18:00:32 #startmeeting snowflake meeting 18:00:32 Meeting started Tue Apr 17 18:00:32 2018 UTC. The chair is t0mmy. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:32 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:00:38 we might learn the lesson that sending out a mtg announcement less than 24 hours from the meeting does not bring all the people :) 18:01:12 hm, maybe, but 7 of the 8 were marked as available on the doodle thing 18:01:14 oh right, irl's mail. it was encrypted so i haven't read it yet. 18:01:28 hi people :) 18:01:32 o/ 18:02:09 sukhe, what's the news on the windows build side? 18:02:22 happy to start :) 18:02:26 o/ 18:02:29 (i know, it's been like 1.5 business days since the last meeting :) 18:02:34 haha 18:03:07 but yeah, the issue still persists but I am trying a few things from our last internal discussion yesterday 18:03:24 in case they don't work, my emails for the "experts" are ready and they go out today :) 18:03:40 because I am at the stage where I have tried all I could so there's nothing much I can do without external input 18:04:06 ok. i'd say don't be shy about bringing in 'the experts'. (who are the experts, anyway?) 18:04:37 also, hm, mozilla has a pile of rust people who are ecstatic that tor is thinking about rust. i wonder how to get the webrtc people to be this excited. 18:04:37 arma4: the webrtc people. some of the issues have been reported on the google group in the past as well but the replies don't seem to work (we have similar issues, if not the same) 18:05:35 the issues persisted when reported and there have been no new follow-ups. not very inspiring to try again, but hey 18:06:09 maybe reach out to the people who asked them the first time, too 18:06:13 * Samdney lurks 18:06:19 in case they found some workaround but didn't say it to the list because the list wasn't useful to them 18:06:33 arma4: yeah, will do that as well, good idea 18:07:09 so I discussed a few things with dcf1 and hooman yesterday, if they don't work, then yes, that's the plan 18:07:40 what is hooman's role here? should we be drawing him into these discussions more thoroughly? 18:07:58 he was..a grad student with nikita? or with ian? or with amir? 18:08:10 arma4: I am not sure about his role so can't comment but he has been trying to tackle this issue for a while now, I am guessing infrequently but nevertheless 18:08:23 Ian's at Waterloo (Masters) but I think he is at Princeton now 18:08:32 oh right, with nick feamster 18:08:42 he is..a postdoc i guess? 18:09:15 arma4: to give an idea, he has been talking about this issue since last year august but as far as my understanding goes, not actively so the duration is not important 18:09:41 the network team has a list called network-team. the browser team has a list called tbb-dev. should we have a list? 18:09:46 but his input is important since he has experience 18:10:26 arma4: maybe 18:10:34 'censor-dev' 18:11:24 obfs-dev 18:11:27 hehe 18:11:39 i hate 'obfs' i can never pronounce 18:11:41 reminds me of the traffic-obf@googlegroups.com that already exists 18:11:59 but we are about more than obfs. we are about access. safe access. 18:12:07 obfs is a method, not an outcome 18:12:34 true 18:12:38 mike already took tor-access, i think, and it was about destination websites. anyway. :) 18:12:58 ok. given the people we have here. how shall we be most productive for the next little while? 18:13:05 one useful thing to do is to lay out next steps we each want to do 18:13:25 (maybe on the pad, no need to do that weird each person types one at a time while we read thing :) 18:13:32 i feel like we should share this job pad with erin 18:13:40 and ask her for questions she has to us 18:13:58 k 18:13:58 makes sense. i don't know what is most useful to contribute. i have a vision but i don't know how to turn it into the text on a job description. 18:14:14 do we want to make it into a more formal thing with roles and responsibilities first? 18:14:21 maybe we use a lot of this mtg for that topic. i had been trying to avoid bothering dcf and arlo with it, but hey, no problem here :) 18:14:24 describing the work part helped i think 18:14:30 to get more around the position 18:15:03 that makes sense but also we have an arlolra now, too 18:15:06 yeah, that paragraph at the bottom hopefully gives some handle 18:15:21 t0mmy: oh! we do! 18:15:38 (do we really? does it speak? :) 18:15:51 it does 18:15:57 woo 18:16:36 to my mind, we have three main areas we want to be pushing on in parallel: 18:16:59 (a) the project coordinator job description, (b) the roadmap of all future snowflake tasks, (c) the short term windows build and tor browser integration 18:17:20 doing 'b' will get us to a place where we can have the conversation with arlo about part-time contracting work 18:17:56 a bit of b exist right 18:17:59 in the spreadsheet 18:18:19 some of it yeah. it would sure be nice to get to the point where we think it's complete-ish though. 18:18:28 rather than "we organized the tickets that somebody had made 2 years ago" 18:18:36 which is more what it feels like right now :) 18:18:41 who would be the people to work on that 18:19:02 i wonder if this comes now or afer we hire some ppl 18:19:08 *after 18:19:24 a good question. i think it has to be iterative. and there's no need to delay the first iteration. 18:19:37 sure 18:19:40 the people would be 'arma' and 'arlo-or-dcf' to start. 18:19:48 cool 18:19:51 sounds good 18:20:15 there are a lot of angles to grab onto at once. like, dcf said we have metrics, and pointed to a graph of total number of snowflake users. but i want to know "per country". i bet we don't pass that through yet. 18:21:17 arlolra: is there anything we should be doing or keeping in mind to help you be more productive? or is 'we continue to get organized' the best way forward for you too? 18:22:33 isabela: the reason i was thinking of a mailing list is to get folks like hooman more connected to us 18:22:35 yes to the second question 18:23:03 if he's at princeton, he never sees any professors and just has to make up his own work schedule and priorities 18:23:13 which means if we're shiny enough, we get more of him 18:23:50 sure, i think that is a good thing to have a list just 'censor-dev' 'censor' is weird 18:23:53 hehe 18:24:09 it is weird. anybody have better colors for the bike shed? :) 18:24:29 i guess another thing i should mention: next week and the week after i'm going to be mostly missing 18:24:48 i will have some down time too 18:25:04 i seem to be leading these meetings, because i say things, but not for any other reason 18:25:23 (cuz you have a vision too :P) 18:25:44 i guess there's a tradeoff between "weekly meetings for sync" and "meetings because you had a topic you wanted to meet about" 18:26:04 here we are synced. we have a pile of stuff we know we need to do. does that mean we finished the meeting? 18:26:29 can you elaborate on what needs to happen to push (c) along, above? 18:26:47 or is it already happening 18:26:49 I think I would like to tackle that, if that's fine 18:26:52 t0mmy: it's what sukhe is working on, with hooman 18:26:54 part of it is already happening 18:27:02 okay perfect 18:27:07 right. step one, get snowflake et al building on windows, 18:27:13 step two, building reproducibly, 18:27:20 step three, being part of tor browser that we ship 18:27:33 and there are a few build enhacements as well but those are for later 18:27:48 arma4, not sure if you saw, but this happened https://trac.torproject.org/projects/tor/ticket/25804 18:27:55 arlolra: do you know if our new snowflake.tp.o is turning people into snowflakes? i think yes? i saw some ticket about making that opt-in? 18:28:16 arlolra: yeah, i heard rumors of that. should we switch to some other thing, like azure or amazon or? 18:28:47 it seems like we're going to have to 18:28:49 arlolra: and that switch of domain front needs to happen in the client side? meaning current snowflake clients are now failing? 18:28:56 yup 18:28:59 good times 18:29:13 dcf1 is struggling with how we present a better UI to all this 18:29:23 * arma4 adds "changes in the technology landscape" to the list of things the project coordinator needs to keep on top of 18:30:03 speaking of, I think our PM notes are close enough to make a formal job posting that I'll share with arma4, isabela, and Erin 18:30:13 (and arlolra if you want to be cc'd? or no need) 18:30:33 r.e. #25804, do we know if that change was announced in advance by Google? How could we be better prepare / more in the loop next time? 18:30:40 *prepared 18:30:49 t0mmy: sure 18:31:04 mcs: it seems Signal knew 18:31:10 arlolra got it 18:31:16 yes, but I wonder how they knew 18:32:08 * t0mmy has one last thing before we wrap up 18:32:13 I'm not sure 18:32:29 t0mmy: i'd be happy to see some transformed version of the job posting 18:32:37 * t0mmy actually two 18:33:06 mcs: maybe they didn't know anything specific, but had a general sense, which it seems was more broadly known 18:33:15 mcs: this is the moat bug? 18:33:21 arlolra: or maybe they know somebody inside google 18:33:30 it just takes one personal connection 18:33:34 the bug effects both moat and snowflake, yes 18:33:57 arma4: also possible 18:34:02 isabela: it's not the moat bug. the moat bug was a thing in tor browser that can be fixed if we try hard enough. this new thing is "google stopped letting people use their cloud service for domain fronting" 18:34:17 yes 18:34:30 that is what i meant / cuz it broke moat 18:34:30 unless you meant a second moat bug :) 18:34:53 well ppl pointed this ticket to me 18:34:54 arma4: Hmm. I guess I misunderstand the situation then. 18:34:55 cuz is breaking moat 18:35:26 arma4, arlolra, mcs: should we also consider having a primary and secondary service (2 cloud providers) for snowflake, so that clients don't need to receive an update to connect to another provider if "next time" happens again? 18:35:30 #25807 18:37:03 dmr: I believe this is what dcf1 is contemplating. how best to present this configuration to the user 18:37:13 arlolra: ah gotcha, thanks 18:38:00 primary and secondary are always mess, because.. do you try your secondary when the primary doesn't work? what if the primary didn't work just because you weren't on the internet then? 18:38:11 (the same terrible choices we're making in the tor guard world) 18:38:43 arma4: yeah, I was mentally heading down that route... 18:39:37 so "two domain fronts" is a much easier goal 18:40:03 t0mmy: ok, did you get your last-thing(s)-before-wrap-up? 18:40:24 yes, so first, was there any action items from irl's email that I forwarded? 18:40:27 arma4: as in: presenting two options to the user in the UI? I think that makes sense to remove the "primary then secondary fallback" automation 18:40:33 action and/or discussion 18:41:10 man, an encrypted email that went to only a subset of us is a crummy way to have a thread 18:41:48 snowflake.tpo has X-Frame-Options: sameorigin set! This means no one can 18:41:48 actually embed the HTML. Should probably fix this, and it'll be an easy f= 18:41:48 ix. 18:42:17 i think that's one we should think through here. the others are variations on "we should collect metrics better" 18:42:43 that said, we might need irl here to tell us what he meant. unless somebody knows. it's yet another web header that tries to protect cross site something, i assume. 18:43:17 it might be that our snowflake.tp.o webserver needs to omit that header 18:43:42 dmr: no, not presenting anything to the user. snowflake should just work. 18:43:57 okay, that sounds like an easy fix 18:44:17 t0mmy: well, the actual act of changing it is easy. but understanding what it is, and what we lose, is harder. 18:44:27 we should get irl to file a trac ticket i think 18:44:44 i think not many / none of us understand the implications of that header 18:44:46 okay I'll follow up 18:45:16 arma4: ok, if I understand this time: have a set of 2 domain fronts, and randomly pick one to try first; then try the second if first doesn't work? (no user interaction involved) 18:45:35 thing 2/2 -- sounds like we don't need to have another meeting for a while, maybe end of next week? I know arma4 is Vancouvering-but-not-that-Vancouver. 18:45:36 dmr: sure. or always try the first one first. anything in that area. 18:46:11 arma4: gotcha; I guess that's what I originally meant but didn't communicate that well - thanks for clarifying! 18:46:52 * arma4 lets other people answer t0mmy's next-week-meeting question 18:47:26 I should also pose this question to folks by email but while I have one or two of you 18:47:37 in case I am very wrong and there is reason to meet earlier 18:48:08 i think until we get our project coordinator going, meeting for needed concrete items could be more fun than meeting each week just to chat 18:48:21 but i know some people are more meeting-people than i am :) 18:48:45 speaking of which.. project manager? project coordinator? project team lead? we might want to settle on a phrase. 18:49:01 i've been saying coordinator but maybe nobody knows what that means. 18:49:13 I vote for not-project manager because we have one of those 18:49:16 i like project manager / team lead 18:49:23 t0mmy: we can have more than one! 18:49:24 would the censorship team be a vegas team? 18:49:27 yes 18:49:30 arma4: maybe these meetings should be working time? 18:49:34 if it works out, yes it would be a vegas team 18:49:42 like get work done for the things we identified we can 18:49:44 t0mmy: and in my world we should have 2 or 3 project managers by now 18:49:45 like roadmap job post 18:49:52 censorship team lead would make sense to me 18:49:59 but I'm not pushed 18:50:14 part of my sneaky plan is to sneak a second general project manager into tor this way 18:50:21 but anyway, the body of the posting is harder to make than the title =) 18:50:29 i am project manager / us team lead 18:50:30 so hehe 18:50:34 x 18:50:38 i think project manager / team lead is ok 18:50:46 “team lead” by itself makes me think we are hiring a senior s/w engineer. But maybe that is not what we are doing? 18:50:47 okay, I'll wrap this up if nobody has anything else?+ 18:50:55 I agree with mcs 18:51:13 it depends 18:51:15 t0mmy: last comment :) i owe for this meeting a branded version of snowflake.tpo 18:51:30 i'll make my best to have it for next week 18:51:30 i think we are hiring a project manager who has some technical clue, and we'll make that person also represent the censorship team to the rest of tor 18:51:48 having “project manager” in the title seems more accurate but it is just my impression. 18:51:54 antonela: irl sent us https://people.torproject.org/~irl/snowflake/ 18:52:06 if that is useful to you in any way 18:52:07 oh <3 18:52:10 thanks irl! 18:52:12 (Tor is not traditional when it comes to titles and positions anyway) 18:52:37 this wont be a dev only team tho 18:52:55 but anw i think is fine PM or PM/team lead or .. 18:53:07 yep, we'll figure it out 18:53:12 yup 18:53:12 and thanks, antonela! and irl 18:53:17 #endmeeting