17:00:52 #startmeeting weekly network team meeting 17:00:52 Meeting started Mon Jun 19 17:00:52 2017 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:52 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:00:55 Hello everybody! 17:01:00 hello! 17:01:04 hi 17:01:09 ayyy 17:01:15 let's take a moment to put our updates on the happy-fun pad that teor started for us at http://5jp7xtmox6jyoqd5.onion/p/csrVSo66GUmG 17:01:20 (also available at pad.riseup.net) 17:02:31 hm 17:03:10 Also, a process hack: If there is something in your update, or somebody else's update, that you would like to discuss: please BOLDFACE the relevant words. 17:04:52 Sebastian, isis, armadev, isabela, meeting in progress if you like. 17:05:05 komlo: ibid 17:06:31 any more updates in progress? If not, let's start! 17:06:45 teor and dgoulet aren't around, so we'll have to hold any questions we have for them. 17:06:52 is there a "batch uncc me and never e-mail me about this again" thing for trac 17:06:54 :( 17:07:00 let's ask each other questions now? 17:07:21 hi, I'm here 17:07:27 Yawning: you can batch-remove yourself from the cc list of anything you're cc'd on, but I odn't know if that's sufficient to make it never email you. For everything else, there's procmail 17:07:42 hi ./ 17:07:46 mmk 17:07:55 hello! 17:07:59 hihi 17:08:10 Yawning: as somebody who's subscribed to tor-bugs@, I am not the best person to ask 17:08:29 ahf: are the log files all in now? How is that going? 17:09:12 o/ 17:10:10 nickm: there is enough for me to analyze them - going to write a small python script that extracts the info from the debug log we added in march/april where we get sizes out of tor. was going to start digging into that after this meeting 17:10:27 Also I wanted to bring up, for anybody who wasn't in wilmington, that I replaced my callgraph generator with a better one. See https://gitweb.torproject.org/user/nickm/calltool.git/ 17:10:41 might not work for others; patches welcome. 17:10:52 ahf: great 17:11:10 one of my tests ran into the issue of crashing with lzma though, so i have restarted that. going to put the logs up in the https://gitlab.com/ahf/tor-sponsor4-compression repo 17:11:19 and make some spreadsheet like we did with compression algorithms 17:11:35 maybe i'll have something for you and me to talk about tomorrow? 17:11:40 yeah. let's do what we can with the info we have, and not block while waiting for more tests to complete? 17:11:43 sounds great 17:11:48 yes, agreed with that 17:12:21 dgoulet, asn are the logical people to answer the "How is prop224 going" question ... 17:12:58 catalyst: did you want to talk about SessionGroup=, or did I boldface it by mistake? 17:13:09 no, i wanted to talk about it briefly 17:13:15 ok, let's go for it! 17:14:19 so there's this bug #22619 about the SessionGroup=N session isolation option being treated as invalid 17:14:45 the validation bug isn't hard to test, but i also discovered that we seem to have no test coverage for the actual session isolation feature it's supposed to control 17:14:54 s/hard to test/hard to fix/ 17:15:18 I can believe that :( 17:15:39 it's quite an old bug, so maybe people don't try to use this feature very often? 17:15:48 I don't understand the point of that 17:16:03 of SessionGroup? 17:16:04 in any case, i'm not sure that coverage testing of the functionality is worth putting in 0.3.1 17:16:06 and till you pointed at the bug, I didn't know that option existed 17:16:07 yes 17:16:26 catalyst: agreed; it's ok to delay all of this IMO for 0.3.2 actually, since it's not an 0.3.1 regression 17:16:30 though a fix might be nice. 17:16:43 as for coverage, I'd say it's worth a day , but not a week? 17:16:44 I assume the people that use it do some placebo bullshit torrc with 3 million untested/buggy options 17:16:56 Yawning: they don't, because the option is rejected :( 17:17:15 if the option doesn't even enable right, does it actually even work 17:17:40 well yeah, it's obviously untested 17:17:52 the benefit of the option is that it lets you do something like say that _this_ DNSPort and _This_ SOCKSPort are associated, and can share circuits. 17:17:55 unless it used to work and then got broken? but there's still no current automated testing 17:18:33 ack 17:19:16 so my preference would be "fix it in 0.3.2.x, include tests, unless it will take more than a couple of days." how does that sound? 17:19:22 (and is that even what you were asking? :) ) 17:19:51 pretty much. though i suspect that testing of the functionality will require chutney 17:20:14 i'll bump the ticket to 0.3.2.x 17:20:31 tricky, and interesting. I think that some of the backend can be tested as unit-tests, like the functions that test whether two streams can share a circuit, etc... 17:20:43 but yeah, it's hard to say if it actually works without an integration test 17:20:57 nickm: that's a plausible unit testing approach; i'll look into it 17:21:28 pastly: fwiw, dgoulet is not online right now, so you'll need to ask him about KIST status when he's back :/ 17:21:49 ack 17:22:11 all my discussion topics are on the topics list at the top, as are the remainging topics i can see. So, anybody have any questions based on anyone else's updates? 17:22:52 wrt 0.3.1 -- we have about 38 open tickets. dgoulet, catalyst, asn, and teor have a small number of tickets. 17:22:56 ahf and I have a huge number of tickets open 17:23:07 there are also 10 "new" tickets for 0.3.1 that are neither assigned nor deferred. 17:23:22 so, I'm hoping to go through the stuff that's assigned to me... 17:23:34 how many of these are in an actionable state? 17:23:36 but as for the stuff in "new" I hope more people pick things up. 17:23:48 most, I think? 17:24:31 depends on what 'actionable' means :) 17:25:05 Would anybody like to take on the job of trying to get somebody attached to each of those tickets? If not I can do it... 17:25:06 needs_information leans towards "not-actionable" 17:25:10 right 17:25:28 I'm talking about the 10 tickets marked "new" in the milestone "Tor: 0.3.1.x-final" 17:26:14 ok, I guess that's me. :/ 17:26:51 one more thing I'll ask for help on: we all share code-review responsibility 17:26:57 i would kind of like to leave the "easy" tickets open for new contributors but i'll jump on them if need be 17:27:28 sometimes, to ensure a loose FIFO, we use a keyword to mark all the reviewable tickets, and try to get those all reviewed before moving on to review new ones. 17:27:34 the current keyword is "review-group-18". 17:27:52 thanks. do we document this practice somewhere? 17:27:54 hi, my internet suddenly crapped out for some reason 17:27:54 I am not going to commit to reviewing any new code until every "review-group-18" ticket has gotten a review. :) 17:28:04 * isis reads backlog 17:28:09 sounds very fair 17:28:09 catalyst: I don't know! maybe on the wiki; does anybody know? 17:29:07 review-group-18 now contains 26 needs_review tickets. Let's all allocate time to get s5tarted on it! In passing, I note that I can't review my own patches. :) 17:29:18 there's https://trac.torproject.org/projects/tor/wiki/org/process/TorOnTrac but wasn't there also a pad that people were writing notes about for revising that wiki page? 17:29:47 i live (trying to keep up with all the backlogs) 17:29:48 catalyst: I think so. anybody have that pad? 17:30:18 I think maybe dgoulet (who is not here right now) was maintaining it. 17:31:18 hello all. i'm jetlagged but back in europe. 17:31:21 noted on the needs_review tickets 17:31:46 nickm: found it: http://5jp7xtmox6jyoqd5.onion/p/trac-hygiene 17:31:53 woo 17:31:55 ty 17:32:09 who wants to start updating the wiki page with that? 17:33:11 I've made a note on the pad with both urls, I hope somebody can pick it up. 17:33:16 if not I'll try to remember. 17:33:48 also, if anybody doesn't have an oniongit.eu account who is participating in tor development, please get one? We're going to start testing it out for our code reviews 17:33:58 to get one, I believe the procedure is "ask hiro" 17:34:10 yep 17:34:13 anything else for this weeK? 17:35:13 ok. I'll be around a few hours more if I can be helpful with anything. Thanks, everybody! 17:35:16 #endmeeting