22:59:49 #startmeeting network team meeting, 7 Aug 2019 22:59:49 Meeting started Wed Aug 7 22:59:49 2019 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:59:49 Useful Commands: #action #agreed #help #info #idea #link #topic. 23:00:01 who is here tonight? 23:00:07 hiya 23:00:10 hi mike! 23:01:03 https://pad.riseup.net/p/tor-netteam-2019.1-keep is our regular pad 23:01:34 hi! 23:01:39 hi gaba! 23:01:48 network-team: anybody else here? :) 23:02:01 hi! swati here. 23:02:08 hello swati! welcome! 23:02:15 Thank you so much! 23:03:15 let's see, catalyst and ahf are busy this month. asn is probably asleep 23:03:43 do we know if dgoulet/teor4 are here? 23:04:15 I am here, just getting breakfast 23:04:22 nice! 23:04:50 so, let's start with a look at 041 status: https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/CoreTorReleases/041Status 23:05:59 mikeperry: is #30992 truly 041-must? 23:06:15 if so we should figure out how to get it done asap; we want to release a stable release on the 20th, and I'm gone next week 23:06:21 right I am not sure actually 23:06:30 Also, can anybody take on a quick review of #31343? It is simple, and it should fix appveyor 23:06:35 there are a couple possible solutions there 23:06:49 the bug itself is not particularly severe.. none of the cases are, I think 23:07:14 but they might cause problems for researchers.. and the warns are noisy 23:07:51 Can we disable the warns, then? Do they do anything useful? 23:08:01 Do we expect to support researchers on stable? 23:08:13 Or disable them in 041, or call them protocol-warn, or something? 23:08:26 #31356 is also related.. its annoying and wastes a cell per circuit but is also not a big deal 23:08:28 I expect researchers to run nearly anything 23:09:08 there is a world where a fix solves both bugs, but that fix would have some risk 23:09:35 (that fix being bump the Padding protover to 2, have 0.4.1 require that to negotiate, and check a sequence number field) 23:09:44 the sequence number field check is the risky bit 23:10:02 that sounds more complex than we should try to do 13 days before release. 23:10:13 It seems that this may not be a must for 041. 23:11:05 I think bumping Padding to 2 is the right (partial) fix. Doing protover + version checks is problematic. 23:11:53 Is this something we should delay the stable release for? Will it take more than a day to do? 23:12:36 hrmm if we wanted minimal possible change, I think that is just demote the logs and not mess with protover at all. 23:13:00 can we do that in 041, and look into doing the "right thing" in 042? 23:13:03 related question is do protocol warn noise affect our CI currently? is that problematic? 23:13:11 yeah I think so 23:13:12 not really 23:13:27 mikeperry: that would be good, then 23:13:30 The protover fix is easy, just change the C constant, Rust constant, and the constant that has the required protover for padding. But we can do it in 0.4.2 and backport to 0.4.1 later. 23:13:39 ack 23:13:45 so thats what I would do then.. make the only actual warn involed into a protocol warn, and have it be something we deal with later 23:13:56 sounds ok 23:14:35 mikeperry: I know you're busy, but could I also ask you to take a look at #31343? It is a simple fix for a 32/64-bit issue on windows, and it's breaking appveyor 23:14:49 Even better, we might want to upgrade the protover declaration before stable, and then change the requirement later? 23:15:00 I mean I don't know what anrea was thinking there with the labs 23:15:17 I don't think the labs makes much sense there. It is a time-traveling labs() 23:15:18 my instinct is that since this is pre-monotime, anrea needed it to account for ntp jumps or something 23:15:18 Then people who upgrade when stable is released will support the updated requirement. 23:15:31 and so I would err on the side of caution and keep it as llabs 23:16:22 So if our clock jumps backwards, we should treat a skew number as plausible? I don't think that's right. 23:16:31 This is about whether we use the skew information from the netinfo cell 23:16:54 if our clock moved backwards, IMO it's a bad time for that, hence the change I made 23:17:03 I also did not look at all the the surrounding code to be honest. I can do that. I have tangled with netinfo-relaed things before 23:17:06 but we don't need to solve this right now; I'm just wondering if I can make you reviewer on this? 23:17:19 thanks, mike! 23:17:49 the cost of this is that it makes geko less likely to get his esr code review by 9/3 23:17:53 wrt #30992, I think advertising an extra protover in 041 would also be okay. Anything much more than that would worry me. 23:18:01 buit maybe I shouldn't rush that anyway 23:18:31 It's a pretty simple patch; if it takes more than a couple hours to review, we can give it to somebody else 23:18:38 but I think it should be fast 23:19:29 Do you think we can get a solution for these two issues in this week or next, so 0.4.1.5 can come out? 23:20:15 (Next topic is the roadmap. Gaba, where is the roadmap now?) 23:21:08 We are trying to get the roadmap in gitlab now. 23:21:17 is it https://dip.torproject.org/torproject/core/tor/boards ? 23:21:20 https://dip.torproject.org/torproject/core/tor/boards 23:21:51 how to I move something from one column to another? 23:21:54 The idea is to test it there and see how we want to use dip (gitlab) for this and if it works. Based on this we will update our plan for migration to move to gitlab when is time :) 23:22:05 I should be able to do the protover bump and log demotion thing, yah. I will also add more circ id lines to the other padding logs to make it easier to spot this kind of stuff. it took some leaps to figure out, even with the new loglines 23:22:13 nickm: you click and move 23:22:29 Most of the stuff is deal by labels. Each column is a filter on a specific label. 23:22:43 We have labels for Tor in general and we have labels specific for Core. 23:22:58 Everybody in the network team is part of Core now and can move stuff around. 23:23:11 I added only the stuff for August for now. 23:23:33 And used the due dates for the month that we said those issues were going to be worked on. 23:23:49 where do I click to move something? It seems to not do it 23:24:08 on the same ticket 23:24:28 you are log in, right? 23:24:33 ohhh! 23:24:37 mikeperry: remember that protover is declared separately in C and Rust 23:24:38 sorry 23:24:39 :-) 23:25:25 I'm moving some tickets around, let me know if I'm doing it wrong 23:26:08 teor: thanks for the heads up. I have not touched the rust pieces of it before.. hopefully I get that bit right (hence my preference for just log demotion but I can give it a shot what could go wrong) 23:27:24 sed -i.bak 's/Padding=1/Padding=2/g' src should do it for both Rust and C 23:27:38 They're just strings 23:27:41 I think you want to advertise padding=1,2 23:27:41 We created a 'The Tor Project' group at the base of all teams/groups with the idea to have one board for everything where people can submit issues. Then each team/project will have its own boards. https://dip.torproject.org/groups/torproject/-/boards 23:27:53 nickm: yes, oops 23:28:58 I also added one more column to the network-team/core board. The column 'next' thinking about the next week of work to be done sort out by priority. 23:29:05 this kanban is pretty neat; I hope everybody takes a minute or two to look it over and advance stuff on it 23:30:32 Next is review assignments; let's look at them quick and move on to discussion? 23:30:41 I really hope this can work out for us and we can have it integrated with the tickets :) 23:31:02 I have a big review backlog, I will try to clear it today and tomorrow 23:31:47 makes sense 23:32:33 For discussion: ... I think we already talked about "dip" a little 23:33:04 yes 23:33:29 the next thing is about our awesome tech writer 23:33:43 :-) 23:33:43 Hi Swati! 23:33:48 hi swati! thanks for waking up so early for this :) 23:33:52 Hello everyone! 23:33:59 No worries at all! 23:35:00 The project starts in September. Swati will be participating in the network team meetings. I'm thinking that she could subscribe to tor-dev mailing list and discussions of her project can be there too. 23:35:15 That sounds good; I also recommend the IRC channel #tor-dev if that's convenient 23:35:20 for quick questions and stuff 23:35:30 yes, sounds good. 23:35:37 Okay sure, I will subscribe to both of these. 23:36:03 There are some people on the tor-dev@ mailing list who are very helpful, and some who have so many ideas as to be a bit confusing. 23:36:06 i need to understand what doc tools have been used, how the source repo is set up, etc. 23:36:10 Sure! 23:36:38 Can I just shoot my questions on any of these 2 channels then? 23:36:44 swati: you might find an IRC bouncer useful, it will keep IRC history when you are offline. pastly can help you set one up. 23:36:54 swati: teor already added some good feedback in the pad. We can talk later to see how to track this project. 23:37:03 The tor manual page is built using an ancient tool called "asciidoc". Its source is in the Tor git repository, in a single file, doc/tor.1.txt 23:37:09 . 23:37:22 Ah asciidoc I see. 23:37:29 okay. 23:37:30 swati: It's better to do questions on #tor-dev than here most of the time: this channel is only used when there is a meeting happening 23:37:46 got it! 23:38:08 we don't love asciidoc, but we _do_ need something that is editable as human-readable text, readable as unix manual pages, and readable as html 23:38:12 teor: that will be helpful.I would want to retain my chat history here. 23:38:43 I think that people who do this kind of thing are mainly using different kinds of markdown now 23:39:03 I understand. So we plan to continue using asciidoc right? 23:39:46 for now, yeah. My understanding is that you want to work on the content and the organization more than on the tooling, and I think that is reasonable 23:39:47 Yes, there are many other markup tools available for this now. 23:40:07 okay. yes. sure. One step at a time :-) 23:40:34 If you run into huge problems with asciidoc, we can revisit that, but it would IMO be better to do this one thing at a time like you say 23:41:01 Yes! makes sense. 23:41:55 before it was in asciidoc, it was in roff, which nobody wants to edit :) 23:42:06 :-) 23:42:22 asciidoc is a bit ancient, but workable. 23:42:59 great 23:44:24 Any other things to ask or talk about right now? I'm on vacation next week, but most of the rest of the team will be here 23:45:21 I think we've answered most people's questions on the pad. 23:45:39 Yes. I think we are fine. 23:45:42 swati: also, if you like, you're welcome to use our meeting-pad format to keep track of what you're doing and planning week to week 23:45:52 mikeperry: did all your questions get answered? 23:45:53 mikeperry: did you want asn and dgoulet to stop assigning you reviews? Or did you want to transfer your current reviews to other people? 23:46:00 I don't have anything right now. But I will have lots of questions, once I start working on the project. 23:46:05 sounds great! 23:46:21 Yes, I will keep my status in there once I begin work. 23:46:44 everyone: Right now I'm planning to prioritize review/merge for others, and things other people ask me to do, so that I have nothing blocking on me when I go offline Saturday morning 23:46:57 If you need to get in touch with me, use signal or personal email: I'll be filtering all lists 23:47:18 mikeperry: I am not sure who can help with your other at risk items, or what tasks we can do to help. 23:47:53 swati: In case you do run into stuff asciidoc can't handle, my sense is that we wouldn't mind moving to markdown, but heavier-weight systems (eg docbook) would be a hard sell 23:48:37 I am not sure what the right answer is here 23:48:38 Okay I will look into options and present them to you guys if need be. 23:48:44 for reviews, etc 23:49:10 mikeperry, teor: please both also feel ok offloading some reviews to me, and I'll look at them tomorrow 23:49:23 I can do these. and the bugfixes. it puts geko's ask of having firefox esr code review done by 9/3 at risk 23:49:41 I plan to discuss that with him tomorrow at the vegas meeting 23:49:51 that's not a hard deadline either btw.. that's just for the alpha 23:49:54 the hard deadline is 10/22 23:49:57 so it's ok 23:50:05 ok 23:50:08 at this point 23:52:31 (as I see it, my top priority for this month is ensuring that circuit padding has no blocking bugs, documenting it, reviewing patches related to it.. so it's all good by me, but geko should have a heads up) 23:53:09 We're coming up on the hour. Anything else for today, folks? 23:54:02 hearing nothing... thanks for coming, everybody! I'll see some of you online tomorrow, and the rest on the 19th! 23:54:09 o/ 23:54:12 peace all! 23:54:13 #endmeeting