I2P dev meeting, June 1, 2004 @ 21:00 UTC

Quick recap

  • Present:

deer, duck, hypercubus, Masterboy, mihi, Nightblade, tessier, wilde,

Jurnal IRC complet

[22:59] <duck> Tue Jun  1 21:00:00 UTC 2004
[23:00] <duck> hi folks!
[23:00] <mihi> hi duck
[23:00] <duck> http://dev.i2p.net/pipermail/i2p/2004-June/000250.html
[23:00] <duck> my proposal:
[23:00] * Masterboy has joined #i2p

[23:00] <duck> 1) code progress
[23:00] <duck> 2) featured content
[23:00] <duck> 3) testnet status
[23:00] <duck> 4) bounties
[23:00] <duck> 5) ???
[23:00] <Masterboy> hi:)
[23:00] <duck> .
[23:01] <duck> since jrandom is off we'll have to do it ourself
[23:01] <duck> (I know that he is logging and verifying our independency)
[23:01] <Masterboy> no problem:P
[23:02] <duck> unless there are problems with the agenda I propose that we stick to it
[23:02] <duck> though there aint much that I can do if you dont :)
[23:02] <duck> .
[23:02] <mihi> ;)
[23:02] <duck> 1) code progress
[23:02] <duck> not much code submitted to cvs
[23:02] <duck> I did win the trophy this week: http://duck.i2p/duck_trophy.jpg
[23:03] * hypercubus has no cvs account yet
[23:03] <Masterboy> and who did submit something?
[23:03] <duck> anybody doing any secret coding?
[23:03] * Nightblade has joined #I2P

[23:03] <hypercubus> BrianR was working on some stuff
[23:04] <hypercubus> i've got maybe 20% of the 0.4 installer hacked out
[23:04] <duck> hypercubus: if you have stuff then provide diffs and $dev will commit for you
[23:04] <duck> ofcourse the strict license agreements apply
[23:05] <duck> hypercubus: cool, any issues / things worth mentioning?
[23:06] <hypercubus> not yet, but i'll probably need a couple of BSD people to test the preinstaller shell scripts
[23:06] * duck turns some stones
[23:06] <Nightblade> is it text-only
[23:07] <mihi> duck: which one is you on duck_trophy.jpg?
[23:07] <mihi> ;)
[23:07] <Nightblade> luckypunk has freebsd, also my isp has freebsd but their config is kind of screwed up
[23:07] <Nightblade> my web host isp that is, not comcast
[23:08] <duck> mihi: left one with the glasses. wilde is the right guy handing me the trophy
[23:08] * wilde waves
[23:08] <hypercubus> you have a choice... if you have java installed, you can skip the preinstaller altogether... 
  if you don't have java installed you can run the linux binary or win32 binary preinstaller (console mode), or a 
  generic *nix script preinstaller (console mode)
[23:08] <hypercubus> the main installer gives you the choice of using console mode or spiffy GUI mode
[23:08] <Masterboy> i will install freebsd soon so in the future i will give a try to the installer too
[23:09] <hypercubus> ok good... didn't know if anyone else besides jrandom was using it
[23:09] <Nightblade> freebsd java is invoked as "javavm" rather than "java"
[23:09] <hypercubus> as built from sun sources?
[23:09] <mihi> freebsd supports symlinks ;)
[23:10] <hypercubus> anyhow the binary preinstaller is 100% complete
[23:10] <hypercubus> compiles with gcj to native
[23:11] <hypercubus> it just asks you for the install dir, and it grabs a JRE for you
[23:11] <duck> w00t
[23:11] <Nightblade> cool
[23:11] <hypercubus> jrandom's packaging a custom JRE for i2p

[23:12] <deer> <j> .
[23:12] <Nightblade> if you install java from the freebsd ports collection you use some wrapper script called 
[23:12] <deer> <r> .
[23:12] <hypercubus> anyhow this puppy will be almost completely automated
[23:12] <deer> <r> .
[23:12] <deer> <r> .
[23:12] <deer> <r> .
[23:12] <deer> <duck> r: cut it
[23:12] <deer> <r> .
[23:12] <deer> <m> .
[23:13] <deer> <m> stoopid irc server, does not support pipelining :(
[23:13] <duck> hypercubus: got any ETA for us?
[23:14] <deer> <m> oops, the problem is "Nick change too fast" :(
[23:14] <hypercubus> i still expect to be finished in under a month, before 0.4 is ripe for release
[23:14] <hypercubus> though at present i'm compiling a new OS for my dev system, so it'll be a couple of days 
  before i get back to the installer ;-)
[23:14] <hypercubus> no worries though
[23:15] <duck> ok. so more news next week :)
[23:15] <duck> any other coding done?
[23:15] <hypercubus> hopefully... unless the power company screws me again
[23:16] * duck moves to #2
[23:16] <duck> * 2) featured content
[23:16] <duck> lots of streaming audio (ogg/vorbis) done this week
[23:16] <duck> baffled is running his egoplay stream and I am running a stream too
[23:16] <Masterboy> and it works quite good
[23:17] <duck> on our site you can get info how to use it
[23:17] <hypercubus> got any rough stats for us?
[23:17] <duck> if you use a player not listed there and figure out how to use it, please send them to me and I'll 
[23:17] <Masterboy> duck where is the link to baffleds stream on your site?
[23:17] <Masterboy> :P
[23:17] <duck> hypercubus: 4kB/s goes pretty well
[23:18] <duck> and with ogg it aint tooooo bad
[23:18] <hypercubus> but that still seems to be the avg. speed?
[23:18] <duck> my observation is that is the max
[23:18] <duck> but it is all config tweaking
[23:19] <hypercubus> any idea why that seems to be the max?
[23:19] <hypercubus> and i'm not just talking streaming here
[23:19] <hypercubus> but downloads too
[23:20] <Nightblade> i was downloading some big files yesterday (a couple megabytes) off of duck's hosting 
  service and i was getting about 4kb-5kb as well
[23:20] <duck> I think that it is the rtt
[23:20] <Nightblade> those Chips movies
[23:20] <hypercubus> 4-5 seems an improvement over the ~3 that i've gotten consistently since i started using i2p

[23:20] <Masterboy> 4-5kb is not bad..
[23:20] <duck> with a windowsize of 1 you dont get much faster..
[23:20] <duck> windowsize >1 bounty: http://www.i2p.net/node/view/224
[23:21] <duck> mihi: maybe you can comment?
[23:21] <hypercubus> but it is a remarkably consistent 3 kbps
[23:21] <mihi> on what? windowsize>1 with ministreaming: you are a wizard if you manage that ;)
[23:21] <hypercubus> no hiccups on the bandwidth meter... a fairly smooth line
[23:21] <duck> mihi: on why it is so stable at 4kb/s
[23:21] <mihi> no idea. i don't hear any sound :(
[23:22] <duck> mihi: for all i2ptunnel transfers
[23:22] <Masterboy> mihi you need to config the ogg streaming plugin..
[23:22] <mihi> Masterboy:?
[23:23] <mihi> no, there is no limit inside i2ptunnel regarding speed. it must be in the router...
[23:23] <duck> my thinking: max packet size: 32kB, 5 second rtt: 32kB/5s =~ 6.5kb/s
[23:24] <hypercubus> sounds plausible
[23:25] <duck> ok..
[23:25] <duck> other content:
[23:25] * hirvox has joined #i2p

[23:25] <duck> there is a new eepsite from Naughtious
[23:25] <duck> anonynanny.i2p
[23:25] <duck> key is commited to cvs and he did put it on ugha's wiki
[23:25] * mihi is hearing "sitting in the ..." - duck++
[23:25] <Nightblade> see if you can open two or three streams at a 4kb speed then you will be able to tell if it 
  is in the router or the streaming lib
[23:26] <duck> Naughtious: you there? tell something about your plan :)
[23:26] <Masterboy> i have read that he provides hosting
[23:26] <duck> Nightblade: I did try 3 parallel downloads from baffled and I got 3-4kB each
[23:26] <Nightblade> i c
[23:27] <mihi> Nightblade: how can you tell that then?
[23:27] * mihi likes listening in "stop&go" mode ;)
[23:27] <Nightblade> well if there is some kind of limitation in the router that only lets it handle 4kb at once
[23:27] <Nightblade> or if it is something else
[23:28] <hypercubus> can someone explain this anonynanny site? i don't have a running i2p router atm
[23:28] <mihi> hypercubus: just a wiki or some sort of it
[23:28] <duck> plone CMS setup, open account creation
[23:28] <duck> allows file upload and website stuff
[23:28] <duck> through web interface
[23:28] <Nightblade> another thing to do would be to test the throughput of the "repliable datagram" which afaik 
  is the same as the streams but without acks
[23:28] <duck> likely a lot like drupal
[23:28] <hypercubus> yeah i've run plone before
[23:29] <duck> Nightblade: I have been thinking about using airhook to manage those
[23:29] <duck> but so far only some basic thought
[23:29] <hypercubus> anything goes for the wiki content, or does it center on something in particular?
[23:29] <Nightblade> i think airhook is GPLed
[23:29] <duck> the protocol
[23:29] <duck> not the code
[23:29] <Nightblade> ah :)
[23:30] <duck> hypercubus: he wants quality content, and lets you provide that :)
[23:30] <Masterboy> upload the best pr0n of yourself you have got hyper;P
[23:30] <duck> ok
[23:30] * Masterboy will try to do that too
[23:30] <hypercubus> yeah, anyone running an open wiki is just asking for quality content ;-)
[23:31] <duck> ok
[23:31] * duck moves to #3
[23:31] <duck> * 3) testnet status
[23:31] <Nightblade> Airhook gracefully handles intermittent, unreliable, or delayed networks  <-- hehe not an 
  optimistic description of I2P!
[23:31] <duck> how has it been going?
[23:32] <duck> lets put the datagram over i2p discussion to the end
[23:32] <tessier> I love to run around to open wiki's and link to this: http://www.fissure.org/humour/pics/squirre
[23:32] <tessier> airhook rocks
[23:32] <tessier> I've been looking at it for building a p2p network also.
[23:32] <Nightblade> it seems to be reliable to me (#3)
[23:32] <Nightblade> best i've seen so far
[23:33] <duck> yeah
[23:33] <mihi> works well - at least for stop&go audio streaming
[23:33] <duck> I see quite impressive uptimes on irc
[23:33] <hypercubus> agreed... seeing lots more blue guys in my router console
[23:33] <Nightblade> mihi: are you listening to techno ? :)
[23:33] <duck> but hard to tell since bogobot doesnt seem to handle connections that go over 00:00
[23:33] <tessier> audio streaming works great for me but loading websites often takes a number of tries
[23:33] <Masterboy> i have an opinion that i2p runs very good after 6 hours of use in the 6th hour i used the irc 
  for 7 hours and so my router was running for 13hours
[23:33] <duck> (*hint*)
[23:34] <hypercubus> duck: er... heheh
[23:34] <hypercubus> i could fix that i guess
[23:34] <hypercubus> do you have the logging set for daily?
[23:34] <duck> hypercubus++
[23:34] <hypercubus> log rotation that is
[23:34] <duck> oh yes
[23:34] <duck> duck--
[23:34] <hypercubus> that's why
[23:34] <Nightblade> I was at work all day and turned on my computer and started i2p and was on duck's irc server 
  in just a few minutes
[23:35] <duck> I have been seeing some weird DNFs
[23:35] <duck> even when connecting to my own eepsites
[23:35] <duck> (http://dev.i2p.net/bugzilla/show_bug.cgi?id=74)
[23:35] <duck> I think that is what causes most problems now
[23:35] <hypercubus> bogoparser will only analyze uptimes that occur wholly within a single logfile... so if the 
  logfile encompasses only 24 hours, nobody will show up as connected longer than 24 hours
[23:35] <duck> Masterboy and ughabugha did also have it I think...
[23:36] <Masterboy> yup
[23:36] <duck> (fix it and you will win next weeks trophy for sure!)
[23:37] <deer> <mihi> bogobot is excited? ;)
[23:37] <Masterboy> i tried my web site and sometimes when i hit refresh it takes the other route? and i have to 
  wait for it to load but i never wait;P i hit it again and it shows instantly
[23:37] <deer> <mihi> oops, sry. forgot that this is gated...
[23:38] <duck> Masterboy: do the timeouts take 61 seconds?
[23:39] <duck> mihi: bogobot set to weekly rotations now
[23:39] * mihi has quit IRC ("bye, and have a nice meeting")
[23:40] <Masterboy> sorry i didn't check it on my web site when i can't reatch it instantly i just hit refresh 
  and it loads instantly..
[23:40] <duck> hm
[23:40] <duck> well, it needs to be fixed
[23:41] <duck> .... #4
[23:41] <Masterboy> i think the route is given not the same eatch time
[23:41] <duck> * 4) bounties
[23:41] <duck> Masterboy: local connections should be cutted short
[23:42] <duck> wilde had some bounty thoughts... you there?
[23:42] <Masterboy> maybe it is a peer selection bug
[23:42] <wilde> I'm not sure that was for the agenda really
[23:42] <duck> oh
[23:42] <wilde> ok but the thoughts were something like:
[23:42] <Masterboy> i think then we go public the bounty system will work better
[23:43] <Nightblade> masterboy: yes there are two tunnels for each connection, or that is how i understand it 
  from reading the router.config
[23:43] <wilde> we could use this month to do some small advertising of i2p and increase the bounty pool a bit
[23:43] <Masterboy> i can see that the Mute project is going good - they got 600$ and they didn't code a lot yet;P
[23:44] <wilde> target against freedom communities, crypto people, etc
[23:44] <Nightblade> I don't think jrandom wants advertising
[23:44] <wilde> not public slashdot attention, no
[23:44] <hypercubus> that's what i've observed as well
[23:44] <Masterboy> i want to push it again - when we go public the system will work a lot better;P
[23:45] <wilde> Masterboy: bounties could speed up myi2p development for example
[23:45] <Masterboy> and as jr said no public till 1.0 and only some attention after 0.4
[23:45] <Masterboy> *wrote
[23:45] <wilde> when we have like $500+ for a bounty people could actually survive for some weeks
[23:46] <hypercubus> the tricky part is, even if we target a small dev community, like *cough* Mute devs, those 
  guys might spread the word about i2p further than we'd like
[23:46] <Nightblade> someone could make a career out of fixing i2p bugs
[23:46] <hypercubus> and too soon
[23:46] <wilde> i2p links are already in many public places
[23:46] <Masterboy> you google and you can find i2p

[23:47] <hypercubus> obscure public places ;-) (i saw the i2p link on a freesite... i'm lucky the damn freesite 
  even loaded!)
[23:47] <wilde> http://en.wikipedia.org/wiki/I2p
[23:47] <Masterboy> but i agree that no advertising till 0.4 is done
[23:47] <Masterboy> wha???????
[23:47] <wilde> http://www.ovmj.org/GNUnet/links.php3?xlang=English
[23:48] <Masterboy> protol0l does a great job;P
[23:48] <Masterboy> ;))))))
[23:48] <hypercubus> nice typo ;-)
[23:48] <wilde> ok anyway, I agree we should still keep I2P private (jr read this log ;)
[23:49] <Masterboy> who did that?
[23:49] <Masterboy> i think the Freenet crew discussion gave more attention..
[23:50] <Masterboy> and jr discussing with toad give a lot info to the big public..
[23:50] <Masterboy> so as in ughas wiki - we can all blame jr for that;P
[23:50] <wilde> ok anyway, we'll see if we can bring in some $ without bringing in /.
[23:50] <Masterboy> agreed
[23:50] <hypercubus> the freenet dev list is hardly what i call the "big public" ;-)
[23:50] <wilde> .
[23:51] <hypercubus> wilde: you'll have a lot of $ sooner than you think ;-)
[23:51] <wilde> oh come on, even my mum subscribe to freenet-devl
[23:51] <duck> my mum reads through gmame
[23:51] <deer> <clayboy> freenet-devl is being taught in schools here
[23:52] <wilde> .
[23:52] <Masterboy> so we will see more bounties after we go 0.4 stable..
[23:53] <Masterboy> that is after 2 months;P
[23:53] <wilde> where did that duck go?
[23:53] <duck> thanks wilde 
[23:53] <hypercubus> though as the only bounty claimant thus far, i have to say that the bounty money had no 
  bearing on my decision to take up the challenge
[23:54] <wilde> hehe, it would if it been 100x
[23:54] <duck> wyou are too good for the world
[23:54] <Nightblade> haha
[23:54] * duck moves to #5
[23:54] <hypercubus> wilde, $100 doesn't mean shit to me ;-)
[23:54] <duck> 100 * 10 = 1000
[23:55] * duck pops("5 airhook")
[23:55] <duck> tessier: got any real-world experience with it
[23:55] <duck> (http://www.airhook.org/)
[23:55] * Masterboy will try dis out:P
[23:56] <duck> java implementation (dunno if it even works) http://cvs.ofb.net/airhook-j/
[23:56] <duck> python implementation (a mess, did work in the past) http://cvs.sourceforge.net/viewcvs.py/khashmir
[23:58] * duck opens the rant-valve
[23:58] <Nightblade> j one is also gpl
[23:58] <duck> port it to pubdomain
[23:58] <hypercubus> amen
[23:58] <Nightblade> the entire protocol doc is only about 3 pages - it can't be that hard
[23:59] <Masterboy> nothing is hard
[23:59] <Masterboy> it's just not easy
[23:59] <duck> I dont think that it is fully specced though
[23:59] * hypercubus takes away masterboy's fortune cookies
[23:59] <duck> you might need to dive into the C code for a reference implementation
[00:00] <Nightblade> I would do it myself but I am busy with other i2p stuff right now
[00:00] <Nightblade> (and also my full-time job)
[00:00] <hypercubus> duck: maybe a bounty for it?
[00:00] <Nightblade> there already is
[00:00] <Masterboy> ?
[00:00] <Masterboy> ahh Pseudonyms
[00:00] <duck> it could be used at 2 levels
[00:00] <duck> 1) as a transport besides TCP
[00:01] <duck> 2) as a protocol to handle datagrams inside i2cp/sam
[00:01] <hypercubus> that's worth serious consideration then
[00:01] <hypercubus> </obvious>

[00:02] <Nightblade> duck: i noticed that the repliable datagram in SAM has a maximum size of 31kb, whereas the 
  stream has a maximum size of 32kb - making me think that the sender's destination is sent with each packet in 
  repliable datagram mode, and only at the beginning for a stream mode -
[00:02] <Masterboy> well airhook cvs is not very updated..
[00:03] <Nightblade> making me think that it would be inefficient to make a protocol on top of the repliable 
  datagrams through sam
[00:03] <duck> airhooks message size is 256 bytes, i2cp's is 32kb, so you need to atleast change a bit
[00:04] <Nightblade> actually if you wanted to do the protocol in SAM you could just use the anoymous datagram 
  and have the first packet contain the sender's destination.... blah blah blah - i have lots of ideas but not 
  enough time to code them
[00:06] <duck> then again you have to problems to verify signatures
[00:06] <duck> so someone could send fake packages to you
[00:06] <Masterboy> topic:::: SAM
[00:06] <Masterboy> ;P
[00:07] <Nightblade> true
[00:08] <Nightblade> but if you sent back to that destination and there was no acknowledgement you'd know it was 
  a faker
[00:08] <Nightblade> there woudl have to be a handshake
[00:08] <duck> but you'll need aapplication level handshakes for that
[00:08] <Nightblade> no not really
[00:09] <Nightblade> just put it in a library for accessing SAM
[00:09] <Nightblade> that is a bad way of doing though
[00:09] <Nightblade> doing it though
[00:09] <duck> you could also use seperated tunnels
[00:09] <Nightblade> it shuold be in the streaming lib
[00:11] <duck> yup. makes sense
[00:12] <duck> ok
[00:12] <duck> I am feeling *baff*-y
[00:13] <Nightblade> ja
[00:13] * duck *baffs*