16:59:12 <nickm> #startmeeting network team meeting, 18 June
16:59:12 <MeetBot> Meeting started Mon Jun 18 16:59:12 2018 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:59:12 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:27 <nickm> Hello everyone!
16:59:33 <nickm> The pad is https://pad.riseup.net/p/ISP7HhTOJePC
16:59:36 <isis> hi!
16:59:41 <nickm> Sorry I didn't send it out till this morning
17:00:03 <isis> i have fast internet via robots with lasers now :)
17:00:20 <catalyst> oooh robots _and_ lasers :)
17:00:27 <nickm> We'll start with a look at the roadmap and the reviews!
17:00:54 <ahf> ack
17:01:28 <nickm> dgoulet: I would love it if you can get through #26383 soon, so I can stop holding back on patches that will conflict with it
17:01:52 <nickm> isis: I'd love it if you can review #26378 so we can get the rust tests one step closer to happiness
17:02:27 <dgoulet> nickm: yes ofc
17:02:31 <nickm> ty
17:03:06 <nickm> as for the roadmap -- are we generally happy with where we're going?  Right now we need to get 0.3.4 out the door, and start on all our 035 things
17:03:07 <isis> nickm: i did review it iirc?
17:03:33 <nickm> isis: I added another patch to make it compile
17:03:37 <nickm> it still doesn't link though
17:03:48 <isis> oh, i see now
17:03:52 <catalyst> nickm: see my discussion item :)
17:04:45 <nickm> and if we're all happy with reviews and the roadmap--
17:05:09 <nickm> let's take a quick look at the remaining 0.3.4 tickets:
17:05:19 <nickm> https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&milestone=Tor%3A+0.3.4.x-final&group=status&order=priority
17:05:19 * dmr is here but going to bounce back and forth for a bit; hopefully full attention soon!
17:05:51 <ahf> https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&milestone=Tor%3A+0.3.4.x-final&group=status&order=priority
17:05:59 <ahf> err, wrong window to paste in
17:06:47 <isis> nickm: #26378 lgtm
17:07:01 <isis> ahf: qubes!! ftw!!
17:07:13 <nickm> mikeperry: are you on track to revise #26214 ?
17:07:23 <mikeperry> dgoulet: where do you want the comment about #26214? do you want it in the code or on the ticket?
17:07:27 <nickm> ahf, dgoulet: What shall we do with #25372 and #26300 ?
17:07:30 <ahf> i don't think we've gotten any extra feedback after #25372
17:07:47 <ahf> we added a patch to get more information out, but i don't think we have gotten any feedback after that
17:07:53 <dgoulet> I'm moving #25976 out of 034 ... I spent time on it trying to how it is possible but ouf difficult without a reprod case :S
17:08:00 <ahf> isis: oh yes, my copy/paste error rate there is quite high :-S
17:08:22 <nickm> ahf: sounds like it should be fine to move it into needs_information + Unspecified ?
17:08:24 <dgoulet> mikeperry: code
17:08:28 <ahf> nickm: agreed
17:08:33 <nickm> dgoulet: sounds ok?
17:09:14 <dgoulet> ahf: yes I think so, fun fact, something is killing the dirauth with huge memory pressure, one theory is the compression + dir port madness so I'll keep an eye out on that
17:09:27 <dgoulet> ahf, nickm: ^
17:09:46 <nickm> good
17:10:12 <mikeperry> dgoulet: ok
17:10:13 <ahf> dgoulet: interesting, they are hammering the dirport with some work that causes us to do (de)compression?
17:11:04 <dgoulet> ahf: maybe, uncertain... I need moria1 data once arma pops online at some point
17:11:34 <nickm> catalyst: do we have a logical next step on the cross-compiling tickets?  Are we going to get the revisions we need, or should we do them ourselves?
17:11:59 <ahf> dgoulet: oki, i'm interested in this if you know more
17:12:02 <nickm> everybody: we need somebody to take on #25976 and #26094
17:13:10 <catalyst> nickm: i guess the cross-compilation stuff partly depends on whether we still want them fixed in 0.3.4
17:13:22 <nickm> How hard is the fix?
17:13:32 <nickm> and how likely to destabilize?
17:14:16 <nickm> Rotations this week are: mikeperry on bug triage, nickm on community, ahf on coverity and asn on ci
17:14:36 <nickm> anything to hand-off there?
17:14:50 <ahf> yes
17:14:54 <ahf> i'm on both coverity and ci
17:14:58 <ahf> asn is afk
17:15:01 <nickm> ok
17:15:30 <catalyst> nickm: oh, it did get revised and i missed that. i'll take a look
17:15:37 <nickm> ty!
17:17:54 <nickm> Here are tickets that are proposed for 0.3.5, but not included.  Let's look at them!
17:18:01 <nickm> https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&keywords=~035-proposed&keywords=~035-roadmap-proposed&milestone=!Tor%3A+0.3.5.x-final&group=status&col=id&col=summary&col=owner&col=type&col=priority&col=component&col=version&order=priority
17:19:03 <nickm> isis: I think #25573 goes in, because it's already on our roadmap?
17:19:57 <dgoulet> #26367 ... so hmmm I think there is at least another ticket about that... and everytime it comes up, huge controversy so I kind of gave up over the years on making it happen
17:20:22 <nickm> isis : oh wait not that one, I meant #24990
17:21:01 <isis> okay, i'll clean up the draft i have
17:21:13 <isis> that's a good airplane task :)
17:21:23 <nickm> isis: no hurry; just moving it into 035
17:22:12 <nickm> Anybody have strong feelings on any of the rest of these?
17:22:39 <nickm> They all seem good-to-have
17:22:52 <nickm> maybe we should revisit when we do 035 triage
17:23:39 <dgoulet> yah
17:23:49 <nickm> ok
17:24:43 <nickm> Let's see -- on discussion, one topic seems mostly answered (who will work on rust build issues)
17:25:02 <nickm> The other one is -- how aggressive should we be with the code-movement refactoring stuff I proposed?
17:25:41 <isis> how bad of an idea would it be to `sudo apt-get dist-upgrade` the bridge authority from the other side of the planet when it has a failing RAID1 array?
17:25:47 <dgoulet> nickm: I say very before we all have big patches for 035 :P ... although your work will make #20700 kind of interesting to refactor but might be ok...
17:25:51 * dmr is giving full attention now
17:25:52 <nickm> isis: scary imo
17:26:13 <isis> nickm: ok, that is my feeling as well
17:26:45 <isis> #7349 is in 035, but it has a child ticket of #17159
17:26:49 <nickm> dgoulet: There will never be a point with no pending patches, I think...
17:27:20 <nickm> isis: if that's the case, please set the milestone?  Currently it's "unspecified"
17:27:36 <isis> the latter is its own whole rabbithole… but i think neither can happen until the bridge auth is running >033
17:27:38 <dgoulet> nickm: no *big* pending ... at least doing it early in 035 allows us to rebase regurlarly if we do have patches
17:28:57 <nickm> ok.  Then I'll plan to revise my refactoring plan thing some time today or tomorrow, get additional comments, and move ahead with the part where we split up src/common ?
17:29:05 <nickm> saving the src/or splitup for 0.3.6?
17:29:28 <nickm> https://pad.riseup.net/p/cJUbdx0c7IzM-keep is the refactoring plan
17:29:40 <nickm> not yet revised to account for feedback
17:29:44 <dgoulet> if 036 becomes our LTS... I think we should do it all in 035 imo but depends ofc on the time it will take to pull it off
17:30:33 <ahf> i started a hack project on the airplane home to europe where i lift sandbox.c into src/sandbox/ and split the linux specifics into individual files - is anybody scared about this being done or?
17:31:28 <nickm> ahf: sounds plausible but I'd like to do something more like the refactoring plan first, to avoid conflicts, if that's ok w you?
17:31:41 <nickm> dgoulet: that's an option; how do other people feel about that?
17:32:07 <ahf> nickm: yes, don't think i'll get back to it this week, so just go ahead with your stuff there - it looks nice!
17:33:32 <nickm> this set of changes, to be clear, would move nearly every C file in Tor.  So if you want to say "no argh no" then maybe this is a good time to shout
17:33:41 <nickm> or to at least say "Hey let's think about it longer"
17:33:46 <nickm> else I can try to rush ahead.
17:34:36 <dgoulet> my two cents: I personally see no point on waiting up tbh... better to do a big change at once instead of many and trailing this refactoring for months and years...
17:35:40 <catalyst> i think we're going to have to do occasional adjustments from time to time no matter what; maybe getting things out of the way sooner so we can more quickly discover if we made any mistakes?
17:35:56 <ahf> i think it looks good. the only thing i think looks a bit icky is that we include 'src/' in all #include statements. think i'd've stripped that with -I$(top_srcdir)/src and use #include <ll/.../foo.h>
17:36:03 <ahf> but that's mostly a bikeshed thing i think
17:36:13 <nickm> Oh!  I did not mean to include src/ in the include statements.
17:36:19 <ahf> ah, ok, good good
17:36:34 <nickm> changed :)
17:36:40 <ahf> great, awesome
17:36:47 <nickm> ok, I'll move ahead ...
17:37:11 <dgoulet> nickm: to be clear, you plan to address comment on the list or just go with the pad?
17:37:11 * catalyst also not fond of "ll" but that's kinda bikesheddy
17:37:19 <dgoulet> nickm: I'm asking because the naming changed on the mailin glist
17:37:31 <dgoulet> nickm: and the structure as well a bit
17:37:35 <nickm> catalyst: see discussion on mailing list -- I'm thinking that will change
17:37:49 <nickm> dgoulet: Yeah, I need to revise the pad _and_ answer on the list
17:37:55 <dgoulet> neat
17:38:03 <nickm> I've already sent some replies
17:38:17 <arma1> a wild arma appears
17:38:48 <arma1> (did i get my cultural reference right?)
17:38:51 <ahf> i'm looking forward to this split up
17:38:53 <nickm> catalyst: you asked about the rust test failure
17:39:15 <nickm> catalyst: I think if we can't get it working again by the end of the week, we should disable that test in 0.3.4
17:39:19 <ahf> arma1: yes, if you were found in the wild and are about to receive a lot of damage before we are going to trap you into a pokeball
17:39:39 <nickm> I'm going to be trying to at least partially solve #25386 this week, though.
17:40:01 <catalyst> yeah i was wondering if disabling the test (#[cfg(broken)] or something) makes sense, given that it seems to be expecting some mocked functions that return all "x" as the hash output
17:40:03 <nickm> I might be enlisting the Usual Rust Suspects to look at my horrible hacks
17:41:45 <catalyst> FYI i get doctest failures if i do the thing i proposed above to disable the test
17:41:58 <isis> nickm: catalyst: is there any way i can help with the rust test stuff?
17:42:16 <nickm> catalyst: for me, the test is failing at the link stage
17:42:34 <catalyst> nickm: yeah i got it to link by disabling mod test
17:42:53 <isis> catalyst: yeah the "xxxxxx" should definitely fail, but if it fails on that then that means the linking is working
17:43:12 <catalyst> isis: yeah we can follow up on #tor-dev after this meeting maybe?
17:43:21 <isis> sure thing!
17:43:26 <nickm> +1; I'd like to make this all happy
17:43:40 <catalyst> getting CI happy again is useful
17:44:08 <isis> fwiw, i just wrote to the noisetor signal group to ask about running the bridgeauth out of the noisetor rack here in the bay
17:44:25 <nickm> arma1: please clarify what exactly you have in mind for #26094 when you can?  It seems like it should be a trivial ticket, but the way you said "if you wanted" makes me wonder if you know some reason not to do it like that
17:44:46 <nickm> any other issues for this meeting today?  time sure has flown!
17:44:55 <isis> so i guess i'll put bridgeauth maintenance/rebuild on my plate for when i'm back from england and canada
17:46:00 <nickm> +1
17:46:08 <nickm> okay, thanks everybody!
17:46:10 <nickm> #endmeeting