I2P dev meeting, August 31, 2004

Quick recap

  • Present:

aum, cervantes, deer, hypercubus, jrandom, kaji, lucky, nicktastic,

تمام وقایع IRC

14:04 < jrandom> 0) hi
14:04 < jrandom> 1) 0.3.4.3
14:04 < jrandom> 2) 0.3.5 and 0.4
14:04 < jrandom> 3) docs
14:04 < jrandom> 4) stasher update
14:04 < jrandom> 5) ???
14:04 < jrandom> 0) hi
14:04  * jrandom waves
14:05 < deer> * Pseudonym waves
14:05  * hypercubus flaps
14:05 < deer> * detonate waves
14:05 < jrandom> weekly status notes @ http://dev.i2p.net/pipermail/i2p/2004-August/000425.html
14:05 < jrandom> moving on to 1) 0.3.4.3
14:06 < jrandom> as it says in the notes, and as you all know from firsthand experience, the net isn't too healthy atm
14:06 < jrandom> lots of messages are lost, and people are often seeing warnings about their leases having expired a while back
14:07 < jrandom> this is unfortunate, and largely addressed in CVS, which will be rolled out when we can (see item 2)
14:07 < kaji> (late) hi
14:08 < jrandom> anyway, i think thats all i've got to mention on 0.3.4.3, beyond whats in the email.  i appreciate your patience as we move forward through the rough patches
14:08 < jrandom> swinging on up to 2) 0.3.5 and 0.4 (unless someone has anything else they'd like to add..?)
14:09 < deer> <oOo> So 90% of broken nodes can knock down the network ^^
14:09 < deer> * Pseudonym eagerly awaits the release of 0.3.5
14:09 < kaji> who was running the dos? they did a good job
14:10 < jrandom> well, I can reach squid consistently from my other CVS HEAD boxes
14:10 < jrandom> so the network isn't 'knocked out' for people on cvs head :)
14:10  * lucky is having partial success with .3.4.3 still.
14:10 < jrandom> but yeah, the old peer selection algorithm did some Stupid Things
14:10 < deer> <oOo> I'm on CVS head and lost suid.i2p a lot of time ;)
14:11 < jrandom> hmm
14:11 < jrandom> what are you seeing for a tunnel failure rate? 
14:12 < jrandom> (total # events at /routerStats.html#tunnel.failAfterTime compared with total # events at #tunnel.buildFrequency )
14:13 < deer> <oOo> lifetime average value: 288 268,91 over 339,00 events
14:13 < jrandom> and tunnel.buildFrequency?
14:14 < deer> <oOo> But you might have been restarting your router too much while repairing thread leaks ;)
14:14 < jrandom> what is your lifetime # of tunnel.buildFrequency?
14:14 < deer> <oOo> 24h frequency: avg per period: (2,76, max 2,76, current is 100,00% of max) strict average per period: 5 645,58 events (averaged using the lifetime of 5 729,00 events)
14:14 < deer> <oOo> 24h ~= router lifetime
14:15 < jrandom> so ~5% tunnel failure
14:15 < jrandom> thats about what i've been seeing on CVS HEAD, as opposed to the 40-60% tunnel failure of 0.3.4.3
14:16 < deer> <oOo> Let's swing on up to 2) then ;)
14:16 < jrandom> consider it swung
14:16 < jrandom> ok, as mentioned in the email, the next rev will be 0.3.5, not 0.4
14:16 < jrandom> it'll have all the goodies y'all have been waiting for, but it won't have the "0.4 stamp of approval" ;)
14:17 < deer> <Pseudonym> 0.4.rc-1
14:17 < jrandom> well, i considered going down the rc road, but I dont want to be overconfident
14:17 < kaji> 0.4.rc-0.9
14:17 < deer> <Pseudonym> heh
14:18 < kaji> beta
14:18 < jrandom> while 0.3.5 is out, I'm going to see if we can mount the DoS again, as well as a variety of new issues that we should be able to come up with
14:18 < lucky> we have to keep DoSing it till it works while being DoSed
14:18 < jrandom> right
14:19 < kaji> dos it till it cant be dosed no more
14:19 < deer> <Pseudonym> but no new features between 0.3.5 and 0.4 right?
14:19 < jrandom> perhaps someone can be inspired to help out with implementing some churn and fail cases in the simulator, so we can test this stuff more easily and automatically... ;)
14:20 < jrandom> correct Pseudonym, I do not expect any significant new features to come during 0.3.5
14:20 < jrandom> at least, from an app user perspective
14:20 < jrandom> perhaps some developer will take this time to improve upon the eepproxy, a transparent webserver, help out aum, etc 
14:21  * jrandom pokes at someone hacking on an irc proxy w/ DCC support ;)
14:21 < deer> <duck> a public inproxy for i2p/tor is in the make
14:21 < jrandom> ah nice, html specific, or bitpipe?
14:21 < jrandom> er, web specific, that is
14:22 < deer> <duck> web specific
14:22 < jrandom> w3rd
14:22 < deer> <duck> the idea being that an ISP can put up some gateways to specific sites
14:22 < deer> <duck> so the world can access alexandria
14:23 < jrandom> ooh, what would *really* rule is if those gateways could act as vhosts
14:23 < jrandom> (maybe thats what you're talking about anyway)
14:23 < deer> <duck> http://anonygateway.com/home.duck.i2p/~alexandria/
14:23 < jrandom> ah ok
14:23 < jrandom> still cool
14:23 < deer> <duck> http://anonygateway.com/6sxoyfb3h2nvok2d.onion/
14:24 < deer> <duck> virtual host is also possible; just for a next iteration
14:24 < jrandom> (though 6sxoyfb3h2nvok2d.onion.anonygateway.com would be cooler ;)
14:24 < jrandom> right right
14:24 < deer> <duck> easy to do with a mod_rewrite ofcourse
14:25 < cervantes> or just set up a subdomain :)
14:25 < kaji> hah vhost a bittorent seed
14:25 < deer> <duck> I am paying dev out of my pocket; patch will be pub domain
14:25 < jrandom> duck++
14:26 < deer> <duck> also talking with an ISP who might want to offer it as a paid service
14:26 < jrandom> nice
14:26 < deer> <duck> ofcourse it is better when anarchistgang.org does so
14:26 < deer> <duck> but you know the stability of those types
14:26 < jrandom> *cough*
14:27 < cervantes> their quackers
14:27 < cervantes> *they're
14:27 < deer> <jon2> hi!!!!!!
14:27  * hypercubus snickers
14:27 < jrandom> hi jon2
14:27 < deer> <jon2> I like meeting >:-D
14:28 < jrandom> i think after the net is settled down a bit more (once 0.3.5 is out there), we'll want to reevaluate some app level activities
14:28 < deer> <duck> *cough* myi2p?
14:28 < jrandom> heh
14:29 < kaji> what about access behind a firewall?
14:29 < deer> <jon2> yes, firewall access :)
14:29 < jrandom> we need something rock solid, usable, *and* secure, that provides functionality that people want (and hopefully, that we can use to encourage community)
14:30 < deer> * duck points at 0.4.2 @ http://www.i2p.net/roadmap
14:30 < jrandom> believe me, i want access behind firewalls / uncontrollable NATs / etc just as much as the rest of you.
14:30 < deer> <jon2> I can do the secure part, I know cryptophagy.
14:30 < jrandom> (someone has to add that as a quote ;)
14:30  * hypercubus wonders what a cryptophage is
14:31 < jrandom> jon2 - we definitely need help on this stuff and would love to snag some of your time!
14:31  * kaji just started back to school, he would like to take i2p with him ;)
14:31 < aum> morning all
14:31 < cervantes> btw I'm wondering if any devs miss their little i2p blogs.... if perhaps they should get devoted forum sections, at least in the short term...
14:31 < cervantes> *if so
14:31 < deer> <jon2> cryptophagy, science of security.
14:31 < jrandom> 'mornin aum
14:32 < hypercubus> jon2: do you also know cryptography?
14:32 < deer> <jon2> Good morning aum.
14:32 < jrandom> cervantes: i'm holding off until i can get a blog of my own, which hopefully wont be too far off
14:32 < deer> <jon2> no :-(
14:33 < cervantes> jrandom: and everyone else?
14:33 < jrandom> nightblade has been using his blog @ cashdollar.org
14:33 < deer> <jon2> I have a blog on blogs.aspnet.com
14:33 < jrandom> though i suppose it'd be cool to have people posting on the forum
14:34 < cervantes> ah good...well it seems most have found alternatives....but it's a same they've become fragmented
14:34 < jrandom> yeah
14:34 < cervantes> *shame
14:34 < cervantes> darn fingerzzz
14:34 < lucky> well, a phage is part oft he immune system.
14:34 < jrandom> i liked having the devblogs on the site.  we'll get something back eventually
14:34 < hypercubus> jon2: funny, blogs.aspnet.com is an unclaimed domain
14:34 < jrandom> ok, anyway, anything else for 2) 0.3.5 and 0.4 ?
14:35 < hypercubus> yeah
14:35 < hypercubus> i've got the firefox problem solved now, in cvs
14:35 < jrandom> w000t
14:36 < deer> <jon2> I am asp developer.
14:36 < hypercubus> reads the default from the registry
14:36 < cervantes> :)
14:36 < deer> <jon2> sorry.. I mean blogs.asp.net
14:36 < hypercubus> no you don't
14:36 < deer> <jon2> weblogs.asp.net
14:36 < jrandom> ah, great hypercubus.  so we're almost there for the 0.3.5 release
14:37 < cervantes> shudder....asp
14:37 < hypercubus> yes i can feel it getting close
14:37 < jrandom> ok, moving on to 3) docs
14:37 < jrandom> well, I dont have anything to add beyond my request in the email
14:38 < jrandom> (send in your questions!  post 'em to the list, send 'em in email, post 'em on the forum)
14:38 < deer> <oOo> Yeah, anonymously use the forum and make Cervantes happy ;)
14:39  * cervantes gets all tingly
14:39  * hypercubus adjusts the rabbit ears
14:40 < nicktastic> haha
14:40 < deer> <jon2> I liked this meeting..
14:40 < cervantes> you said that...
14:40 < cervantes> <deer> <jon2> I like meeting >:-D
14:40 < hypercubus> great, you get to buy the donuts next time ;-)
14:40 < jrandom> ok, if there's nothing else, 4) stasher update
14:41 < jrandom> aum seems to have awoken early... you still 'round?
14:41 < deer> <jon2> GREAT MEETING!
14:41  * hypercubus wonders if dm has children
14:41 < jrandom> heh, yeah, he's back ;)
14:41 < cervantes> I'd think it's an impossibility
14:42 < hypercubus> guess aum missed that first cuppa
14:42 < jrandom> ok, maybe he'll swing back to the term
14:42 < jrandom> anyway, his general update was posted in the email
14:42 < jrandom> looks like there's lots of progress going on
14:43 < jrandom> some questions remain, but ever onward 
14:43 < deer> <oOo> But no release date given ;)
14:43 < hypercubus> how many people are testing it atm?
14:43 < jrandom> i dont know if the code he has now w/ the things mentioned is public yet
14:43 < hypercubus> ah
14:44 < deer> <jon2> BAF BAF BAF BAF BAF
14:44 < kaji> whats new about stasher?
14:44 < jrandom> kaji: see the http://dev.i2p.net/pipermail/i2p/2004-August/000425.html
14:45 < deer> <oOo> It now use less water to wash the dishes
14:45 < hypercubus> i've been waiting for that feature
14:45  * jrandom too
14:45 < jrandom> ok
14:45 < jrandom> if aum is still afk, swinging on to 5) ???
14:45 < jrandom> does anyone else have something they want to bring up?
14:45  * cervantes puts on a tin hat
14:46 < lucky> How's jetta for serving web pages coming along?
14:46 < jrandom> i dont know of anyone working on an app to safely allow people to host pages with jetty
14:46 < jrandom> (host pages that can be served as an eepsite, that is)
14:47 < jrandom> jetty does allow people to deploy client applications (though i dont know anyone working on a web based app yet either)
14:47 < hypercubus> i'd like to say something about systray4j vs. SWT
14:47 < jrandom> mmhmm?
14:47 < hypercubus> the cost of ditching systray4j for SWT: we'd be dropping systray4j.jar and systray4j.dll, shedding 147 KB from our distribution size -- and replacing that with swt.jar (885 KB) + native libs (332 KB on Win, 639 KB on *nix), a net difference of 1.2-1.5 MB, but with that we gain systray icons on KDE, Gnome, and OS X as well as Win32, and also launch icons for plain X environments a la NextStep/GNUstep
14:48 < hypercubus> and this will give us the ability to add other GUI components later, independent of the JRE the user has (otherwise, accomodating Kaffe users would limit us to using AWT only)
14:48 < hypercubus> just food for thought... maybe down the road
14:48 < jrandom> worth discussing, down the road, as users demand it
14:49 < jrandom> if the value is there, the value is there
14:49 < deer> <oOo> Web interface is intend to be the GUI, isn't it ?
14:49 < hypercubus> cervantes had a cool idea to make further use of SWT
14:49 < hypercubus> an I2P dashboard ;-)
14:49 < jrandom> yes oOo
14:49 < hypercubus> oh, and skins! j/k
14:49 < jrandom> i'd really much rather have that sort of functionality built into the router console, if you mean what i think you mean
14:50 < hypercubus> point is...
14:50 < cervantes> it might also encourage application development if i2p comes with a nice set of SWT libraries
14:50 < hypercubus> it seems development on systray4j is winding down or otherwise mired
14:50 < deer> <oOo> As long as systray and GUI stuff aren't mandatory to have a fully working router...
14:50 < jrandom> right oOo
14:50 < hypercubus> i don't see them fixing the KDE version anytime soon
14:51 < hypercubus> correct, we could just add a hook in the router's systray class
14:51 < hypercubus> and the user could optionally download the systray/SWT stuff
14:51 < jrandom> hypercubus: personally, i'm not entirely 100% sure the user base will even need a systray.  i think we need to deploy it and get feedback to know the value
14:51 < jrandom> cervantes: client application developers can absolutely bundle SWT with their app
14:51 < jrandom> (or say "get SWT")
14:51 < hypercubus> i suspect we'll get requests for expanded systray options
14:52 < jrandom> and if a client app dev gets something we want to bundle with the router, we'll deploy swt with the bundle
14:52 < jrandom> (etc)
14:52 < deer> <oOo> Too late to split the console/status monitor/whatever from the really-routing-stuff ?
14:52 < jrandom> really routing stuff?
14:52 < jrandom> the router console is a fully seperate client application
14:53 < jrandom> (apps/routnerconsole/)
14:53 < deer> <oOo> The stuff needed to have the bytes anonymously flowing
14:53 < jrandom> i do think down the line we will want to have a minimal-router install as well
14:53 < jrandom> (with nothing in clients.config, etc)
14:53 < jrandom> but we dont have the developer hours to maintain multiple sets of things
14:55 < jrandom> ok, anyone else have anything they want to bring up?
14:57 < jrandom> if not
14:57  * jrandom winds up
14:57 < deer> <oOo> 0.3.5, when ? ;)
14:57 < jrandom> it'll be out hopefully this week
14:57 < jrandom> (in the next day or two if all goes well)
14:57 < deer> <oOo> Ok ^^
14:57  * jrandom stops winding up
14:57  * jrandom *baf*s the meeting closed