Timestamps are in GMT/BST.
[0:01] * nextgens hmmmz too
[0:02] * Fennes is now known as Fennes[zzzz]
[0:02] * Fennes[zzzz] is now known as Fennes[zzzzz]
[0:02] * railk (n=railk@) Quit ("Leaving")
[0:08] <toad_> rehi
[0:09] * toad_ d'oh @ breaking old plugins
[0:09] <toad_> not intended!
[0:11] <toad_> means we'll probably have to roll out the new plugin API ...
[0:11] <toad_> or exclude the recent changes from 1008
[0:11] <toad_> hmmm
[0:11] * toad_ needs to talk to Bombe ...
[0:11] <toad_> probably not right now but soon
[0:11] <toad_> Bombe: can we talk tomorrow?
[0:13] <toad_> and anyone else who's interested in plugin APIs
[0:13] <toad_> or plugin deployment
[0:13] <toad_> or any other aspect of plugins...
[0:13] * toad_ supposes existing plugins on existing node builds will continue to work...
[0:13] <toad_> hmmm
[0:16] <nextgens> hmm
[0:16] <nextgens> toad_> well I can manually force them to the previous version
[0:17] <nextgens> it will work until you commit on their path again
[0:17] <nextgens> is that fine ?
[0:17] <nextgens> or shall I really prevent the auto-builder from updating them?
[0:17] <toad_> hmmm
[0:17] <toad_> might be best to turn off auto-building of plugins for now
[0:18] <nextgens> ok
[0:18] <nextgens> well, anyway you haven't broken anything as your commit wasn't building
[0:18] <toad_> hehe
[0:18] <toad_> it works here ... i wonder why
[0:18] * nextgens disables them anyway
[0:18] <nextgens> I know why
[0:18] <nextgens> because of timing
[0:19] <toad_> ok
[0:19] <toad_> ok
[0:19] <nextgens> your commit were requiering latest trunk
[0:19] <nextgens> and trunk got build afterwards
[0:19] <nextgens> ;)
[0:19] <toad_> i need to talk to Bombe about plugins, we need to sort out the API
[0:19] <toad_> also with anyone else who cares about plugins
[0:20] <nextgens> ok, I've disabled the auto-builder for plugins
[0:20] <toad_> what APIs are necessary for plugins? i suppose i should go over the existing plugins and find out what they need?
[0:20] * earthsound (n=another1@) has joined #freenet
[0:21] <toad_> Librarian requires the ability to fetch stuff and to be accessed via HTTP (returning simple short HTML)
[0:21] <toad_> that we have now
[0:21] <nextgens> that's probably a good idea yeah
[0:21] <toad_> TestGallery is much the same
[0:21] <toad_> JSTUN uses datagram sockets (!) but that will be taken care of by Bombe's java.security work
[0:21] <toad_> if Bombe turns up, that is
[0:21] <toad_> so there will be a permissions interface for plugins that claim to need that
[0:22] <toad_> and also there will need to be an interface for feeding detected IPs to the node
[0:22] <toad_> Librarian ideally would have persistent config
[0:22] <toad_> SNMP needs persistent config
[0:23] <nextgens> we need versionning facilities
[0:23] <nextgens> ie : the node ought to ensure the plugin doesn't require a newer version
[0:23] <toad_> and access to DataStatisticsInfo
[0:23] <nextgens> before loading it, that is
[0:24] * nextgens was thinking about writing a bonjour discovery plugin
[0:24] <toad_> nextgens: FreenetPlugin exposes plugin version and internal version; the plugin is passed a FreenetPluginManager, which exposes node version and internal version
[0:24] <nextgens> to ease the wirering up of nodes behind the same nat
[0:24] <earthsound> can anyone in here swap refs w/ me? i tried freenet-refs, but it's quiet. particularly, I'm looking for ppl who've run a node for more than a couple weeks and will keep theirs running a good while. I find myself culling my peer list down by a couple dozen nodes every few weeks
[0:24] <toad_> well, nodes behind the same NAT can connect to each other easily enough (over LAN IPs)
[0:25] <nextgens> toad_> the plugin would be about discovering nodes
[0:25] <toad_> the only problem is for security reasons we don't include local LAN addresses normally, nor do we connect to them
[0:25] <toad_> nextgens: hmm, interesting
[0:25] <nextgens> or automating ... in case someone wants to setup a cluster of freenet node
[0:26] <nextgens> +s
[0:26] <toad_> well, we probably want to provide routing table access, like we do in FCP?
[0:26] <toad_> of course that means there will be refbot plugins etc :|
[0:27] <nextgens> I don't think we should overengeener our API for now
[0:27] <toad_> we can add more Needs/Provides pairs later, it's not a big deal
[0:27] <nextgens> I suggest you provide what actual plugins need ... and that's all
[0:27] <toad_> nextgens: what do you mean about versioning?
[0:27] <nextgens> the node ought to detect that some plugins might require "newer interfaces"
[0:27] * toad_ thinks we need to put some thought into the basics before we cast it in stone anyway
[0:28] <nextgens> an higher version number of the API
[0:28] <nextgens> new functionnalities
[0:28] <toad_> okay, how about we require FreenetPlugin's to implement minimumNodeVersion() ?
[0:28] * nextgens didn't know english say "cast into stone"
[0:28] <nextgens> we do have the same in french :)
[0:28] <toad_> :)
[0:28] <toad_> how about we require FreenetPlugin's to implement minimumNodeVersion() ?
[0:29] <toad_> that's the obvious thing
[0:29] <nextgens> good idea imo
[0:29] <toad_> another thing that would be relatively easy but ugly would be looking for any missing APIs via reflection
[0:29] <toad_> actually
[0:29] <toad_> if it needs Needs*'s that we don't provide, won't it just not load anyway?
[0:29] <toad_> won't it fail at the class loader stage? or are unprovided interfaces not a fatal error?
[0:30] * nextgens doesn't know
[0:30] <toad_> hmmm
[0:30] <nextgens> but I know we do need to have some kind of mechanism to handle that
[0:30] <toad_> nextgens: other point: do we need to worry about other implementations of fred?
[0:30] <nextgens> no
[0:31] <toad_> is it okay to assume that a freenet build number is fine for the foreseeable future?
[0:31] * toad_ supposes we can just do 10000000, 1000010000 in the future
[0:31] <nextgens> I don't think we should ease anyhow the "opennet forking movment"
[0:31] <toad_> for 1.0, 1.1
[0:31] <toad_> nextgens: i was thinking of impl's in C etc
[0:31] <toad_> for the node we have a string for version
[0:32] <nextgens> well, the versionning will have changed before 1.0 anyway :)
[0:32] <toad_> no reason to break existing plugins
[0:32] <toad_> which will hopefully be numerous
[0:32] <nextgens> and Ian will be willing us to give birthnames to it
[0:32] <nextgens> as it's better from a PR pov
[0:32] <toad_> birthnames?
[0:32] <nextgens> again, KISS
[0:33] * toad_ supposes so
[0:33] <toad_> like i said we can always use Big Constants for released versions
[0:33] <toad_> okay so we call this after loading but before starting
[0:34] <toad_> that's the official build number, not the SVN number
[0:34] * nextgens heads to bed
[0:34] <toad_> as we probably don't know the SVN number anyway
[0:34] <nextgens> bbl
[0:34] <toad_> good night, i should go to bed soon (early start...)
[0:34] <toad_> cya
[0:34] * phrosty (n=phrosty@) Quit ("PostQuitMessage(0);")
[0:34] <CIA-14> toad * r11429 /trunk/freenet/src/freenet/plugin/api/FreenetPlugin.java: minimumNodeVersion()
[0:36] <CIA-14> toad * r11430 /trunk/freenet/src/freenet/plugin/api/ (4 files): GPL headers
[0:37] <CIA-14> toad * r11431 /trunk/freenet/src/freenet/support/api/ (HTTPReply.java HTTPRequest.java HTTPUploadedFile.java): More GPL headers
[0:37] * earthsound (n=another1@) has left #freenet
[0:43] <CIA-14> toad * r11432 /trunk/ (36 files in 10 dirs):
[0:43] <CIA-14> Extract an interface for a SubConfig.
[0:43] <CIA-14> Move *Callback into freenet.support.api.
[0:43] * _ph00 (n=z@) has joined #freenet
[0:48] * IMCensored_ (n=KMIntern@) has joined #freenet
[0:56] * IMCensored1 (n=KMIntern@) Quit (Read error: 145 (Connection timed out))
[0:56] * IMCensored_ is now known as IMCensored1
[1:01] <CIA-14> toad * r11433 /trunk/freenet/src/freenet/ (6 files in 5 dirs): StringArrCallback/StringArrOption should use String[].
[1:03] * toad_ (n=toad@) Quit (Remote closed the connection)
[1:18] <JustMe> Your browser sent a request that Freenet could not understand.
[1:18] <JustMe> http://127.0.0.1:8888/
[1:20] <JustMe> r11433 won't even start.
[1:25] <JustMe> r11424 works but upload resumes are broken
[1:25] <JustMe> :(
[1:30] <JustMe> t11427 is where the Your browser sent a request that Freenet could not understand. starts on http://127.0.0.1:8888/
[1:30] <JustMe> r11427 is where the Your browser sent a request that Freenet could not understand. starts on http://127.0.0.1:8888/ ;)
[1:38] <JustMe> Upload resumes seem to be broken if uploading more than one file at a time it seems. The first file keeps its upload status the second two go back to zero with a restart.
[1:43] * Wayn2 (n=wscott@) has joined #freenet
[1:48] * oCc (n=nodeOne@) Quit (Read error: 131 (Connection reset by peer))
[1:58] * phrosty (n=phrosty@) has joined #freenet
[2:09] * PraiseChaos (n=kcecil@) Quit (Read error: 104 (Connection reset by peer))
[2:38] * Wayn2 (n=wscott@) has left #freenet
[3:02] * phrosty (n=phrosty@) Quit ("PostQuitMessage(0);")
[3:12] <_ph00> JustMe: I have two uploas n progress, and the second one is stuck to zero. the current download #1 was also stuck at zero while the previous upload was running. (looks like we can upload one file at the time)
[3:27] * jaja (n=jaja@) has joined #freenet
[3:27] <jaja> Anyone know what programming language Frost and Fuqid are written in?
[4:00] * jaja (n=jaja@) has left #freenet
[4:08] * vfervers_ (n=vfervers@) Quit (Read error: 113 (No route to host))
[4:08] * vfervers_ (n=vfervers@) has joined #freenet
[4:14] * ATGeek (n=ATG@) has joined #freenet
[5:59] * timmy2chk (n=violent@) Quit (Read error: 110 (Connection timed out))
[6:14] * boshaus2 (i=nobody@) has joined #freenet
[6:28] * Kata_Pult (i=Kata@) has joined #freenet
[7:05] * creativename (i=blahmore@) has joined #freenet
[7:05] * creativename (i=blahmore@) has left #freenet
[7:07] * creativename (i=blahmore@) has joined #freenet
[7:08] * Zelda-Niamb (i=KataPult@) has joined #freenet
[7:21] * Zelda-Niamb (i=KataPult@) Quit (Read error: 104 (Connection reset by peer))
[7:23] * Kata_Pult (i=Kata@) Quit (Remote closed the connection)
[7:24] * JustMe (i=JustMe_@) Quit ()
[7:46] * creativename (i=blahmore@) Quit (Client Quit)
[8:15] * KataPult (i=KataPult@) has joined #freenet
[8:41] * IMCensored1 (n=KMIntern@) Quit (Read error: 104 (Connection reset by peer))
[9:10] * sanity (n=ian@) Quit ()
[9:17] * whiterabbit (n=Whiterab@) has joined #freenet
[9:24] * railk (n=railk@) has joined #freenet
[9:41] * sandos (n=sandos@) Quit (Read error: 131 (Connection reset by peer))
[9:43] <- *sleon* hi
[9:43] <- *sleon* help
[9:43] <- *sleon* list
[9:59] * bright_fame (n=robertvi@) has joined #freenet
[10:00] * iwantnode (n=iwantnod@) has joined #freenet
[10:01] <bright_fame> anyone else listen to this podcast about freenet and tor? http://www.grc.com/SecurityNow.htm#70
[10:03] <sleon> bright_fame: oh interesting
[10:04] <bright_fame> they spend more time talking about tor i guess. but give a fair (from my limited knowledge of freenet!) overview of freenet
[10:06] <iwantnode> download the mp3 at min bright will listen in car
[10:17] * KataPult (i=KataPult@) Quit (Remote closed the connection)
[10:20] <iwantnode> Does any one worry about conecting to the universitys
[10:21] * KataPult (i=KataPult@) has joined #freenet
[10:22] <bright_fame> iwantnode: you mean they run freenodes and you're concerned about them?
[10:23] <iwantnode> bright : i am not but wondered if others were
[10:27] * iwantnode (n=iwantnod@) Quit ()
[10:35] * phrosty (n=phrosty@) has joined #freenet
[11:09] <phrosty> lol @ onion routing on numb3rs
[11:21] <_ph00> I cant't upload more than one file at the time to freene, if I have mor4e than one in upoad queue, the second one will stick to zero until the first one is done: another user suggested to switch all the options to SOFT to make it possible to download multiple files at the same time, but before I do that I want to know what that does exactly,
[11:32] <nextgens> setting insert schedulers to SOFT is probably a good idea
[11:32] * rasco (i=rasco@) Quit (SendQ exceeded)
[11:32] <nextgens> for requests on the other hand, you should keep the default value : HARD
[11:33] <_ph00> what if I run uploads and downloads at the same time?
[11:33] <_ph00> what's the best setting then?
[11:33] <nextgens> http://emu.freenetproject.org/cgi-bin/viewcvs.cgi/trunk/freenet/src/freenet/client/async/ClientRequestScheduler.java?rev=11432&view=markup
[11:33] <nextgens> to know how it works, see that file
[11:33] <_ph00> k
[11:33] <nextgens> as I told you : keep your requests schedulers to hard
[11:34] <nextgens> otherwise browsing fproxy is gonna be slow
[11:34] <_ph00> but if uploading only, SOFT is best
[11:34] * _ph00 goes read that file
[11:35] <nextgens> what you're looking for is removeFirst()
[11:36] <_ph00> k
[11:36] * timmy2chk (n=violent@) has joined #freenet
[11:36] <nextgens> bbiab
[11:37] * bright_fame (n=robertvi@) Quit (Read error: 110 (Connection timed out))
[11:40] <_ph00> oh.
[11:40] <_ph00> ok, I *think* I got it. thx nextgens
[11:43] <_ph00> well.. not really.
[11:44] <_ph00> but OK, soft when uploading only and hard when downloading/freensites browsing
[11:44] <_ph00> freesites*
[11:44] * bright_fame (n=robertvi@) has joined #freenet
[11:54] * NullAcht15 (n=NullAcht@) has joined #freenet
[11:57] <_ph00> oh.. now I go it. hard for req. and softy for inserts (duh)
[12:00] <_ph00> (softy?)
[12:04] * JRVC (i=JJK@) has joined #freenet
[12:13] * NullAcht15 (n=NullAcht@) Quit (Remote closed the connection)
[12:15] * JRVC (i=JJK@) Quit (Remote closed the connection)
[12:22] * timmy2chk (n=violent@) Quit ("so what?")
[12:24] <_ph00> huh...
[12:25] <_ph00> my bit torrent client is informing me that tne NAT settings are OK... which is *weird*, given the limitation that I talk about all the time (ISP doing NAT, no access to the router, etc)
[12:25] <_ph00> yet, I'm getting inbound connections now
[12:25] <_ph00> maybe they changed policy, and allow inbound connections without charging extra, now...
[12:26] <_ph00> there is one way to test it (one I can think of) I'll try eMule (or a *nix clone of it): if I get "high" ID, the most hated of my ISP's restricions is *gone*
[12:28] <bright_fame> _ph00: does your ISP give you a non-routable IP (like 192.168.xxx.xxx)?
[12:28] <nextgens> they are simpler ways
[12:28] <nextgens> give us your ip and one open port
[12:29] <_ph00> nextgens: I'd ask you to port scan my external IP for me but I heard that some tools are illegal in Frence
[12:29] * JRVC (i=JJK@) has joined #freenet
[12:29] <_ph00> I get a non routable address, but it's on 39.5 range, no 192 or 10
[12:30] <nextgens> cool, I'm logged on irc from England
[12:30] <_ph00> you are? OK, I'll PM you
[12:30] <nextgens> well, they could be doing NAT
[12:30] <nextgens> not PAT
[12:30] * sandos (n=sandos@) has joined #freenet
[12:31] * nextgens sets mode +v JRVC
[12:33] <JRVC> Hi, I just tried to join freenet-refs using tor, but I see it's blocked. Can anyone tell me how to exchange refs whil using tor?
[12:33] <railk> JRVC: refs contain your IP, so exchanging them with tor is pointless
[12:34] <JRVC> railk, so I should connect without using tor and exchange them?
[12:34] <railk> yep, pretty much...
[12:35] <bright_fame> _ph00: you can get a port scan from grc.com's shieldsup page
[12:35] <JRVC> Well, okay. I just didn't notice the IP address in the ref.
[12:38] <railk> JRVC: it may not be in there at the moment because your node hasnt been able to detect your ip yet
[12:38] <railk> it'll be in the line physical.udp=
[12:38] * railk was kicked from #freenet by FreenetLogBot
[12:38] <nextgens> :p
[12:38] * railk (n=railk@) has joined #freenet
[12:38] <nextgens> :p
[12:38] <railk> ...
[12:39] <nextgens> FreenetLogBot good dog : stay quiet now :)
[12:39] <railk> well from now on i'll use phy51c41.udp= :D
[12:41] <JRVC> railk, It was detected already, I see
[12:41] <_ph00> ralthe word phisical [dot] udp tricggers auto-kick
[12:42] <_ph00> oh, you got that already
[12:42] <railk> yup ;)
[12:42] <_ph00> (usually, I write 'physical udp' with a space instead of the dot)
[12:42] <railk> of course, if you put that many typoes into it, the bot probably wont auto-kick ;)
[12:42] <_ph00> one typo is enough
[12:43] <JRVC> lol
[12:43] <nextgens> there is no point in auto-kicking missspelt words
[12:43] <nextgens> the word isn't banned
[12:44] <_ph00> no, ref sending is. and the line *won't* have any typo in a ref
[12:44] * whiterabbit (n=Whiterab@) Quit ("KVIrc 3.2.5 Anomalies http://www.kvirc.net/")
[12:57] * CIA-14 (i=cia@) Quit ()
[13:05] * Kata_Pult (i=KataPult@) has joined #freenet
[13:08] <_ph00> my node crashed
[13:08] <_ph00> I installed the latest testing r, restarted, now I wanted to have a look and it was down. I can paste the wrapper.log on a pastebin if someone is intrested (or paste what you need)
[13:09] * CIA-14 (i=cia@) has joined #freenet
[13:10] <_ph00> it keeps crashing
[13:10] * nextgens sets mode +v Kata_Pult
[13:12] * alongtimeago (n=anthony@) has joined #freenet
[13:18] <alongtimeago> #leave
[13:18] * alongtimeago (n=anthony@) Quit ("using sirc version 2.211+KSIRC/1.3.12")
[13:21] <_ph00> here's the last hour or so of wrapper.log => http://pastebin.ca/281057
[13:21] <_ph00> I'm switching back to latest stable
[13:25] <JRVC> railk, I'm connected now
[13:25] <JRVC> railk, Does freenet automatically add new nodes?
[13:26] <bright_fame> JRVC: not afaik
[13:27] <_ph00> OK, I'm back on 11307 (I had 11399 before and it was working, then I updated to latest testing and it kept crashing)
[13:27] <JRVC> That would be a great feature imo. It makes the network more stable, because when a node quits, freenet can connect to others, like tor does :P
[13:28] * CIA-14 (i=cia@) Quit ()
[13:30] <_ph00> nextgens: the one-more-click thingy to remove peers is preventing from removing multiple peers, maybe after answering 'yes remove' it should go forthy to the next remove-or-cancel instad of removing (or canceling) and then go back to darknet page without doing anything more. Or it should have an option to disable the 'are you sure' thing
[13:30] <_ph00> forth*
[13:31] * Muixirt (n=trixium@) has joined #freenet
[13:31] * atlantide (n=10402A70@) has joined #freenet
[13:39] <railk> JRVC: that feature will never be added because you are not anonymous to your immediate peers. thus you should only add people you trust (at least to a certain extent)...
[13:39] <JRVC> railk, Okay.
[13:39] <railk> JRVC: there are however some websites that try and emulate this feature
[13:39] <railk> iirc
[13:42] * lattt (n=gerhard@) has joined #freenet
[13:49] <Muixirt> any devls here?
[13:54] <_ph00> try asking anyway: they read the backlogs (I'm no dev, but yes, they usually hang around here)
[13:54] * CIA-14 (i=cia@) has joined #freenet
[13:56] * nux (n=noaddres@) has joined #freenet
[13:58] <Muixirt> ok _ph00
[13:58] <Muixirt> please fix StringArrOption.java
[13:59] <Muixirt> java.lang.StackOverflowError
[14:02] * atlantide (n=10402A70@) has left #freenet
[14:04] <railk> Muixirt: they're gonna need more information than that...
[14:04] * nux (n=noaddres@) Quit ()
[14:05] * CIA-14 (i=cia@) Quit ()
[14:06] <Muixirt> railk, it seems getValueString calls getValue calls getValueString calls getValue ... :-)
[14:06] <railk> in class?
[14:07] <Muixirt> freenet.config.StringArrOption
[14:08] <railk> nextgens: you around?
[14:08] <railk> hmm...
[14:09] <railk> Muixirt: try updating your node
[14:09] <railk> maybe theres a fixed version available
[14:12] <Muixirt> railk, fyi i got that though svn so i asked for the trouble :-)
[14:12] <railk> lol... true.. but it should get fixed anyway :D
[14:19] * bright_fame (n=robertvi@) Quit (Read error: 110 (Connection timed out))
[14:20] * bright_fame (n=robertvi@) has joined #freenet
[14:22] * Gryvon (i=whoosh@) has joined #freenet
[14:24] <nextgens> railk> yep
[14:24] <nextgens> what's up ?
[14:24] <railk> read up a bit...
[14:26] <nextgens> _ph00> no; that's a feature
[14:26] <nextgens> not a bug
[14:27] <nextgens> ok, I'll fix it
[14:31] <nextgens> hmm
[14:32] <nextgens> that will require some testing
[14:33] <nextgens> Muixirt> grab r11434 once built
[14:35] <nextgens> hmm
[14:35] <nextgens> it doesn't compile
[14:36] <nextgens> Muixirt> r11435 maybe ?
[14:37] * bright_fame (n=robertvi@) has left #freenet
[14:37] <nextgens> "Your browser sent a request that Freenet could not understand."
[14:37] <nextgens> hmm
[14:38] <nextgens> either I've broken something or toad did :'(
[14:38] <Muixirt> yeah that was good too ;-)
[14:39] <nextgens> pluginmanager2.loadedPlugins=[Ljava.lang.String;@1827284;
[14:39] <nextgens> Muixirt> don't use what I've commited
[14:39] <nextgens> it's gonna screw your config
[14:40] <Muixirt> ok nextgens
[14:40] <nextgens> where did he put the stringarraycallback ?
[14:40] <nextgens> hmm
[14:41] <nextgens> freenet/support/api/StringArrCallback
[14:41] <nextgens> hmm
[14:41] * Gryvon (i=whoosh@) Quit ("<LSnK> I printed out 'Ritz' in 1,000,000 point font and stuck it all up my asshole.")
[14:44] * nextgens tests a fix
[14:50] * JustMe_ (i=JustMe_@) has joined #freenet
[15:10] <nextgens> I'll let toad fix the welcome toadlet
[15:14] * lattt (n=gerhard@) Quit ("baba")
[15:31] * mrflibble (n=mrflibbl@) has joined #freenet
[15:46] * toad_ (n=toad@) has joined #freenet
[15:46] * ChanServ sets mode +o toad_
[15:46] * MADDY (n=MADDY@) has joined #freenet
[15:46] <MADDY> hello
[15:50] <toad_> hi
[15:52] * vfervers (n=vfervers@) has joined #freenet
[15:52] * vfervers_ (n=vfervers@) Quit (Read error: 110 (Connection timed out))
[15:54] <nextgens> hey toad_
[15:54] * toad_ will ignore OOMs for now; they don't seem to appear on demand, profiling isn't producing useful results ...
[15:54] <toad_> nextgens: hey
[15:54] <nextgens> toad_> it seems that you have broken the trunk
[15:54] <toad_> oh?
[15:54] <nextgens> I've fixed something in the StringSubConfig
[15:55] <nextgens> but I'm not quite sure it ought to be done that way
[15:55] <nextgens> ...
[15:55] <toad_> StringArrOption?
[15:55] <toad_> what file did you change?
[15:55] <nextgens> and the welcome toadlet is producing an error
[15:55] <toad_> okay so it compiles but it doesn't run?
[15:55] <nextgens> yes StringArrOption
[15:55] <nextgens> now it runs
[15:55] <nextgens> I belive I've fixed it
[15:55] <nextgens> but I'm not 100% sure
[15:56] <toad_> i don't see a commit on the -cvs list
[15:56] <toad_> and yesterday CIA was broken
[15:56] <toad_> is something wrong?
[15:56] <nextgens> [13:58] < Muixirt> | please fix StringArrOption.java
[15:56] <nextgens> [13:59] < Muixirt> | java.lang.StackOverflowError
[15:56] <nextgens> [14:06] < Muixirt> | railk, it seems getValueString calls getValue calls getValueString calls getValue ... :-)
[15:56] <nextgens> cia is down
[15:56] <toad_> ok
[15:56] <nextgens> I dunno about @cvs
[15:56] <toad_> you said you committed a fix
[15:57] <toad_> it should have shown up on cvs, it didn't
[15:57] <nextgens> indeed
[15:57] <nextgens> what's the latest commit you got from @cvs ?
[15:57] <toad_> r11428
[15:58] <nextgens> the archiver might be out of date too
[15:58] <nextgens> ok
[15:58] * nextgens will check
[15:58] <toad_> 11425 is on the modqueue
[15:58] <toad_> etch is frozen btw
[15:58] <nextgens> since two days iirc :)
[16:02] <toad_> nextgens: looks okay to me, but i need to see the diff
[16:02] <nextgens> I'm on it
[16:02] * toad_ wonders if we can use java.compiler in the future :)
[16:03] <toad_> in updating! :)
[16:03] * JRVC (i=JJK@) Quit ("Ik ga weg")
[16:03] <nextgens> DOH
[16:03] <nextgens> I got the problem
[16:06] <toad_> fixed?
[16:06] <nextgens> not yet
[16:07] * toad_ supposes build security is ultimately a job for a distributed RCS
[16:07] <nextgens> :'(
[16:07] <nextgens> you've commited ?
[16:07] <toad_> no
[16:08] <nextgens> hmm
[16:08] <nextgens> toad_> ok, it's fixed now
[16:08] <nextgens> tell me if a diff is missing
[16:08] <toad_> nextgens: you forced it to send the missing notifications?
[16:08] <nextgens> yes
[16:08] <toad_> nextgens: you committed something, that should be sent at least
[16:08] <toad_> ok
[16:08] <toad_> 1sec
[16:11] <nextgens> ok, Jflesch has commited to /apps/Thaw in the meantime :/
[16:14] * MADDY (n=MADDY@) Quit (Read error: 110 (Connection timed out))
[16:14] * rebo123 (n=rebo123@) has joined #freenet
[16:15] * iwantnode (n=iwantnod@) has joined #freenet
[16:15] * toad_ wonders if he will be able to talk to Bombe about plugins at some point
[16:16] <nextgens> Bombe> around ?
[16:19] <_ph00> <nextgens> _ph00> no; that's a feature <= that's the most annoying featire ever
[16:20] <toad_> _ph00: what feature?
[16:21] <railk> <_ph00> nextgens: the one-more-click thingy to remove peers is preventing from removing multiple peers, maybe after answering 'yes remove' it should go forthy to the next remove-or-cancel instad of removing (or canceling) and then go back to darknet page without doing anything more. Or it should have an option to disable the 'are you sure' thing
[16:21] <railk> that one
[16:21] <_ph00> toad_: when I try to delete multiple peers, and more than one of them happen to have been inactive for less than a week..... (uh... OK, thx railk )
[16:22] <_ph00> that's no feature if you ask me, only an annoying windows-like thing
[16:24] <nextgens> indeed
[16:24] <nextgens> a "usefull" feature as such
[16:24] <nextgens> it it was depending only on me, you wouldn't be able to remove any peer at all :)
[16:26] <_ph00> well, it's not useful at all, and only annoying for me. I get the point, tho: it was to reduce indiscrimined deletions... but making it diable-able for "advanced" users would be nice
[16:26] <nextgens> it's usefull
[16:26] <nextgens> not to you but to the network
[16:26] <_ph00> nextgens: if you could decide it alone, we couldn't do *anything*
[16:27] <nextgens> making it boring enough to discourage you to "prune" peers is good for network health
[16:27] <nextgens> that's not true
[16:27] <nextgens> but you wouldn't do nor be allowed to have any anti-social behaviour
[16:27] <_ph00> nextgens: you never thought that maybe I'm trying to do something? and if I want to delete multyiple peers maybe I know what I'm doing? do you really think that all the non-dev users are morons? try and trust your fellow human beings a bit more
[16:28] <iwantnode> 1 of my peer is at the uni of callifornia should i remove
[16:28] * Gasi (n=chatzill@) has joined #freenet
[16:29] <nextgens> [10:01] < bright_fame> | anyone else listen to this podcast about freenet and tor? http://www.grc.com/SecurityNow.htm#70
[16:29] <_ph00> iwantnode: no reason to remove a peer only because it's based at univ. of ca. probably some student or prof experimenting with freenet: that's a *good* thing
[16:29] <nextgens> toad_> in case you hadn't seen it
[16:29] <toad_> _ph00: hmmm
[16:29] <toad_> nextgens: i think we should confirm _all_ the nodes ...
[16:29] <toad_> if you submit a multiple node deletion, we should have one confirm page for all of them
[16:30] <nextgens> a separated one then ?
[16:30] * nextgens doesn't want the user to be able to prune more than one ref in two clicks
[16:30] <toad_> nextgens: is it (podcast) interesting?
[16:30] * sanity (n=ian@) has joined #freenet
[16:30] * ChanServ sets mode +o sanity
[16:30] <toad_> nextgens: it will be more than two clicks anyway
[16:30] <toad_> nextgens: they have to select them in the first place
[16:30] <nextgens> well, it's speaking about anonymity and freenet
[16:31] * sanity (n=ian@) Quit (Client Quit)
[16:31] <nextgens> I wouldn't say it's interresting for a dev :)
[16:31] <toad_> how do they deal with freenet vs tor then?
[16:31] <_ph00> toad_: nextgens: no, really, I think that this ' you shouldn't be allowed to...' way of thinking is completely wrong, and OK, we should make people aware of the fact that too aggrressive deletion is not a good thing... but "you shouldn't be allowed tp" id the wrong way to fix it. Making people understand *why* they shouldn't be too aggressive on deletion would be the right way if you ask me
[16:31] <_ph00> (plz don't say "no one is asking you...."
[16:32] <_ph00> )
[16:32] <nextgens> toad_> IIRC I wrote that part of the code ... and it was on purpose I've broken it 'cause I wanted to simplify the code
[16:32] <toad_> _ph00: don't we tell them why?
[16:32] <_ph00> yes
[16:32] <_ph00> and I also do
[16:32] <toad_> _ph00: is there some part of the warning page's message that needs rewriting?
[16:33] <nextgens> toad_> they don't ... they do present things honnestly, stating that freenet is a "virtual network" whereas tor is a tool to access the regular internet
[16:33] <toad_> nextgens: well it looks to me like we should have one warning page
[16:33] * nextgens is against that
[16:33] <_ph00> but what I was talking abou was only one thing: the "feature" that implements "one more click" before deleting a peer (which is no big deal per se) is also preventing me from deleting multiple peers at once, which I think is *very* annoying
[16:33] <nextgens> we do provide the functionnality
[16:33] <toad_> for however many non-"NEVER CONNECTED" peers
[16:33] <nextgens> he can remove as many peers as he wants ...
[16:34] <nextgens> but he is too lazy to click :/
[16:34] <nextgens> his bad
[16:34] <toad_> obviously it should provide a list of peer names etc to be deleted
[16:34] * sanity (n=ian@) has joined #freenet
[16:34] * ChanServ sets mode +o sanity
[16:34] <nextgens> and yes that's not good in terms of usability ... and that's ON PURPOSE :p
[16:35] <toad_> nextgens: why the extra sabotage?
[16:35] <toad_> nextgens: why not just show the user all the peers they are about to delete, and explain why that's not a good idea?
[16:35] <_ph00> toad_, nextgens: the problem is not the "are you sure" message, that's fine. The problem is that if I mark more than one peer for deletion, it asks me if I want to delete *one* (the first in queue), then, after I've done my choice, it won't continue with the other deleting requests, but go back to he darknet page intead, after only deleting *one* peer
[16:35] <nextgens> toad_> because the user won't care
[16:36] <toad_> nextgens: so?
[16:36] <nextgens> whereas if we require him to click ten times on the warning, he will read
[16:36] <_ph00> implementing multiple peer deletion, vene *with* "are you sure" message, would be a good thing
[16:36] <nextgens> at least he ought to end up reading it
[16:36] <nextgens> and hopefully understand why it's bad
[16:36] <toad_> nextgens: well then maybe we should have it remember all the other peers, and show one warning message per peer?
[16:36] <toad_> nextgens: losing information isn't really on
[16:36] <toad_> once the user has chosen to drop N peers, we should remember that fact
[16:37] <toad_> and either ask them once for all of them, or ask them N times
[16:37] <_ph00> nextgens: plz don't try to guess what the users will think. A warning may be OK, as information, but crippling users' control over their own nodes can be very annoying
[16:37] <toad_> but certainly we shouldn't pick one and then force the user to go back and select them all over again
[16:37] * toad_ would be happy with either a) single warning, or b) N warnings
[16:37] <toad_> but certainly not with pick one node, warn user and ignore the rest
[16:37] * nextgens doesn't consider a
[16:38] <_ph00> one single warning and an "OK, delete them all" option would be fine
[16:38] <nextgens> and wouldn't be convince that implementing b has a high enough priority to be bothered for
[16:38] <nextgens> +d
[16:38] <toad_> nextgens: losing information is almost always bad
[16:38] <nextgens> yes that's a bug ... fill in a ticket on the BTS :p
[16:38] <_ph00> even a delete (1), OK; delete (2), OK; etc would be fine. just dfon't force me to delete one peer at the time
[16:39] <nextgens> _ph00> in any case you will have to delete them one by one
[16:39] <toad_> nextgens: i'm sure sanity would say that we should let the user shoot himself in the head if he wants to
[16:39] <_ph00> nextgens: I was thinking about keep clicking OK's in a raw
[16:39] <sanity> toad_: its not our job to prevent suicide
[16:39] * nextgens is glad he is missing :)
[16:39] <toad_> unfortunately he also shoots us in the head
[16:40] <toad_> lol
[16:40] <sanity> ah, you see, just when you least expect it :-)
[16:40] <toad_> well, because of its ramifications, I might be persuaded to show the warning N times once for each peer
[16:40] <_ph00> nextgens: but the *best* solution for me, would be an "OK, delete them all" option
[16:40] <nextgens> _ph00> then I'll implement "randomization" of the button order :p
[16:40] <toad_> but I DEFINITELY WOULD NOT support the current behaviour
[16:40] <toad_> because it loses the user's selection
[16:40] <toad_> which is a bad thing
[16:41] <toad_> nextgens: better suggestion: have a one click button to drop all NEVER CONNECTED peers over 2 weeks old
[16:41] <toad_> or maybe a two click one
[16:41] <_ph00> nextgens: I really undersand your concerns, but I think tyou should't make it too difficult for the user; if a user decides to delete, I would assume that he knows what heì's doing... at least after *one* warning. no need tor epeat it for each deletion
[16:42] <toad_> inline it from the you-have-never-connected-peers user alert
[16:42] <nextgens> return; // FIXME: maybe it breaks multi-node removing
[16:42] <nextgens> http://emu.freenetproject.org/cgi-bin/viewcvs.cgi/trunk/freenet/src/freenet/clients/http/DarknetConnectionsToadlet.java?rev=11427&view=markup
[16:42] <nextgens> "known issue"
[16:42] <toad_> nextgens: a FIXME doesn't make it a known issue; a bug ticket makes it a known issue
[16:42] <nextgens> expect fixing before 1.0
[16:43] <toad_> nextgens: how come Thaw commits are showing up on -cvs?
[16:43] <nextgens> 'cause I've forced them
[16:43] * iwantnode (n=iwantnod@) Quit ()
[16:43] <nextgens> it's my bad
[16:43] <_ph00> how about: multiple peer selection ==> delete ==> "are you sure?" ==> list of "to be deleted peers" ==> select which ones to actually delete ==> "OK delete those ones"
[16:43] <toad_> _ph00: that's not bad
[16:43] <toad_> _ph00: care to submit a patch? ;)
[16:44] * nextgens will ensure it's merged before 1.0
[16:44] <nextgens> I promise
[16:44] <toad_> lol
[16:44] <toad_> if there's a good patch i'll merge it
[16:44] <toad_> anyway in the meantime we need to have a bug filed
[16:44] <toad_> also we haven't decided the appropriate behaviour
[16:45] <_ph00> of course I've got no idea about how to implement that, as I can't code... but maybe the ground idea is not so bad after all: you get to review your choice before clicking OK, but you only need *one* OK once you've made up your mind
[16:45] <toad_> I think what _ph00 said last is probably the best option
[16:45] <_ph00> "befoer 1,0" sounds like a bad joke to me :P
[16:45] <_ph00> before 1.0 *
[16:45] <toad_> either that, or manual confirmation of each, BUT it remembers them, and automatically moves on to the next after each
[16:46] <nextgens> toad_> second option ; manual confirmation of each
[16:46] <_ph00> manual confirmation of each, implemented in a list with chackboxes, and ONE final OK
[16:46] <_ph00> checkboxes*
[16:46] <nextgens> no.
[16:46] <_ph00> damn
[16:46] <_ph00> (OK, I tried...)
[16:46] <toad_> _ph00: how do checkboxes imply that the user reads the text above?
[16:46] <toad_> :)
[16:46] <toad_> i suppose they do mean more effort => more likely to read text...
[16:47] <_ph00> make the checkboxes checkable only after some tect file has been displayed?
[16:47] <_ph00> text
[16:47] <toad_> but really we should let the user do what they want to do, at least in advanced mode
[16:47] <nextgens> that requires javascript
[16:47] <_ph00> right...
[16:47] <_ph00> :(
[16:47] <nextgens> toad_> they can
[16:48] <nextgens> they just have to workaround a bug :)
[16:48] <_ph00> but if in doubt, I would leave the final decision to the user
[16:48] <nextgens> most users are used to anyway
[16:49] <_ph00> you cvan't keep assuming that 99.9 of users are morons... not even if they *are*. that would be annoying for that .01% non-moron users, who are actually your testers, your bug reporters, etc
[16:49] <nextgens> well, I don't mind you fill in a ticket on mantis :)
[16:49] <_ph00> :)
[16:50] <_ph00> ok, I think you know what I'm talking about
[16:50] <toad_> well I've explained my preferred solutions
[16:50] <toad_> there are two
[16:50] <nextgens> it will be processed as its priority stands : according to the percentage of users it will be usefull to : .01%
[16:50] <toad_> _ph00: if you would file bugs that would actually be really helpful
[16:50] <_ph00> making it a bit less easy to delte peers has a pint, but making it unnecessarily difficulòt is only annoying
[16:51] <_ph00> I never did that, I don't even know what app I'm supposed to use
[16:51] <nextgens> the bug tracker ?
[16:51] <nextgens> https://bugs.freenetproject.org/
[16:52] <_ph00> anyways, I find the nextgens solutions a bit too 'hard' towards 'regular 6-pack-joe users'
[16:52] <_ph00> I tink you should trust your users' judgement a bit more
[16:52] <_ph00> think
[16:52] <toad_> regular 6-pack user would have opennet, opennet, and opennet :)
[16:52] * _ph00 checks out the big tracker
[16:53] <toad_> no, /me is confusing most common with most vocal
[16:53] <nextgens> and wouldn't bother with managing their peer connections anyway
[16:53] <nextgens> maybe too :)
[16:53] <toad_> well some idiot told them they needed to...
[16:53] <toad_> brb
[16:53] <_ph00> zothar's bot2bot thingy is working: if you set up a bot-dedicated channel, you could implemet a "click here to autp add peers" feature
[16:54] <_ph00> ...without even needing to tell them that they're using IRC
[16:54] <_ph00> just "click here to auto add peers"
[16:55] <_ph00> [and a "how does that work" link, for those who bother to read)
[16:55] <toad_> yeah, and we can add a "Connect me to the NSA" button too...
[16:56] <_ph00> you really think that no one of our most knweon and *trusted* users are actually DoD undercover guys?
[16:56] <_ph00> can you ruile that out?
[16:56] <sanity> toad_: what did I tell you? if we don't implement opennet, other people will implement it for themselves in a crap way
[16:56] <_ph00> no, you cant, of course
[16:56] * rebo (n=rebo123@) has joined #freenet
[16:57] <toad_> of course not but it's more expensive to infiltrate a social network at the social level than to infiltrate an electronic network at the purely electronic level
[16:57] <_ph00> so the "connect me to NSA" problem is not a real problem, *any* freenet user *could* be an agent
[16:57] <toad_> that's the whole basis of darknet
[16:57] <_ph00> yes
[16:57] * MikeW (i=Mike@) has joined #freenet
[16:57] <toad_> _ph00: #freenet-refs does indeed suck for that reason
[16:57] * rebo123 (n=rebo123@) Quit (Read error: 104 (Connection reset by peer))
[16:57] <_ph00> but, if I choos to run true darklnet, I'm able to do that
[16:57] <toad_> _ph00: it's just the same as opennet, it's just harder to use
[16:57] <toad_> anyway
[16:57] <toad_> one thing we do know
[16:58] <toad_> a lot of users will put significant effort in to "getting the bloody thing to work"
[16:58] <toad_> usually they do that on the basis of dubious third hand theories
[16:58] <nextgens> until we have StS deployed and a solid, working load-balancing algorithm, implementing opennet would be assimilable to suicide
[16:58] <_ph00> but as many ppl want opennet, the bot2bot thingy with a "click here to auto add peers" weould work just as fine as connecting to strangers on IRC
[16:58] <toad_> nextgens: why does MITM/impersonation matter on opennet? they can just connect to you
[16:59] <toad_> i suppose it matters for hybrid nodes
[16:59] <nextgens> ;)
[16:59] <toad_> it matters for darknet really, but you're busy :(
[17:00] <toad_> sanity: any thoughts on the plugin API?
[17:01] <_ph00> <_ph00> even a delete (1), OK; delete (2), OK; etc would be fine. just dfon't force me to delete one peer at the time
[17:01] <_ph00> <nextgens> _ph00> in any case you will have to delete them one by one
[17:01] <_ph00> at least I could keep hitting OK, much faster than actually select and delete one by one
[17:01] <toad_> _ph00: like i said, losing the original selection is unacceptable
[17:02] <toad_> _ph00: file a bug
[17:02] <_ph00> (yet, the "delet all" option is my favorite)
[17:02] <_ph00> ok
[17:02] <_ph00> file a bug for "losing the original selection"
[17:02] <_ph00> I'll check bugs.freenetproject.org out
[17:03] <toad_> well the behaviour is that it deletes one peer and only one peer
[17:03] <toad_> and it forgets about the rest
[17:03] <toad_> what it should do is either ask you to confirm all of them or ask you to confirm each one one at a time
[17:03] <toad_> _ph00: i am curious why you are bulk deleting btw?
[17:05] <nextgens> actually it would be faster to implement it rather than debating about it
[17:06] * einstein14 (n=einstein@) has joined #freenet
[17:06] <einstein14> hi
[17:07] <_ph00> toad_: I've been testuing zothar's bot, and that gave me more peers than I actually need, so I'm deleting some.
[17:07] <_ph00> testing
[17:09] <_ph00> toad_: for me, the best option would be a list of the peers.to-be-deleted to show up, where the user has to re-select (checkboxes) what peers to actually delete, and then do that with *one* OK button
[17:09] <_ph00> like "there peers will be deleted <list> <checkboxes> <OK>"
[17:10] <_ph00> these*
[17:10] <_ph00> but then again, you are the coders... I only say what woyld look good to me
[17:10] <_ph00> would*
[17:10] * toad_ can see why nextgens would regard that as abusive :)
[17:14] <_ph00> I tink *abusive* is keep all the control fro the devs and making it difficult for users to decide anything. I don't know how much I'm right, but if all users thaought like I do, that attitude could even decide the fate of freenet: failure
[17:14] <_ph00> (well, I hope not any user thinks like I do, then...)
[17:15] <sanity> toad_: my main thought on the plugin API is that it should be a low priority relative to opennet for reasons that are becoming increasingly apparent (ie. now that my prediction that people would build their own opennet if we don't provide one is coming true)
[17:15] <toad_> well adding hundreds of peers from an irc channel through a bot is a little anti-social that's all
[17:15] <_ph00> there's the warnig at 30 connected
[17:16] * Caco_Patane (n=caco@) has joined #freenet
[17:17] <toad_> sanity: well, I finally managed to get hold of Bombe, and we decided on a new (securable, self-contained) architecture; plugins are being used more (e.g. there's a Librarian index being maintained by Entry Point)
[17:17] <_ph00> "click here to auto add peers" button that runs zothar's bot2bot thing on a bot-dedicated channel would be teh solution, opennet people should stop bitching and darknet people can still run darknet
[17:17] <toad_> sanity: so it seemed to make sense to try to move forward a bit
[17:17] <toad_> sanity: but if Bombe disappears back into the woodwork, I'm not going to do it all myself
[17:17] <sanity> toad_: ok, but what about opennet?
[17:18] * MikeW (i=Mike@) Quit (Read error: 113 (No route to host))
[17:19] <toad_> sanity: we don't understand the load balancing system yet
[17:19] <sanity> toad_: that shouldn't prevent you from *beginning* work on opennet while we work on load balancing
[17:19] <toad_> it's being worked on, but simulations aren't really my job
[17:20] <_ph00> Caco_Patane: you node has been disconnected from mine for over two days; is it up? are you having any problems? (my nodename: zaphod)
[17:20] <sanity> toad_: this IRC pseudo-opennet should serve as a wake up call as to the fact that if we don't provide opennet, people will provide if for themselves, and it will likely be centralized and not maintain network topology, and consequently the entire network will fail.
[17:20] <Caco_Patane> hi _ph00
[17:20] <toad_> yeah and 5 people will use it
[17:20] <sanity> toad_: we have a short window of opportunity to provide a robust opennet before people start adopting their own
[17:20] <Caco_Patane> i have to migrate my node out of my workstation
[17:20] <_ph00> Caco_Patane: OK...
[17:20] <sanity> toad_: you want to bet the entire project on that?
[17:20] <toad_> the whole point with opennet is it makes the node work out of the box
[17:20] <Caco_Patane> because i'm turning it off when i'm not home
[17:21] <sanity> toad_: exactly
[17:21] <toad_> a third party opennet will NEVER work out of the box
[17:21] <_ph00> Caco_Patane: do I have to r4e-add you, only wait, or what?
[17:21] <Caco_Patane> no
[17:21] <Caco_Patane> i'm just turn it on again
[17:21] <Caco_Patane> *i
[17:21] <sanity> toad_: which means there is an even worse danger - that Freenet is re-bundled by the third party and we lose all hope of control
[17:21] <_ph00> only wait then
[17:21] <toad_> sanity: that's a danger?
[17:22] <sanity> toad_: of course it is, how could our Freenet possibly compete with a "Freenet++" provided by a third party that worked out of the box?
[17:22] <Caco_Patane> it up, isn't connected?
[17:22] <_ph00> Caco_Patane: looks disconnected to me
[17:22] <sanity> toad_: i'm just surprised that it hasn't happened already
[17:22] <toad_> except that it wouldn't, because it wouldn't scale beyond a few thousand nodes
[17:22] <_ph00> Caco_Patane: connected now
[17:22] <Caco_Patane> great!
[17:23] <sanity> toad_: perhaps not, or perhaps they will be smart enough to make the network topology good enough that it would scale
[17:23] <toad_> sanity: and if they did, more power to them, what's the problem?
[17:23] <sanity> toad_: IT WOULD BE CENTRALIZED
[17:23] <toad_> sanity: it's all open source
[17:24] <toad_> sanity: creating a small world topology in a centralized way would be at least as hard as building a true opennet fork
[17:24] <sanity> toad_: the problem is that they would be exploiting the reputation we have built up for producing a relatively secure decentralized product to hawk a completely insecure centralized product
[17:24] <toad_> no, not a completely insecure product
[17:24] <toad_> a slightly-less-secure-than-opennet product
[17:25] <sanity> toad_: assuming that the people that build it even care about security
[17:25] <sanity> toad_: the bottom line is it would be the end of *this* project
[17:25] <toad_> if they don't care about security they'll just build Napster
[17:26] <sanity> toad_: they can't build napster because they would get sued
[17:26] <toad_> everyone cares about security for that exact reason
[17:26] <toad_> right
[17:26] <toad_> that's why everyone DOES care about security
[17:26] <sanity> everyone doesn't care about security
[17:27] * Gasi (n=chatzill@) Quit ("Chatzilla 0.9.67+ [SeaMonkey 1.0.6/2006103003]")
[17:27] <sanity> toad_: so, anyway, you are saying that you don't mind if a group of people you don't even know, effectively take over Freenet?
[17:27] <toad_> i don't believe it would work
[17:27] <toad_> not without forking and writing a real opennet
[17:28] <sanity> toad_: you are wrong
[17:28] <sanity> toad_: a centralized opennet will be adequate for many people
[17:28] <toad_> they could do it in a centralized way but they'd need an Oskar
[17:28] <sanity> toad_: and by the time you accept it, it will probably be too late
[17:29] <sanity> toad_: oh come on, an oskar or someone with secondary school honours maths skills
[17:29] <sanity> toad_: maintaining the 1/d property in a centralized way is not deep science
[17:29] <toad_> sanity: I have secondary school maths skills, and I don't see an easy way to build a small world network in a centralized way with constant node churn
[17:29] * railk (n=railk@) Quit ("Leaving")
[17:29] <toad_> sure it's easy to build one if everything is fixed
[17:30] <sanity> toad_: trust me, it wouldn't be that hard
[17:30] * nextgens is worried by that too ; the churn
[17:30] <sanity> toad_: and it certainly wouldn't require a PhD in math
[17:30] <toad_> whatever
[17:30] <toad_> so I build an opennet
[17:30] <toad_> and load balancing isn't fixed
[17:30] <toad_> we deploy it
[17:31] <toad_> and it performs significantly worse than 0.7 does now, and far worse than 0.5 does
[17:31] <toad_> then what?
[17:31] <sanity> toad_: except i never said that you should build it AND DEPLOY IT before load balancing is fixed
[17:31] <sanity> toad_: all i said is that you should start it, and work on it in parallel with the opennet stuff?
[17:31] <nextgens> sanity> following your points, I would say that people will do it for us
[17:31] <sanity> nextgens: do what?
[17:31] <toad_> sanity: work on what in parallel with opennet?
[17:32] <sanity> toad_: load balancing
[17:32] <nextgens> if opennet is ready but not deployed, people will bundle an installer with opennet enabled
[17:32] <toad_> fair point
[17:32] <nextgens> even if we tell them that's a bad idea, not ready, ...
[17:32] <toad_> if we build it somebody will deploy it even if we don't
[17:32] <sanity> nextgens: better that than them building an installer which connects to a centralized matchmaking service such as that which has now been built
[17:32] <toad_> but a true opennet would be a separate network
[17:32] <toad_> well ... not necessarily a separate network, it's designed to be a hybrid
[17:32] <toad_> so it still stings us
[17:33] <sanity> for all i care, toad can keep the code branch on his computer and not share it with anyone until load balancing is ready (although that would be entirely contrary to the spirit of free software)
[17:33] <toad_> okay so you ARE arguing for me to deploy it before it is ready
[17:33] <toad_> hmm ok
[17:33] <sanity> toad_: no i'm not
[17:33] <sanity> toad_: and i have said so explicitly
[17:34] <toad_> sanity: you know what would help with getting more users?
[17:34] <toad_> sanity: MORE SPEED
[17:34] <sanity> toad_: so unless you are accusing me of being a liar, please accept that i don't think opennet should be deployed before load balancing is fixed
[17:34] <toad_> sanity: we bleed users at some significant rate to 0.5, simply because 0.7 is significantly slower - especially for inserts
[17:34] <toad_> people will put the effort in if they get something out
[17:34] <sanity> toad_: because of load balancing, I GET IT, load balancing is the highest priority, but you aren't working on that
[17:35] <sanity> toad_: so i'm suggesting that you work on the #2 priority which is opennet
[17:35] <sanity> so that when load balancing is fixed, opennet is good to go
[17:35] * nextgens doesn't think it's that simple
[17:35] <sanity> nextgens: what isn't that simple about it?
[17:35] <nextgens> plugins are usefull to opennet as well
[17:35] <nextgens> to some extends
[17:35] <toad_> okay... /me: why not just try load balancing on its own? sanity: more speed = more users on #freenet-refs , worse topology - we need opennet
[17:36] <nextgens> as they provide STUN support for instance
[17:36] <toad_> nextgens: no, STUN works now
[17:36] <toad_> nextgens: other plugins would be useful mostly for pseudo-opennet
[17:36] <toad_> nextgens: your view is that if we have opennet, nobody will use darknet, right?
[17:36] <nextgens> sure, but don't we want other plugins -wich will be developped in the meantime- to be using a newer api ?
[17:36] <sanity> toad_: we don't just try load balancing on its own because if we don't provide an opennet other people will build their own opennets and they are very unlikely to be secure, decentralized, or resultant in a good network topology
[17:36] <toad_> nextgens: they won't, the plugin API is horrible, hardly anyone uses it
[17:37] <nextgens> :)
[17:37] <nextgens> hence we dont' have that much plugins btw ;)
[17:37] <toad_> nextgens: your view is that if we have opennet, nobody will use darknet, right?
[17:37] <toad_> sanity: or they'll just use 0.5, or Tor, or I2P
[17:37] <nextgens> well my view is even more pessimistic than that
[17:38] <toad_> nextgens: hmm?
[17:38] <nextgens> for opennet to be usefull it has to be enabled by default
[17:38] <toad_> nextgens: I don't see how we can move forward without opennet, frankly
[17:38] <sanity> we WILL have an opennet, the only question is whether it will be built by us, or by someone else - and if it is built by someone else expect it to be seriously problematic in all the ways i mentioned previously
[17:38] <toad_> nextgens: right, opennet by default, and people get darknet connections later for more security
[17:38] <toad_> the only problem with the theory is people instinctively know that darknet is LESS secure
[17:38] <nextgens> as soon as an opennet enabled node connects, its reference will be harvestable
[17:38] <sanity> nextgens: you can't force people to use darknet
[17:38] <toad_> because they know the people they are connecting to
[17:38] <nextgens> so, to me it's already too late to go dark afterwards
[17:38] * toad_ agrees with sanity; more speed, more users, more idiots on #freenet-refs !
[17:39] <sanity> nextgens: we have already seen that they will work around darknet unless they really *want* to do darknet properly
[17:39] <sanity> nextgens: denying them an opennet option will backfire badly
[17:39] <sanity> nextgens: indeed, i believe it has already backfired
[17:39] * toad_ apologizes to #freenet-refs members for calling them idiots
[17:39] <toad_> but you get the idea
[17:39] <toad_> more people who can't do darknet on #freenet-refs
[17:39] * thrownb (n=d0wb02m@) has joined #freenet
[17:40] <toad_> idiots only in nextgens' eyes :)
[17:40] <thrownb> Does anyone know why Frost doesn't show any boards in the "known boards" list?
[17:40] <nextgens> héhé
[17:40] <toad_> nextgens: #freenet-refs refs are harvestable too
[17:40] <nextgens> toad_> I don't mind taking credits or your phrasing :)
[17:40] <nextgens> of
[17:41] <toad_> nextgens: opennet is a problem to the network, but we have to build a real opennet
[17:41] <nextgens> sure but we don't force the user to go there
[17:41] <toad_> we do
[17:41] <toad_> we advertise freenet via slashdot, via the web site, via all our media coverage
[17:41] <toad_> people download it, and come to us for references
[17:41] <toad_> and we direct them to #freenet-refs because it's the ONLY option available to them
[17:42] <toad_> much better to provide a real opennet; a lot more users would stick with it, apart from anything else
[17:42] <nextgens> on a "real" opennet, where is the initial rendez-vous ?
[17:42] <toad_> and fight the necessary battles on "darknet is less secure" from a position of strength (having a working, large opennet to hybridise from)
[17:42] <toad_> nextgens: it's automatic
[17:42] <toad_> nextgens: we maintain a list of seed nodes
[17:42] <nextgens> I mean who/what is gonna be asked first for reference ?
[17:42] <toad_> a peer can announce through any of them
[17:42] <nextgens> ok, so we do maintain a list
[17:43] <toad_> and it gets a bunch of noderefs to connect to
[17:43] <toad_> yes, it's still centralized (oh no!)
[17:43] <nextgens> like on .5/tor/...
[17:43] <nextgens> ;)
[17:43] <toad_> right
[17:43] <toad_> somewhat like that
[17:43] <toad_> only nodes which are "open" can be seednodes
[17:43] <toad_> for NAT reasons
[17:43] <toad_> i mean open to the world, not NATed
[17:44] <Jflesch> Do you plan to activate opennet by default at the node installation ?
[17:44] <toad_> nextgens: you can see my reasoning, no? we have a sucky (inconvenient, easier to harvest) opennet right now
[17:44] <toad_> Jflesch: we have to, otherwise people will come here and ask how to connect, and we'll tell them to turn it on
[17:44] <nextgens> I'm not convinced
[17:45] <toad_> Jflesch: which is a rather pointless ritual
[17:45] <nextgens> it's not trivial to harvest an irc chan.
[17:45] <toad_> nextgens: yes it is
[17:45] <Jflesch> toad_: why not simply ask at the installation if the user want to use it ?
[17:45] <toad_> nextgens: they paste their refs to code.bulix.org
[17:45] <nextgens> most people are swapping references off chan
[17:45] <toad_> nextgens: all you have to do is lurk on the channel, and read all the code.bulix.org's
[17:45] <toad_> nextgens: I doubt that very much
[17:45] * thrownb is now known as rgm
[17:45] <Jflesch> toad_: explaining the differences between the two, and saying, "If you understand nothing : just click on opennet"
[17:45] <toad_> all newbies are advised to paste to code.bulix
[17:45] <nextgens> toad_> FYI, dark-code.bulix.org has been created for that purpose
[17:45] <_ph00> toad_: re-reading it, I have some doubts... it it clear enough? ==> https://bugs.freenetproject.org/view.php?id=1021
[17:46] <toad_> Jflesch: yes we can have a switch on install
[17:46] <nextgens> toad_> it's not havestable anymore
[17:46] <_ph00> nextgens too ==> https://bugs.freenetproject.org/view.php?id=1021
[17:46] <nextgens> not that way at least
[17:46] <toad_> nextgens: it is if people paste the ref to the channel
[17:46] * rgm (n=d0wb02m@) Quit ()
[17:46] <nextgens> _ph00> stop repeating !
[17:46] <toad_> nextgens: the ref URL
[17:46] <nextgens> of course they shouldn't ...
[17:46] <_ph00> nextgens: I wasnt repeating
[17:46] <Jflesch> toad_: I would really prefer this option, because I plan to make a separate darknet with other students in some months
[17:46] <nextgens> and they ought to do it off-chan
[17:46] <toad_> nextgens: they ought to get real darknet peers, but they don't
[17:46] <Jflesch> toad_: a darknet on a LAN
[17:46] <toad_> nextgens: because in real life they can't
[17:47] <toad_> nextgens: they don't exchange them off channel because they don't know any better
[17:47] <toad_> until they've been here for quite some time
[17:47] <toad_> and even if they did they don't know to register with nickserv so they can send privmsg's
[17:47] <toad_> nextgens: would it be possible to ask whether you want the node to be opennet in the installer?
[17:47] <nextgens> I've offered to you to use/advertise an alternate irc server :)
[17:48] <nextgens> sure
[17:48] <nextgens> it could be a "pack"
[17:48] <toad_> nextgens: so then the bad guys only have to root one server instead of however many freenode has
[17:48] <nextgens> and we would present opennet as a poisoned gift
[17:48] <toad_> hehe
[17:48] <nextgens> a "feature" like _ph00 loves
[17:49] <_ph00> no
[17:49] <nextgens> toad_> r00ting one server on freenode is probably much harder than spoofing the seednode file from emu/one of our mirros
[17:49] <_ph00> that was not the point
[17:49] <nextgens> +r
[17:49] <anonymouse> can't we all just get along? =P
[17:49] <toad_> WARNING: You are about to enable opennet. If your node runs in opennet mode, then the Bad Guys can identify your node and even connect to it! If you know people who run Freenet already, then please DO NOT USE OPENNET. Instead connect to them manually via Darknet.
[17:49] <nextgens> _ph00> I was just j/k :)
[17:50] <toad_> nextgens: true
[17:50] <_ph00> ok
[17:50] <toad_> nextgens: ok, on freenode they just have to lurk
[17:50] <toad_> nextgens: if we host our own server, they have to root the server ... IF people don't just paste their ref url's, which they probably will still do
[17:51] <nextgens> we could auto-kick people before they proceed
[17:51] <anonymouse> If you control the server software, you could filter stuff from being posted
[17:51] <toad_> Jflesch: what do you think?
[17:51] <nextgens> we could have services reconize bulix's refs as "badwords"
[17:51] <anonymouse> heck why not all URLs
[17:51] <toad_> nextgens: it's still centralized, and it's still fairly easy to harvest, or to get lots of people to connect to you
[17:51] <anonymouse> er
[17:51] <anonymouse> all paste site url's
[17:51] <toad_> nextgens: and it would have fairly high maintenance costs
[17:52] <nextgens> no much than #freenet-refs
[17:52] * Urs_ShPo (n=gaim@) has joined #freenet
[17:52] <nextgens> wich is more a support channel than anything else lately
[17:52] <toad_> nextgens: so you're still opposed to opennet?
[17:52] <toad_> nextgens: why? we have an opennet already, it sucks
[17:53] <nextgens> I consider time spent on implementing opennet "now" as wasted
[17:53] <toad_> nextgens: why?
[17:53] <toad_> nextgens: it only needs to be done once
[17:53] <toad_> yes it will need to be adapted to load balancing (and vice versa)
[17:53] <toad_> but that can be done later
[17:54] <nextgens> 'cause the "smoothless" transition to the "next generation" load-balancing algorithm will require us to mix up the basis of what we based our design of opennet upon
[17:54] <toad_> i disagree, I don't think that there is that much of an interface between the two
[17:55] <sanity> nextgens: opennet only really replaces what people are doing manually with #freenet-refs - i don't see why any change would be needed to it as a result of changes to load balancing
[17:55] <toad_> there are certainly interactions, but I don't see why there would be a lot of code rewriting needed
[17:55] <nextgens> when you build a house, have you ever wondered why workers don't start deploying the roof before digging ?
[17:55] <nextgens> that would be usefull to the extend they wouldn't be bothered by weather anymore ;)
[17:55] <toad_> nextgens: what we are proposing to do is replace a tent with a big plastic sheet
[17:56] * toad_ thinks we will need to simulate the new load limiting scheme with opennet before we deploy opennet
[17:56] <nextgens> as long as you don't spend more time deploying the sheet than building the roof/ the entire building it's probably wise
[17:56] <nextgens> wise enough to be considered at least
[17:56] <toad_> although we may have to choose between deploying opennet and deploying new load limiting...
[17:57] <nextgens> in wich case I would definitly be for the second option
[17:57] <nextgens> opennet will raise us a bunch of new problems
[17:57] <toad_> well it would have much greater connection churn, that much is clear
[17:57] <toad_> which may have bad interactions with load limiting
[17:58] <toad_> that's something we'll have to face in any case though, unless we decide not to do opennet, which is insane
[17:59] * nextgens does think that we could avoid doing opennet if we had speed
[17:59] <sanity> toad_: why would the two be mutually exclusive? they should be quite separate and relatively independent of each-other
[17:59] * mrflibble (n=mrflibbl@) Quit (Read error: 131 (Connection reset by peer))
[17:59] <toad_> nextgens: if it was faster, we'd get more people _STAYING_
[17:59] * Apophis2_ (n=Apophis@) Quit ("Verlassend")
[17:59] <toad_> that's good
[17:59] <toad_> HOWEVER
[17:59] <nextgens> people are getting used to "swap references"
[17:59] <toad_> and yes we would get more people arriving virally
[17:59] <toad_> BUT
[17:59] <toad_> we would also have to do a release at some point
[17:59] <toad_> and we'd probably get more media coverage
[17:59] <toad_> this means more people using bogus opennet systems
[18:01] <toad_> publicity kills viral growth :|
[18:01] <nextgens> on .5 we were fearing the /. effect
[18:01] <toad_> nextgens: /. kills the network for some time afterwards
[18:02] <nextgens> imho .7 has proven it can handle it well
[18:02] <toad_> nextgens: on a 0.5 opennet, and probably on a 0.7 opennet too
[18:02] <nextgens> I do regard opennet as an avowal we shouldn't have done .7
[18:02] <toad_> nextgens: yes, that's because 80% of people who download freenet uninstall it (or forget about it) when they realise what hoops they'll have to jump through to get it working
[18:03] <toad_> didn't your stats show 1 in 4 uninstalling it ?
[18:03] * nextgens watche