Files
i2p.www/www.i2p2/pages/meeting132.html
2008-02-04 18:22:36 +00:00

365 lines
29 KiB
HTML

{% extends "_layout.html" %}
{% block title %}I2P Development Meeting 132{% endblock %}
{% block content %}<h3>I2P dev meeting, March 8, 2005</h3>
<div class="irclog">
<p>13:06 &lt;@jrandom&gt; 0) hi</p>
<p>13:06 &lt;@jrandom&gt; 1) 0.5.0.2</p>
<p>13:06 &lt;@jrandom&gt; 2) mail.i2p updates</p>
<p>13:06 &lt;@jrandom&gt; 3) i2p-bt updates</p>
<p>13:06 &lt; legion&gt; so it's related to the irc servers?</p>
<p>13:06 &lt;@jrandom&gt; 4) ???</p>
<p>13:06 &lt;@jrandom&gt; 0) hi</p>
<p>13:06 &lt;@jrandom&gt; weekly status notes up @ http://dev.i2p.net/pipermail/i2p/2005-March/000633.html</p>
<p>13:07 &lt; fedo&gt; hi</p>
<p>13:07 &lt;+postman&gt; hi</p>
<p>13:07 &lt; frosk&gt; goodday</p>
<p>13:07 &lt;@jrandom&gt; legion: no, related to i2p bugs, being worked on</p>
<p>13:07 &lt; bla&gt; hi</p>
<p>13:07 &lt; legion&gt; ok</p>
<p>13:07 &lt;@jrandom&gt; speaking bugs being worked on, lets jump on in to 1) 0.5.0.2 :)</p>
<p>13:07 &lt; cervantes&gt; 'lo</p>
<p>13:07 &lt; cervantes&gt; -- Disconnected</p>
<p>13:08 &lt;@jrandom&gt; heh</p>
<p>13:08 &lt; ant&gt; &lt;mihi&gt; hi all</p>
<p>13:08 &lt;@jrandom&gt; 0.5.0.2 is out, and while your irc connection may lag at times, it'll recover ;)</p>
<p>13:08 &lt;@jrandom&gt; woah heya mihi</p>
<p>13:09 &lt; cervantes&gt; hey mihi</p>
<p>13:09 &lt;@jrandom&gt; the status notes give a general overview of where things are and the most immediate priorities</p>
<p>13:10 &lt;@jrandom&gt; the scary thing I'm trying to track down can be seen on http://localhost:7657/oldstats.jsp#router.invalidMessageTime</p>
<p>13:10 &lt; bla&gt; As for me, I can say that 0.5.0.2 already improved realiability _vastly_ compared to 0.5.0.1: errors where destinations couldn't be contacted almost don't occur anymore </p>
<p>13:10 &lt;@jrandom&gt; those numbers should be very very small, but they're not, unfortunately</p>
<p>13:10 &lt;@jrandom&gt; wikked bla </p>
<p>13:11 &lt;@jrandom&gt; yeah, the 0.5.0.2 is definitely an improvement, and everyone should upgrade ASAP </p>
<p>13:11 &lt; bla&gt; 375,932.22 in the last 10 minutes here....</p>
<p>13:11 &lt;@jrandom&gt; well, the particular value isn't really the problem, its their frequency</p>
<p>13:11 &lt;@jrandom&gt; (events per period)</p>
<p>13:12 &lt;@jrandom&gt; those messages can likely be attributed to 0.5 routers, and some of it to 0.5.0.1 routers, which is why I want people to upgrade ASAP</p>
<p>13:12 &lt;@jrandom&gt; it may be the case that its something else though, but I'd like to rule it out</p>
<p>13:12 &lt; bla&gt; jrandom: I get about 200 per hour here</p>
<p>13:13 &lt;@jrandom&gt; bla: i've currently got 93 this hour, but peak count much higher (thousands)</p>
<p>13:13 &lt;@jrandom&gt; anyway, this particular stat is published in the netdb</p>
<p>13:13 &lt; bla&gt; jrandom: How about excluding 0.5-0 from the net in software when releasing 0.5.0.3?</p>
<p>13:14 &lt;@jrandom&gt; so we can all look around and see what values other people have ;)</p>
<p>13:14 &lt;@duck&gt; 309,854.24 peak 5,473,314.59</p>
<p>13:15 &lt;@duck&gt; pasting the wrong one, huh</p>
<p>13:15 &lt;@jrandom&gt; bla: definitely. I added some code in the 0.5.0.2 rev to do soem forward compatability that 0.5.0.1 and 0.5 don't have</p>
<p>13:16 &lt;@jrandom&gt; duck: hard to have a nonintegral # of events ;)</p>
<p>13:16 &lt; bla&gt; jrandom: Good. At least that allows you to test your invalid-messages-are-due-to-0.5-0 hypothesis in a controlled manner</p>
<p>13:16 &lt;@jrandom&gt; bla: aye, though it'd be great if people updated before then ;)</p>
<p>13:17 &lt;@jrandom&gt; (so for those reading at home: http://www.i2p.net/download is your friend ;)</p>
<p>13:17 &lt; maestro^&gt; jr: those numbers for router.invalidMessageTime deviations in ms?</p>
<p>13:17 &lt;@jrandom&gt; maestro^: yes</p>
<p>13:18 &lt;@jrandom&gt; (aka some really insanely skewed values)</p>
<p>13:18 &lt; legion&gt; Here is a little network report [version|Number of nodes][0.5|6][0.5.0.1|39][0.5.0.2|107]</p>
<p>13:18 &lt;@jrandom&gt; yeah, y'all have been great about updating</p>
<p>13:18 &lt; legion&gt; So there is still a few people running 0.5 and many people running 0.5.0.1</p>
<p>13:18 &lt; maestro^&gt; so any idea where they might be lagging?</p>
<p>13:18 &lt; bla&gt; jrandom: Freenet has a flag in each release that specifies the minimum node version it will communicate with. Is the new forward-compat. code something like that?</p>
<p>13:19 &lt;@jrandom&gt; maestro^: many, many ideas for why 0.5 and 0.5.0.1 users are lagging.</p>
<p>13:19 &lt;@jrandom&gt; bla: similar</p>
<p>13:19 &lt; maestro^&gt; or is it clock drift on nodes?</p>
<p>13:20 &lt;@jrandom&gt; maestro^: clock skew, some serialization bugs, the 100% cpu bug</p>
<p>13:20 &lt;@jrandom&gt; ok, thats generally my focus atm, trying to get the message reliability back up</p>
<p>13:21 &lt;@jrandom&gt; anyone have any questions/comments/concerns on 0.5.0.2?</p>
<p>13:21 &lt; ant&gt; * mihi has a 0.4.2.5 router here on hd not started since dec 22th... but he thinks he'd better delete it...</p>
<p>13:21 &lt;@jrandom&gt; heh</p>
<p>13:21 &lt;@jrandom&gt; yeah, that wont talk to too many routers ;)</p>
<p>13:21 * postman got a backup copy of his last 0.4 installation :)</p>
<p>13:21 &lt; ant&gt; &lt;mihi&gt; question for me 'd be upgrade or delete.</p>
<p>13:22 &lt;@jrandom&gt; delete</p>
<p>13:22 &lt;@jrandom&gt; (backing up any destination keys)</p>
<p>13:22 &lt;@jrandom&gt; there is no upgrade procedure from pre-0.5 anymore</p>
<p>13:22 &lt; legion&gt; Perhaps releasing another update say 0.5.0.2-1 that only allows connections from 0.5.0.2 or newer, would be good?</p>
<p>13:22 &lt;@jrandom&gt; legion: that would segment the network</p>
<p>13:22 &lt;@jrandom&gt; people should juts upgrade.</p>
<p>13:23 &lt;@jrandom&gt; (and we should work around those that dont)</p>
<p>13:24 &lt; legion&gt; yeah until the people running outdated nodes updated ;)</p>
<p>13:24 &lt;@jrandom&gt; segmenting the network hurts us all, not just them</p>
<p>13:25 &lt; legion&gt; Maybe if there was a update notification in the router console or something that let them know they are running outdated versions?</p>
<p>13:25 &lt;@jrandom&gt; yeah, that'd certainly be pretty cool</p>
<p>13:25 &lt;@jrandom&gt; hopefully that can get tied in with the updater as well</p>
<p>13:26 &lt; legion&gt; yeah, I know, segmentation is bad...</p>
<p>13:26 &lt;@jrandom&gt; smeghead is working on some of the key components of that, though not sure if that includes the notification / download</p>
<p>13:26 &lt;@jrandom&gt; (so if anyone wants to help work on that, get in touch!)</p>
<p>13:27 &lt;@jrandom&gt; ok, movin' on to 2) mail.i2p updates</p>
<p>13:27 &lt;@jrandom&gt; postman: ping</p>
<p>13:27 &lt;+postman&gt; yes</p>
<p>13:27 &lt; bla&gt; jrandom: smeghead was doing some signing-related stuff IIRC (so that when you get an update notice, you at least know it's real, and not a phishing/spyware/crap thing)</p>
<p>13:28 * postman takes over the mike</p>
<p>13:28 &lt; legion&gt; hmm, maybe if there was a autoupdate feature built in, where updates would be downloaded through i2p and the nodes would simply download the update, then do a graceful restart.</p>
<p>13:28 &lt;@jrandom&gt; right bla</p>
<p>13:28 &lt; ant&gt; &lt;Gatak&gt; Oh, btw. Would I2P work behind nat even if you cannot open a port?</p>
<p>13:28 &lt;@jrandom&gt; Gatak: not yet. some people will be able to at 0.6, others at 2.0</p>
<p>13:29 &lt;@jrandom&gt; legion: patches welcome</p>
<p>13:29 &lt; ant&gt; &lt;Gatak&gt; 2.0 heck, that is far on the future =)</p>
<p>13:29 &lt;@jrandom&gt; (http://www.i2p.net/roadmap#2.0 ;)</p>
<p>13:29 &lt;+postman&gt; erm, shall i start now?</p>
<p>13:29 &lt; aum&gt; morning all</p>
<p>13:30 &lt;@jrandom&gt; mic is all yours postman (sorry ;)</p>
<p>13:30 &lt;@jrandom&gt; 'lo aum, made it for the meeting</p>
<p>13:30 &lt;@jrandom&gt; (d'oh! /me shuts up again)</p>
<p>13:30 &lt; cervantes&gt; Gatek: http://www.i2p.net/roadmap</p>
<p>13:30 &lt;+postman&gt; first, i wanted to say, that we reached 300 accounts registered at postman.i2p already</p>
<p>13:30 &lt;@jrandom&gt; w00t</p>
<p>13:30 &lt;+postman&gt; the number of mails from/to internet is growing steadily and once more proves that we need to move further</p>
<p>13:31 &lt; cervantes&gt; *squeeeel*</p>
<p>13:31 &lt;+postman&gt; after talking to jr some weeks ago we agreed upon the the release of v2mail together with I2P 1.0</p>
<p>13:31 &lt;+postman&gt; recent status is: the java based smtp proxy designed to run on every node is finished</p>
<p>13:31 &lt;@jrandom&gt; nice!</p>
<p>13:32 &lt;+postman&gt; the java based POP3 proxy is at 80% with just the maildir engine missing</p>
<p>13:32 &lt;+postman&gt; there will be a webmanager that needs some heavy tweaking still (15% done)</p>
<p>13:32 &lt;+postman&gt; the inter node communication is at 40% - we tested some datarecord exchanging with HTTP/XML</p>
<p>13:33 &lt;+postman&gt; seems to work quite well and fast even</p>
<p>13:33 &lt;+postman&gt; even if a relay node fails/was powered off for a few days, it'll be synced within a few minutes after going back onlione again</p>
<p>13:33 &lt;@jrandom&gt; wikked</p>
<p>13:33 &lt;+postman&gt; i think we're quite n track</p>
<p>13:34 &lt;+postman&gt; one thing is noteable</p>
<p>13:34 &lt; bla&gt; postman: Nice work man! One question: Many nodes cannot receive or send data on port 25 (not directly, anyway). Will node-owners be able to specify this (or will this be auto-detected)?</p>
<p>13:34 &lt; cervantes&gt; cool</p>
<p>13:34 &lt;+postman&gt; bla: later</p>
<p>13:34 &lt;+postman&gt; in v2mail there will be a locally run webapp</p>
<p>13:34 &lt;+postman&gt; with this you can manager your local proxies AND apply for an "relayaccount"</p>
<p>13:35 &lt;+postman&gt; this relayaccount will then be used to associate your addess/domain to the relays</p>
<p>13:35 &lt;+postman&gt; the relays will sync the information automatically</p>
<p>13:35 &lt;@jrandom&gt; cool</p>
<p>13:35 &lt;+postman&gt; even features like the addressbook / public keys and stuff will work with the LOCAL interface</p>
<p>13:36 &lt;+postman&gt; so the idea is to have one centralized manager where you can do all your mailstuff</p>
<p>13:36 &lt;+postman&gt; relevant data is transferred to ONE of the relays and then being synced between the relays</p>
<p>13:36 &lt;+postman&gt; and this webbased manager will run on your very node</p>
<p>13:37 &lt;+postman&gt; when your node is online, the relays will deliver mails queued for your destination/domain/address</p>
<p>13:37 &lt;+postman&gt; it will be delivered to your local smtp proxy</p>
<p>13:37 &lt;+postman&gt; you can even trigger the whole thing with ETRN :)</p>
<p>13:37 &lt; aum&gt; hi again</p>
<p>13:37 &lt; aum&gt; i'd like to raise a discussion point in this meeting, if it's ok</p>
<p>13:37 &lt;+postman&gt; so much for the future folks :)</p>
<p>13:37 &lt;+postman&gt; .</p>
<p>13:38 &lt;@jrandom&gt; sound bitchin postman </p>
<p>13:38 * postman hands back the mike</p>
<p>13:38 &lt;@jrandom&gt; aum: great, should be some time at 4) </p>
<p>13:38 &lt;+postman&gt; yeah, iam ecstatic :)</p>
<p>13:38 &lt;@jrandom&gt; postman: so for the normal user, the smtp proxy will have the local maildir, and the pop3 proxy will read/etc, right?</p>
<p>13:39 &lt;+postman&gt; yeah, the smtp proxy got a MDA</p>
<p>13:39 &lt;+postman&gt; and will deliver the mail into local maildirs</p>
<p>13:39 &lt;+postman&gt; even several accounts/users can be created locally</p>
<p>13:39 &lt; cervantes&gt; postman: will the relays keep track of your quotas etc and propogate such info between each other?</p>
<p>13:39 &lt;+postman&gt; and mapped to accounts of your domain</p>
<p>13:39 &lt;+postman&gt; cervantes: yes, they will</p>
<p>13:39 &lt; septu_ssh&gt; sorry, can I ask postman about payment/anti-spam mechanisms in the new model?</p>
<p>13:40 &lt;+postman&gt; septu_ssh: have you read any of the documents on the webpage?</p>
<p>13:40 &lt;+postman&gt; cervantes: it's not perfect real time</p>
<p>13:40 &lt;+postman&gt; cervantes: but i am fine with a few minutes update of quota information exchange</p>
<p>13:40 &lt; septu_ssh&gt; postman: in the queue for reading :/</p>
<p>13:40 &lt; septu_ssh&gt; but if it's doc'd, then it's fine</p>
<p>13:40 &lt; cervantes&gt; postman: yeah I figured</p>
<p>13:41 &lt;+postman&gt; septu_ssh: www.postman.i2p/inout.html</p>
<p>13:41 &lt;+postman&gt; septu_ssh: www.postman.i2p/mailv2.html</p>
<p>13:41 &lt;+postman&gt; cervantes: this is no drama really - the quota is a sane limit</p>
<p>13:41 &lt; cervantes&gt; postman: even someone being able to send nrelays * quota recipients is no bad thing</p>
<p>13:41 * septu_ssh is bungle</p>
<p>13:41 &lt;+postman&gt; cervantes: yep</p>
<p>13:42 &lt;+postman&gt; the goal is just to stop anybody from really abusing the service</p>
<p>13:42 &lt;+postman&gt; in the tests i had 3 relays have been really fast </p>
<p>13:42 &lt;@jrandom&gt; postman: i forget, will this have support for the local smtp relay talking directly to someone else's smtp relay, rather than bouncing through your nodes?</p>
<p>13:42 &lt;+postman&gt; cervantes: within 10 secs they have been synced :)</p>
<p>13:43 &lt;@jrandom&gt; (or perhaps thats just for later)</p>
<p>13:43 &lt;+postman&gt; jrandom: the i2p mail relays will be operated by several ppl and are the preferred dests for routing mail</p>
<p>13:43 &lt; cervantes&gt; postman: you could introduce an exponential delay to the send queue</p>
<p>13:43 &lt; cervantes&gt; if it becomes an issue</p>
<p>13:43 &lt;+postman&gt; jrandom: so sending to other destinations could be handy under certain circumstances</p>
<p>13:44 &lt;@jrandom&gt; aye, though dangerous under others</p>
<p>13:44 &lt; cervantes&gt; so the more mail you send the greater the time the mail gets queued for...should give the relays time to catch up</p>
<p>13:44 &lt;+postman&gt; jrandom: but if a node's owner discloses his IMIO destination he could be spammed w/o control :)</p>
<p>13:44 &lt;@jrandom&gt; exactly</p>
<p>13:44 &lt;@jrandom&gt; otoh, same goes if the i2p mail relays are hostile</p>
<p>13:45 &lt;+postman&gt; jrandom: indeed, it's a WOT like construction</p>
<p>13:45 &lt;@jrandom&gt; &lt;/tinFoil&gt;</p>
<p>13:45 &lt;+postman&gt; jrandom: i cannot stop a relay operator from distributing a quota of 0 for your address</p>
<p>13:45 &lt;@jrandom&gt; 'k great. yeah, no need to worry about it for now</p>
<p>13:45 &lt;+postman&gt; :)</p>
<p>13:46 &lt;+postman&gt; ok</p>
<p>13:46 &lt;+postman&gt; .</p>
<p>13:46 &lt;@jrandom&gt; ok cool, thanks for the update. some really exciting stuff</p>
<p>13:46 &lt;@jrandom&gt; ok, swinging on to 3) i2p-bt updates</p>
<p>13:46 &lt;@jrandom&gt; duck: ping</p>
<p>13:46 &lt;@duck&gt; hi</p>
<p>13:47 &lt;@duck&gt; Yesterday BitTorren 4.0.0 was released</p>
<p>13:47 &lt; ant&gt; &lt;dm&gt; sounds german</p>
<p>13:47 &lt;@duck&gt; which we more or less waited for before starting on 0.2</p>
<p>13:47 &lt;@duck&gt; wrote a tasklist / todo: http://pastebin.ca/raw/7037</p>
<p>13:47 &lt;@duck&gt; (sorry my www is currently down)</p>
<p>13:48 &lt;@jrandom&gt; cool</p>
<p>13:48 &lt; legion&gt; what sort of timetable are we talking about for 0.2?</p>
<p>13:48 &lt;@duck&gt; the goal was 4 weeks</p>
<p>13:49 &lt; legion&gt; cool</p>
<p>13:49 &lt;@duck&gt; as you can see RawServer (the part that communicates with i2p) is the biggest task</p>
<p>13:50 &lt;@duck&gt; .</p>
<p>13:50 &lt;@duck&gt; a quick poll:</p>
<p>13:50 &lt; legion&gt; yeah, I'm well aware of that :)</p>
<p>13:50 &lt;@duck&gt; who is planning to create an i2p-bt fork?</p>
<p>13:50 &lt;@jrandom&gt; cool, is there anything people can do to help?</p>
<p>13:50 &lt;@jrandom&gt; heh</p>
<p>13:51 &lt; ant&gt; &lt;dm&gt; i</p>
<p>13:51 * jrandom grabs a spoon</p>
<p>13:51 &lt; ant&gt; &lt;dm&gt; m wiling to hepl</p>
<p>13:51 &lt; legion&gt; i</p>
<p>13:51 &lt; ant&gt; &lt;dm&gt; m gay</p>
<p>13:51 &lt; legion&gt; I'm working on a fork</p>
<p>13:52 &lt;@duck&gt; good, then I know who not to take serious.</p>
<p>13:52 &lt;@duck&gt; really, I think it is silly; pooling resources might get you much further</p>
<p>13:53 &lt;@jrandom&gt; or perhaps if there are better ways to go, you can convince duck to work that way?</p>
<p>13:53 &lt; named&gt; I'm going to write a fork in qbasic, please take me seriously.</p>
<p>13:53 &lt;@duck&gt; I'll try to have the process more open, so others can see what is planned etc</p>
<p>13:53 &lt; ant&gt; &lt;dm&gt; your openness is not swaying us. FORK! FORK! FORK! FORK!</p>
<p>13:53 &lt;@duck&gt; if you have any other suggestions</p>
<p>13:54 &lt; ant&gt; * dm raises legion onto his shoulders.</p>
<p>13:54 &lt; legion&gt; hmm, that may be true, though with what I'm doing I doubt you want me polluting the main i2p-bt development process ;)</p>
<p>13:54 &lt; ant&gt; &lt;dm&gt; FORK! FORK! FORK! FORK!</p>
<p>13:54 &lt;@jrandom&gt; legion: what are you doing that duck wouldn't want to support?</p>
<p>13:55 &lt;@duck&gt; legion: congrats, if you google for 'i2p bittorrent', then an announcement of "Windows I2P Bittorrent Version 1.0" is #1</p>
<p>13:55 &lt;@jrandom&gt; jesus</p>
<p>13:56 &lt; bla&gt; jrandom: Yes?</p>
<p>13:56 &lt;+postman&gt; jrandom: yeah, they will rip this network's ass open soon :)</p>
<p>13:56 &lt; bla&gt; ;)</p>
<p>13:56 &lt; named&gt; 1.0? Damn, I'm using 0.1.8!</p>
<p>13:56 &lt; Ragnarok&gt; oy</p>
<p>13:57 &lt; legion&gt; omfg, really?! I cannot believe it... that's insane.</p>
<p>13:57 &lt;@duck&gt; anyway, I dont think that there is much new to say on this</p>
<p>13:57 &lt; legion&gt; my 1.0 release is based on 0.1.8 if you're running 0.1.8 you're fine.</p>
<p>13:58 &lt;@jrandom&gt; (and the 1.0 release is a .exe that no one has reviewed, ymmv)</p>
<p>13:58 &lt; legion&gt; I poorly named and numbered it sorry, again about that.</p>
<p>13:58 &lt; ant&gt; &lt;dm&gt; 1.0 &gt;&gt; 0.1.8</p>
<p>13:58 &lt; ant&gt; &lt;dm&gt; any day of the week</p>
<p>13:59 &lt;@duck&gt; slightly related:</p>
<p>13:59 &lt;@jrandom&gt; ok, anything else on 3) i2p-bt, or shall we move on to 4) ???</p>
<p>13:59 &lt;+postman&gt; legion: when there will be sourcecode downloadable?</p>
<p>13:59 &lt; frosk&gt; "I2P-BT 0.1.8 works pretty fine and stable so far. I personally see no reason to update to I2P-BT 1.0" (seen on forum)</p>
<p>13:59 * jrandom sighs</p>
<p>13:59 &lt;@duck&gt; last month bram cohen held a talk about bittorrent on some university</p>
<p>14:00 &lt;@duck&gt; quite interesting: http://netnews.nctu.edu.tw/~gslin/tmp/050216-ee380-100.wmv.torrent</p>
<p>14:00 &lt;@duck&gt; (learned lessons about big p2p programs, plus some bittorrent details explained)</p>
<p>14:00 &lt;@duck&gt; .</p>
<p>14:01 &lt;@jrandom&gt; word</p>
<p>14:01 &lt;@duck&gt; postman: legion has released some sourcecode</p>
<p>14:01 &lt; ant&gt; &lt;dm&gt; is he the inventor of BT?</p>
<p>14:01 &lt;@duck&gt; but according to smeghead it isnt the same as the .exe</p>
<p>14:01 &lt;@jrandom&gt; dm: yes</p>
<p>14:01 &lt; legion&gt; There is a developer source you can download from http://legion.i2p/archives/Itorrent_1_x_Developer_Source.zip.bz2</p>
<p>14:02 &lt;+postman&gt; k, will have a look</p>
<p>14:02 &lt; ant&gt; &lt;dm&gt; is the exe a direct compile of that source?</p>
<p>14:03 &lt; legion&gt; though really the 1.0 source is really just 0.1.8 with a patch from smeghead, compiled and nicely packaged.</p>
<p>14:04 * cervantes walks over to 4)??? and waits for everyone to catch up</p>
<p>14:04 &lt; ant&gt; &lt;dm&gt; the question remains unanswered</p>
<p>14:04 &lt; ant&gt; &lt;dm&gt; Legion, did you or did you not, order a code red???</p>
<p>14:04 &lt;@jrandom&gt; *cough*</p>
<p>14:04 &lt; legion&gt; Perhaps we should get back on topic, my bt client discussion moved to #itorrent</p>
<p>14:05 &lt;@jrandom&gt; ok, 4) ???</p>
<p>14:05 &lt;@jrandom&gt; anything else people want to bring up?</p>
<p>14:05 &lt;@jrandom&gt; aum: you had something?</p>
<p>14:06 &lt; ant&gt; &lt;dm&gt; stasher is back?</p>
<p>14:06 &lt; legion&gt; I'm just seeing some funky behavior with 0.5.0.2 periods of heavy traffic...</p>
<p>14:06 &lt; aum&gt; yes</p>
<p>14:06 &lt; aum&gt; i'd like to raise the question of automated tunnel creation/management</p>
<p>14:07 &lt; ant&gt; &lt;dm&gt; go on</p>
<p>14:07 &lt;+detonate&gt; there's a null pointer exception in the systray thing in windows, i just noticed</p>
<p>14:07 &lt; aum&gt; it's 1337 that the web console now allows for humans to manually create/delete/manage tunnels</p>
<p>14:07 &lt;@jrandom&gt; detonate: could you toss 'er on the bugzilla?</p>
<p>14:07 &lt; aum&gt; but I also strongly believe that there should always be a reliable and convenient way for programs to manage tunels as well</p>
<p>14:08 &lt;@jrandom&gt; aum: no one disagrees. we need it, and we will have it. just not yet.</p>
<p>14:08 &lt; ant&gt; &lt;dm&gt; can't you do that through SAM?</p>
<p>14:08 &lt; aum&gt; i noticed on my recent return to i2p that the pysam library is no longer working</p>
<p>14:08 &lt; septu_ssh&gt; I have a quick question as well after aum</p>
<p>14:08 &lt; aum&gt; which was a disappointment</p>
<p>14:08 &lt;@jrandom&gt; the SAM protocol works, pysam doesnt</p>
<p>14:08 &lt; Ragnarok&gt; did it ever work?</p>
<p>14:09 &lt; aum&gt; correct</p>
<p>14:09 &lt; aum&gt; pysam used to work brilliantly</p>
<p>14:09 &lt; legion&gt; During such periods there are 1000+ tunnels my node is participating in and several seconds of lag and delay.</p>
<p>14:09 &lt;@jrandom&gt; legion: aye, the # of tunnels is because of older builds</p>
<p>14:09 &lt; cervantes&gt; ah mymodesty</p>
<p>14:09 &lt; cervantes&gt; eerm pymodesty</p>
<p>14:09 &lt; aum&gt; i'm presently writing a module 'i2ptunnel.py', which defines classes allowing easy tunnel management</p>
<p>14:10 &lt; legion&gt; so if older builds were not being connected to, the networking would be much smoother?</p>
<p>14:10 &lt;@jrandom&gt; 'k, i don't know if thats the right long term solution, but if it bridges the gap for you now, cool</p>
<p>14:10 &lt;@jrandom&gt; legion: those tunnels aren't the problem</p>
<p>14:11 &lt; aum&gt; well, the class interfaces can remain even though the underlying mechanism changes</p>
<p>14:11 &lt;@jrandom&gt; 'k</p>
<p>14:11 &lt; legion&gt; aren't they?</p>
<p>14:12 &lt; legion&gt; When there a few tunnels, there is very little lag and delay...</p>
<p>14:12 &lt; cervantes&gt; legion: sorry aum is just raising some questions, if you hang fire a minute</p>
<p>14:12 &lt; legion&gt; it just seems odd to me.</p>
<p>14:13 &lt; legion&gt; ok</p>
<p>14:13 &lt;@jrandom&gt; i just worry that we need to take into consideration whats been successful in the past - the web config works and is maintained because everyone uses it. perhaps it'd be best to get whatever app you're working on working with manual tunnel creation *first*, that'd be more efficient?</p>
<p>14:13 &lt;@jrandom&gt; just so that there is always something that is using i2ptunnel.py, to stress it</p>
<p>14:13 &lt; aum&gt; we seem to be deadlocking</p>
<p>14:13 &lt;+detonate&gt; jrandom:sure</p>
<p>14:14 &lt; ant&gt; &lt;dm&gt; let's move on then</p>
<p>14:14 &lt; aum&gt; i don't want to invest time in developing my app till I've got a tunnel mgmt API I can rely on</p>
<p>14:14 &lt; septu_ssh&gt; \o. - point to raise </p>
<p>14:14 &lt; cervantes&gt; realistically I can't imagine the tunnel interface will be revamped in the next couple of months though...</p>
<p>14:14 &lt;@jrandom&gt; but surely you see that we can add one trivially</p>
<p>14:14 &lt; cervantes&gt; so a stopgap solution is viable</p>
<p>14:15 &lt; named_&gt; Couldn't the web config have some kind of api that aum's program manipulates?</p>
<p>14:15 &lt;@jrandom&gt; named_: yes</p>
<p>14:16 &lt;@jrandom&gt; its trivial to add something in to allow safe control via URLs, but only makes sense if there's something that needs it</p>
<p>14:16 &lt;@jrandom&gt; otherwise it'll just rot</p>
<p>14:16 &lt; aum&gt; named_: that would be nice, and could work if there were a hardcoded password in config that client progs need to POST in along with tunnel control fields</p>
<p>14:16 &lt; cervantes&gt; personally I'd like to see the whole tunnel system completely revamped, if you include a tunnel management interface from the start then you won't have to worry about the extra effort needed to maintain a seperate interface</p>
<p>14:17 &lt;@jrandom&gt; aye, the proxies do need a bunch of work, which i've been hiding from as much as possible :)</p>
<p>14:17 &lt; aum&gt; SAM is good for some situations, bad for others</p>
<p>14:17 &lt; cervantes&gt; but that's somewhat down the line...</p>
<p>14:17 &lt; fedo&gt; (</p>
<p>14:18 &lt;@jrandom&gt; aum: but as a stopgap, couldn't you just use one of the three available methods?</p>
<p>14:18 &lt; cervantes&gt; ie if the webinterface itself uses the api then there's no maintenance overhead</p>
<p>14:18 &lt;@jrandom&gt; right. the web interface uses the TunnelControllerGroup</p>
<p>14:19 &lt; aum&gt; SAM usage gets difficult when one wants to use existing libs that are extensively dependent on standard TCP sockets</p>
<p>14:19 &lt; aum&gt; jrandom: the I2PTunnel CLI doesn't work for opening server tunnels, so i'm presently writing code for using TunnelControllerGroup</p>
<p>14:19 &lt;@jrandom&gt; aum: exising libs need to be carefully audited. for instance, the gzip utility itself exposes sensitive data</p>
<p>14:19 &lt; aum&gt; coding as we speak</p>
<p>14:21 &lt;@jrandom&gt; I'm certain that the CLI works for server tunnels, but using the TunnelControllerGroup is preferred, if you need it that way</p>
<p>14:21 &lt;@jrandom&gt; ok, anyone else have anything to bring up?</p>
<p>14:22 &lt; septu_ssh&gt; My question pertains to a distributed version of the hosts.txt, a DHT table is used currently for routerInfo, could this not be extended to a distributed version of DNS? The DNS DHT could contain mappings from www.bla.i2p to the eepsite SHA, and the entries would be signed by an 'I2P registrar'... comments? rebuttals?</p>
<p>14:22 &lt; mancom&gt; a question concerning the roadmap: is 0.6 still scheduled for april?</p>
<p>14:22 &lt;@jrandom&gt; septu_ssh: non-routing data goes in the netDb over my dead body ;)</p>
<p>14:23 &lt; septu_ssh&gt; jrandom: not the same db</p>
<p>14:23 &lt; septu_ssh&gt; a different distributed db</p>
<p>14:23 &lt; aum&gt; jrandom: did you see my bug report? the CLI 'server' command /does not work/</p>
<p>14:23 &lt; maestro^&gt; septu_ssh: there isnt any i2p registrar</p>
<p>14:23 &lt;@jrandom&gt; septu_ssh: there are many dangerous aspects of naming, with a few key tradeoffs. have you seen the naming discussion on ugha.i2p?</p>
<p>14:24 &lt;@jrandom&gt; septu_ssh: ah, a DHT on top of I2P could certainly be used to distribute entries, though those names would not be secure, if they were treated as global entries</p>
<p>14:26 &lt;@jrandom&gt; aum: i used it daily up through a few weeks ago, did you see my reply?</p>
<p>14:26 &lt;@jrandom&gt; maestro^: thats the plan</p>
<p>14:26 &lt;@jrandom&gt; er, mancom:</p>
<p>14:26 &lt; cervantes&gt; aum: I have a reply to that i2plist mail from jr, has it not reached you yet, or does the issue remain?</p>
<p>14:26 &lt; septu_ssh&gt; the only reason why I suggest a 'registrar' is because collisions can take place otherwise</p>
<p>14:26 &lt;@jrandom&gt; septu_ssh: embrace collisions :)</p>
<p>14:26 &lt;@jrandom&gt; globally unique, human readable, distributed, and secure naming doesnt exist</p>
<p>14:27 &lt; septu_ssh&gt; it can also happen in host.txt if it is manually edited, but the problem remains the same</p>
<p>14:27 &lt;@jrandom&gt; drop the first parameter, and you're golden</p>
<p>14:27 &lt; aum&gt; jrandom: i did see your reply - and I /do/ have streaming.jar in my cp</p>
<p>14:27 &lt; septu_ssh&gt; postman manages a central node for mail, so there is some element of trust within the network, surely someone would trust a registrar to manage namespace?</p>
<p>14:27 &lt;@jrandom&gt; ok cool, and it still comes back with that stacktrace aum?</p>
<p>14:28 &lt; aum&gt; yes</p>
<p>14:28 &lt;@jrandom&gt; septu_ssh: postman only acts as a central element for postman's outproxies and inproxies</p>
<p>14:28 * Ragnarok really need to get around to writing that addressbook doc...</p>
<p>14:28 &lt; aum&gt; this is when i manually run the cli, do a genkeys, then do a 'server' using the privkeyfile generated by genkeys</p>
<p>14:28 &lt;@jrandom&gt; septu_ssh: no one will trust anyone to manage a namespace. censorship == exert presure on that registrar.</p>
<p>14:28 &lt; maestro^&gt; everyone is really their own registrar</p>
<p>14:29 &lt; maestro^&gt; you trust your friends and they trust you</p>
<p>14:29 &lt; aum&gt; oh shit, i picked up an old classpath</p>
<p>14:29 * aum tests again</p>
<p>14:30 &lt; ant&gt; &lt;dm&gt; ok, I'll be the registrar.</p>
<p>14:31 &lt; ant&gt; &lt;dm&gt; I'll be as unbiased as I can... cool?</p>
<p>14:31 &lt; septu_ssh&gt; hmmm, ok, back to the proverbial drawing board then...</p>
<p>14:31 &lt;@jrandom&gt; septu_ssh: a good place to review is http://zooko.com/distnames.html :)</p>
<p>14:32 &lt;@jrandom&gt; everyone wants it, but what they want just isn't secure. we have a solution that is, but we give up global uniqueness</p>
<p>14:33 &lt; septu_ssh&gt; hmmm, ok</p>
<p>14:33 &lt;@jrandom&gt; ok, anyone else have anything else to bring up for the meeting?</p>
<p>14:33 &lt; cervantes&gt; septu_ssh: http://forum.i2p.net/viewtopic.php?t=134</p>
<p>14:33 &lt; aum&gt; jrandom - ok, cli 'server' now works, but i never got a 'job number' for the tunnel</p>
<p>14:34 &lt;@jrandom&gt; hmm right, it runs forever</p>
<p>14:34 &lt; aum&gt; oh, i gotta do 'list' to get the job num</p>
<p>14:36 &lt;@jrandom&gt; ok cool, if there's nothing else...</p>
<p>14:36 * jrandom winds up</p>
<p>14:36 * jrandom *baf*s the meeting closed</p>
</div>
{% endblock %}