I2P dev meeting, December 21, 2004

Quick recap

  • Present:

ant, bens, bob, detonate, dm, duck, Frooze, frosk, jrandom, kaji, Madman2003, modulus, mule, mule2, orion, Ragnarok, redzog, scintilla, susi23, ugha2p,

完全な IRC ログ

13:05 <@jrandom> 0) hi
13:05 <@jrandom> 1) &
13:05 <@jrandom> 2) 0.5 strategy
13:05 <@jrandom> 3) naming
13:05 <@jrandom> 4) eepsite roundup
13:05 <@jrandom> 5) ???
13:06 <@jrandom> 0) hi
13:06  * jrandom waves
13:06 <@jrandom> weekly status notes posted a lil while ago @ http://dev.i2p.net/pipermail/i2p/2004-December/000528.html
13:07 <@jrandom> lets jump on in to 1) &
13:08 <@jrandom> for those of you who have already upgraded to - a good 1/3 of the network so far - thanks!
13:09 <@jrandom> i do try to keep a more calm pacing to the releases, but there were some things in that really needed wider deployment
13:10 < Madman2003> is working well for me when it comes to disconnects, but i don't run i2p 24/7(i had quite a few irc disconnects lately) and it's only been a few hours after the release
13:10 <@jrandom> as mentioned later on in the email, i dont have a planned date for when the next bugfix release will be, but we shall see
13:10 <@jrandom> ah great Madman2003 
13:10 <@jrandom> yeah, its definitely too early to tell about
13:11 < frosk> i used to see periods of high lag on .4, so far none of those with .5, but again, a bit early
13:11 < frosk> (i'm talking about irc lag, of course)
13:11 <@jrandom> the dns bug fixed could manifest itself in large numbers of peers running older releases failing at once, so the sooner people upgrade, the better
13:12 <@duck> is that related with the failures on those manually entering a hostname?
13:12 <@jrandom> yep
13:12 < dm> how useless is the windows system tray I2P icon!?!?
13:12 <@duck> ah, so that is why config.jsp is still friendly
13:13 < Madman2003> anyone have a clue why some still run pre routers?(it's been out a while)
13:13 <@jrandom> dm: its more of a placeholder right now, plus a status icon saying "i2p is running"
13:13 < dm> They have a life? :)
13:13  * jrandom should resent that...
13:14 < redzog> is there a way to do soft-restarts from the command line?
13:14 <@jrandom> redzog: unfortunately not
13:14 < redzog> hmm, pity
13:14 <@jrandom> other than with wget, perhaps
13:14 < redzog> would make it easier to do automatic updates
13:14 <+detonate> i2prouter stop && i2prouter start :)
13:14 <@jrandom> no, nm, wget wouldnt work either
13:14 <@jrandom> (as the form requires interaction)
13:14 < Madman2003> i generally update trough cvs several times in between releases(at best once a day), only takes a few minutes
13:15 < redzog> lwp::simple could manage it
13:15 < redzog> just a POST
13:15 <@jrandom> redzog: support for that would be pretty cool
13:15 < redzog> I'll try to whip something up
13:15 <@jrandom> well, its more than just a post, you need to read the form presented to you then post those fields back
13:16 <+detonate> eventually releases will be further spaced though.. right?
13:16 <@jrandom> (there's a hidden flag to prevent people from doing things like <img src="/configservice.jsp?action=restart">
13:16 < redzog> heh, right
13:16 <@jrandom> right detonate, t'wasn't planned to be this rapid, once a week at most
13:16 < redzog> does the nonce value change?
13:17 <@jrandom> if it didn't, it wouldnt be a nonce ;)
13:17 < redzog> hmm, seems so
13:17 < redzog> well, between sessions, between pageloads... ;)
13:17 < redzog> pageloads it is
13:17 <@jrandom> right
13:17 <@jrandom> ok, anyone have anything else wrt
13:18 <@jrandom> i'm sur there'll be more discussion later after we've burnt in the new release more
13:18 < dm> oh, is this a meeting?
13:18 <+detonate> starting up seems to be a lot less smooth
13:18 <+detonate> than 2.3
13:18 <@jrandom> oh?  in what way detonate - cpu, lag, memory, time?
13:19 <+detonate> the list of peers takes forever to populate
13:19 <+detonate> and i get a huge number of shitlisted peers
13:19 <+detonate> also the i2ptunnel stuff sometimes hangs, but generally seems to take at least 2x as long to actually start up
13:19 <+detonate> once it's started things smooth out
13:19 <+detonate> it's odd
13:19 <@jrandom> hmm, what does it list for the cause on /logs.jsp#connectionlogs ?
13:20 < ant> <BS314159> I just did a graceful restart into It took 120s to have Local Destinations
13:20 < ant> <BS314159> seems good
13:20 <@jrandom> cool BS314159 - thats pretty much the bare minimum, as we don't start i2ptunnel until 2 minutes after startup :)
13:20 <+detonate> there's nothing out of the ordinary
13:20 <+detonate> a shutdown exception
13:21 <+detonate> but i think i caused that
13:21 < mule> i have pulled over 300M through fcp for a movie with the last release. never been that good before. top rates beyond 40k. great work.
13:21 <@jrandom> wow, nice mule!
13:21 < mule> however i still have serious trouble with recovering from an IP change
13:21 <@jrandom> detonate: hmm, ok, i'd love to debug this further after the meeting or another time you have available
13:22 <+detonate> yeah
13:22 <+detonate> ok
13:22 < dm> tunnel lag: 364ms. What the fuck is going on , the tunnel lag is dropping 100-200ms on each release!
13:22 <@jrandom> ah mule, ok
13:22 <@jrandom> i've got an idea for how we could deal with those hung tcp connections - just toss on a 5m keepalive
13:23 <@jrandom> heh dm, dont worry, it'll get back up again ;)
13:23 < frosk> wow, i only have 261ms here :)
13:24 <@jrandom> ok, if there's nothing else, lets jump on to 2) 0.5 strategy
13:24 < dm> That can't be right...
13:25 <+ugha2p> Looks like I'm late for the meeting again.
13:26 <@jrandom> there's still a lot of work to be done with 0.5, but a broad outline of the process was included in that email
13:26  * jrandom sends ugha2p to the principal's office
13:27 <@jrandom> there are still some details left to be worked out on the tunnel pooling and creation, but i think we've got a few different offerings that will meet the needs of various user bases
13:28 <@jrandom> there'll be some good ol' fashioned documentation posted once most of the kinks in the design are hammered out for y'all's review
13:28 <@jrandom> (currently its taking up ~8 pages in the notebook, should compress well though)
13:29 < kaji> has the meeting started yet?
13:29 <@jrandom> but another one of the tasks listed for 0.5 is "deal with the bandwidth needs of the network", and i have no idea how to plan for that, so we'll play that by ear
13:29 <@jrandom> yes kaji, we're on 2) 0.5 strategy
13:30 <@jrandom> well, thats all i have to say about that at the moment, unless anyone has any questions/comments/concerns?
13:31 <+ugha2p> Wow, most of the routers have already upgraded.
13:31 <+detonate> is filtering http traffic to strip out javascript/etc in the roadmap?
13:31 <+detonate> for 0.5
13:31 <+ugha2p> detonate: No.
13:31 <@jrandom> detonate: 0.6
13:31 < ant> <cat-a-puss> WRT bandwidth, should we enable probabilistic tunnel length, and/or local biased tunnels, for bittorrent as in general BT users have a weaker threat modle?
13:32 <@jrandom> cat-a-puss: yes, definitely.  thats one of the big parts of the 0.5 release
13:32 <+ugha2p> detonate: Unless you implement it first. ;)
13:32 <+detonate> i was thinking about it
13:33 < ant> <cat-a-puss> will html filtering be conducted in a seperate process?
13:33 <@jrandom> i think michelle is looking at that too, so if you two wanted to work together (michelle is learning java) that'd rule
13:33 <+detonate> ok
13:33 <@jrandom> cat-a-puss: i know not.  
13:34 <+ugha2p> cat-a-puss: Why should it?
13:35 < ant> <cat-a-puss> (I ask because I was thinking of making a proxy that ran all incomming brouser traffic through clamav) That is GPLed so if we could include that in the filter, it would probably be good.
13:35 <@jrandom> cool cat-a-puss!
13:35 <+ugha2p> Some people already use Privoxy for I2P.
13:36 < bens> in general, I'm anti-including-stuff
13:36 < susi23> I would rather see people configuring their browsers right than promising to protect them from malicious code.
13:36 <@jrandom> susi23: no one configures their browser properly
13:36 <@jrandom> especially not joe sixpack
13:37 < frosk> one can wonder if Joe is even able to set a proxy for his browser
13:37 <@jrandom> my personal view is that something cgi-proxy like would be ideal
13:37 <@jrandom> exactly frosk
13:37 <@jrandom> with a cgi-proxy like interface (filtering according to their preferences, safe by default), even a drooling moron could use it
13:38 < bens> I suppose I2P needs multiple versions for multiple markets even more than MS Office
13:38 <@jrandom> 'tis why we have small components and push this stuff out of the router bens ;)  
13:38 < Ragnarok> a proxy auto config file would help
13:39 <@jrandom> Ragnarok: we have one, but there are still dangerous things that can be done with it
13:39 < frosk> maybe a specialized i2p browser even (if someone is drowning in free time ;)
13:39 < susi23> ragnarok: that one? http://dev.i2p.net/cgi-bin/cvsweb.cgi/i2p/apps/proxyscript/i2pProxy.pac
13:39 <@jrandom> frosk: on the specialized i2p OS and hardware too, i suppose
13:40 < frosk> hehe, perfect
13:40 < Ragnarok> that's not in the install, though
13:40  * jrandom implements those in the specialized i2p universe
13:40 < susi23> . o O ( perhaps we should try to find a dedicated i2p planet too )
13:40 < susi23> . o O ( damn, too slow )
13:40 < mule> ok, we'll sell the hardware :)
13:40 < frosk> you know what they say, to create something from scratch, first create the universe
13:41 <@jrandom> w00t, now all we need are some investors..
13:41 < bens> seriously, a firefox autoconfigurator might be reasonable
13:41 <@jrandom> bens: the .pac susi linked to above should do the trick
13:41 < bens> not just for proxy; also for the security settings, homepage, etc.
13:41 <@jrandom> we can ship that with the install too, but its insufficient for people who need anonymity (and are not ubergeeks already)
13:42 <@jrandom> hmm, perhaps that sort of thing could go into cervantes' i2p xul app
13:43 <@jrandom> but thats getting further off topic from the 2) 0.5 strategy
13:43 <@jrandom> anyone else have anything for that, or shall we move on to 3) naming?
13:44 -!- Irssi: #i2p: Total of 40 nicks [2 ops, 0 halfops, 6 voices, 32 normal]
13:44 <@jrandom> consider us moved
13:44 <@jrandom> ok, aparently i kind of jumped the gun w/ the 2.0.1 ref of addressbook - Ragnarok, want to give us an update?
13:44 <+ugha2p> jrandom: Can we expect the dates on the roadmap to be correct?
13:45 <@jrandom> ugha2p: they currently reflect my best estimate
13:45 <+ugha2p> jrandom: Ok, right.
13:45 < Ragnarok> it's released now
13:45 <@jrandom> w00t
13:45 < Ragnarok> check ragnarok.i2p
13:45 < Ragnarok> I wasn't planning on releasing it yet, but jrandom forced my hand :)
13:46 <@jrandom> hehe
13:46 <+ugha2p> Ragnarok: Btw, you're missing a link from the homepage. :)
13:46 < Ragnarok> it's just a few bug fixes, nothing major, but it should deal better with some corner cases
13:46 <@jrandom> its on the top right ugha2p 
13:47 < Ragnarok> ugha2p: it's on the sidebar
13:47 < Ragnarok> I'll add links to the post as well, though :)
13:47 < mule2> "that'll be the day when i die". daily IP change to set the clock after.
13:48 < Ragnarok> anyway, if everyone could try it out, that'd be nice.  Bug reports are always appreciated
13:48 <+ugha2p> Ragnarok: Oh, that sidebar is seriously fucked in Opera.
13:48 < mule2> Lease expired 12773d ago
13:49 <+ugha2p> Ragnarok: Well, not really fucked, but just located at the end of the page.
13:49 <@jrandom> cool Ragnarok, thanks
13:49 < Ragnarok> your window's probably not wide enough
13:49 <+ugha2p> Ragnarok: Right, but it should work with any window size.
13:50 <+ugha2p> So you might want to fix it in the future. :)
13:50 < Ragnarok> ugha2p: should is an interesting choice of words :)
13:50 < Frooze> ah, wrong in mozilla 1.7 too.  my window is small though.
13:50 <+ugha2p> Why's that?
13:50 < Frooze> thanks ragnarok.  cool stuff.
13:51 < Ragnarok> I may fix it in the future, but it's really low on my priorities
13:51  * jrandom prefers addressbook updates to html fixes
13:52 < Ragnarok> anyhoo, any questions?
13:53 < frosk> thanks for addressbook, Ragnarok, sounds very useful
13:54 <+ugha2p> Is the documented way of loading addressbook the only way, or are there any less intrusive ones?
13:54 < kaji> i just installed it, it rocks
13:54 < Ragnarok> you can start it by hand using "java -jar addresbook.jar <path to i2p/addressbook>"
13:54 < Ragnarok> thank you :)
13:55 < kaji> oh, and i dled version 2.0.0 is there an update someware?
13:55 < Ragnarok> ok, I fixed the column, it was just a stupid mix of absolute and realitive sizes
13:56 < Ragnarok> yep, there's 2.0.1 up now on ragnarok.i2p
13:57 <+ugha2p> I'm getting "Failed to load Main-Class manifest attribute from" now, but never mind, I'll do a restart later.
13:57 < Ragnarok> whoops
13:58 < Ragnarok> that's my bad
13:58 < Ragnarok> I'll try to fix that soon
13:58 <+ugha2p> Ah, okay. :)
13:58 < Ragnarok> there will also be an easy to install .war version soon
13:59 < dm> jrandom: you are a machine
14:00 <@jrandom> wikked, thanks Ragnarok 
14:00 <@jrandom> susi23: ping?
14:00 < susi23> 1200ms
14:01 <@jrandom> !thwap
14:01 <@jrandom> anyway, wanna give us a rundown on whats up w/ susidns?
14:01 <@jrandom> or should that wait for later?
14:01 < susi23> do we have time for a more general discussion about naming stuff?
14:02 < susi23> what features we want in the future?
14:03 <@jrandom> some of my thoughts are posted up on http://dev.i2p.net/pipermail/i2p/2004-February/000135.html
14:03 <@jrandom> (for what general features)
14:04 <@jrandom> i think the hardest thing will be weaning people off globally unique human readable names, but with some good interfaces it should be doable
14:04 < Ragnarok> implementing the data structures you outlined in xml is one of my next goals
14:04 < susi23> ok, there is a small writing about attributes at http://susi.i2p/removablekeys.html
14:05 < ant> <Jnymo> wow.. pretty crowded in here tonight
14:05 < bens> ragnarok: have you checked out YAML? Might be easier
14:05 <+ugha2p> Jnymo: Yeah, we're trying to hold a meeting here.
14:05 < Ragnarok> YAML's name is far too apt
14:05 <@jrandom> cool susi23, though i think we'll definitely want to migrate away from the plain hosts.txt format
14:05 < ant> <Quadn-werk> addition of a command line graceful restart?
14:06 < ant> <Jnymo> ugha2p: ah
14:06 < susi23> are there any ideas how to keep names unique in the long run?
14:06 <@jrandom> one of the important parts of the data to be managed in the naming service is for an entry to be signed, requiring some hard structure (or careful xml)
14:07 <@jrandom> i dont believe in globally unique human, human readable, and secure names.
14:07 <@jrandom> (i bundle centralized & secure together)
14:07 <@jrandom> susi23: have you seen http://zooko.com/distnames.html ?
14:07 < Ragnarok> I think using an addressbook like system, the names will end up being mostly unique, since it's in the interest of the person claiming a name not to choose one that's already in use
14:08 <@jrandom> Ragnarok: we'll see.  perhaps
14:08 < susi23> i'll check this out
14:08 < bens> I suspect trusted authorities will emerge
14:08 < Ragnarok> well, there already is one
14:08 < frosk> hosts.txt? :)
14:09 < Ragnarok> jrandom's, yeah
14:09 <@jrandom> or if not trusted authorities, names that include the path to uniquely identify it
14:09 <@jrandom> (e.g. "the site orion.i2p calls 'frosk.i2p'")
14:10 <@jrandom> Derek Eddington had some posts along those lines in september - http://dev.i2p.net/pipermail/i2p/2004-September/000432.html
14:10 < bens> frosk.orion.i2p
14:10 <@jrandom> smtp.frosk.ns.orion.i2p
14:11  * jrandom starts building uucp bang paths
14:11 < frosk> hah
14:12 < susi23> ok, what now... how about a "naming roadmap"? :)
14:12 < ant> <Jnymo> you guys have swayed me away from an absolute distributed dns for i2p.. somewhat.. but ducks ideas started me thinking that a trust system might work.. like, a lookup could bring back a list of sites/files, and each could be listed with the amount of trust that the network gives it
14:12 < susi23> once we agreed what to do
14:12 <@jrandom> thats a good idea susi23, wanna write one up?
14:13 <@jrandom> trusting other people's trust has potential, but needs to be done very carefully
14:13 < susi23> I could do this, but I still have no idea WHAT we want to do. There are some decisions to make.
14:14 <@jrandom> (aka only according to the terms that you trust the peers along the chain to the trust author)
14:14 < modulus> there is or there should not be a "network trust" of a site, trust has to be user-centric always
14:14 <@jrandom> susi23: roadmap step 1: decide among $featureset
14:14 < susi23> Or at least we have to develop all ideas into a more precise concept.
14:14 < ant> <Jnymo> well, if it was explicitly simple.. like, if files i2p listed how many sites linked to siteinquestion.i2p
14:15 < Ragnarok> ok, the I've updated the addressbook package with an executable jar.
14:15 < ant> <Jnymo> er, files.i2p
14:15 <@jrandom> jnymo: that turns into a centralized authority - files.i2p
14:15 < modulus> not to say that you could poison the pool of links by establishing a shitload of sites.
14:16 < modulus> googlebombing on i2p
14:16 < ant> <Jnymo> true.. but files i2p could be decentralized
14:16 < susi23> ok, how about collection ideas/information/concepts until, lets say january
14:16 < orion> 'lo all. I see naming is on the table.. *again* :)
14:16 < susi23> then comes the decision phase, ok?
14:16 <@jrandom> sounds good - will you be the point of contact to gather that together?
14:16 < Ragnarok> sure
14:16 < modulus> doesn't matter if the trust aggregate is descentralized, trust has to emanate from the user. anything else can be poisoned imo.
14:17 < susi23> can't we take the mailinglist for this?
14:17 < bob> or perhaps ugha's wiki?
14:17 < ant> <Jnymo> agreed.. but what how to do that? put a little trust meter bar at the top of the web browser?
14:18 <@jrandom> the wiki would be good, we can gather links to all the previous discussions there
14:18 < modulus> jnyo: probably the most feasible solution is to bind to the first name encountered or something.
14:18 < dm> let's all give a hand of applause to jrandom for his wonderful project management
14:18 < susi23> fine
14:18 < modulus> but there are more ways than sausages.
14:19 < susi23> url to the wiki? (for the records)
14:19 < ant> * Jnymo claps
14:19 <@jrandom> ugha.i2p
14:19  * dm claps
14:19 < susi23> ok
14:19 < susi23> then I'm done and ping jrandom back ;)
14:20 < ant> <Jnymo> modulus: so, if i refer a link to someone else, i'm refering them to the site i first binded to.. that might work.. 
14:20 <+ugha2p> Looks like jrandom has ping timeouted.
14:20 <@jrandom> ok cool, anything else on nami^W nm, no more on naming.  on to the wiki
14:20 < modulus> anyway, if you're linking you'll probably want to put an absolute path in the link, not just a name
14:21 <@jrandom> moving forward to 4) eepsite roundup
14:21 < dm> dm.i2p is up and running
14:21 <@jrandom> cool
14:22 <@jrandom> ok, i dont really have much to add beyond whats mentioned in the mail
14:22 < bob> nice to see an influx of sites!  all speedy to access as well!
14:22 <@jrandom> aye, agreed bob
14:22 < bob> orion, thanks for your work.. I use your site daily.
14:22  * jrandom too, the 'last updated' is especially helpful
14:23 < bob> dm: :-)
14:24 <@jrandom> ok, if there's nothing more on that, we can jump to 5) ???
14:24 <@jrandom> is there anything else people want to bring up for the meeting?
14:24 < ant> <Jnymo> hows the net status?
14:24 < ant> <Jnymo> wrt 4.2.5?
14:25 <@jrandom> its looking good, but the release is only a few hours old, so too soon to tell
14:25 < ant> <Jnymo> oh, heh
14:25 < ant> <Jnymo> any fusenet news?
14:26 <@jrandom> (http://piespy.i2p/i2p/i2p-current.png heh)
14:26 < frosk> my work on i2pcontent has been largely put aside for some weeks, but the latest version of the document can be read at http://frosk.i2p/i2pcontent.html . if anyone is interested, do read, and do comment harshly if needs be (om irc when i'm not /away or mail to frosk@mail.i2p)
14:26 < frosk> i2pcontent/fusenet/anything ;)
14:26 < ant> <Jnymo> wordicus
14:28 <@jrandom> ok, if there's nothing else...
14:28 < mule2> tons of applause for all the excellent contributions
14:29 <@jrandom> aye, y'all are doing some kickass shit
14:29 < frosk> you too, jrandom :)
14:29 < orion> word.
14:29 < orion> yes, very much so, you too jrandom.
14:29 < scintilla> hear hear!
14:29 < ant> <Jnymo> yea, i noticed on the site, theres less info on how to help out
14:29 <@jrandom> sometimes kickass, sometimes ass kicked ;)
14:29 < orion> HIP HIP
14:30 < ant> <Jnymo> HORRAY
14:30  * orion smiles
14:30 < Frooze> downloaded eclipse today, to learn java over holiday, because y'all are so impressive.
14:30 <@jrandom> jnymo: many of the small easy-to-accomplish tasks have been done
14:30 <@jrandom> ooh wikked Frooze 
14:31 < Frooze> so, trouble on horizon.  heh
14:31 <@jrandom> jnymo: i really should collect some more of them though and post 'em
14:31 < ant> <Jnymo> jrandom: You still looking for someone to help out on alexandria.i2p?
14:31 <@jrandom> (take cover arizona!)
14:31  * jrandom is not involved in alexandria, but yes, i believe they are still looking for a librarian
14:31 < ant> <Jnymo> learn to swim, folks ;)
14:31  * orion loves pump up the volume references. Vague though they may be.
14:31 <@duck> yes we do
14:31 <@jrandom> :)
14:31 < Ragnarok> jrandom: where is the war actually supposed to go?
14:31 <@jrandom> (orion++)
14:32 <@jrandom> Ragnarok: i2p/webapps/addressbook.war
14:32 <@jrandom> (then restart the router)
14:32 < ant> <Jnymo> duck, you talkint to me?
14:32 < Ragnarok> cool.  I shall commence testing
14:32 <@jrandom> r0x0r
14:32 < ant> <Jnymo> duck: is alexandria on your site?
14:33 <@duck> duck.i2p/alexandria/
14:33 < ant> <Jnymo> word
14:34 <@jrandom> ok, if thats all, we can slide out of here @ the 90m mark..
14:34  * jrandom winds up
14:34  * jrandom *baf*s the meeting closed