13:28:59 #startmeeting 13:28:59 Meeting started Wed Jun 17 13:28:59 2015 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:28:59 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:29:11 I've got to refill my coffee; let's start by checking in! Someoby go first? 13:29:14 *somebody 13:30:27 ok back now 13:30:30 I guess I can start 13:30:34 (this is the regular time, right?) 13:30:59 oh hi! 13:31:05 I've been running around doing organizational things over the last week and having conversations. With help from athena we got an 0.2.6 release out. Lots of patches to review and merge. I hope I can rescue some time for hacking 13:31:10 (over the next week or so) 13:31:21 but wow, so much random not-really-programming stuff all the time :/ 13:31:35 hi 13:31:41 nickm: is this going to slow down a bit soon? 13:31:58 I also see people doing all kinds of work and discussion on patches, and am glad that others are making progress on things 13:32:13 dgoulet: I'm hoping that after contracts are out and the ED search process is launched I will have a few more brain cycles 13:32:25 great :) 13:32:29 I hope 13:32:38 anyways, who's next? 13:33:02 * dgoulet can go 13:33:43 go for it 13:34:04 Got a review from asn on #4862 which was very useful, that branch fixes lots of stuff but while testing we discovered #16381 which spawned #16389 13:34:17 I did a patch for #16260 that is awaiting review 13:35:10 I'm currently thinking about a nice design for 16389, that's about it for me on little-t tor 13:35:41 * teor can go 13:36:26 go 13:36:33 I did some very early drafts of fuzzing and released a draft of chutney bandwidth testing 13:36:57 Fuzzing in #16301 has bugs I have to work through, and it only does torrc 13:37:31 teor: it's potentially valuable, but I think expanding beyond torrc might be sensible. Remember, torrc is trusted. 13:37:48 Yes, torrc is a toy example 13:38:10 Which is easy to read if something goes wrong! 13:38:30 dgoulet: just put a comment for you on 16260 13:38:37 The idea is to have a framework for fuzzing, and then write a small amount of code for each feature we want to fuzz 13:38:37 nickm: omg! 13:38:52 nickm: I was about to click submit for a comment with a new branch that adds the assert you proposed :P 13:38:57 nickm: what a timing 13:39:21 But I found #16366 and #16360 doing torrc 13:39:36 yeah, #16360 is important 13:39:38 Once I have the bugs out I want to do cell parsing or something 13:39:56 nickm: there 13:40:13 16360 is a nice catch! 13:40:23 teor: it would be nice to fuzz the generated trunnel code 13:40:32 Yes, I was thinking that 13:40:57 I trust it more than non-trunnel generated code, but I think it's getting reviewed less 13:41:14 Just need an interface that is basically parse(data, len) and there will be very little code to write 13:41:34 asn and I are working on getting bandwidth weights working with chutney in #16386 13:41:39 great 13:42:19 I released a large lump of chutney code in #14175 that I need to split into commits 13:42:29 But I wanted to check it on Linux first 13:42:57 This was trying to duplicate #8902, which I can on both OS X and Linux 13:43:59 So I see a pattern where the HS is fine for the first few connections, then starts lagging badly around the 40-60 mark. 13:44:10 It gets there, but it hangs for 10 seconds or more at times. 13:44:48 interesting. I wonder what it's doing. 13:44:49 And I stalled on fallback directories #15775 and other tasks. Might need some help prioritising. 13:44:59 teor: maybe comment:10 there could be it, when I did this experiment, my guard was the one failing 13:46:40 anybody else around today? 13:46:46 shall we move on to discussion? 13:47:32 I think so 13:47:33 dgoulet: I haven't had a chance to check in detail. I think it might be IPs in my case 13:48:43 I am running 60 tor clients, with 1 connection each, and my guards seem fine. But the HS is dying, and maybe IPs/RPs. Can't quite remember 13:49:19 nickm: anything you have in mind for discussion? 13:49:34 teor: interesting, profiling would help I think as a good starting point 13:49:59 Yep, wanted to get the code out there so I could try on OS X and Linux, and others could try too. 13:50:43 dgoulet: not especially; I think I need to do a lot of ramping up of my Tor game this next week and try to be a programmer for a while. 13:50:44 teor: FYI, 15775 is still on my "to review" list, 13:50:58 dgoulet: I'm worried about the pace of our progress on 0.2.7, but what's new :) 13:51:15 isabela: do you need anything from me on the contracts process btw? Just ask if so. 13:51:25 Shall we endmeeting, or does somebody else have a topic 13:51:26 dgoulet: yes, I have been making notes about extra features required to make it happen, but they're minor 13:51:39 nickm: that's the thing right, seems little-t tor has more woman/man power now which pills up patches/review everywhere :) 13:51:57 nickm: clearly a sign of health I think even though we are behind :) 13:52:03 nickm: do we need to start shifting effort into reviews, or tasks into 0.2.8? 13:52:10 instead of a man/woman, I'm an attack helicopter 13:52:15 nickm: is all looking good :) thanks! 13:52:29 Yawning: hi! 13:52:30 Yawning: ;) 13:52:35 (busy in pt land, not much tor stuff, also wrist being fucked kills my typing) 13:52:52 ouch 13:52:53 :( 13:53:09 it's k, it's recovering 13:54:44 though I might get it looked at, not sure yet 13:55:31 (pain's subsiding, but I still hear audible grinding/clicking/popping whenver I bend that wrist, which indicates a decent amount of badness) 13:55:37 get a brace if you haven't already. 13:55:41 imo 13:55:43 not a doctor 13:56:09 last time I went to a dr for this crap, they wanted me to do an endoscopic carpal tunnel release 13:56:14 and ice 13:56:25 sounds complicated 13:56:48 a friend did soemthing like that, had 3 screws on his risk for a while 13:56:49 teor: fairly routine, recovery time kind of sucks 13:58:48 I'm just getting off a six-week recovery time, it does suck 13:59:50 do they still make twiddlers for one-handed typing? 14:00:13 nickm: probably 14:00:29 but I'm fairly sure I'll be ok by the begining of next week 14:00:37 well, back to normal, which isn't ok, bu tmanagable 14:01:41 nickm: re 0.2.7, which areas need attention? Reviews? Triage? 14:01:49 I know I keep shoving urgent tasks in 0.2.7 as I find them, which probably doesn't help. 14:01:56 remember that ibuprofen is not just a pain reliever but also an anti-inflammatory 14:01:59 (still not a doctor) 14:02:17 teor: code review needs help; so do basically all of the items in "assigned" or "accepted" not currently being worked on 14:02:28 triage in "new" might be helpful 14:02:32 I'm not sure that #15775 or #14882 will make it 14:03:01 Yeah, I tried that with #14882, and never got around to unit tests :-) 14:03:04 oops 14:04:57 I think I might have to work on what I have already 14:05:12 Unless we want to bump it for something more important 14:05:55 hm 14:05:58 up to you I'd say 14:07:41 I think #15775 is most important, as it makes it possible to bootstrap when the authorities are blocked 14:07:53 And the addition of a whilelist and blacklist is simple 14:09:50 Then we can work on the opt-in process 14:10:11 agreed 14:10:53 *Anyone* can work on that process, so that helps, too 14:11:52 nickm: replied #13642 14:12:01 thx, merging, let's see what breaks 14:13:01 #endmeeting