17:59:17 #startmeeting weekly network team meeting, 26 Feb 2018 17:59:17 Meeting started Mon Feb 26 17:59:17 2018 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:59:17 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:59:19 Hi all! 17:59:22 hello :) 17:59:40 buenas 17:59:53 The pad is at https://pad.riseup.net/p/DP6HmRXvVW6w 17:59:58 im checking out pad 18:00:21 Only a couple of quick announcements from me: 18:00:40 * isabela is writing 18:00:41 hello 18:00:44 1) I'm on call for jury duty tomorrow. In 3-4 hours I find out whether they need me to come in. If so, I'm offline tomorrow, maybe more. 18:01:17 2) Please let me know if anything should go into the upcoming releases that I have not already merged, other than the TROVE-2018-001 patches under discussion 18:01:36 I'm planning to release between now and wednesday 18:01:39 s/planning/hoping/ 18:03:18 o/ 18:03:27 * isis is writing 18:03:49 ahf: should I expect more data coming from the wakeup experiments you did, or should I write my tickets with that in mind? 18:04:44 nickm: there is more coming, this is the start. i'm trying to do one with some active data being downloaded and then the stacks. i have the stacks from "entering a system call" and when we get taken off the CPU by the scheduler 18:05:02 but i'm trying to do the fanout data which is "just" the stacks that were around right before we returned back to the event loop 18:05:14 ok 18:05:16 if you can flesh out some tickets from what is there now, i think that is good, but there is more on its way 18:05:25 I think I know a bunch of those, based on the code and on the data you have 18:06:07 ok! 18:06:14 second_elapsed_callback and refill_callback should account together for about 11 wakeups per second = 660 wakeups per minute 18:06:41 yes, they pop up frequently in my sampled traces 18:06:52 ahf: is there a master ticket i should make items for those under? If you don't remember, I'll look for one. 18:06:53 i don't think there is, no 18:06:59 there is for CPU performance, but that is unrelated 18:08:27 (I am here, I keep getting disconnected from the pad; will cut+paste update as soon as I am connected long enough) 18:10:01 announcements: everybody please read the announcements, and fill out the spreadsheets 18:10:13 #action please do the heat-mapping exercise today if you haven't already 18:10:28 #action please do the subsystem maintainer comfort exercise today if you haven't already 18:11:16 (who are we still missing on those?) 18:11:25 eveyrone has done heat-mapping 18:11:36 isis is the only one who still hasnt done subsystem maintainer 18:11:48 isis: ^ 18:11:58 so we pretty good but not 100% yet 18:12:18 k, will do 18:12:28 I don't see pastly and armadev on the heatmap thing. 18:12:34 arg this pad is determined to keep me disconnected, lol 18:12:37 isabela: the ping is the email today, right? or did i overlook something? 18:12:50 #action isis does subsystem maintainer thing 18:12:50 ahf: today :) 18:12:55 good, good 18:12:57 #action armadev does the heat-mapping spreadsheet, if he wants. 18:13:00 sorry i left the comment on the last week item 18:13:04 #action pastly does the heat-mapping spreadsheet, if he wants. 18:13:05 but it was done today :) the email 18:13:17 thought i had missed something. last week asn had to ping me because i had overlooked one of the sheets :-S 18:13:19 thanks, good to know 18:13:48 after dgoulet asked us to be more willing to score stuff as 3 in the subsystem sheet, i'm wondering if i should change my answers 18:14:46 catalyst: being more bold miht be a good idea 18:15:05 also because this spreadsheet might help us assign reviews/etc. in the future 18:15:11 isis: maybe wrt the rust sha2 thing, put up the code you have on the ticket, share the errors, and take a break to see if anybody can help figure it out? 18:15:19 i was close to writing 2.5 many places, but then i know how developers dislike floats 18:15:35 so if you care about a subsystem and you are not sure if you are 3 yet, but you could be or you want to be, perhaps putting 3 is fine! 18:15:41 +1 18:16:04 the idea is to be a PoC about the subsystem not the massive line-by-line expert 18:16:18 (but enough also to be able to do reviews ofc :) 18:16:21 isis: ping on the subsystem spreadsheets -> https://docs.google.com/spreadsheets/d/1Ufrun1khEo5Cwd6OwngERn829wU3W3eskdrriaYfUBQ/edit#gid=1974867547 18:16:26 re roadmapping, are we supposed to ignore the older roadmap spreadsheet that was standalone (not consolidated with the latest heatmap and subsystem stuff)? 18:16:46 isis: and check the tabs for the areas of work one too - so you can add your votes 18:17:28 catalyst: yes, we will build a new one for march-nov - the email i sent to you and ahf and nick today is related to that 18:17:59 catalyst: and another one where is focus on 2019-2021 18:19:05 isabela: would you like to talk about what you need from people for roadmapping and other stuff between now and Rome? 18:19:45 yes! 18:20:02 my goal is to get as much of the 6 months roadmap done before rome 18:20:16 so we can leave more of our f2f interactions to answer those questions for the other roadmap: 18:20:28 how the network team will look like in jan 2019 18:20:38 cool 18:20:38 and what this team will do from jan 2019 till dec 2021 18:21:18 the heatmap should guide us on these stuff - but of course wont be the only factor, we are doing exercises now 18:21:22 so we learn what others are thinking 18:21:27 and understand where others are coming from 18:21:35 so the idea is to come to rome with our roadmaps begun -- at least the sponsor-driven parts? 18:21:39 will we get some info about this before rome? if we have any grants send out that we are waiting to get a response on and so on? 18:21:49 this way i hope our discussions are more productive with these contexts of where folks are at 18:21:53 (for the 2019 to 2021 part) 18:22:15 regarding sponsor related work 18:22:27 we will only plan for it for the shorter roadmap 18:22:41 i would prefer the second (longer one) we plan in a different way 18:23:05 more vision/priorities focus - independented of sponsor 18:23:12 neat 18:23:29 this will guide us 18:23:59 as we continue to have 6months roadmaps (or roadmaps we create every tor meeting) 18:23:59 how are we doing so far on the roadmapping? I filled out what I could for the sponsor8 stuff that I know most about. What should I and others still do, and when should we do it by? :) 18:24:28 nickm: yes! let me give a short report on that 18:24:57 right now i am asking forlks involved with sponsor work to drop what they think the work for those deliverables should be in the 6 months roadmap 18:25:02 (lol at no one having a '3' on the rust subsystem) 18:25:17 oh wow, there _are_ tabs, you were not kidding 18:25:23 so if i ping you please follow up with me on that :) 18:25:24 i put 3 on rust… 18:25:40 once i have this done i will ask folks to give me tickets related to the work they wrote in the pad 18:25:57 ok now we have one 3! nice 18:26:03 oh, it changed my answers? maybe i bumped the scroll wheel 18:26:09 we have at least one expert on every subsystem 18:26:12 it changed a bunch of my answers! grrr 18:26:18 once we have tickets, they will be: 1. placed in the roadmap in the month we will work on them 2. have owners :) 18:26:29 i will write this all out via email to the list too so everyone has it 18:26:52 isabela: it's *future* work? so it's for the tickets for the next 6 month period, right? 18:26:55 (sorry, what has tabs exactly? I worry that maybe I didn't see the tabs) 18:27:17 i kind of want to see the totals per column per developer? like to see if i feel as confident as other people 18:27:36 isis: scroll horizontally? 18:27:45 isis wants to see the total per column 18:27:46 ahf: yes, keep in mind that sponsor8 contracts ends in this period too 18:27:54 not per row 18:27:57 yes, in november, right? 18:28:00 or earlier? 18:28:01 ahf: so we should be thinking what we want as a goal and what needs to be done to get there 18:28:09 i can do that easy, i've become an expert in google spreadsheets. ready for my accountant gig. 18:28:10 i can't remember if it's november or august 18:28:14 ahf: officially earlier but I already told them i will ask for extension 18:28:18 ok! 18:28:20 ahf: cuz mobile started late too 18:28:26 :) 18:28:28 * ahf nods 18:28:29 cool! 18:28:35 since i feel like i'm being pretty conservative and giving "1" in places where maybe i am a "2"? 18:28:45 asn: lol 18:28:57 isis: i just didnt want to give total numbers to people. might make people feel bad or sth. 18:29:04 i will do this tomorrow morning - fits ok with going over where we are now to prepare for the mobile session in rome 18:29:06 we have 3 for every subsystem so that is a good news 18:29:17 yes :D 18:29:48 I should probably be lower on a lot of subsystems 18:31:41 this is spreadsheet is very cool 18:31:49 very very cool :) 18:31:49 asn: yeah, i also thought about that it might make people feel bad, but i don't think there's anything wrong with being less confident… i just want to know if i'm using the same relative scale of confidence i guess? 18:31:54 it seems like we're in a pretty good place this week. After everybody's done with the stuff we need to do for roadmapping, please remember to focus on getting 033 out the door: fixing 033-must bugs, reviewing code, etc... 18:32:02 yeah, this data is rad 18:32:10 do we have more discussion things we haven't gotten to at this meeting? 18:32:10 i like how it's colour coded too 18:32:30 i would also be interested in seeing per-developer totals 18:32:39 ill do per-developer totals guys 18:32:39 isis: i started out with 3 in compress, 2 a few places, 1 a lot of places and some zero's and have slowly bumped the values up every time i've looked at the sheet :-S 18:32:47 i guess we can kind of visually infer it because of the color coding 18:32:52 i cant stop writing google doc summations. 18:32:57 hahaha 18:33:19 need to create a twitch channel with google sheets advice for developers, asn 18:33:34 ahf: yeah same! i looked at other people's answers and was like "oh i guess if so-and-so is an expert on that, then i probably am not a '0' like i thought" 18:34:02 asn: you should hang out with Sue from accounting in rome (spreadsheet master) 18:34:22 haha yes i have lots to learn 18:35:34 what are the third/fourth roadmapping tabs for? do i need to fill them out somehow? 18:35:51 there 18:36:10 isis: just the areas of work and subsystem ones for now 18:36:14 isis: you can ignore the rest 18:36:49 isis: each person has 5 votes for priorities (in general) tor network has at the moment 18:37:09 ok cool 18:37:29 i did the vote thing last week, just didn't realise you could have tabs in a Google doc 18:37:49 :) 18:38:08 coolio o/ 18:38:47 do we have more discussion topics for this week? 18:38:57 more ideas / issues / generally interesting things? 18:39:05 i have a question about the build system 18:39:25 but that could be one on one with someone who knows the build system really well 18:39:39 (also build system is maybe an area of code that we forgot) 18:39:39 i can try ... is this the rust thing you mentioned? 18:39:42 yeah 18:39:53 I also have questiosn about control port+bw stats stuff 18:40:09 komlo's Rust code links to the C code to call the C function tor_version_as_new_as 18:40:29 which is built into libtor 18:40:35 * catalyst had a Q in the pad about prep for Rust stuff in Rome 18:40:51 my code links to crypto_digest.h, which is linked into libor-crypto 18:40:54 I will ask in #tor-dev after, though, since it will require some code rummaging 18:41:19 mikeperry: sounds cool 18:41:21 libor-crypto.a and libtor.a and libtor_rust.a all get linked together at the end 18:41:48 does anybody know the answer to catalyst's rust question ? 18:41:50 i need libor-crypto to get built first, then made available to libtor-rust as it's getting built 18:42:27 idk about Rust stuff at rome, i won't be able to be there 18:43:03 last time i invited the Rust people, so i don't know if they got invited? i could ask komlo if she had plans 18:43:33 komlo is on the wiki as a "yes" arriving 9 mar 18:43:53 isis: is this a circular dependency thing? 18:44:30 there is a thread on the net team list about pre-meeting hackfest in Rome on the 10th 18:44:32 catalyst: ^ 18:44:50 I think last email was from teor responding to mine iirc 18:44:50 nickm: i am worried about that in general moving forward yes, but in this specific case i don't think the graph is circular 18:45:30 catalyst: but it could benefit now for a bit more structure for the 10th 18:45:43 it might be a good idea to ask our rust contacts how they handle this kind of thing. Is the problem explained on a ticket anywhere? 18:45:46 unless there is something in libor-crypto calling some Rust code, but there shouldn't be (at least not yet) 18:45:55 catalyst: oh it is an email for the 11th from isabela but blended in the 10th a bit 18:46:00 or is this a build issue where we need to build some C stuff before rust and some C stuff after rust? 18:46:13 nope, i can explain it on #24659 and link to my code 18:46:27 yes, it is a problem of some C, then some Rust, then some C 18:47:06 isis: that would be great. We might need to change how our build system builds C and invokes Cargo. 18:47:29 okay, will explain everything i understand so far on the ticket! thanks! 18:48:25 did we get to everything? We have 10 min left :) 18:49:52 hearing no additional topics, I'll call the meeting adjourned. 18:49:54 Thanks, everyone! 18:50:01 bueno 18:50:10 o/ 18:50:20 please remember to poke me today about anything you need me to do ; I might not be online tomorrow 18:50:23 peace! 18:50:24 #endmeeting