16:59:31 <nickm> #startmeeting network team meeting, 2 Oct 2017
16:59:31 <MeetBot> Meeting started Mon Oct  2 16:59:31 2017 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:59:31 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:58 <ahf> hello
17:00:09 <isabela> oi
17:00:13 <catalyst> hi
17:00:28 <asn> good $time_of_the_day
17:00:31 <komlo> hello!
17:00:33 <nickm> hello all!
17:00:43 <nickm> our pad is over at https://pad.riseup.net/p/QsVn0OkRRsWF or its onion-flavored equivalent
17:01:12 <nickm> Remember to use BOLDFACE for stuff you'd like to discuss
17:01:30 <nickm> (other than isis, are we missing anyone?)
17:02:11 <dgoulet> hi
17:02:59 <nickm> dgoulet: do you and teor and asn think that you'll likely want to get IPv6+SoS in 0.3.2?
17:03:11 <nickm> or for later?
17:03:29 <dgoulet> nickm: matter of opinion so far but in my case, later
17:03:36 * nickm nods
17:03:45 <asn> i think so as well
17:04:07 <isis> o/
17:04:19 <nickm> hi!
17:04:27 <nickm> https://pad.riseup.net/p/QsVn0OkRRsWF is our pad this week
17:04:35 <isis> is this because of the meetbot brokenness?
17:05:02 <nickm> brokenness/offensiveness : see https://lists.torproject.org/pipermail/tor-project/2017-September/001459.html
17:05:53 <nickm> isis, ahf: Do you think today is a good target for, or is another day more likely?
17:05:58 <nickm> and, how is it going?
17:08:49 <ahf> my aim was for today, yes.   we got all the changes in (i believe) last week, isis started going over the changes files. i've been testing it on a mac + my bsd machine before the meeting.  we need to cleanup the old changes files and do the bumping in the configure.ac script and create the tarballs
17:08:54 <pastly> dgoulet: nearby now. assuming you want to notify me about meeting place
17:08:54 <ahf> then write the announcement mails
17:09:16 <ahf> nickm: anything you've noticed that worries you?
17:09:37 <nickm> ahf: nope!  Did the changelog scripts turn out to be usable for you?
17:09:49 <nickm> And how shall we handle signing/releasing/announcing?
17:09:56 <nickm> isis, ahf: ^
17:10:52 <ahf> signing and tagging i'm not sure about
17:10:53 <nickm> stuff should still probably go out signed by my key, I think
17:10:58 <ahf> yes :/
17:10:59 <nickm> if that makes sense to you
17:11:11 <isis> yeah
17:11:18 <isabela> isis: can you stick around tb meeting again? like you did last week to follow up on the moat stuff with them
17:11:32 <ahf> it does. i just don't like the idea of us generating something and you signing it - maybe we sign the tarballs, you check that, test it out and then sign it with your key?
17:11:41 <nickm> komlo: are you going to have time to circulate some rust stuff before the meeting, or should we wait for Monday? :)
17:12:00 <isis> isabela: i think so? not sure? i think we had planned to do this release right after the meeting
17:12:10 <nickm> ahf: Or you can make the tarballs, I can re-make the tarballs, and compare them to the ones you made, and sign the ones I made? :)
17:12:31 <isabela> isis: ok maybe just give an update here and i can pass it along to tb team later?
17:12:38 <komlo> nickm- it is going to be a bit tight for me before Monday, but if there are things that will be helpful for Monday, i can prioritize that!
17:12:40 <ahf> then let's make very sure that we have the same versions of the autotools generated files - otherwise we are going to have very big diff's ?
17:13:05 <nickm> komlo: okay; I wasn't sure whether you were going to send out your exercise first or not. up to you!
17:13:18 <ahf> err, let's make sure we have the same versions of the autotools tools such that the generated files *should* be identical
17:13:26 <ahf> danglish is hard.
17:13:39 <nickm> I have autoconf 2.69; automake 1.15.
17:14:16 <nickm> specifically, the fedora versions
17:14:26 <ahf> ok, same versions in my debian stretch
17:14:39 <ahf> cool
17:14:42 <nickm> ok. We'll learn something intersting either way
17:14:59 <ahf> yes.
17:15:13 <nickm> I have a montreal request for everyone which I'd like to highlight:
17:15:21 <nickm> please don't go into a huge crunch right before montreal!
17:15:32 <nickm> The meeting will go better if you are rested, happy, and healthy.
17:15:38 <nickm> Please try to arrange that if you can :)
17:15:45 <komlo> that is a good recommendation!
17:16:00 <ahf> yes <3
17:16:02 <nickm> at what time are we starting on Monday morning?
17:17:14 <isis> any time that people are all awake?
17:17:30 <isis> it will be thanksgiving that day, so many things are closed i believe
17:17:32 <ahf> what time did we start in wilmington? i remember that as being quite OK
17:17:47 <isis> (just as a warning to people who expect to find food/coffee)
17:17:51 <isabela> isis: do you want to give an update? I remember mcs asking for something to interact with, with mock data...
17:17:52 <nickm> People who arrive before Monday should pick a time?  I'm going to try to arrive as early as I can, but I have a 5.5 hour drive
17:18:08 <asn> 10:00 arrive, 10:30 start?
17:18:12 * asn went for it
17:18:16 <ahf> wfm
17:18:20 <nickm> sounds fine to me
17:18:28 <isis> isabela: i updated the spec a bit but did not finish the mocked server yet
17:18:39 <komlo> nickm: sorry, i think i misunderstood you! i sent out the rust exercise to the team, but not the solution.
17:18:50 <nickm> oh!
17:18:51 <komlo> 10/10:30 sounds good!
17:19:00 <nickm> subject line?
17:19:08 <isabela> isis: do you have an eta for the mocked server?
17:19:14 <isis> https://github.com/chelseakomlo/rust_kata
17:19:24 <isis> isabela: by mid this week?
17:19:33 <isabela> tx
17:19:47 <isis> isabela: mocking the server actually just means writing the whole server since it's twisted
17:19:51 <nickm> Oh, it's on the thread!
17:20:08 <komlo> nickm: "Thread for organising team days and keeping others in the loop" but i am not realizing that it is not super clear
17:20:10 <nickm> okay if I link your exercise from the notes?
17:20:21 <komlo> we can send out a "Rust hackday prep" email!
17:20:28 <nickm> +1
17:20:30 <komlo> nickm: yes, please do
17:20:41 <isis> isabela: or rather, "not actually serving a captcha/bridges" is about the same amount of work as "actually serve them"
17:20:54 <isis> in terms of being able to interact with it
17:21:51 <isabela> ok
17:21:57 <nickm> also everyone: teor sent out a note to the network-team list about a team dinner. please RSVP if you can, so that we know how many people to expect. and let us know who else to invite that isn't on the network-team list
17:21:58 <isabela> i mean we do need it done anw right?
17:22:11 <nickm> anyone else have discussion topics for this week, or tuff we should talk about before montreal?  If so please add it to the pad and/or boldface it
17:22:24 <nickm> if not, I want to talk about where we are at with unit test coverage
17:22:30 <ahf> nice with team dinner!
17:24:35 <isis> isabela: yeah, i need to finish it asap, but i accidentally signed myself up for too much with also meeting planning and learning to od releases :/ sorry
17:25:16 <isabela> isis: could you please prioritize this tho?
17:25:22 <nickm> okay. Seeing no extra topics, let's talk about test coverage a little?
17:25:35 <ahf> yep
17:25:53 <nickm> The good news is that our line coverage rate has been steadily increasing.
17:25:54 <isis> isabela: yes
17:26:13 <nickm> (I'm looking at the results of "make check" in the stuff I note on the pad, since I think integration tests are a separate deal)
17:26:26 <isabela> isis: thank you, we are in october already and the longer this takes the harder will be for TB team to meet their deadline
17:26:50 <nickm> the less good news is that we are not descreasing the amount of untested code; it is just growing more slowly than the amount of covered code.
17:27:26 <nickm> If you have a look at https://people.torproject.org/~nickm/volatile/coverage-diff-summary -- it shows the number of untested lines in each module that are new or modified in 0.3.2 vs 0.3.1
17:28:22 <nickm> I wrote a bunch of tests last week, and will try to write more, but perhaps we should think about whether any of the stuff towards the bottom of this list can get more unit tests than it has now.
17:28:40 <nickm> you can see nice visualizations of our test coverage at https://people.torproject.org/~nickm/tor-auto/coverage/
17:28:58 <nickm> the cov_check directories have the results of "make check";
17:29:22 <nickm> the cov_full directories have the results of make check, make test-stem, and make test-network-all
17:29:36 <catalyst> is there a quick summary of how to read the coverage report? like i have no idea what the "-:" "1:" "#####:" etc prefixes mean
17:29:48 <isis> +1
17:29:51 <nickm> are you looking at coverage-diff.xz ?
17:29:59 <catalyst> yeah
17:30:14 <ahf> if we have a day in montreal like the last day in amsterdam then perhaps it could make sense to sit and hack on this together to get some more tests in?
17:30:33 <nickm> https://lists.torproject.org/pipermail/tor-dev/2017-September/012462.html describes the format briefly; I don't know if it makes sense from that, though
17:30:37 <nickm> does that explain it for you?
17:31:10 <nickm> basically, that file was made by taking the gcov output from each version, and then converting all lines with nonzero coverage count to "1"
17:31:20 <nickm> and then making a unified diff
17:31:40 <catalyst> i can see it's a unified diff.  does "#####:" mean uncovered?
17:32:00 <nickm> yes.  That's a gcov thing.
17:32:07 <nickm> And 1: means covered
17:32:20 <nickm> and -: means "no code was generated for this line"
17:32:38 <catalyst> thanks
17:32:42 <nickm> np
17:33:03 <catalyst> it's surprisingly hard to search for "how do i read the gcov output format?"
17:33:04 <nickm> oh, and x: means "this region was marked as unreachable using the LCOV_EXCL annotations"
17:33:17 <dgoulet> test-network-all must not have HS v3 integrated? hs_client.c should have WAY MORE lines covered by it :S
17:33:30 <nickm> This is 'make check'
17:33:32 <nickm> not make test-network-all
17:33:36 <dgoulet> 13:30 <+nickm> the cov_full directories have the results of make check, make test-stem, and make test-network-all
17:34:17 <nickm> dgoulet: I'm not sure.  The tor-auto coverage is made using the latest tor and the latest chutney each night, so...
17:36:09 <nickm> If anybody needs help looking at test coverage themselves, I can walk you through the relevant part of doc/HACKING/HelpfulTools whenever you want
17:36:35 <nickm> ahf: I like the idea of seeing if we can grab some time to write tests;
17:36:52 <nickm> and maybe we can make it a default "finger exercise" when we're not sure what to do for a while
17:37:23 <catalyst> in case anyone else is looking, https://gcc.gnu.org/onlinedocs/gcc/Invoking-Gcov.html has some documentation about how to interpret the gcov output format buried near the end
17:38:08 <nickm> anyway, it's something to keep an eye on.
17:38:27 <nickm> we should probably talk more about improving our unit tests at montreal as part of our other meetings, too
17:39:00 <ahf> nickm: i think it sounds like a good idea for late in the week. i think it is something that is possible to do even after many days with a lot of people around you
17:40:19 <isis> ah yeah, that would be a good task for once we're in the overwhelmed stage of the week
17:40:39 <ahf> yes
17:41:55 <nickm> ok. sounds like we're low on topics. anything else for today, or shall we call the meeting over?
17:42:25 <asn> good over here. busy with guard tickets today.
17:42:38 <ahf> sounds good. looking forward to see you all in a week!
17:42:53 <asn> yes yes :)
17:42:58 <komlo> see you soon!
17:43:28 <nickm> thanks, everyone!
17:43:31 <nickm> #endmeeting