a bad thing?
[9:18] <toad_> if that's not a bad thing, we have to severely limit the rate at which we add them to the RT
[9:19] <toad_> which is possible
[9:19] <toad_> also...
[9:19] <toad_> we need them to learn pretty fast
[9:19] <sanity> which they will do if they are getting lots of requests
[9:19] <toad_> yes but we don't want them stuck around the init point
[9:19] <sanity> "init point"?
[9:20] <toad_> the global averag
[9:20] <toad_> e
[9:20] <toad_> if the global avg is 0.95, and a node gets 10 failures in a sector, what value should that sector return/
[9:20] <toad_> ?
[9:21] <sanity> close to 0.95 i would say, given that 10 failures in a row isn't that unlikely with a pDNF of 0.95
[9:22] <sanity> in fact, there is a 60% chance that a given 10 requests will fail with a pDNF of 0.95
[9:23] <toad_> well what would the formula be?
[9:23] <sanity> formula for what?
[9:24] <toad_> for the sector return value
[9:24] <sanity> same thing it is now, except where the sector was initially initialized to 0.95
[9:24] * Iakin thinks we should introduce some kind of 'lock' file..
[9:24] <toad_> how would that happen?
[9:25] <toad_> I mean, currently we use SimpleBinaryRunningAverage's
[9:25] <toad_> initialization to 0.95 would just mean that if there are NO reports it returns 0.95
[9:25] <sanity> ok, well the question is now many node-specific reports do we need before we think that we know enough to deviate from the global pDNF
[9:25] <toad_> that's an odd way of looking at it
[9:25] <toad_> Iakin: well volunteered :)
[9:25] <Iakin> This after having spent some time debugging the cause of 'could not write out node'-file, when the actual cause seems to have been me having spawned two nodes...
[9:26] <sanity> brb - phone call
[9:26] <toad_> we can hack SBRA up to support a single non-binary report
[9:26] <toad_> if it's absolutely necessary
[9:27] <toad_> if we did, the above would return (10*1.0 + 1*0.95)/11 = 0.995
[9:27] <toad_> yay, I replicated the SBKE bug!
[9:27] <Iakin> Has anyone an idea of how this is done? Do I create a file that contains the PID of the node and when checking for node-file validity checking wheter or not the process still exists?
[9:28] <toad_> sanity: please summarize your current thinking on this on devl anyway
[9:28] <toad_> bbl
[9:28] <toad_> Iakin: can we get the PID from java? if so that's sensible
[9:28] <toad_> also I get the node rename errors on linux too :|
[9:28] <toad_> bbl
[9:29] <Iakin> Or hmm. maybe I could let the creating node keep the file open..
[9:29] <Iakin> And check wheter or not it is openable in exclusive mode to verify that the owner still exists..
[9:51] <lolo-laptop> mang, my logs are just flooded...
[9:52] <toad_> bbl
[9:52] * toad_ (toad@82-32-16-91.cable.ubr03.azte.blueyonder.co.uk) Quit ("using sirc version 2.211+KSIRC/1.3.10")
[9:55] <lolo-laptop> Iakin: hmm... what do I need to do to get jcpuid working on linux? the Contrib comes with the windows version ... I'm confused
[10:02] <Iakin> lolo: Just compile the two files together
[10:02] <Iakin> (i haven't yet written up a build.sh for linux)
[10:02] <lolo-laptop> hmm... ok
[10:02] <Iakin> The actual code in the .cpp and .h file should compile nicely on linux
[10:02] * hirvox would prefer a Makefile
[10:02] <Iakin> hirvox: Feel free :)
[10:03] <lolo-laptop> I'm really dumb ... I have like no idea where to include the native FEC or native jbigi or anything so that it works, blah.
[10:03] <Iakin> It would be good if it could be used to produce .dll's in mingw and .so's in linux
[10:03] <hirvox> I have only basic knowledge about c++, and even less knowledge about Makefiles
[10:04] <Iakin> ll: Fir jbigi, either put it in your LD_LIBRARY_PATH or in the net/i2p/Util/ folder in freenet-ext.jar
[10:04] <Iakin> I recommend the second
[10:04] <Iakin> But it wont work without jcpuid..
[10:05] <Iakin> Or well.. you will then get the generic jbigi-none.so
[10:05] <Iakin> instead of one that is better for your CPU
[10:06] <lolo-laptop> nod.
[10:13] <lolo-laptop> src/jcpuid.cpp:12: error: `_asm' undeclared (first use this function)
[10:18] <Iakin> ll: _that_ was interesting.. now, how do you do that in GCC?
[10:18] <lolo-laptop> what is it that I need to be doing in GCC?
[10:19] <lolo-laptop> hmm...
[10:19] <lolo-laptop> never done embedded assembly...
[10:24] * Iakin committed build 60165.. Now nodes will create a lock-file that prevents the user from accidentially spawning multiple instances of a node
[10:24] <lolo-laptop> what causes the lock file to be deleted?
[10:24] <hirvox> iakin:isn't that the job of startup scripts?
[10:25] <Iakin> hirvox, there aren't any in use on windows really
[10:25] <Iakin> ll: try removing the '_' from the '_asm
[10:25] <Iakin> '
[10:25] <hirvox> not even freenet.exe?
[10:26] <Iakin> Mmmmm.. well, I am not really into that part of the code..
[10:26] <lolo-laptop> src/jcpuid.cpp:12: error: parse error before `{' token
[10:26] <lolo-laptop> hmm...
[10:26] <Iakin> Is it _very_ wrong to let the application itself do it?
[10:26] <Iakin> I thing eclipse does it that way
[10:27] <hirvox> Iakin:is the location/use of the lockfile configurable?
[10:27] * tyesinclair (~tyesincla@170.252.64.1) has joined #freenet
[10:27] * tyesinclair (~tyesincla@170.252.64.1) has left #freenet
[10:28] <hirvox> my startup script already makes a lockfile in /var/lock/subsys/fred, which contains the node's pid to detect whether it's running or not
[10:28] <Iakin> yuck, GCC inline assembly sucks
[10:28] <Iakin> hirvox, not right now..
[10:29] <Iakin> hirvox: maybe I should create a config param for enabling disabling of the internal lock file?
[10:29] <hirvox> Iakin:and does it handle stale lockfiles?
[10:30] <Iakin> hirvox: The code I wrote used the OS's file locking capabilities to handle that situation
[10:30] <hirvox> I mean that if (when) the JVM crashes, will the locking code stop the node from restarting?
[10:30] <Iakin> It dont care about the presence of the file really, it cares about wheter or not it can lock it for exclusive use
[10:31] <Iakin> That _that_ is something that the OS manages
[10:31] <Iakin> So.. the answer to the question is.. no.. it will still work ok if the JVM chrases or is killed by the user
[10:31] <Iakin> ok?
[10:32] <lolo-laptop> Iakin: what's iFunction?
[10:32] <hirvox> what about freenet.node.Main --export or --manual while the node is running?
[10:32] <Iakin> ll: the CPUID instruction has multiple modes of operation.
[10:33] <Iakin> the 'iFunction' integer is put into EAX to choose the desired mode of CPUID
[10:34] <hirvox> afk
[10:35] <lolo-laptop> got it compiling
[10:35] <Iakin> hirvox: That I haven't tried really.. but the code is executed during Node.init(), so I think that should work
[10:36] <Iakin> Node.Main() handles the --export & co
[10:36] <Iakin> ll: care to commit a fix?
[10:36] <lolo-laptop> replace _asm with asm, replace {} with (), turn the assembly into strings, and put %s in front of register names, and use ;s
[10:36] <lolo-laptop> Iakin: it'll probably almost definitely break vc++,eh?
[10:39] <Iakin> Heh.. probably
[10:39] <Iakin> grr.. but it aint _that_ much asm in there, right?
[10:40] <Iakin> Maybe one could do an #ifdef _WINDOWS
[10:41] <Iakin> toad: Seems like something causes the node-file to be rewritten quite often..
[10:41] <Iakin> There might be a need for a if(!oldAddress.equals(newAddress)) somewhere in the autodetenction code
[10:44] <lolo-laptop> I don't seem to have write access to Contrib ...
[10:44] <lolo-laptop> oh, I got it as anonymous ...
[10:44] <lolo-laptop> hold up
[10:47] <lolo-laptop> ok
[10:59] <lolo-laptop> woot, I think I did it
[10:59] <lolo-laptop> or not
[10:59] <lolo-laptop> java.lang.UnsatisfiedLinkError: /var/freenet/store/temp/jcpuid12765lib.tmp: /var
[10:59] <lolo-laptop> /freenet/store/temp/jcpuid12765lib.tmp: undefined symbol: iFunction
[11:10] <lolo-laptop> Iakin: http://www-106.ibm.com/developerworks/linux/library/l-ia.html hehe, cpuid example *chuckle*
[11:16] <hirvox> hmm, 4 hours uptime with 128MB of max allocatable memory (61MB in use), and no problems so far
[11:17] <lolo-laptop> INFO: Native CPUID library jcpuid loaded from resource
[11:17] <lolo-laptop> INFO: Non-optimized native BigInteger library 'jbigi-linux-none.so' loaded from resource
[11:17] <lolo-laptop> :-( all that work and I must have done something wrong
[11:23] * taltipoa is now known as SadPeng
[11:24] * SadPeng is now known as taltipoa
[11:24] <TheSeeker> which .so did you copy for bigi?
[11:24] <TheSeeker> %*$!@ I just list 7 TINY screws >.o
[11:24] <TheSeeker> lost
[11:25] * edt (~knoppix@dsl-139-28.aei.ca) Quit (Read error: 110 (Connection timed out))
[11:26] * edt (~knoppix@dsl-130-81.aei.ca) has joined #freenet
[11:28] <TheSeeker> yay, found them all. I took my watch apart to change the battery :P
[11:34] <lolo-laptop> yikes
[11:34] <lolo-laptop> I copied all of the jbigi libraries into the thingus
[11:42] <TheSeeker> only copy the one you want to use
[11:47] * raVen| (~foo@80.65.240.198) has joined #freenet
[11:47] * raVen| is now known as raven
[11:50] <raven> i've had my node running for 24 hours now and i'm still getting "Route Not Found" for all the bookmarks. Is this normal?
[11:51] <Iakin> OR all, that should work too
[11:51] <TheSeeker> stable? unstable? how many nodes in your RT?
[11:51] <raven> stable
[11:52] <Iakin> lll: so.. something went wrong with the compilation
[11:52] <raven> i'm running the 1 node on my computer at the moment
[11:52] <raven> i am behind a firewall
[11:53] <raven> does freenet need me to open up the firewall
[11:53] <Iakin> yes.. the 'listenPort' should be opened
[11:53] <Iakin> (for inbound TCP connections)
[11:54] <Iakin> And, if the FW is on another machine than freenet, you need to forward that port to that machine
[11:58] <raven> yer i set up the correct firewall rule yesterday. but im still not getting any "routes found"
[11:58] <raven> the thumbnails are loading however...
[12:00] <raven> i'll try unstable i thinl
[12:00] <raven> think*
[12:00] <TheSeeker> do you have any incoming connections on your OCM page?
[12:02] <raven> a few
[12:02] <raven> i think it probably needs more time to find routes i guess
[12:12] * vivimul (~ke_navo@ARennes-303-1-25-191.w81-53.abo.wanadoo.fr) has joined #freenet
[12:16] * vivimul (~ke_navo@ARennes-303-1-25-191.w81-53.abo.wanadoo.fr) Quit ()
[12:20] <TheSeeker> well, incoming connections means you did get your port forwarding done right at least...
[12:45] * toad_ (toad@82-32-16-91.cable.ubr03.azte.blueyonder.co.uk) has joined #freenet
[12:45] <Iakin> How interesting that my node shows the highest probability of success for an incoming request at the very same area in the keyspace as for which it has revieved the least requests
[12:45] <Iakin> bbl
[12:45] <toad_> <Iakin> How interesting that my node shows the highest probability of success for an incoming request at the very same area in the keyspace as for which it has revieved the least requests
[12:45] <toad_> hmmm
[12:45] <Iakin> toad, I committed the 'lockfile.lck' code
[12:45] <toad_> I've heard this before... many times
[12:45] <toad_> I wonder why
[12:45] <toad_> Iakin: how does it work?
[12:46] <Iakin> Note! The node has been up for < 20 minutes
[12:46] <Iakin> new FileOutputStream('lockfile.lck').getChannel().tryLock()
[12:47] <Iakin> So.. it lets the OS handle staleness of lockfiles..
[12:47] <Iakin> bbl
[12:48] * robilad is now known as robilad[afk]
[13:05] * toad_ debugs inconsistent errors in estimators
[13:08] <TheSeeker> fun fun
[13:10] <toad_> hmmm
[13:11] <toad_> i think it's bogus
[13:11] <toad_> and debugging can indeed be fun
[13:12] <toad_> # START CENTRE
[13:12] <toad_> 0 739cfb76fd9d1aab49083fdac2cee20f75ab36c5a9b65 ba5b5dfba75527f611770ff911bec2a97b820b21601a8
[13:12] <toad_> 1 10119c080510d3540d9e5e01760aea3438158df7d167ec 141aaf75a0eed606861a167470e06932f64aaf514e3cfe
[13:12] <toad_> 2 1823c2e33cccd8b8fe95cee76bb5e831b47fd0aacb1211 1e6b4b0ac18b345e205cf7f072ac792f952fe5472f4c54
[13:12] <toad_> 3 24b2d33246499003422420f979a30a2d75dff9e3938697 33e74ca2705b277ceecdf3c403e97ab245ca2c3abbf1c5
[13:12] <toad_> 4 431bc6129a6cbef69b77c68e8e2feb3715b45e91e45cf3 48e7068284f10045b421ea9cded7cc246d93ebc08e9196
[13:12] <toad_> ....
[13:12] <toad_> 14 e000000000000000000000000000000000000000000001 ec9bb61fdc47f837c72545002f65bf36526a502423d88a
[13:12] <toad_> 15 f9376c3fb88ff06f8e4a8a005ecb7e6ca4d4a04847b114 389dfb9434e10d216d86ff057c3646ce17a9da51263c
[13:12] <greycat> my god! it's full of hex!
[13:12] <toad_> view on fixed space font with really wide screen for it to make sense
[13:13] <toad_> the problem here is that it seems ordered by start rather than by centre
[13:13] <toad_> otherwise it'd be consistent
[13:15] <toad_> woah
[13:15] <toad_> that was dumb
[13:15] <toad_> we were never calling correctSlide()
[13:18] * Hory (~Miranda@81.196.25.110) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[13:26] * toad_ starts localtestnet
[13:28] <toad_> ooooh
[13:28] <toad_> Caught java.io.IOException: already closed in process()
[13:29] <toad_> on the localtestnet!
[13:29] <toad_> where there can't possibly be any link congestion
[13:31] <toad_> hmmm
[13:31] <toad_> i suppose it might be a collision
[13:32] <toad_> yeah, it must be
[13:36] * toad_ runs some more tests
[13:38] <toad_> so far success ratio around 85% just as last time
[13:38] <toad_> hrrm
[13:38] * toad_ suspicious...
[13:38] <toad_> hmmm, not this minute
[13:41] * toad_ committed 60166
[13:46] * toad_ changes topic to 'Upgrade to 5084 | Unstable: Upgrade to 60166 (60160: network level IP detection should work now); network reset at 60151 so 60151 is mandatory | Unstable nodes must be updated daily | Stable nodes should generally be updated weekly | #freenet-chat and #freenet-politics are open for business! Off topic discussions may be directed there. | Logs: http://newton.matcmp.ncc.edu/~lockej/freenet/chanlog/'
[13:46] <toad_> hmmm
[13:48] <toad_> if I implement sanity's suggestion... and do 1000 simulated reports of the average on the estimators on creation...
[13:48] <TheSeeker> 85% sounds like an improvement...
[13:48] <toad_> TheSeeker: it's a pretty contrived test scenario
[13:48] <TheSeeker> yeah, but I thought you were getting ~45% before on the testnet
[13:50] <toad_> yup
[13:50] <TheSeeker> anyway, you were in the middle of a thought, sorry.
[13:50] <toad_> it was 40%
[13:50] <toad_> then it was 70%
[13:50] <toad_> then it was 85%
[13:50] <toad_> when it was 70%, it would get up to 70% and then fall
[13:51] <toad_> well..
[13:51] <toad_> firstly, it will be a problem on testnets, and completely new nodes
[13:54] <vsalento> hirvox: what is the status of the templates?
[14:10] * greycat waves to toad
[14:10] <toad_> hi greycat
[14:11] <toad_> have you tried the latest unstable builds yet?
[14:11] <greycat> no, I haven't tried unstable for a very long time
[14:11] <toad_> yesterday's work fixed some ooold insert bugs you were grumbling about
[14:12] <lolo-laptop> hmm... Iakin's lockfile code creates the lockfile in cwd ... ugh?
[14:13] * Toast (~k@82-34-170-133.cable.ubr03.maid.blueyonder.co.uk) has joined #freenet
[14:15] <toad_> lolo-laptop: where else should it be created?
[14:16] <toad_> routingDir perhaps...
[14:16] <lolo-laptop> hmm... not sure, but cwd is annoying on systems which run freenet as a service and therefore dont' expect cwd to be used at all...
[14:16] <lolo-laptop> yeah, routingDir would make much more sense
[14:20] <TheSeeker> what's a cwd?
[14:20] <TheSeeker> :P
[14:21] <lolo-laptop> Current Working Directory
[14:21] <TheSeeker> ah
[14:31] * Hory (~Miranda@81.196.25.110) has joined #FreeNet
[14:34] * toad_ is going to make it cwd
[14:34] <toad_> err
[14:34] <toad_> routingDir
[14:34] <lolo-laptop> kewlies
[14:41] <toad_> hmmm
[14:41] <toad_> how long will it typically take for a node to get 1,600 hits?
[14:42] <toad_> a node in my routing table?
[14:43] * toad_ suspects a looooooong time
[14:43] <toad_> I have 200 nodes in my RT
[14:43] <toad_> I send maybe 1000 requests per hour
[14:43] <toad_> assuming a node got even 10% of that... 100 requests per hour... it would take nearly a day (!!)
[14:44] <toad_> if we're dealing with that sort of timescale we probably want to reintroduce ARKs, although bidi serves a similar function unless both sides change address
[14:50] <toad_> hmmm
[14:50] <toad_> wonderful
[14:50] <toad_> my top nodes' routinghits: 536, 375, 1551, 974
[14:51] <toad_> however if you add in all the other failure modes...
[14:51] <toad_> then they're all well above the threshold
[15:10] <leexgx> you talking to your self
[15:11] <leexgx> :)
[15:11] <toad_> :)
[15:13] <verl> hrm! please continue :)
[15:13] <leexgx> toad_ i think that in 6 months stable is now alot faster (as in time i have not used it) freesites load faster adn i only need to run it for an day before it runs smoth
[15:14] <leexgx> i agree :)
[15:14] <toad_> leexgx: good
[15:15] <leexgx> toad_ you know why i have to wipe out my freenet folder to get rid of ipadress *.*.*.*:-1 error in the log (right click on icon and look at log right at the start)
[15:15] <toad_> leexgx: huh?
[15:16] <leexgx> when i upgrade some times i get that error
[15:16] <leexgx> even when i put the stuff in the free net ini file
[15:17] <leexgx> the bit that gets my att is the {:-1} on the end of the ipadress
[15:17] <leexgx> this has been an bug i have had from last time of useing it
[15:17] <toad_> can you show me exactly what it said?
[15:17] <leexgx> 1 sec
[15:17] <leexgx> looks tho logs
[15:18] <leexgx> it mosty did it when i whent from unstable to stable but last time i did it it buged me for 3 hrs before i just wiped out the freenet folder and reinstalled (aka not the log)
[15:19] <TheSeeker> is that the listenport?
[15:19] <leexgx> i see if i can make it do it
[15:20] <leexgx> its in two lines as well
[15:21] * toad_ discovers another argument for all-files-are-16k
[15:21] <toad_> or 4k or whatever
[15:21] <toad_> fixed file size
[15:22] <toad_> 1) simplify routing slightly, 2) make traffic analysis a lot harder, and now 3) make nodes learn faster and therefore cease to be newbies faster
[15:23] * toad_ thinks there's a reasonable case
[15:23] <toad_> the question is do we do it now or do we do it in 2.0?
[15:24] <toad_> or in 0.7?
[15:25] <hirvox> aaargh
[15:25] <toad_> hirvox: hmm?
[15:25] <hirvox> (freenet.node.Node, main, ERROR): Lockfile '/etc/fred/lock.lck' doesn't exist after we tried to create it. Permissions problem? Shutting
[15:25] <hirvox> down.
[15:26] <toad_> hehe
[15:26] <toad_> i have local code to put it in routingDir
[15:26] <toad_> i assume you would like me to commit it?
[15:26] <hirvox> yes
[15:27] <toad_> do you have any opinion on the edge kludging debate?
[15:27] <toad_> there's been little input probably because it's Deep Magic
[15:28] <hirvox> not really, I haven't actively followed the mailing lists in the last few days
[15:28] * toad_ thinks ideally we'd have all files are 4kB.. but I don't think that's feasible with current CPUs and current lack of fred optimization
[15:34] * toad_ commits 60167
[15:35] <toad_> CANNOT RENAME NODE FILE node-temp TO node
[15:35] <toad_> still getting it on the testnet
[15:35] <toad_> nodes
[15:35] <toad_> despite the locking code...
[15:35] <toad_> anyway, committed
[15:36] <hirvox> thx
[15:40] * lolo-laptop updated.
[15:40] * KenManAway is now known as KenManAlive
[15:41] <toad_> there are more bugs to fix of course...
[15:41] <toad_> but merging is more important
[15:41] <toad_> KenManAlive: hi
[15:41] <KenManAlive> greetings...
[15:41] <KenManAlive> how are things ?
[15:41] <toad_> KenManAlive: somebody else said that after returning from the dead :)
[15:41] * KenManAlive is now known as KenMan
[15:42] <toad_> KenManAlive: well, I need your input on the edge kludging debate on devl
[15:42] <toad_> and your opinion on whether and what we should merge to stable
[15:42] <KenMan> okay, okay. I just returned from the dead - give me an hour or a day ;)
[15:44] <KenMan> anybody here have a node that has been running for more than a day ? if so, how many incoming xfers do you have ?
[15:44] * toad_ thinks we should merge most of the non-routing changes to stable, at least
[15:44] <toad_> KenMan: no, sorry
[15:44] <toad_> just started it up
[15:44] <KenMan> (hint) keep your eye on it over time.
[15:46] <KenMan> only 110 new devl mails in my (devl)Inbox... oh well, activity is good :)
[15:48] <toad_> woah
[15:48] <toad_> we have a LOT of stuff to merge
[15:48] <KenMan> not surprising. how much of it can be done without a reset ?
[15:48] <KenMan> i mean, is that a significant barrier ?
[15:48] <toad_> oooh
[15:48] <toad_> looks like it's going to 100%!
[15:49] <toad_> the testnet test...
[15:49] <toad_> or perhaps not...
[15:49] <toad_> last hour, 42 out of 44 successes
[15:50] <toad_> 7/21/04 7:00:00 PM BST 104 78 0.75
[15:50] <toad_> 7/21/04 8:00:00 PM BST 476 414 0.8697478991596639
[15:50] <toad_> 7/21/04 9:00:00 PM BST 44 42 0.9545454545454546
[15:50] <KenMan> wow!! what kind of configurations have you been trying ??
[15:50] <toad_> that's the testnet with 1 requestor node and 8 pre-specialized nodes
[15:50] <toad_> i'm not convinced that we should just merge the lot though
[15:51] <toad_> not until we have sorted out the edge kludging issue
[15:51] <KenMan> ahh, looks like things have gotten better... are you able to barcodize the datastores for a visual ?
[15:51] <toad_> hmm?
[15:51] <toad_> ah
[15:51] * toad_ checks
[15:51] <lolo-laptop> hmm...
[15:51] <lolo-laptop> I'm getting a lot of NRFs
[15:51] <lolo-laptop> RNFS
[15:52] <toad_> very impressive
[15:52] <toad_> not perfect but impressive
[15:52] <toad_> lolo-laptop: unstable?
[15:52] <lolo-laptop> yah ...
[15:52] <lolo-laptop> Jul 21, 2004 3:10:11 PM (freenet.node.http.DistributionServlet$DistributionRequest, YThread-98, NORMAL): RouteNotFound Fetching (running) freenet:CHK@7fWN2TWdHoIODoxCeD1OebVdUB8QAwI,uLq58BoeKATpIJFa8PCiJA as freenet.exe for DistributionServlet
[15:52] <toad_> brb
[15:57] * toad_ back
[15:57] <toad_> lolo-laptop: you just started?
[15:57] <toad_> if so, perfectly normal
[15:58] <toad_> so here is the question: merge what we have, with radically improved routing, or wait until we solve the edge kludging problem?
[15:59] <toad_> the problem with the edge kludging thing is that it will mess up the discard sort order...
[15:59] <toad_> this matters a lot on the stable network because the stable network is vast
[15:59] <greycat> if it can insert FEC splitfiles, make it go into stable please.
[16:00] <toad_> the insert fixes are definitely needed
[16:00] <toad_> ditto for on-net IP detection
[16:02] * toad_ thinks he will merge everything except the routing changes
[16:06] <KenMan> how soon do you anticipate having suitably fixed routing changes available for stable ? more than a week ?
[16:07] <toad_> could do it now if I wasn't worried about edge kludging
[16:07] <KenMan> ahh, well, then push everything else, and we will focus on the edge problem(s)... meaning , do the reset now.
[16:08] <KenMan> or is that part of edge kludging ? perhaps i don't understand it well enough :(
[16:08] <toad_> no, we probably don't need a reset
[16:09] <KenMan> fair enough. We should guesstimate that it will take 2 to 4 weeks for the majority of stable to upgrade past any reset point...
[16:09] <toad_> well
[16:09] <toad_> we may need a reset
[16:09] <toad_> but it'd be a big deal on stable
[16:10] <KenMan> best to get it over with now and start that user switch-over process asap... that's my simple opinion.
[16:10] <KenMan> if the reset comes first, then users will see that a newer version is available, when it is ready.
[16:12] <KenMan> that assumes that many nodes will make the switch to the reset network (they won't see the new version on the 5084 net, right? ) !!
[16:13] <KenMan> how could we advertise it ? run a handful of nodes that claim to be 5085 ? thus encouraging an upgrade. i imagine that will create user troubles too.
[16:13] <toad_> or we could just not reset, and up the lastGoodBuild after a week or so
[16:14] <KenMan> yeah... not a bad idea. "See" how many we get initially, percentage-wise ...
[16:14] <toad_> bbl
[16:14] * leexgx back
[16:15] <leexgx> cant seemt to make it happen
[16:16] <toad_> bbl
[16:16] <leexgx> {IPadress=<number here>:-1 is not an vaild ipadress } plus the rest that goes on about put it in the freenet.ini file and it is rare but
[16:16] <leexgx> cya
[16:16] <leexgx> bug
[16:22] <Iakin> Hey hirvox, at least I made a good log message when lockfile creation fails :)
[16:22] <hirvox> I warned you, didn't I?-)
[16:23] <Iakin> toad, re CANNOT RENAME...
[16:23] <Iakin> cant it be due to no synchronization on writeNodeFile() or something?
[16:25] * lscd (~lscd@adsl-213-180-182-6.cybernet.ch) Quit ("later")
[16:27] <Iakin> hirvox: You did :) Well.. always good that someone sees the potential problems :)
[16:30] <hirvox> Iakin:I had my share of problems when fred assumed that the config file always lived in the home directory :p
[16:35] <Iakin> toad: How would I do to get my nodeto open up more outbound connections?
[16:39] <leexgx> modify the freenet.ini file
[16:40] <leexgx> look for max connecions some where
[16:40] <leexgx> i reloading my windows now
[16:40] <leexgx> or will be in an sec
[16:41] <Iakin> Well.. I have upped it to around 1000..
[16:42] <Iakin> But my node still refuses to open more than a few (around 50 or so)
[16:43] * robilad[afk] (~topic@mpiat2304.ag2.mpi-sb.mpg.de) Quit (Remote closed the connection)
[16:43] * greycat (~wooledg@192.35.79.70) Quit ("Lick Bush in '04")
[16:44] <lolo-laptop> Iakin: what processors should doohickus detect ...
[16:44] <lolo-laptop> it seems to not work right on my axp ...
[16:45] <Iakin3> athlon, p2, p3 and p4
[16:45] <Iakin3> (as of now)
[16:45] <lolo-laptop> athlon includes athlon-xp
[16:45] <lolo-laptop> ?
[16:45] <lolo-laptop> INFO: Native CPUID library jcpuid loaded from resource
[16:45] <lolo-laptop> INFO: Non-optimized native BigInteger library 'jbigi-linux-none.so' loaded from resource
[16:45] <Iakin3> yup
[16:45] <lolo-laptop> :-(
[16:45] <Iakin3> Hmmm..
[16:46] <Iakin3> That really ought to work
[16:46] <lolo-laptop> did you look at how I modified the cpp and how I built it in CVS for Contrib?
[16:46] <Iakin3> public boolean IsAthlonCompatible()
[16:46] <Iakin3> {
[16:46] <Iakin3> return getCPUFamily() >= 6;
[16:46] <Iakin3> }
[16:47] <lolo-laptop> it works on your xp right?
[16:47] <Iakin3> On my MP at least
[16:47] <TheSeeker> Athlon Compatible ... :D
[16:47] <Iakin3> But that should be the same thing
[16:47] <lolo-laptop> yeah
[16:47] <TheSeeker> my XP gets reported as an MP for some reason
[16:47] <lolo-laptop> TheSeeker: they are the same processor... MP is tested better and labelled differently
[16:48] <lolo-laptop> Iakin3: hmm... I must have broken it then :-(
[16:48] <Iakin3> Dont you linux people have some command for displaying the CPU family
[16:48] <lolo-laptop> cpuinfo
[16:48] <Iakin3> maybe dmesg or something
[16:48] <Iakin3> What does it say?
[16:48] * robilad[afk] (~topic@mpiat2304.ag2.mpi-sb.mpg.de) has joined #freenet
[16:48] <lolo-laptop> lost freenet # cat /proc/cpuinfo |grep -i family
[16:48] <lolo-laptop> cpu family : 6
[16:49] <Iakin3> Then the code should work really
[16:49] <lolo-laptop> computers are dumb.
[16:49] <Iakin3> let me check your changes
[16:49] <lolo-laptop> okies
[16:50] * robilad[afk] is now known as robilad
[16:50] <Iakin3> Ehhh...
[16:51] <Iakin3> That was quite a different syntax
[16:53] <lolo-laptop> yeah, I thought it was simple at first, but then I had to dig deeper
[16:54] <lolo-laptop> http://www-106.ibm.com/developerworks/linux/library/l-ia.html
[16:57] <Iakin3> yup.. reading there now
[16:57] <Iakin3> maybe you should try modding the java source to print out the return values?
[16:57] <Iakin3> Try executing CPUID.main()
[16:57] <lolo-laptop> yeah, I'm going to shorly
[16:57] <Iakin3> It should do that for you
[16:58] <Iakin3> It contains my testing cpde
[16:58] <Iakin3> s/cpde/code
[16:58] <Iakin3> but I have to say that your changes _look_ alright
[16:59] <Iakin3> But... you overwrote the MSVC syntax
[16:59] <lolo-laptop> you could ask CVS for the old way
[17:00] <Iakin3> do an #ifdef _MSC_VER
[17:00] <lolo-laptop> **CPUInfo**
[17:00] <lolo-laptop> CPU Vendor: AuthenticAMD
[17:00] <lolo-laptop> CPU Family: 6
[17:00] <lolo-laptop> CPU Model: 6
[17:00] <lolo-laptop> CPU Stepping: 2
[17:00] <lolo-laptop> CPU Flags: 58981375
[17:00] <lolo-laptop> Exception in thread "main" freenet.support.CPUInformation.UnknownCPUException: Failed to read CPU information from the system. The CPUID instructio exists on x86 CPU's only
[17:00] <lolo-laptop> at freenet.support.CPUInformation.CPUID.getInfo(CPUID.java:119)
[17:00] <lolo-laptop> at freenet.support.CPUInformation.CPUID.main(CPUID.java:349)
[17:00] <Iakin3> _that last part was interesting
[17:00] <lolo-laptop> yeah
[17:01] <Iakin3> It asks the JVM for which architechture it is running on
[17:02] <Iakin3> isX86 = System.getProperty("os.arch").toLowerCase().indexOf("x86") != -1;
[17:03] * Hory (~Miranda@81.196.25.110) Quit ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org")
[17:03] <Iakin3> You can check the environment page to see what the "os.arch" property contains
[17:03] <Iakin3> bbl
[17:04] <lolo-laptop> kk
[17:04] <Iakin3> about 30 mins
[17:04] <lolo-laptop> i386
[17:05] <lolo-laptop> hrmph
[17:05] <lolo-laptop> will fixor
[17:05] <Iakin3> Heh.. _that_ was strange
[17:05] <Iakin3> ok, cu in a while
[17:12] <lolo-laptop> w00t.
[17:14] <lolo-laptop> INFO: Optimized native BigInteger library 'jbigi-linux-athlon.so' loaded from resource
[17:18] <toad_> blah
[17:18] * toad_ back
[17:19] <toad_> <lolo-laptop> TheSeeker: they are the same processor... MP is tested better and labelled differently - oh, and it has a bridge set differently
[17:20] <lolo-laptop> :)
[17:21] <toad_> :)
[17:22] <toad_> who's this Brandon Low who's committing now? haven't seen him committting for aaaaages ;)
[17:23] <KenMan> take it where you can get it !!
[17:24] <KenMan> now see, Brandon, I suggested (last september) that maybe I could help out by drawing pictures... hopefully I am doing that now...
[17:24] <toad_> awww
[17:24] <lolo-laptop> :)
[17:24] <toad_> it went back to 87%
[17:24] <toad_> :<
[17:25] <KenMan> 87% sounds like a good number ! for your local testnet ?
[17:25] <toad_> 7/21/04 7:00:00 PM BST 104 78 0.75
[17:25] <toad_> 7/21/04 8:00:00 PM BST 476 414 0.8697478991596639
[17:25] <toad_> 7/21/04 9:00:00 PM BST 44 42 0.9545454545454546
[17:25] <toad_> 7/21/04 10:00:00 PM BST 627 543 0.8660287081339713
[17:25] <toad_> yeah
[17:25] <KenMan> well, look at the total volume and keep that in perspective.
[17:25] <toad_> yup
[17:25] <toad_> that's with the totally contrived can-it-detect-a-forced-specialization test, of course
[17:26] * toad_ is waiting for sanity to give some more input on edge kludging
[17:26] <KenMan> well, we are hopeful that it can, because life can't get any easier for NGR :( ...
[17:26] <lolo-laptop> ok, I'm not sure how the freenet-ext.jar is created for the snapshots, but I think it is about time to put the jcpuid and jbigi libraries in it.
[17:26] <toad_> in the meantime, I'm going to try to merge the non-routing changes to stable
[17:27] * KenMan begins sniffing at the code, searching for a native DS barcode representation...
[17:27] <vsalento> what do I need to build freenet-ext by myself?
[17:28] <toad_> vsalento: check out freenet, then check out the Contrib module into a directory adjacent to the dir you put freenet in
[17:28] <toad_> then run ant
[17:30] * thelema is back
[17:30] <lolo-laptop> vsalento: and then build and add the jcpuid and jbigi libraries to the freenet-ext.jar so that you have native big int support :-P ...
[17:30] <thelema> hi toad
[17:30] <toad_> hi thelema
[17:30] <thelema> I got your email...
[17:30] <toad_> what is your opinion on the edge kludging debate?
[17:30] <toad_> which one?
[17:30] <Iakin3> [23:43] <toad_> that's with the totally contrived can-it-detect-a-forced-specialization test, of course
[17:31] <thelema> the edge-kludging one.
[17:31] <Iakin3> If it cannot, then it for sure cannot keep it
[17:31] <thelema> (you cced me)
[17:31] <Iakin3> So it is still _very_ relevant
[17:31] <toad_> did i? i thought it sent it to devl
[17:31] <toad_> anyway
[17:31] <toad_> what's your take on ian's reply?
[17:31] <thelema> oh, n/m, your message went out wierd and didn't get caught by my filter for -devl
[17:31] <thelema> anyway... let me read ian;s reply first
[17:32] <KenMan> "The Squirrelly Graph" -> http://mywebpages.comcast.net/jkcorson/FreeNet/60158.png I just want the orange line to stay in the backs of people's minds...
[17:32] <toad_> well, it's my message to devl
[17:32] <toad_> well to him
[17:32] <toad_> bcc'd to devl
[17:33] <toad_> in response to his message which was replying to mine
[17:33] <thelema> ah, bcced to devel. that's why it went through wierd.
[17:33] <toad_> KenMan: yikes
[17:33] * toad_ thinks that's some sort of bug
[17:33] <toad_> in reporting
[17:33] <thelema> I think this whole debate just shows the weakness of the current model for estimating future performance.
[17:33] * KenMan does too...
[17:34] <KenMan> which is okay, if it opens up debate and development
[17:34] <toad_> thelema: then what hope is there?
[17:35] <KenMan> we still don't know, but the optimistic answer is "there is always hope for improvement" :)
[17:35] <thelema> toad_: maybe we could go back to CPRouting and use more alchemy to factor in routing times.
[17:35] <toad_> thelema: umm, and that would work better?!
[17:35] <toad_> I doubt it...
[17:35] <KenMan> yeah, and let's remove rate limiting too !! :p
[17:35] <thelema> maybe sliding bucket routing would work.
[17:35] <toad_> eh?
[17:35] <thelema> routing doesn't need to be that complicated.
[17:36] <KenMan> let's finish analyzing the current model first. Okay ?
[17:36] <toad_> we have sliding bucket routing
[17:36] <thelema> no, we have sliding bucket estimators.
[17:36] <Iakin3> KenMan: That orange line *could* tie in pretty we to all those 'timed out waiting for next chunk'
[17:36] <toad_> ah
[17:36] <thelema> sliding bucket routing takes the keyspace and divides it up into buckets; one (maybe more?) for each node.
[17:36] <KenMan> Iakin3 I hope it is something with a basic explanation such as that !
[17:36] <thelema> to route a request, see which node's bucket that request goes into.
[17:37] <KenMan> then how does discovery ever happen ?
[17:37] <thelema> if it succeeds, give the node a bigger bucket.
[17:37] <thelema> if not, shrink the bucket.
[17:37] <thelema> discovery happens independent of routing; nodes keep track of other nodes that they know about seperately
[17:38] <thelema> when a bucket gets too small, they choose a new node to replace it.
[17:38] <thelema> I'm not saying this would work 100%
[17:38] <thelema> it's missing some nice properties
[17:38] <KenMan> just to be clear, i meant keyspace discovery for a single route, not node discovery...
[17:38] <toad_> well, the fact that thelema has invented it doesn't give me more confidence than the fact that ian has
[17:38] <leexgx> Caught freenet.node.rt.EstimatorFormatException: Estimate 1646462 over max 1000000 reading tcp/193.206.155.34:30002, sessions=1, presentations=3, ID=DSA(9219 00c1 9a78 19aa ef41 ab0f 330d fcac 49f8 c24e), version=Fred,0.5,STABLE-1.50,5084 from FieldSet :(
[17:38] <leexgx> freenet.node.rt.EstimatorFormatException: Estimate 1646462 over max 1000000
[17:38] <leexgx> Caught freenet.node.rt.EstimatorFormatException: Estimate 1646462 over max 1000000 reading tcp/193.206.155.34:30002, sessions=1, presentations=3, ID=DSA(9219 00c1 9a78 19aa ef41 ab0f 330d fcac 49f8 c24e), version=Fred,0.5,STABLE-1.50,5084 from FieldSet :(
[17:38] <leexgx> freenet.node.rt.EstimatorFormatException: Estimate 1646462 over max 1000000
[17:38] <thelema> toad_: I just thought it up on a whim.
[17:39] <thelema> toad_: but I think it's a nice idea, worth at least simulating.
[17:39] <leexgx> Caught freenet.node.rt.EstimatorFormatException: Estimate 1646462 over max 1000000 reading tcp/193.206.155.34:30002, sessions=1, presentations=3, ID=DSA(9219 00c1 9a78 19aa ef41 ab0f 330d fcac 49f8 c24e), version=Fred,0.5,STABLE-1.50,5084 from FieldSet :(
[17:39] <leexgx> ffs
[17:39] <leexgx> stuped mirc
[17:39] <leexgx> what is that error any way
[17:40] <leexgx> in stable
[17:40] <KenMan> i think it involves reading in seed nodes, not a major problem...
[17:40] <leexgx> sory about flood
[17:42] <thelema> KenMan: keyspace discovery could handled by having shrinking eliminate buckets easily
[17:42] <KenMan> we always have the opportunity to try multiple routing algorithms on a small local testnet, i'm not quite to the point of writing alternate NGR implementations (wish I was!)
[17:42] <KenMan> but thelema might be skilled enough to try... it would be useful to compare several algorithms anyway
[17:50] <thelema> SlidingBucketRouting wouldn't be NGR; it'd be hackish to make it fit the paradigm
[17:51] <KenMan> well, it could be instructive to try it out on a smaller scale - with 2 or 3 machines netted together, we can do some really neat experiments...
[17:52] <KenMan> just one modern machine is proving useful enough to gather some metrics anyway...
[17:53] <KenMan> thelema - do you think the APIs are sufficiently supportive of 'pluggable' , multiple, routing implementations that can be swapped easily ?
[17:54] <Ash-Fox> No.
[17:54] <KenMan> from what I can remember, it seems to be 'not too bad' , when you examine how CPA and NGR were simultaneously supported.
[17:54] <thelema> It's a bit rough developing a routing implementation
[17:54] <KenMan> ignore the announcements for now...
[17:54] <thelema> but it's not to unreasonable.
[17:54] <thelema> The hardest part is doing all the walkable stuff
[17:55] <KenMan> perhaps we could 'fork' the project in a beneficial way, without truly forking it !!
[17:55] <thelema> it takes a bit of code to be able to give out all the references that are needed.
[17:55] <KenMan> i know it is not trivial, in any case...
[17:55] * thelema was tempted to just run a node with a different routing table implementation and see how well it worked.
[17:55] * guido^pe (~unknown@dsl-213-023-232-066.arcor-ip.net) Quit (Remote closed the connection)
[17:55] <thelema> a heterogenous network wouldn't be bad.
[17:56] <KenMan> i can agree with the logic in that , until we have so many different versions of things that nothing works :( !! depends on whose hands and braincells are at the wheel...
[17:57] <thelema> toad_: reedge kludging; I have no problems with mixing in the global pDNF to partial estimators
[17:57] <KenMan> i mean, we can distill the logic down to summarizing it as 'all just a guessing game' :)
[17:57] <vsalento> now that I build freenet-ext.jar by myself it contains fec dll/so files that weren't there before. Is there reason why they aren't part of 'official' release?
[17:58] <KenMan> vsalento- take it up with human resources !
[18:01] * KenMan thinks we need more humans to resource.
[18:02] <toad_> thelema: even given the consequences?
[18:02] <toad_> what's your opinion on the newbieness issue related to it?
[18:04] <toad_> Iakin: should I keep the lock file change out of stable for now?
[18:05] <Iakin3> toad: Do so until ithas been somewhat more tested
[18:11] <KenMan> I have wondered whether DNFs can really be used for an equivalent 'amount' of information as a success can...
[18:12] <KenMan> if we were only trying to guess key specialization (no time measurements) then a DNF might not be worth as much 'learning' as a success is.
[18:13] <toad_> bbl
[18:13] <KenMan> see you later...
[18:15] <KenMan> in a great big world of keys, and a smaller universe of nodes, (even with superb specialization) what are the odds of a random key existing on any 'well specialized' node ?
[18:16] <thelema> toad_: we have to deal with when to expire; I don't like the code effects of the timed expiration.
[18:16] <KenMan> seems like we would need to know X Y & Z - collective DS size, degree of redundancy, and the popularity coefficient of a specific key !
[18:17] <KenMan> expiration of queries (around 18s) or expiration of transfers ??
[18:19] <thelema> expiration of node references
[18:19] * KenMan is only guessing that queries expire after 18 seconds, and transfers expire after 5 minutes
[18:19] <KenMan> oh. that's a good one.
[18:19] <thelema> 18 seconds for messages that only go one hop
[18:19] <thelema> much longer for requests/inserts
[18:19] * Pascal666 (Pascal@c-24-13-17-191.client.comcast.net) Quit (Read error: 113 (No route to host))
[18:19] <KenMan> ugh ... still learning here ;)
[18:22] * thelema is now known as thelema|aikido
[18:23] * Toast (~k@82-34-170-133.cable.ubr03.maid.blueyonder.co.uk) Quit ()
[18:26] * Pascal (Pascal@c-24-13-17-191.client.comcast.net) has joined #Freenet
[18:32] * raven (~foo@80.65.240.198) Quit ()
[18:44] <toad_> <thelema> toad_: we have to deal with when to expire; I don't like the code effects of the timed expiration. - huh?
[18:51] * jay (jcl@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[19:14] * oierw (mathew@cpe-69-75-109-200.hawaii.rr.com) Quit ("hmm.")
[19:22] * jay (jcl@ool-18bf6dac.dyn.optonline.net) Quit (Read error: 60 (Operation timed out))
[19:31] <TheSeeker> yey, I resurected my dead timex ironman watch...
[19:31] <TheSeeker> those screws are a pita
[19:36] * oierw (mathew@cpe-69-75-109-200.hawaii.rr.com) has joined #freenet
[19:39] * jay (jcl@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[19:39] * jay (jcl@ool-18bf6dac.dyn.optonline.net) Quit (Client Quit)
[19:39] * jay (jcl@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[19:43] <TheSeeker> about two-thirds of the 4.3 billion Internet addresses currently available were used up, adding that IPv6 could magnify capacity by some "25,000 trillion trillion times."
[19:48] <TheSeeker> hmm, that's a lot :P
[19:51] * KenMan fires up a 60167 unstable node...
[19:52] * KenMan remembers someone correcting him that IPv6 uses 128-bit addresses, not 64
[19:59] <KenMan> anybody here who remembers the barcode graph ? I can't seem to find it in the source, yet :(
[19:59] <lolo-laptop> I remember it, but I have no idea where in source it would be ... :-(
[19:59] <lolo-laptop> why not check out the source from 9 months ago and fire it up to find it or something
[20:00] <KenMan> heh, that's an acceptable answer ;)
[20:04] <KenMan> nice class name: "SnapshotCachingLesserLockContentionHashTable.java"
[20:05] <lolo-laptop> hahaha
[20:05] <KenMan> sounds like an Iakin creation to me
[20:05] <lolo-laptop> possibly ejhuff
[20:05] <lolo-laptop> *shrug*
[20:05] <KenMan> @author Iakin
[20:05] <lolo-laptop> heh
[20:06] <KenMan> it makes me almost feel like I know *something* anyway ;)
[20:07] <lolo-laptop> nicely done
[20:13] * KenMan reads his 10/02/03 devl post stating ... 'I noticed that the "barcode" graph went away' ...
[20:13] <KenMan> so we have an approximate time of death
[20:24] * KenMan is frightened that 6122 still builds without errors, but now using SunJava 1.4.2_03
[20:28] <TheSeeker> builds without errors? that can't be right :P
[20:29] <KenMan> that was my first reaction ...
[20:32] * jmercier (~jmercier@4.10.164.111) Quit (Remote closed the connection)
[20:34] <KenMan> this is nostalgic: takes me back to when we thought FreeNet was working much, much better than it really was
[20:35] <KenMan> it is ironic that I feel compelled to draw a blue box around the barcodes I've been generating recently (that's how they used to be shown)
[20:36] * KenMan figures it is more useful if the internal barcode generator can be brought back to life, for all to use
[20:37] <TheSeeker> I loved the barcodes!
[20:37] <leexgx> lol
[20:37] <leexgx> hmm
[20:37] <leexgx> sleep
[20:46] * robilad (~topic@mpiat2304.ag2.mpi-sb.mpg.de) Quit (Remote closed the connection)
[20:50] <KenMan> awww, barcodes (RTInfolet / rthist) have been expunged from Fred ...
[20:52] <TheSeeker> :(
[20:53] <KenMan> it would be useful to add the ability to draw the various histograms graphically, as that offers more detail than the ASCII versions
[20:54] <KenMan> i wonder why ASCII was desired... low bandwidth is my best guess <shrug>
[20:58] <TheSeeker> most likely
[21:01] <KenMan> of course, people are free to generate their own graphics from the raw data available through fproxy. Best to avoid code bloat at this point !! ! !! ! !!!
[21:02] <KenMan> although I guess the cat is already out of the bag, regarding code bloat...
[21:03] <jay> oof node crash
[21:15] <KenMan> it sure was a big bag too...
[21:15] <KenMan> by the way, does anybody really keep cats in bags anymore these days ?
[21:15] <jay> only to piss them off
[21:15] <KenMan> heh heh heee !!
[21:16] * KenMan prefers dogs, and has no sympathy for cats
[21:17] <jay> cats rock.. i hate dogs
[21:17] <jay> esp small dumb dogs
[21:17] <jay> actually i like bigger dogs like pit bulls
[21:17] <KenMan> my big dogs will eat your small cats anyday
[21:17] <jay> perhaps but my cat is funnier
[21:17] <jay> she was climbing the wall yesterday trying to get a moth
[21:18] <jay> and she's 16 years old
[21:18] <KenMan> i don't like little dogs myself, and I don't like pit-bull rottweiler poodle attack dogs . Just affectionate 'mid-to-full size' dogs. No compacts.
[21:18] <jay> i like rotts, dobermans, pit bulls..
[21:19] <jay> i like attack dogs that are tamed to be nice
[21:19] <jay> labs are nice
[21:19] <KenMan> cats are okay, but they always start to use your leg as a scratching post, should you dare to scratch them...
[21:19] <KenMan> labradors are one of the best breeds, especially when mixed
[21:20] <jay> cats are just more girly.. if they don't want to be touched, *and* you do anyway.. oof
[21:20] <jay> but when cats feel like being affectionate they're great
[21:20] <jay> you can't just get it out of them the same way you can with dogs
[21:21] <jay> i really like both for opposite reasons :)
[21:21] <KenMan> i know a girl who absolutely abuses her cats - she thinks they are there for HER to pet them, no matter what. She is always sporting a new set of scratch-marks :o
[21:21] <jay> yeah exactly
[21:21] <jay> my cats listen to me, obey my commands (within reason) because i never pet them when they would rather not
[21:21] <jay> they follow me around and sit near me while they clean themselves
[21:22] <KenMan> any cats that want to occupy my yard, so long as they eat mice, are welcome in my yard. But it stops there.
[21:22] <jay> theyre like tiny peeps
[21:22] <jay> cats need to be taught how to catch mice and birds, otherwise they can't do it
[21:23] <jay> mine can't, but the first one i had killed everything it could catch
[21:23] <KenMan> i can't teach them. I can offer them some food (I have a bag of cat food upstairs), but that's it. If they look hungry, maybe they will repay the favor and catch me some mice.
[21:23] <jay> the other 2 just run after birds and they fly away
[21:23] <jay> they need to be taught by their mother actually
[21:24] <jay> it's odd.. if they don't learn from her they typically can't hunt
[21:24] <KenMan> yes. My neighbor had an awesome mouser. He would start with the skull, chew and crush it real good, and the rest just slipped right down... i really adored that cat when he came around with a tail hangin' outta his little mouth.
[21:25] * KenMan has a small rodent problem...
[21:25] <KenMan> things have gotten a *little* better since I imported some rodent constrictor snakes, but I need more of them.
[21:26] <jay> shit where do u live? the amazon?
[21:26] <jay> the city?
[21:26] <jay> eheh
[21:26] <KenMan> even opposum (the only land-based US marsupial) come round here, sniffin for mice. They are welcome too, even though they look like giant rats...
[21:26] <KenMan> I am on the eastern US seaboard.
[21:27] <jay> me too i suppose
[21:27] <KenMan> the chipmunk population 'round here has exploded - they are just large mice too.
[21:28] <jay> but they're cute
[21:29] <jay> you only kill the ugly rodents right?
[21:29] <KenMan> one or two are cute. I have so many they tease me... anything that might get in the house and keep me up all night is fair game.
[21:29] <KenMan> and when I say they tease me, they all come close around me and just stare at me. sometimes in an actual circle !
[21:30] <KenMan> and their little high-pitched shrieks really piss me off as well.
[21:30] <jay> that's funny
[21:30] <jay> we don't have chipmunks where im at
[21:31] <KenMan> if I get my air-rifle sight tuned well, I might just make them be quiet...
[21:31] <jay> they'rsland
[21:31] <jay> err
[21:32] <KenMan> but I need a .22 for the groundhogs living underneath my yard... the deer are generally repelled by the air-rifle though, and without killing them.
[21:32] <KenMan> the turkeys only come through once or twice a year... the geese constantly fly between the two ponds in my area, night and day.
[21:32] <jay> yer in the sticks eh>
[21:33] <jay> ive seen deer on the beach but that's it
[21:33] <KenMan> A neighbor de-barked their dog, but he still hewwws a lot anyway, sounds like a rabid animal !
[21:33] <jay> debarked?
[21:33] <jay> that's terrible
[21:33] <jay> why not train him to just not bark?
[21:33] <KenMan> I really appreciate the shrill call of a hawk and the hoot of an owl, they are most welcome.
[21:34] <KenMan> i don't think they remember he is outdoors from november to march. Or maybe they vacation for those months...
[21:35] <jay> maybe one of those dog owners who just leaves the dog outside all day?
[21:35] <KenMan> so, no food, lots of cold weather, general neglect... he can't help barking now and then. Originally I thought he was a den of foxes...
[21:36] <KenMan> somebody has a donkey around here, but his distant braying is infrequent enough to be more of a curiousity than anything else...
[21:37] <KenMan> the hunting club's 100+ hounds live over 2 miles away, and they only group-yelp the morning before a hunt... so they are okay.
[21:38] <jay> yeah that's the sticks
[21:38] <jay> i just hear cars where i live
[21:38] <jay> rolling wheels
[21:39] <KenMan> but the most annoying animal of all, is the young teenager with his ATV. They don't have enough land for him to let loose, so he can only accelerate for 2 seconds max. Every 9-15 seconds, it is brr-r-r-eee-aa-hh.
[21:40] <KenMan> and he can do this for hours. I think it is cruel to own an animal if you cannot provide sufficient living space for him/her.
[21:41] <jay> hah
[21:41] <jay> bbl.. leaving work
[21:41] <KenMan> see ya !
[21:41] * jay (jcl@ool-18bf6dac.dyn.optonline.net) Quit ("!")
[21:42] * KenMan is done talking animals for now... and bestiality jokes never even crossed his chatter !!
[22:13] <KenMan> that must be because soros and his sexy grey kitty never visit us anymore :o
[22:29] <Pascal> how do I get ls to tell me how many files it has listed?
[22:37] <TheSeeker> I'm sure there's a command line switch ... good luck finding it :P
[22:37] <KenMan> ls -l | wc -l
[22:38] * salahx (salahx@sc1-24.217.174.147.charter-stl.com) has joined #freenet
[22:38] <Pascal> KenMan: that counts both files and directories the same
[22:38] <KenMan> sorry, that will likely be off by at least one. Try ls -C1 | wc -l ... hmmm, now, about those directories being in the count...
[22:39] <salahx> how abotu `find /var/freenet -type f | wc -l` ?
[22:39] <Pascal> got it: ls -la |grep "^-" |awk 'END {print NR}'
[22:39] <TheSeeker> http://lists.debian.org/debian-user/2002/06/msg04141.html
[22:39] <Pascal> thanks though
[22:40] <KenMan> good sol'n
[22:40] <salahx> ok
[22:40] <salahx> I haven't noticed with NTPL here though....
[22:42] <KenMan> isn't the internet just like a great big reusable brain ?? ;)
[22:42] <salahx> ion crack, yes
[22:42] <KenMan> one with a few too many memories though. Praise the Great Chinwah for Google...
[22:43] <KenMan> or whomever your particular god may be...
[22:44] <KenMan> i liked the use of find with -maxdepth 1 in that suggestion, it earns points for cleverness
[22:47] * jay (jcl@ool-18bf6dac.dyn.optonline.net) has joined #freenet
[22:47] <Pascal> salahx: you put fred in /var ?
[22:47] <KenMan> the content of the internet is merely a reflection of the millions of people who put the content there. Unfortunately, many of those people surely were on crack !
[22:48] <salahx> In /var/freenet actually
[22:48] <Pascal> why?
[22:48] <salahx> THat where Gentoo put it :)
[22:49] <KenMan> ask a stupid person... get a stupid answer !
[22:49] <salahx> THe jar itsefl is in /usr/lib/freenet
[22:49] <salahx> the config file is in /etc
[22:49] * KenMan jests in kind. No offense directed at salahx !!
[22:49] <salahx> and the datastroe, rotuign stable ,etc; is in /var/freenet
[22:49] <salahx> lol
[22:49] <salahx> it also runs as its own user and group, too
[22:50] <salahx> I think teh FHS-complaint way to do it
[22:50] <Pascal> fred has an install program for Gentoo?
[22:50] <KenMan> other way round.
[22:50] <salahx> its a Gentoo ebuild
[22:50] <salahx> emerge net-p2p/freenet
[22:50] <salahx> The ebuidl is probabyl smei-broken if you use unstable though
[22:51] <salahx> But if you are usign both unstable and Gentoo you probabyl knwo how to fix it :)
[22:51] <Pascal> ah. cool
[22:52] <salahx> ACutally it looks liek they already know about bug i ran into: http://bugs.gentoo.org/show_bug.cgi?id=45004
[22:53] <Pascal> one of these days I need to figure out how to create users and groups. I prob shouldn't run fred as root
[22:53] <jay> oh my
[22:53] <salahx> uhh defintely NOT
[22:53] <jay> init should run as root and that's about it :)
[22:53] <salahx> Well fi you're using Gentoo jsut emrege the ebuiodl a at it wil create it for you :)
[22:53] <KenMan> it doesn't really matter. Until you determine who has already subverted the JVM.
[22:55] <KenMan> and assuming that you place total faith and trust in The Toad ... and any other FreeNet hackers...
[22:55] <salahx> at least ehre we cna look at teh source. Not understand it, as that is impossible, but we CAN look at it :)
[22:56] <salahx> There might be a backdoor in the Sun code - there was one in the Firebird (the database, not the browser now known as Firefox)
[22:57] <Pascal> salahx: has toad seen that bug? Seems to say the opposite of what he keeps yelling about.
[22:57] <KenMan> there may even be something that might be exploitable under highly specific (or completely generic) circumstances :o !
[22:57] <salahx> i don;t think so
[22:57] <jay> the JVM bug?
[22:59] <Pascal> toad: see what you think of http://bugs.gentoo.org/show_bug.cgi?id=45004
[22:59] <salahx> it think he's asleep
[22:59] <jay> he must be sleeping
[23:00] <Pascal> prob. But he'll see it when he wakes up
[23:00] <jay> it's after 3am where he's at
[23:00] <jay> yeah
[23:00] <Pascal> basically if I'm awake hes asleep and vice versa
[23:01] <KenMan> did brits just have a clock change recently ? I'm only 4 hours off GMT now, it was 5 not too long ago.
[23:01] <KenMan> GMT has a summer / winter adjustment, I think... but not UTC.
[23:01] <jay> there was an email to dev about NPTL being broken still in 1.5 (from toad)
[23:01] <salahx> a saaw that, but it doesn';t SEEM broken here...
[23:01] <jay> KenMan: it's our daylight savings time thingie in the states
[23:01] <Pascal> jay: yes, but that gentoo bugs says it's not
[23:02] <jay> i don't know anything about it.. but if it works on gentoo and nowhere else it's basically not working
[23:02] <KenMan> jay - it is not synchronized with ours, and I don't know if GMT incorporates a twice yearly adjustment !
[23:02] <Pascal> maybe its OS specific. what NPTL code is in Gentoo vs whatever toad is testing on
[23:03] <jay> KenMan: we're synced with UTC time.. but we have daylight savings time here im sayinh
[23:03] <KenMan> that's quite possible. NPTL itself is still extremely highly alpha experimental code. Probably the widest used 'highly experimental alpha' stuff out there !
[23:03] <jay> Pascal: dunno
[23:04] <KenMan> i think toad uses a fedora
[23:04] <Pascal> salahx: what OS and NPTL ver are you running?
[23:04] <jay> nah debian i believe
[23:04] <KenMan> no, debian. One or the other, anyway.
[23:04] <salahx> Gentoo, Linux 2.6.7
[23:06] <KenMan> so "the current GMT time" (as seen by anyone in any country or zone) does not incorporate a daylight adjustment period ? I feel really dumb for not knowing this stuff anymore.
[23:06] <Pascal> I'm building a new node right now on Fedora Core 2. I'll try it with NPTL enabled and see what happens
[23:06] <KenMan> I don't think "GMT" skips ahead or backwards an hour at any point in time. But I'm not sure.
[23:06] <jay> it seems that sun is not interested in perfecting the NPTL on linux.. since thread handling on sparc/solaris works well already
[23:07] <jay> KenMan: GMT and UTC are equivalent afaik
[23:07] <Pascal> GMT aka UTC never jumps ahead or back
[23:07] <jay> GMT was the old name
[23:07] <KenMan> that's what I thought, but Perl is reporting GMT time as +4 now, last week it was reporting as +5 !!! wtf ?
[23:07] <jay> KenMan: your locale could be off
[23:08] <jay> UTC/GMT does not change according to daylight savings rules
[23:08] <KenMan> most likely. I've never gotten it right for more than 6 months at a time :(
[23:08] <jay> KenMan: if you dual boot with windows the problem is more complex
[23:08] <KenMan> but nothing should have changed for me in the past week, unless a hacker has altered my locale
[23:08] <jay> since windows uses local time
[23:08] <KenMan> I don't have windows, only a penguin
[23:09] <jay> KenMan: what's in /etc/timezone ?
[23:10] <lostlogic> /etc/localtime
[23:10] <KenMan> /etc/localtime -> /usr/share/zoneinfo/EST5EDT
[23:11] <jay> my localtime is linked to /usr/share/zoneinfo/America/New_York
[23:11] <KenMan> we last turned our clocks around here in April or May... and I did it too.
[23:11] <KenMan> I have never gotten this correct in 10 years of linux... :(
[23:12] <KenMan> should I crontab the change ? does New_York automatically make the system adjust for each spring forward fall back ?
[23:13] <jay> KenMan: for me it does.. but i also sync using NTP
[23:13] <KenMan> has linux ever had problems in this area (meaning some distro's may not have gotten it right in the past) ?
[23:13] <KenMan> yes, NTP usually does warm fuzzy things for me, but not lately
[23:14] <KenMan> what happens if I set the TZ variable ? some tools respect the POSIXly correct manner, some do penguin-specific things, and some just do the WRONG thing...
[23:14] <jay> TZ is outdated i think
[23:14] <KenMan> I'm almost ready to throw my hands up and just use some wonderful distribution to take care of exactly half of everything for me.
[23:14] <jay> it isn't set here
[23:14] <jay> what distro?
[23:15] <KenMan> and spend exactly all of my time trying to fix the other half.
[23:15] <KenMan> I use 'KenMan Linux'
[23:15] <jay> can u run tzconfig ?
[23:15] <KenMan> i have a tic
[23:15] <jay> tick?
[23:15] <KenMan> no, tic
[23:16] <KenMan> oops, i meant a zic
[23:16] <jay> dunno what either is
[23:16] <KenMan> zic - time zone compiler ... well, that's not what i need either
[23:16] <jay> oh right
[23:16] <salahx> Aren't timezone's wonderful?
[23:16] <salahx> Stupid farmers....
[23:16] <KenMan> i hate them all. Without discrimination.
[23:17] <KenMan> anyway, i have some tool that is supposed to set the system timezone... but I've never gotten it right, so I'm almost resigned after 10 years of trying.
[23:17] <KenMan> That's only about 7 - 12 attempts (those times when I noticed it was wrong)...
[23:18] <jay> what is kenman linux?
[23:18] <jay> slak?
[23:18] <KenMan> And there was some fighting over the proper way to handle it a few years ago. All that matters is that all your tools in your particular chosen distro is consistent in use
[23:18] <KenMan> No , i build everything myself. After reading the docs and compiling the crap.
[23:18] <jay> debian did it fine for me automatically
[23:19] <jay> KenMan: what source code base?
[23:19] <jay> you're getting the source from... ?
[23:19] <KenMan> a very interesting mix of things !!
[23:19] <hirvox> Redhat/Fedora also handled Finland's DST automagically
[23:19] <KenMan> but mainly original authors
[23:19] <jay> hirvox: for me i had to set it to america/NY so that the daylight savings crap worked correctly
[23:20] <KenMan> well, your hat is off to me (that is a play on an american saying)
[23:20] <jay> KenMan: how can someone help you tune your distro if it's completely custom built?
[23:20] <KenMan> "my hat is off to you" - basically translates to "I respect you"
[23:20] <hirvox> jay:yeah, that's what I did, too. setting the timezone to Europe/Helsinki was sufficient
[23:20] <jay> and who are the original authots?
[23:21] <KenMan> i talk to people and read what other people have done (mainly the folks who build these various distros)
[23:22] <jay> like?
[23:22] <jay> debian, redhat, suse, ...
[23:22] <KenMan> a year or two back I found a neat site called 'linux from scratch' and said "hey! that's what I've been doing for years. These guys have created the illusion that THEY have it under control. Neat."
[23:22] <jay> oh i see
[23:22] <jay> i did that years ago
[23:22] <KenMan> it is kind of fun to see how people resolve some of the types of issues that arise. And disappointing to see a 'fedora fix' for this and a 'debian fix' for that.
[23:22] <jay> when distros came out i welcomed them
[23:23] <KenMan> slackware was great, especially when that was the only one :)
[23:23] <jay> yeah but upgrading was a bitch
[23:23] <jay> this was before packaging existed
[23:23] <KenMan> slackware has always had packaging.
[23:23] <jay> i ran slakware 2.0
[23:23] <jay> nope
[23:24] <KenMan> although it wasn't the medusa that most use today
[23:24] <jay> 2 didn't, 3 didn't
[23:24] <jay> they introduced packaging in 5.0 i think
[23:24] <KenMan> hmmm.... i remember a series, then d series, then n ...
[23:24] <jay> prior to 1.0 ?
[23:25] <KenMan> i don't know, first picked it up around '94 ... it was easy enough to install, after a week or a month of total confusion :)
[23:25] <jay> i recall buying slak 3.0 at a computer show around 1993
[23:25] <KenMan> okay, you got me then :)
[23:25] <jay> heh barely
[23:25] <jay> but that was crazy times for linux
[23:26] <jay> just getting X to run was a skill
[23:26] <salahx> Too bad the Gentoo appopach doesn't work so well for slow/non-internet-connect machines...
[23:26] <KenMan> oh man, that was like 2 or 3 years later for me :)
[23:26] <jay> i remember when i got the stupid 3COM card driver working..
[23:26] <KenMan> oh, did you work with what's his name ?
[23:26] <jay> first thing i did was connect to efnet/#linux in console mode and brag to the noobs
[23:27] <jay> noobs back then.. eheh
[23:27] <jay> we were all noobs
[23:27] <KenMan> I'm only kidding. It is a prime opportunity for conspiracy to note that American NSA wrote most of the linux net drivers...
[23:27] <KenMan> and this IS Freenet, after all !!
[23:27] <jay> is that true>?
[23:27] <KenMan> Donald Becker, that's the guy.
[23:28] <jay> ahj right
[23:28] <jay> he wrote a lots
[23:28] <KenMan> yeah, it is true. And I don't smell anything but good old linux coding...
[23:28] <jay> he wrote tons of net drivers back then
[23:28] <mikeDOTd> he's at NASA, not NSA
[23:28] <salahx> THer Tulip driver in particular
[23:28] <jay> well im sure the code's been looked at a few times
[23:28] <jay> mikeDOTd: yeah i was about to say that
[23:28] <KenMan> No, NSA left THERE NAME in the code. just take a look some time
[23:28] <KenMan> THEIR :)
[23:29] <TheSeeker> THEY'RE
[23:29] <jay> no it's their ;)
[23:29] <TheSeeker> heh
[23:29] <jay> KenMan: becker's email address was at nasa i believe
[23:30] <KenMan> i know. Do this and shit your pants : find /usr/src/linux -name \*.c -exec grep -i "national security a" {} \; -print
[23:30] <jay> i wrote a doc on booting linux with NTLDR and in the doc i use linux 1.1 as an example
[23:30] <KenMan> because we all need a decent conspiracy theory. I get around 50 hits at least...
[23:31] <KenMan> OH MY GOD. NSA BUGGED LINUX WAY BACK THEN. RUN FOR THE HILLS !!!
[23:31] <KenMan> you know, they actually do some beneficial things in this world, but I am not here to promote them.
[23:31] <jay> as long as they released the damn source, i don't care who they are ;)
[23:32] <KenMan> that is at least more open-minded than many of #freenet participants.
[23:32] <KenMan> and American :)
[23:32] <jay> ew they wrote the Appletalk driver
[23:33] <jay> KenMan: i have friends who work in the government.. met some people who claimed to work there
[23:33] <jay> some crypto geeks
[23:33] <jay> i know who they are.. but i won't go into it in an irc channel (esp since it's now being logged by me)
[23:33] <jay> zero politics from me
[23:33] <KenMan> and did they have purple eyes and did their breath hint slightly of gasoline ??
[23:34] <jay> wtf is that?
[23:34] <jay> hah
[23:34] <KenMan> good. No politics from me either. Just more shit to feed to those who need to be fed.
[23:34] <jay> i could stand on a soapbox and preach the word for hours.. if i wanted to ;)
[23:35] <KenMan> I hear that many techno-spooks live deep beneath the earth's surface, and only climb out when they can't ...
[23:35] <salahx> But that; what #freenet-(chat|politics) is about :)
[23:35] <jay> salahx: it's still traced and logged by (...)
[23:35] <KenMan> oh yeah, that why i get enjoyment from some of it.
[23:35] <jay> i send anon emails to usenet with rants
[23:36] <jay> i haven't done one in a while
[23:36] <KenMan> NSA not the only player in their space , other governments are beginning to learn to crack a code themselves. Imagine !
[23:36] <KenMan> anyway, i best stop now ... lest I never will.
[23:36] <jay> KenMan: exactly.. i almost retorted but then said "fuck it"
[23:37] <jay> eheh
[23:37] * oierw` (mathew@cpe-69-75-126-12.hawaii.rr.com) has joined #freenet
[23:38] <KenMan> greetings to hawaii !
[23:38] <jay> longtime freenet user i believe
[23:38] <jay> or his infobot
[23:39] <oierw`> having a bot would be too much work :)
[23:39] <jay> oierw`: u set up irc logs way back.. right?
[23:40] <oierw`> err... what kind of irc logs are we talking about?
[23:40] <jay> http://newton.matcmp.ncc.edu/~lockej/freenet/chanlog/
[23:40] <jay> i set up a logger recently with pircbot
[23:40] <oierw`> my irc client logs everything I see int his channel, but it's not in an easily exportable system. and the bot that used to run in here and do logs isn't here
[23:40] <KenMan> BTChanLog ?
[23:41] <KenMan> where did s/he go anyway
[23:41] <jay> oierw`: i recall them existing when i first came to this project at http://espnow or something like that
[23:41] <jay> hapi is the pircbot
[23:42] <oierw`> i'll go a hunting
[23:42] <oierw`> but I haven't had luck with the esp irc logs lately
[23:42] * oierw (mathew@cpe-69-75-109-200.hawaii.rr.com) Quit (Nick collision from services.)
[23:42] * oierw` is now known as oierw
[23:44] <KenMan> so Perl's time() function has begun returning times that are +4 for me, whereas they were +5 a week or so back...
[23:45] <KenMan> or actually, I guess that would be -4 and -5 (hours)...
[23:45] <KenMan> bbt
[23:46] <Pascal> whatis the point of preconfig.sh? --config will dot he same thing
[23:51] <oierw> err... no luck jay
[23:52] <oierw> i'd have to get access to the comptuer that runs the bot that used to log the channel to see if the files are laying around, but going through the directory structure I could find, they don't exist in a broken down format anymore
[23:53] <Pascal> jay: anything in particular you were looking for? I have about a years worth.
[23:54] <jay> Pascal: nothing precise.. i have logs in xchat too that are in a different format
[23:54] <jay> id like to see them integrated, but i have bigger freenet fish to fry atm
[23:56] <jay> oierw: well if you come up with anything email it.. gzipped if it's not over a meg or so
[23:56] <jay> Pascal: how far do your logs go back?
[23:57] <Pascal> jay: 2002 off hand. Maybe further on other discs
[23:57] <Pascal> what is the block size for ext3?
[23:58] <mikeDOTd> 4K, IIRC
[23:58] <Pascal> thanks
[23:59] <jay> my earliest chatlogs go back to Feb 2003
[23:59] * e1ven (Kemrain@pool-68-163-146-123.bos.east.verizon.net) has joined #freenet
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.
