16:59:41 #startmeeting network team meeting, 23 september 2019 16:59:41 Meeting started Mon Sep 23 16:59:41 2019 UTC. The chair is ahf. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:59:41 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:59:46 hello everyone o/ 16:59:55 o/ 16:59:56 our pad for today is at https://pad.riseup.net/p/tor-netteam-2019.1-keep 17:00:18 hi all! 17:00:20 hi!! 17:00:22 do we have everyone minus david today? 17:00:26 and tim 17:00:29 * asn writing report 17:00:40 * catalyst is here 17:01:08 that leaves mikeperry ? 17:01:13 we have a long list of things todo today, so let\s start with the usual stuff 17:01:29 https://pad.riseup.net/redirect#https%3A//dip.torproject.org/torproject/core/tor/boards 17:01:32 whoops 17:01:38 https://dip.torproject.org/torproject/core/tor/-/boards 17:01:41 is our roadmap 17:02:09 seems plausible to me 17:02:31 i see that one of my tickets is missing there 17:02:38 but does it look OK to everyone else? 17:02:43 ahf: maybe is because is from the anti-censorship? 17:02:47 which one? 17:03:20 #31810 is one i grabbed friday 17:03:42 do you want me to just create a redirection issue on torproject/core/tor and add it? 17:03:48 ahh, it was not on the roadmap. You are getting it because 0.4.2 17:03:54 yes please 17:04:05 or I can do it 17:04:10 i can do it after the meeting 17:04:14 ok 17:04:23 ok, everyone else is OK with the state of the roadmap? 17:05:01 let's go to the 042status page at https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases/042Status 17:05:34 we do have a discussion point related to this moved forward from last week 17:05:44 (note that I removed a bunch of items from the branch last week; see my status for the keyword I used to tag them) 17:06:09 Is dgoulet the logical reviewer on #31614? He's not around till next week. 17:06:14 iiuc 17:06:20 in that our list of -must tickets with no assignee got shorter, but the list of taken -should tickets are still pretty high 17:07:12 hmmm 17:07:23 * nickm adjust all "must" priorities to "very high" to make them stand out 17:07:46 david seems like he already reviewed parts of this thing 17:08:02 but only with an indentation comment 17:08:46 I'll take the review on it this week 17:08:52 ok 17:09:37 so we dont have any unassigned very high tickets as far as i can tell? 17:09:50 but the list of unassigned other tickets are still quiet high 17:09:57 yeah; 19 of them right now 17:10:57 let's work on assigning them during update time 17:11:10 for now: is anybody not making progress or not likely to make progress with any 042-must/should tickets? 17:11:14 we should adjust if so 17:11:21 update time? 17:11:40 oh hey 17:11:48 o/ mikeperry 17:11:58 I got confused wrt the Oct 2 patch meeting time this week rather than next 17:11:59 sorry about that 17:12:06 no worries 17:12:28 mikeperry: when you do your updates to the pad, can you revise the list of things you might have to drop and maybe have dropped? 17:12:52 nickm: what did you mean with update time here? 17:13:05 sorry, discussion time 17:13:10 ahf: yeah I did. all of those things will continue to be ijn some state of ris until Oct 22 17:13:18 ris=risk 17:14:00 mikeperry: okay! cool, please update items there if there is anything to add, otherwise it gets hard to continue to have an overview of 8) 17:14:03 nickm: oki, cool! 17:14:56 we have some announcements that a new this week: nick has gotten a new phone and new signal security number and tonight at 22 we have the ED tea time with isa 17:15:11 i think otherwise we can start with discussion items? 17:15:24 wait 17:15:25 no! 17:15:30 are the reviewer assignments looking OK? 17:15:50 there were lots of reviews this week 17:15:55 so everyone got about 2.5 of them 17:16:05 except from david and mike 17:16:26 thanks asn 17:16:39 cheers 17:16:42 anybody who have gotten something they dont think they will get around to do already? 17:17:14 i take that as a no 17:17:46 lets do discussion items 17:18:28 first one is related to 042 status that we just talked about: how do we distribute the load of the current unassigned tickets for 042 to the team 17:18:47 i see some natural owners to some of them. david is good at scheduler/kist, but is also not around right now, so i think those can wait untill he is around 17:18:54 there is some windows stuff that i am the natural person for 17:19:32 are there some tickets in this set that people think they would be the best for, and want to take ownership of now, so others can see that they intent to work on it sometime in the future? 17:19:52 imo they should be assigned, similarly to how reviews are assigned 17:20:09 that is, if targetted assignment does not work, pseudorandomly 17:20:23 agree 17:20:46 ok w me 17:20:48 we can do that. would people be OK with me (and maybe gaba?) trying to do an assignment on this work? 17:20:55 yes 17:20:57 I can do #31653 quick enough 17:21:05 mikeperry: assign to yourself then? 17:21:06 mikeperry: please assign yourself then :-) 17:21:13 aassinging it now 17:21:20 there should also be a clear exceptation of up to when they should be done 17:21:31 in terms of timeline 17:21:47 nickm: when do you wnat things done at latest? i guess that goes for all the mergers really 17:23:17 so we want to do a stable release on 15 Dec. To do that, we need to get alphas out earlier, leading to an rc with no known bugs. It helps if the rc with no known bugs is out around 15 Nov 17:23:47 To do that, it would really help if we had our next alpha in a couple of weeks 17:24:04 okay, so a bit over a month and a half to do these 17:24:09 and get them merged 17:24:12 and revised 17:24:14 yeah 17:24:20 and fix whatever problems they turn up 17:24:57 oki, cool, gaba and i will try to distribute the load of these and i hope people wont get stressed from it even though it is a few bugs per person. it seems like we do have some time to do them, but not a long time 17:25:40 ok 17:25:42 If you get anything you don't know how to do fast, please talk to other people, and don't just sit on it 17:26:01 yes, next monday we will probably talk about it and people can swap if they need to 17:26:10 there is no next monday 17:26:13 eek 17:26:14 oh is there? 17:26:17 good catch, next wednesday meeting 17:26:18 it is 17:26:24 next monday is 30th 17:26:31 (sorry for the bikeshedding) 17:26:54 we have in our schedule that the next meeting is wednesday the 2nd of october though. is there some US holiday coming up? 17:26:57 ok 17:27:36 i see arguments either way for Sep 30 vs Oct 2 being the next meeting 17:27:58 let\s stick to the schedule so we dont confuse people who arent here, i think the schedule have been there for a while 17:28:22 ok 17:28:29 (It's Rosh Hashanah, but I don't know if that's a company holiday.) 17:28:59 ok 17:29:45 (last time this would have come up is Apr 29 vs May 1, i think) 17:30:07 next item is also a follow up item from last week: we discussed the PR policy that i got the feeling that most people seemed happy with. mikeperry expressed some concerns about these guidelines. i have the feeling the people who often reviews very large patches do agree that being able to have a guideline to point people to when they get very large PRs would make them happier 17:30:42 so i am a bit unsure what next natural step would be. i think it would be to turn it into a policy that we can try out and vote on, but i also don't know if others share mikeperry's concern here? 17:30:54 well, if we have a -1, then the policy doesn't pass 17:31:07 so it's probably not something we can propose as-is 17:31:07 I can produce an alternative, in response to dgoulet, who also had issues with this specific aspect of the prposal 17:31:09 right, so would anybody vote -1 on it right now? 17:31:21 mikeperry: okay! that would be interesting 17:31:42 I believe I would vote -1 on this. it feels railroaded and I also believe dgoulet would -1, until we figure out this piece 17:31:53 so I can try to propose such an alternative 17:32:30 whoa, that's serious. What do you mean by "railroaded"? 17:32:41 cool. i think that would be interesting. it is something we should try to make progress on, so do you think it could happen this week? 17:32:44 we're not waiying for dgoulet to vote before adopting it? 17:33:04 we should, yeah, but i think david is back on monday 17:33:09 (also our policy process explicitly deals with people going away) 17:33:27 (a policy cannot become adopted until it has been provisionally adopted for a while, and nobody has -1'd it.) 17:33:46 the reason i ask about next step is because if i know someone already know want to -1 it, then its current shape is not good enough to be turned into a policy 17:35:54 mikeperry: could you give a short summary of your alternative? 17:37:17 large PRs must but split into small reviewable commits, and must have an early, quick architectural review. additionally, they should be exposed to a special CI queue that gives them more CI exposure, to catch rare CI flapping bugs 17:38:03 we don't have money i think to have such a queue? 17:38:26 i think everyone wants faster CI, but right now i don't think that is gonna happen 17:38:42 one possibility I thought of is that if you and dgoulet are the only people who want to write huge branches, and the only people who think it is reasonable to review huge branches, you could just reveiew all each other's huge branches. ;) 17:38:47 *sp 17:39:03 though probably that wouldn't be as pleasant as it sounds, and there are also the mergers to think of 17:39:11 nickm: that is not a good solution though to build silos on policy #2 though :S 17:39:20 and yeah, the mergers still need to look at this 17:39:23 nickm: that sounds like a "split the party" scenario 17:39:40 yup; not actually a good idea 17:39:48 and the rest of us have to deal with the resulting code quality 17:40:50 "fast(er) CI" on /some/ platforms is a discussion i am very open for when we get passed the gitlab migration, since i think we might be able to do something then 17:41:12 but until then i think it is very unrealistic, unless we find a sponsor who is willing to make our current solutions faster with a near zero effort change from our side 17:41:49 i think the actionable item right now is that mikeperry writes to network team with their suggestion to this within the next week or so, so we can try to make some progress to this? 17:42:17 ok 17:42:19 sounds good as a next step 17:42:33 another actionable thing I am doing, and others can do too, voluntarily: 17:42:42 I am trying to make all my PRs comply with this policy 17:42:47 so I get a feel for how hard it is 17:43:08 (I won't get a feel for the "big branches" thing again till 043 is open) 17:43:20 i think experimenting with how you want this to happen is great 17:43:49 ok, cool! 17:43:59 retrospective 17:44:04 gaba: do you want to say something here? 17:44:22 mostly that we need to find a date 17:44:24 * ahf is fine with moving it to 1st of october 17:44:30 this week doesn't work for asn 17:44:35 so I'm proposing next week 17:45:12 ack 17:45:24 anybody who cannot make 1st of october, 20 UTC? 17:45:30 i think it is two hours after our gitlab meeting 17:45:39 i can do it 17:45:55 catalyst, nickm, mikeperry: OK with you? 17:45:58 maybe david is also gonna be around 17:46:20 that should be fine 17:46:28 works for me, but if we need to postpone, then next time let's do a doodle-style poll instead of a boolean query 17:46:33 asn: yep! 17:46:39 that time works for me 17:46:48 nickm: good point 17:46:57 gaba: lets hear teor later if they can make it too 17:47:01 we had a doodle when we set up retrospectives. The idea is the last tuesday of the month for what I understood 17:47:04 ok 17:47:28 are people OK with doing jitsi? i am a huge fan of these jitsi meetings over irc, but that might just be me 17:47:38 ok 17:49:19 ... i see no objections 17:49:25 :) 17:49:27 ok, 10 min. left and we have one item left 17:49:45 how do we handle reviews that are needed to be done during the week (after monday) 17:49:53 right now asn and dgoulet does assignments monday each week 17:50:06 but sometimes we all have things we really want to get in even though it goes into needs_review after monday 17:50:22 we are not going to ask asn and david to do more of this during the week and as it is already very nice they do this for us each monday 17:50:53 so one thing i think we can do, is that people prod around in an open manner in #tor-dev if they have something. if they get no response, they come to me and try to find someone who is able to review it? 17:51:16 usually when this happens in s27 with david, we just assign the extra review mid-week to each other and we let the other person know it's urgent 17:52:03 who is the other person here? 17:52:15 david 17:52:16 for me 17:52:18 if you assign to each other 17:52:19 ah! 17:52:20 and me, for david 17:52:20 right 17:52:31 asking directly to the person that you know can review is a good idea 17:53:09 yeah, with fallback to going to me, if you don't want to ask directly or if you think anybody can do it 17:53:12 right 17:53:18 everybody OK with that? 17:53:19 i think that's reasonable 17:54:46 nobody have anything to add? 17:55:03 not from me 17:55:28 does anybody have anything else we need to discuss in the last five minutes ? 17:55:42 not I! 17:56:01 im good 17:56:26 great. i am gonna write a summary of what we have talked about here and post on tor-project@ - the summaries should allow people who arent here to quickly get what we have talked about and figure out the actionable items from it. just remember that if you want the raw dirty details of our meetings you still need to read our irc logs 17:56:31 cool! 17:56:38 i am gonna end the meeting then. remember tea time later today 17:56:41 #endmeeting