I2P dev meeting

Quick recap

  • Present:

BrianR, cervantes, deer, dm, duckie, fvw, jrandom, mihi, Nightblade, thecrypto, wilde,

Jurnal IRC complet

14:00 < jrandom> 0) hi
14:00 < jrandom> 1) status update
14:00 < jrandom> 2) cvs HEAD
14:00 < jrandom> 3) 0.4
14:00 < jrandom> 4) 1.0
14:00 < jrandom> 5) other activities
14:00 < jrandom> 6) ???
14:00 < jrandom> 0) hi
14:00 < dm> 7) Profit!!
14:00 < mihi> hi
14:00 < cervantes> hello
14:00 < dm> hi
14:00 < fvw> hi dr nick
14:01 < jrandom> for all those playing at home, we've got the status notes up @ http://dev.i2p.net/pipermail/i2p/2004-June/000286.html
14:01 < jrandom> as we've done in the past, that contains the actual content - we get all meta here and discuss stuff
14:01  * dm is playing at home. WINK WINK WINK.
14:01 < jrandom> BrianR: this is where the LART comes in handy
14:02 < jrandom> anyway, jumping into 1) status update
14:02 < thecrypto> let's see if i can connect to an eepsite
14:03 < jrandom> the other day we had a larger than usual # of routers jumping off the network, which i'm hoping is because a lot of the services on the network were b0rked due to my stupid bugs
14:03  * mihi fires up his router
14:03 < thecrypto> yah!
14:03 < jrandom> if people are taking down their router for another reason (bugs, cpu/memory usage, etc) i'd really love to hear about it
14:04 < dm> ok
14:05 < BrianR> I must say, i2p is currently more well behaved than freenet on my machine... Both in bandwidth hogging and cpu/ram hogging.
14:05 < jrandom> there's a really bad bug in cvs at the moment, but fixed locally - i'll have that committed later tonight (not sure if i'll do so before or after getting my new machine together)
14:05 < jrandom> ah wikked
14:06 < jrandom> well, to be fair, i2p does a lot less than freenet - once we have thousands of routers we'll see how it goes
14:06 < dm> Why don't operating system's have per-application bandwidth limiting built-in? Is it a hard task?
14:06 < BrianR> I'd like to do some stress testing with i2p's built in bandwidth limiting stuff though. How far along is that?
14:06 < dm> All these projects wasting time implementing their own bandwidth limiting, seems silly.
14:06 < fvw> dm: Linux does it, as does freebsd no doubt.
14:07 < dm> Do they?
14:07 < cervantes> netlimiter on windoze does it
14:07 < BrianR> dm: It's not a hard task. There's lots of system level bandwidth limiting stuff available, for 'doze, linux, *bsd...
14:07 < dm> Makes sense to put it in the OS if you ask me.
14:07 < fvw> dm iptables marking + tc under linux; Using it right now, works like a charm.
14:07 < jrandom> BrianR: its in pretty good shape - it'll be in the next release (which i'm hoping to have out this week)
14:07 < dm> But anyway...
14:07 < BrianR> thing is, it often makes more sense to do limiting at the application level so you can make intelligent decisions about where to spend bandwidth first
14:08 < jrandom> exactly
14:08 < dm> That's a different problem.
14:08 < jrandom> with bandwidth limiting within i2p, we know when we're saturating our limit and can refuse to participate in tunnels (etc)
14:08 < BrianR> At the application level you have no idea which one of your TCP connections is going to lose a packet when you're above the system imposed TX limit.
14:08 < fvw> actually, best would be to have the OS give proper feedback; But that's not relevant for i2p right now.
14:09 < jrandom> true fvw, once we integrate i2p with tun/tap ;)
14:09 < BrianR> fvw: You do get some feedback in terms of backpressure.. But it's hard to tell if backpressure is due to the remote side's link being slow or the local side..
14:09 < dm> Put a bounty on patching windows to have per-application bandwidth limiting. Thanks.
14:09 < jrandom> heh
14:09 < fvw> "my stuff isn't arriving" and "this connection is being throttled now" are two entirely different things.
14:09 < fvw> they sometimes overlap ofcourse.
14:10 < dm> (extra $20 for having a small graph with bandwidth usage of each window in the window title bar... Next to the minimize buttons, etc)
14:11 < jrandom> ok does anyone else have any network status related things to bring up?
14:12 < BrianR> Oh.. That IRC connection monitoring thing..
14:12 < duckie> I appologize for updating to cvs
14:12 < jrandom> hehe
14:12 < BrianR> Is there a MRTG graph of that or logs or something somewhere?
14:12 < jrandom> bogobot right?  duck.i2p/bogo/ iirc?
14:13 < duckie> something like that
14:13 < jrandom> (its linked to off duck.i2p, my routers are down atm so i cant check)
14:13 < duckie> linked to on http://duck.i2p/ uder the stats section
14:13 < duckie> http://duck.i2p/bogobotlogs/
14:14 < BrianR> it might be useful to extend that thing to act as a general reachability survey tool..
14:15 < dm> You guys type reaaaaaaaaaaaaaaaally slowly.
14:15 < deer> <cervantes> hypercubus has some enhancements on the backburner, but he's pretty busy with installer stuff atm
14:16 < jrandom> BrianR: bogobot is just for passive irc tracking, but someone could modify i2ptunnel's ping command to support periodic checking?
14:16 < jrandom> or someone could whip up a perl/SAM script ;)
14:16 < BrianR> maybe some sort of script that produces a MRTG graph of endpoint reachability....
14:16 < BrianR> jrandom: speaking of perl/sam, bring that up again when we get to the appropriate number in the agenda
14:16 < duckie> doesnt the heartbeat thing do that?
14:17 < BrianR> If it does, I couldn't make it work :(
14:17 < jrandom> kind of - heartbeat tests heartbeat servers, not generic apps
14:18 < BrianR> Yes... Like I want to see if I can successfully do a "GET /" on duck.i2p, and how long it took.
14:18 < jrandom> but yes, once the net is in good shape, lots of people can run a heartbeat server and people can ping each other to measure network performance
14:18 < Nightblade> pingflood too
14:18 < dm> Is there a limit on GET strings?
14:18 < dm> like if you have a really long URL, because of cgi parameters?
14:19 < cervantes> 2047 characters iirc
14:19 < jrandom> that wouldnt be too hard to put together
14:19 < BrianR> dm: Depends on the web server... I think the RFC also specifies a minimum which must be suported.
14:19 < jrandom> (er, an http tester)
14:19 < dm> cervantes: thanks
14:19 < dm> (just realized it might affect something I've been working on)
14:20 < cervantes> you run into 2 problems, webserver support and client support
14:20 < BrianR> have we stumbled off topic?
14:20 < cervantes> both IE and Mozilla are 2047
14:20 < cervantes> eer yep :)
14:21 < dm> I got my answer, we can move on :)
14:21 < jrandom> ok, i think that covers 1), now briefly to 2) cvs HEAD
14:21 < jrandom> there's a nasty bug in there currently, and i apologize for not fixing it sooner
14:22 < jrandom> in general, if you try out cvs HEAD and it doesnt work well, just roll back (download jars / pull from the tag)
14:22 < jrandom> anyway, 'nuff 'bout that
14:23 < jrandom> jumping forward to 3) 0.4
14:23 < jrandom> the email has my viewpoint / rational for the release - anyone have any concerns / questions / ideas / suggestions?
14:24 < Nightblade> looks good to me
14:24 < duckie> how does it change the time estimates?
14:24  * fvw nods. Sane and everything.
14:25 < jrandom> it actually moves 0.4 closer, as we're pushing the scary parts of 0.3.3 into 0.4.1 and 0.4.2
14:25 < jrandom> i dont think it'll change the 1.0 dates, just moving things around before it
14:26 < jrandom> the sim has shown our main problem with tunnel diversification can be easily remedied by active peer testing
14:26 < jrandom> we had some of this in the pre 0.2.3 revs, but stopped that since it seemed to add too much overhead
14:27 < jrandom> (when in fact lots of shit was just plain broken, causing the overhead)
14:27 < deer> <cervantes> what does active peer testing involve?
14:29 < jrandom> it just means we'll build some secondary tunnels through peers we wouldnt normally use - specifically, i'm going to have it pull random peers from the 'reliable' set who arent in the 'fast+reliable' set
14:29 < jrandom> (and if that intersection is too small, extend it to the 'not failing' set)
14:29 < jrandom> i dont know how many tunnels or peers to test yet, we'll probably keep that as a tunable parameter
14:30 < jrandom> the issue manifests itself in the peer profiles - you'll see lots of values for your favorite peers, but lots of 0s or other low numbers for other peers
14:30 < jrandom> (which may be due to them being shitty, or more likely due to them not being tested)
14:31 < jrandom> ok, anything else on 0.4?  or should we move on to 4) 1.0?
14:32 < duckie> yes
14:33 < jrandom> 4) 1.0
14:33 < jrandom> i dont see 1.0 as a big scary asymptotically approached release.  its just a release where things work and people can use it
14:34 < jrandom> that said, anyone have any concerns / ideas / suggestions/ questions wrt whats in the mail?
14:35 < dm> The rate at which the versioning is increasing would indeed indicate that it is an asymptotically approached 1.0
14:36 < dm> We're gonna need some big jumps!
14:36 < jrandom> 0.4--> 1.0
14:36 < dm> Sweeeeeeeet....
14:36 < jrandom> and we're only a few weeks from 0.4
14:37 < jrandom> but, of course, we could use some help to get things moving faster :)
14:37 < dm> Are we happy with stability then?
14:37 < jrandom> no, stability right now sucks
14:37 < cervantes> myi2p could be a fairly hefty bit of development to get implemented properly, since it ideally relies on other projects coming to fruition
14:37 < dm> Okay, haven't been following closely.
14:38 < cervantes> such as DHT and datagrams etc
14:38 < jrandom> agreed cervantes, but if we just have it with the address book and distributed blogging, that'll be good enough for me.
14:38 < cervantes> right
14:38 < jrandom> nightblade is working on a dht that we can plug in whenever its ready, or we could add hooks for referencing freenet:CHK@ or SSK@
14:39 < cervantes> private messaging is just a small hop from the syndicated address book too
14:39 < jrandom> right, its all doable
14:40 < cervantes> live streaming cam girl desktop might take longer...
14:40 -!- Irssi: #i2p: Total of 21 nicks [0 ops, 0 halfops, 0 voices, 21 normal]
14:40 < jrandom> heh
14:40 < jrandom> actually, as I mentioned to Sonium before the meeting, one of the things people can help out with is the documentation front -
14:41 < jrandom> defining what docs we will need, starting to do some outlines of how those docs will fit together, and perhaps even tossing open an editor and writing 'em
14:42 < cervantes> ah I meant to enquire earlier about enhancing the console GUI too
14:42 < jrandom> docs are an essential part of the releases, both for tech docs (since without good tech docs, no devs will get involved) and for user docs (since without good user docs, users will leave)
14:42 < jrandom> that we'll want for the 0.4 release, and will be jsp/servlet driven
14:43 < jrandom> i havent done any mockups or workflows for that even, so if someone wants to jump on that, we could use the help
14:43 < Nightblade> console GUI?  what is that - SVGALIB? :)
14:43 < jrandom> heh
14:43 < mihi> Nightblade: TWIN / curses
14:43 < cervantes> is the data set the same or are there new and improved stats for 0.4
14:44 < dm> I want a console gui in WIN32/MFC
14:44 < Nightblade> the http console on 7655 right
14:44 < cervantes> ie the duration that elephants stay on the moon, and whether they return freuqently
14:44 < cervantes> yup
14:44 < jrandom> all of the things on the console now reflect data that can be made available to a 0.4 console, but we should totally throw out its existing design
14:45 < jrandom> (e.g. no one cares what the base64 of a routerIdentity's public key is)
14:45 < cervantes> although I suggest that info should still be available
14:46 < jrandom> yeah, i think we may keep the existing console as an option - no need to throw it away
14:46 < jrandom> (but it'd be something like http://localhost:7655/uglyConsole/)
14:46 < cervantes> ./forlynxusers/
14:47 < jrandom> the new admin console should work for lynx users too
14:47 < dm> Does anyone really use lynx?
14:47 < cervantes> eeew cross browser compatibility
14:47 < jrandom> i do
14:47 < jrandom> (at times)
14:47 < dm> but.... why?
14:47 < fvw> for things like brief config edits, definately.
14:47 < cervantes> dm: if you're stuck at a command prompt you don't have much choice
14:47 < fvw> it's quick and you don't need X
14:48 < Nightblade> i use it for web browsing on headless servers
14:48 < deer> <Pseudonym> I use links browser
14:48 < jrandom> well, yeah, links is preferable to lynx
14:48 < mihi> anyone using w3m?
14:48 < dm> It's quite cool.
14:48 < jrandom> good point - anything else wrt i2p 1.0?
14:49 < dm> Won the browser war of 1992-1998
14:49 < jrandom> or moving on to 5) other activites?
14:49 < dm> 3 medals of honor.
14:49 < Nightblade> yes
14:49 < Nightblade> I am hoping to release a new version of libsam tomorrow but my home computer is broke so I don't know
14:49 < Nightblade> I am going to also upload Nickster's threading library
14:49 < jrandom> awesome!/doh!
14:50 < jrandom> nice!
14:50 < Nightblade> which he said was public domain - i am just going to upload it exactly as he sent it - i haven't had time to do much with it
14:50 < jrandom> once is out, i'll start running the cvs server on i2p as well
14:50 < jrandom> cool
14:51 < jrandom> (so nickster will be able to make edits/updates to his code in cvs without exposing his ip)
14:51 < jrandom> whats new in the next rev of libsam?
14:51 < Nightblade> some fixes to datagram send/receive which were broken
14:51 < Nightblade> and some other stuff i don't remember
14:51 < Nightblade> it iwll be in the changelog
14:51 < jrandom> cool, sounds good
14:52 < jrandom> btw, if you want a mailing list for libsam (-announce,-dev,-users,etc) lemmie know
14:52 < jrandom> (but posting on i2p@i2p is fine too)
14:52 < Nightblade> overkill
14:52 < jrandom> hehe
14:53 < jrandom> yeah prolly
14:53 < cervantes> he can have a section in i2pforum
14:53  * cervantes moves smoothly onto his agenda
14:53 < Nightblade> maybe once I get that DHT working - but that is a long way off, and today and yesterday I hvane't had time to do much with it
14:53 < Nightblade> cervantes: there is already an application development forum or somethihng like that iirc
14:53 < Nightblade> also Connelly needs an account on i2p.net if someone has gotten him one yet
14:53 < cervantes> specifically in relation to i2p?
14:54 < Nightblade> http://www.i2p.net/forum/3
14:54 < jrandom> oh, i havent added an account for connelly, i'll chat with 'im next time i see him
14:54 < fvw> dang, got to run. See you all next week, keep up the good work.
14:54 < jrandom> ugh, yeah, i dont know about the www.i2p.net/forum/s
14:54 < cervantes> ah...drupal...
14:55 < jrandom> cool fvw, glad you could make it
14:56 < cervantes> the i2pforum is pretty much set up
14:56 < cervantes> it basically just needs content/users to fill it :)
14:56 < jrandom> i'm not sure about the drupal forums, or whether they really fit with what i see the website being.  i kind of like the idea of a seperate site thats dedicated to being a user/dev forum
14:57 < jrandom> cool cervantes - is that as an eepsite, website, or both?
14:57 < cervantes> I've picked arbitrary forum topics/sections so if people have any ideas or want to write stuff for FAQs etc them that would be appreciated
14:57 < cervantes> it's both
14:58 < cervantes> currently I have the i2p tunnel offline while my router is so unstable
14:58 < jrandom> cant say i blame you :)
14:58 < Nightblade> what is the url
14:58 < cervantes> but people can start playing on outweb
14:59 < cervantes> currenlty you'll have to add an entry to /etc/hosts c:/windows/system32/drivers/etc/hosts
14:59 < cervantes> i2pforum.cervantes.not2p
14:59 < jrandom> heh
15:00 < cervantes> I'll configure it with a proper hostname soon
15:00 < cervantes> or if someone wants to add a subdomain pointer to the php.net domain record...that would be better
15:00 < jrandom> lemmie know if you want something under the i2p.net domain or if you're going with another domain thats cool
15:00 < cervantes> blah
15:00 < cervantes> i2p.net
15:00 < cervantes> <-- php on the brain
15:01 < jrandom> do you want that to be e.g. forum.i2p.net?
15:01 < cervantes> that would be cool ;-)
15:01 < jrandom> w3rd, i'll update the records tonight
15:01 < cervantes> I've tweaked standard phpbb2 code so it doesn't require a valid email to become a member
15:01 < dm> "The site you requested is no longer hosted here"
15:02 < cervantes> if you forget your pass then you have to contact an admin
15:02 < cervantes> dm: did you enter that into your hosts file or are you just browsing to the IP
15:03 < dm>                                   Invalid URI in request GET i2pforum.cervantes.
15:03 < dm> not2p/<P>

15:03 < dm> I don't use browsers, I'm using telnet.
15:04 < jrandom> i did it through telnet just now.  works fine.
15:04 < dm> hmm forgot the HTTP 1.1 I think.
15:04 < jrandom> GET / HTTP/1.1\nHost: i2pforum.cervantes.not2p\n\n
15:04 < jrandom> anyway
15:04 < jrandom> BrianR: ping
15:05 < dm> That worked. Your telnetting skills are sharp, young one.
15:07 -!- Irssi: #i2p: Total of 19 nicks [0 ops, 0 halfops, 0 voices, 19 normal]
15:07 < jrandom> ok, anyone else working on i2p related projects they want to discuss?
15:07 < jrandom> if not, 6) ???
15:07 < dm> cum!
15:07  * wern put a hand in the air
15:07 < jrandom> hi^2, que tal?
15:08  * jrandom calls on wilde
15:08 -!- wern is now known as wilde
15:08 < wilde> I just want to thank our new sponsors, cervantes and hypercubus
15:08 < wilde> :)
15:09 < wilde> we got two generous donations to the general fund
15:09 < cervantes> wilde: did they convert ok?
15:09 < dm> how much?
15:09 < wilde> cervantes: will exchange them tomorrow
15:09 < cervantes> ah
15:09 < wilde> (hopefully)
15:10 < cervantes> so there still might be a runner-up prize ;-)
15:10 < jrandom> awesome, thanks cervantes &amp; hypercubus :)
15:10 < cervantes> heh...well save praise until the money is in the bank I think ;-)
15:11 < dm> twas a good meeting. I'm off to bed.
15:11 < dm> Nite!
15:11 < jrandom> 'night dm
15:11 < wilde> yeah, I'll do the full praising next time
15:11 < jrandom> ok, anything else to add wilde, or someone else?
15:11 < wilde> yeah
15:12 < wilde> I'll take the i2p.net site down tomorrow for a while (intentionally this time ;)
15:12 < wilde> to switch servers
15:12 < jrandom> heh ok cool
15:12 < wilde> hold on to your posts, they may disappear
15:12 < jrandom> will we have the same IP, or should I update the records?
15:12 < dm> can you make it seamless?
15:12 < wilde> if you post tomorrow that is
15:13 < wilde> dm: no way ;)
15:13 < wilde> not me
15:13 < dm> okay.
15:13 < wilde> jrandom: same IP
15:13 < jrandom> ok cool
15:13 < jrandom> wilde++
15:13 < wilde> over and out
15:14 < jrandom> ok, anyone else have something they'd like to discuss?
15:14  * cervantes puts on his metal "ming the merciless" skullcap
15:14 < jrandom> in that case
15:15  * jrandom winds up
15:15  * jrandom *baf*s cervantes on the head, closing the meeting