17:59:27 #startmeeting weekly network team meeting, 12 Nov 2018 17:59:27 Meeting started Mon Nov 12 17:59:27 2018 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:59:27 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:59:32 hi folks! I think it's time! 17:59:36 hello o/ 17:59:45 o/ 18:00:04 o/ 18:00:12 hi 18:00:19 let's start with the roadmap as usual. how's everybody doing? 18:00:49 progressing down the road 18:00:52 Personally, I'm pretty swamped with stuff, but I think I'll still make progress on #27359 and #28335 this week 18:01:10 wrt wtf-pad i think we've done pretty good progress on review and testing. i think the next plan is to figure out what needs to be done to get it merged. mike has a big TODO file that we need to triage. 18:01:16 catalyst: what do you think the odds are that you'll be done with the pubsub review soon? I'd like to use it in #28335 18:01:39 i spoke with HC about ARM64 builds, he'd like us to do a profile of 32-bit vs. 64-bit ARM performance before they include 64-bit builds in their orbot 18:01:54 i'm going to try to hear how many devices we have using 64-bit android from our google play store people (isa/matt i think) 18:02:02 to see if they think it's worth including in orbot 18:02:08 they = TGP 18:03:01 asn: it does need a lot more test coverage. that is the bulk of the TODO now 18:03:04 nickm: i'd also be interested in your opinion about getting the wtf-pad branch merged. where do you think we are? i know we need to write some unittests. what else? 18:03:15 mikeperry: agreed 18:03:16 that plus addtl github comments 18:03:17 nickm: looking 18:03:27 mikeperry: i can write some unittests this week 18:03:46 asn: ok ool. maybe we can divide them up? 18:03:47 I think I need to review it again for control flow. It would be really helpful if the module documented all of the in-points and out-points for its calls. 18:03:49 possibly as part of triage 18:03:51 mikeperry: yes sounds good. 18:03:54 And it needs testing like crazy 18:04:08 unit & Integration 18:04:09 nickm: it does.. in the comments and the header. I can add more comments 18:04:12 it has a lot... 18:04:27 maybe it got lost in the noise of all the other comments ;) 18:04:39 maybe; I'll look again 18:04:41 o/ 18:04:48 it's been a big review and you've been doing a great job responding, mikeperry 18:04:50 also the interfaces are a separate commit 18:05:04 so you can review just the in-call points 18:05:10 mikeperry: did it make sense what I said about the tor_assert() usage? That has hurt us a *lot* in the past 18:05:16 that's why I ordered the coimmits the way I did and recommended that review order 18:05:31 nickm: I think I did not get to that point of the code review yet 18:05:38 I'm just fixing them up in order 18:05:53 ok 18:06:11 short version: tor_assert() is best used only for truly unrecoverable errors where it is better to crash 18:06:18 mikeperry: i might not have time to do triaging tonight, but i can do unittests and coverage tomorrow all day. 18:06:30 since otherwise it tends to make us have DOS bugs 18:06:37 mikeperry: so perhaps i can start tomorrow and see where it takes me, and maybe also find a way to divide the remaining work? 18:06:51 agreed about asserts i also find many asserts that were not quite necessary 18:06:55 *found 18:07:00 yah I thought I used the non-crash assert macros like BUG() and tor_nonfatal_assert() 18:07:06 maybe I missed one 18:07:24 search the branch for instances; I think I saw a lot, but that was last week and time passes 18:07:44 dgoulet, ahf, catalyst, teor4 (in absentia) -- how are with with the s8-bootstrap stuff? 18:08:09 dgoulet is on some sort of canadian public day holiday 18:08:18 just fwiw 18:08:21 ah 18:08:35 progress on #28179, landed some patches to my GH, need a look at somet hings i broke with rebasing last night, but i think david gets to look at it tomorrow and then we can start looking at #28180 18:08:36 * catalyst plans to do some #27167 work this week 18:08:41 and then after that i'll jump on #27100 18:08:59 (i'm not sure if we have merged the PT parts of s8 into s8-bootstrap now, or?) 18:09:15 I am just bad at keeping them separate in my mind :/ 18:09:24 right, ok, i think it's fine to group them 18:09:45 gaba: anything else on the roadmap? 18:09:53 no, you covered it all 18:10:05 catalyst: anything else about the review of the pubsub ticket? 18:10:18 catalyst: (did you come up with any ETA guesses on the pubsub review?) 18:10:47 nickm: i've broken the review work into subsections; i can comment on the other smaller sections by tomorrow maybe 18:11:21 great. is that the review done, or will there be more after that? 18:11:58 (this is a good time to segue into looking at the review assignments) 18:12:33 (is everybody okay with their assignments this week? I see that nobody has more than 3, which is good 18:12:36 ) 18:12:38 nickm: i think the meat of the dispatch code will take longer; should have a better idea soon 18:13:07 hm, okay. Please keep me updated 18:13:36 ok 18:13:43 Any comments on reviewr assignments? If not, rotations! 18:13:58 we have dgoulet on bug triage and nickm on CI 18:14:02 I'll do my best :) 18:14:26 it seems like appveyor is busted again, though: #28399 18:14:45 Who understands the appveyor build environment well enough to help me fix that one? 18:14:53 It looks like an appveyor issue, not an issue with tor 18:16:26 anyone? :) 18:16:35 I'll see if teor has a guess; they've done work there before 18:16:47 i wont say that i know it, but i might be someone who can try out some commands locally too 18:17:11 okay 18:17:33 looks odd that we find it in configure script 18:17:48 autoconf looks at what will link, not what the headers have 18:17:54 if they don't match, though, there's a problem 18:18:02 ye 18:18:20 what happens if we don't install an openssl lib? 18:18:29 and just use the that is on the system? 18:18:36 (we can talk about this afterwards or later) 18:18:46 let's try after :) 18:19:24 please have a look at the reminders: it's not too early to book Brussels flights! 18:19:35 (and they get more expensive if we wait too long) 18:19:58 also please remember that s8 gets priority till it's done 18:20:02 on to announcements? 18:20:04 nod 18:20:15 yep 18:20:28 The upcoming releases we need to do have gotten complicated enough that I tried to make a timeline 18:21:05 I don't think we can do a stable on Dec 15, but early Jan is plausible 18:21:42 if no questions there, on to discussion. 18:22:00 I think we talked about O2.5 a little, and ahf scheduled the snowflake meeting. any more on those?? 18:22:10 oops, only one "?" intended 18:22:18 yes, that is an anouncement and not discussion really 18:23:04 yeah, the kick off meeting that i now call it :-S 18:23:08 in lack of a better name 18:23:13 cool 18:23:20 :) 18:23:51 Next discussion item is -- I've marked about 3-4 tickets with "035-rc-blocker?". By the end of the week at latest, can people tag any other stuff that might be 035-rc-blocker? 18:24:14 The meaning here is "If this is not addressed, we should not call our next release a 'release candidate'." 18:25:09 also, "If this is not addressed, we should not release stable." :) 18:25:31 Is end-of-week reasonable for everybody scanning the 035 tickets for that? Or I could do it if everybody trusts me, but we might miss something 18:26:00 it would be great if people nod :) 18:26:13 it's good with me :) 18:26:19 ty 18:26:31 ok. hearing no objections... 18:26:36 there is no ticket that is urgent for 035 on my side, but i will also re-triage to make sure 18:27:01 yah same afaik 18:27:05 my next discussion question is "do we need a different process for 'proposed' tickets?" 18:27:16 the list gets long, but we seem to skip over them at meetings from lack of time 18:27:36 should we do something here to change our process? 18:28:08 mmm, not sure what you mean here. 18:28:37 for me, if something is 'proposed', but does not get promoted to 'must', it can be ignored for the release 18:28:41 like, should we ignore xxx-proposed? Should we ask people to advocate particular items they care about at the meeting? 18:29:12 so right now all 035-roadmap-proposed that do not get promoted to must, can IMO be dropped from the release 18:29:19 without further discussion maybe 18:29:30 I agree ... I'm asking, do we need any process to make sure we look at them from time to time? 18:29:41 ahh, i see, yes, we should review the proposed each time to see if we are not missing a must 18:29:51 there is a step in the meeting pad saying we should do that, but we seem not to be getting to it. 18:29:58 Maybe we need some other process? 18:30:41 OTOH, I am pretty happy about how 035 and 040 have gone... 18:30:43 oops, i missed that 18:31:50 ok, let's think about it a little for next meeting. For the 035 it was not so many tickets 18:31:58 ack 18:32:51 Any other discussion things or help requests? (I see that teor has a query for me.) 18:32:52 fwiw i usually look at the proposed stuff as part of the monday meetings (e.g. i saw the url in the pad today). if that doesn't work very well, we can consider sending an email to the network-team list periodically. 18:33:08 for people to retriage their tickets 18:33:15 plausible 18:33:25 it's good somebody is looking at it 18:33:30 altho usually the tickets i really care about i monitor them and try to have them triaged properly 18:33:45 sorry for the times i havent done it well enough :) 18:34:51 hey, thanks for doing it at all 18:35:26 any more for this week, folks? We've gone through this pretty fast 18:35:37 yah nice 18:35:44 :-) 18:35:54 yep 18:37:13 okay, hearing nothing else, we can call this done. Woot! 18:37:19 See you online, everybody! 18:37:21 #endmeeting