17:30:10 <sysrqb> #startmeeting Tor Browser 23 September 2019
17:30:10 <MeetBot> Meeting started Mon Sep 23 17:30:10 2019 UTC.  The chair is sysrqb. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:30:10 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:30:26 <sysrqb> anyone around?
17:30:36 <antonela> hi 👋
17:30:45 <pospeselr> hi
17:30:48 <tjr> o/
17:30:55 <Jeremy_Rand_Talos_> hello!
17:31:00 <sysrqb> o/
17:31:08 <boklm> hi
17:31:21 <sisbell> hi
17:31:28 <sysrqb> okay the pad is here (as usual): https://storm.torproject.org/shared/tHoN4Ii7rLSjPE0OP4gydX4cMGadsXmRQNc-6lwru0N
17:31:45 <sysrqb> we'll get started in about 2 minutes
17:32:20 <GeKo> hi
17:33:05 <sysrqb> pospeselr: i don't see any items for this week :)
17:33:16 <antonela> wait wait, we are arriving sysrqb
17:33:25 <pospeselr> i'm getting there sysrqb ;)
17:33:29 <pospeselr> i'm very very busy lol
17:33:40 <sysrqb> okay, sure :)
17:33:40 <sysrqb> i'll be patient
17:33:41 <pospeselr> and distractable
17:34:26 <sysrqb> okay, let's start in the meantime
17:34:38 <sysrqb> GeKo: i think you're first
17:35:05 <tjr> I bolded that
17:35:13 <sysrqb> okay, tjr, you're up :)
17:36:09 <GeKo> i am currently testing a patch according to what i've said in #29013
17:36:29 <GeKo> mainly because we are building those libs anyway
17:37:04 <GeKo> but if there are concerns i am fine to use the approach of llvm-mingw
17:37:36 <GeKo> which means just building the libssp bits during the mingw-w64-clang toolchain assembly
17:37:50 <tjr> Ah okay I think I thought this was harder than I thought it was
17:37:57 <tjr> er
17:38:05 <tjr> I think I thought this was harder than it is
17:38:16 <GeKo> mabye it is! :)
17:38:50 <pili> hi, sorry I'm late
17:38:56 <tjr> I was going to say it might not be worth a huge investment of time if there are other high priorty things to do, since we haven't seen many stack-related bugs lately
17:39:25 <GeKo> maybe
17:39:27 <_hc> sorry I'm late, kids bedtime chaos
17:39:43 <sysrqb> welcome, welcome
17:40:21 <GeKo> tjr: doing something like
17:40:24 <GeKo> -            if target.os != 'WINNT' or c_compiler == 'gcc':
17:40:43 <GeKo> should be enough to trigger what we need on firefox's side, right?
17:41:17 <GeKo> (in the block wher fstack-protector-strong gets specified)
17:41:22 <GeKo> *where
17:42:16 <tjr> You mean to include the flags in the compilcation steps?
17:42:22 <tjr> You would remove that if statement (https://searchfox.org/mozilla-esr68/source/build/moz.configure/toolchain.configure#1680)
17:43:17 <GeKo> yeah, that's what i meant
17:43:21 <GeKo> thanks
17:44:19 <sysrqb> okay, good?
17:44:25 <tjr> yup
17:44:38 <sysrqb> cool
17:44:57 <sysrqb> okay, i see pili has some bbolded items
17:45:01 <sysrqb> *bolded
17:45:07 <sysrqb> but she;s also still typing :)
17:45:23 <pili> sneaky last minute items ;)
17:45:24 <pili> I can go
17:45:25 <pili> and carry on typing later ;)
17:45:29 <pili> the stuff I'm typing is more informational :)
17:45:37 <sysrqb> okay :)
17:45:38 <pili> shall I go? :)
17:45:43 <sysrqb> yes, please
17:45:54 <pili> the first thing is a reminder that we have the Tor Browser release meeting this week
17:46:22 <pili> the second reminder is to please update the actual points for closed issues :) (I have added in a query so you can check which issues have missing points)
17:46:44 <pili> and I quickly want to add points estimates for a few other tickets that are missing them: #19417
17:46:52 <pili> #30036
17:46:59 <pili> #30460
17:47:23 <pili> #30461
17:47:25 <pili> #31778
17:47:26 <pili> and #31616
17:48:24 <sysrqb> #30036 - 0 points because this was removed during the 68esr rebase
17:48:48 <pili> hehe
17:49:02 <sisbell> #30460 - 1 point, remaining issue kinda trivial
17:49:12 <sysrqb> #31616 - probably 1.5 points between GeKo and me, if i take a guess
17:49:24 <pili> ok, so maybe I'll remove the keyword for the month or something
17:49:26 <GeKo> i already added my points :)
17:49:40 <GeKo> so just add up yours
17:49:46 <GeKo> and we are done
17:50:12 <sysrqb> oh, i see!
17:50:17 <antonela> i talked with acat about #31778 last week, not sure if he is already working on it GeKo
17:50:50 <sisbell> I'm not sure why #30461 is still open
17:51:46 <sisbell> I'll check it again and put it for review
17:52:36 <pili> can anyone help with points for #19417?
17:53:22 <GeKo> sisbell: i am stil waiting on a patch for #31568 for that
17:53:36 <pili> anyway, you can also update offline :) (that's all from me)
17:53:37 <sisbell> What more do we need for that one?
17:53:38 <GeKo> as i want to o be sure we have the right deps merged
17:53:45 <GeKo> a patch
17:53:49 <GeKo> against master
17:53:50 <sysrqb> pili: we may need acat for that
17:54:01 <pili> ok
17:54:04 <GeKo> and the ticket set into needs_review
17:54:15 <GeKo> i still don't know whether my idea is working for you e.g.
17:54:54 <GeKo> and then the latest thoughts on that ticket needs to get put into a patch i or someone else can review
17:55:08 <sisbell> I see comment #18 - okay I'll verify
17:55:25 <GeKo> if there is a patch already existingm please point to it
17:55:33 <sisbell> I'll generate everything from sratch, eaxactly like the comments
17:56:09 <GeKo> thanks
17:56:19 <sysrqb> pospeselr: i guess #31778 needs points, too
17:57:03 <GeKo> pili: acat would know re #19417
17:57:22 <GeKo> i think 1 point for everything is not unreasonable as a start
17:57:42 <GeKo> (we need a patch there, too and test it etc.)
17:57:45 <pospeselr> sysrqb: handled
17:57:47 <pili> ok, thanks :)
17:57:52 <sysrqb> oh, i didn't mention the ticket for "we may need acat for that".
17:57:54 <pili> thanks everyone! :)
17:58:05 <pili> and please don't forget to update the actual points also!
17:58:05 <sysrqb> (i meant #19417, too)
17:58:07 <pili> (I'm done)
17:58:12 <sysrqb> thanks!
17:58:27 <sysrqb> okay
17:58:30 <sysrqb> _hc: you're up
17:58:49 <_hc> so, shoudl I give background again?
17:58:59 <sysrqb> a little background will help
17:59:06 <sysrqb> you don't need to go into to omuch detail
17:59:25 <_hc> basically, I'm working on making standlone builds of tor for Android, then also making a native Android TorService instead of running tor as a UNIX daemon
17:59:52 <_hc> then working on making torified apps all play nicely together on a single device
18:00:07 <_hc> so first up is the libtor.so binaary
18:00:33 <_hc> right now, I think all the major proects are using some version of n8fr8's tor-android binaries for that
18:00:59 <_hc> GeKo said that they ultimately want to build libtor.so as part of tor-browser
18:01:10 <_hc> so ideally, we'd share as much of that build setup as possible
18:01:19 <_hc> since the config for Android is pretty involved
18:01:51 <_hc> I recently posted my first patch to tor/configure.ac to move the most obvious bits of that config to a central location
18:02:28 <_hc> so I'm trying to figure out which parts will still be useful for tor-browser and the rest
18:02:56 <_hc> I'm guessing that the only real difference between the tor-browser builds and the rest is the potential use of NSS instead of openssl
18:03:31 <_hc> otherwise, they'll all want the same configuration, and include lzma and zstd
18:03:42 <sysrqb> (for reference https://trac.torproject.org/projects/tor/ticket/28766#comment:6)
18:03:55 <_hc> thanks
18:04:14 <_hc> at this point I'm awating feedback from y'all about this approach
18:04:37 <_hc> and also: will tor-browser also switch to libtor.so and TorService? Or is there some blocker there?
18:05:52 <sisbell> I don't see any problem with switching to using those
18:06:01 <sysrqb> we'll want to ship libtor.so in the near future
18:06:10 <sysrqb> so this works is definitely helpful for us, too
18:06:38 <sysrqb> i don't see us using a central TorService directly, when it's available
18:07:06 <_hc> you mean the binary?
18:07:19 <sysrqb> but, as i mentioned on the mailing list, i can see us using it as a proxy
18:07:54 <sysrqb> yes, i mean the binary shared library
18:08:03 <_hc> I need better names for all this ;-)  TorService would be a Java subclass of android.app.Service
18:08:29 <_hc> then there will be a TorServices.apk which will be a minimal app built with that TorService
18:08:48 <sysrqb> oh. i see.
18:09:28 <_hc> so I could see TBB using TorService directly, either built from source or via mavenCentral, where we will publish it
18:09:34 <sysrqb> that sounds very similar to tor-android-service
18:09:53 <sysrqb> (with less functionality, which is what we want, too, ideally)
18:10:00 <_hc> right, it is the modern version of that
18:10:07 <sysrqb> heh
18:10:10 <_hc> yeah, totally stripped down
18:10:13 <sysrqb> tor-android-service was supposed to be modern :)
18:10:20 <sysrqb> was/is
18:10:30 <sysrqb> maybe you and sisbell should discuss this offline
18:10:36 <sisbell> There is a minimal tor service in TOPL
18:10:41 <_hc> to be literally onl tor daemon
18:10:42 <_hc> I already have it building reproducibly in a docker image
18:10:43 <_hc> one thing I know little about is loading PTs in this arrangement
18:10:46 <sysrqb> and you can work together on this
18:10:55 <_hc> but I know its possible since its done that way on iOS OnionBrowser
18:10:56 <sysrqb> it seems wasteful for us to competing projects
18:11:09 <sysrqb> okay
18:11:20 <_hc> as far as I know TOPL still runs tor as a daemon, that's the key part that needs to be changed
18:11:39 <_hc> UNIX daemons were never fully supported by Andorid, and they are getting ever more constricted
18:12:46 <_hc> oh, sisbell what time zone are you in?  I'm in central european time currently UTC+2
18:13:08 <_hc> I work pretty normal hours, m-f, 9-6
18:13:11 <_hc> ping me any time
18:13:17 <_hc> but I can schedule meetings as needed
18:13:32 <sisbell> PDT
18:14:19 <sysrqb> okay, i thik this discussion can continue after the meeting
18:14:28 <_hc> ok
18:14:42 <sysrqb> but it is good to talk about this
18:14:45 <_hc> offline for that part, any other feedack?
18:15:10 <_hc> like rust? Is that coming soon for Android?
18:15:15 <_hc> I haven't touched it yet
18:15:30 <sysrqb> not in the near future
18:15:41 <GeKo> when we build tor nightlies
18:15:50 <GeKo> we want to test that there as well :)
18:16:03 <GeKo> but i think there is not even a ticket for that yet in trac
18:16:09 <_hc> ok then I'll skip rust for now
18:16:16 <GeKo> yeah
18:16:17 <sysrqb> but is that a blocker for this?
18:16:19 <_hc> #28766 mentions rust
18:16:34 <_hc> I don't need rust for anything I'm working on
18:17:17 <GeKo> _hc: oh, while you point out that ticket
18:17:37 <GeKo> it seems you want to upstream that patch or something similar?
18:18:09 <_hc> yes, I posted it there to start the feedback
18:18:10 <GeKo> or what are we supposed to do with it?
18:18:17 <GeKo> i see
18:18:28 <GeKo> but i doubt the network folks are seeing that ticket
18:18:33 <_hc> the idea is to get as much of the Android config in the --enable-android flag so its easily shared
18:18:41 <GeKo> it is more for tor build integration into tor-browser-build
18:18:53 <_hc> OK, I can post it elsewhere once its vetted by y'all
18:19:21 <GeKo> sounds good
18:19:37 <GeKo> thanks for doing all that work, really appreciated
18:20:00 <_hc> thank you Handshake Foundation for paying me to do it :)
18:20:16 <sysrqb> :)
18:20:19 <GeKo> :)
18:20:22 <_hc> I'm all about intgeation :)
18:20:24 <_hc> integration
18:20:38 <sysrqb> okay, any other feedback for _hc before we move on?
18:20:53 <_hc> also, FYI, I have  barebones gitlab-ci build of tor for Android, that could be expanded to run the test suite in the emulator https://trac.torproject.org/projects/tor/ticket/28704#comment:6
18:21:26 <_hc> anyone know with the gitlab migration, is the idea to also migrate to gitlab-ci? or still use travis? or both?
18:21:27 <sysrqb> neat!
18:21:48 <sysrqb> i don't know
18:21:59 <sysrqb> these are some questoins for the network-team
18:22:00 <pili> _hc: I think we still want to figure that out
18:22:03 <GeKo> _hc: ahf knows most about it
18:22:15 <pili> we might move some stuff to gitlab-ci for some teams
18:22:23 <pili> but maybe just websites stuff
18:22:30 <_hc> with fdroid, we've been using travis for osx and gitlab-ci for all the GNU/linux flavors
18:22:37 <_hc> works well enough
18:23:21 <ahf> _hc: we are not far enough to have discussed CI much, but yes, i do believe we will need infrastructure at some point with CI runners, since that is also useful for pages and so on
18:23:50 <ahf> _hc: a big part of CI for us is also related to windows
18:23:53 <_hc> I maintain some runners for fdroid, which I can share to torproject, to get the ball rolling
18:24:07 <ahf> and we have jenkins that needs to be integrated somehow in this
18:24:08 <_hc> openssl uses travis for osx/linux and appveyor for windows, so it seems there is a windows possibility there
18:24:19 <ahf> yeah, we use appveyor and travis too right now for core tor
18:24:28 <_hc> all three can be run in parallel even
18:24:41 <ahf> yep, that is why the gitlab ci runners are not so high on our list right now
18:25:13 <sysrqb> okay, i think this can move to #tor-dev
18:25:29 <sysrqb> for all the CI related detail questions :)
18:26:18 <sysrqb> anyone else have something they want to discuss from the weekly updates?
18:26:35 <GeKo> who is working on #31192 next week?
18:26:41 <GeKo> sisbell?
18:26:44 <GeKo> sysrqb?
18:26:53 <sysrqb> i saw sisbell was working on it
18:26:54 <GeKo> or better: this week
18:27:07 <sysrqb> hrm, but it's not on his list for this week
18:27:10 <sisbell> I have a working implementation of that issue
18:27:32 <GeKo> i'd like to have a patch for this merged by friday
18:27:47 <GeKo> as this is one of the big parts we want to have as m any alphas to test for as we can
18:27:51 <GeKo> get
18:28:15 <sysrqb> sisbell: i posted a comment on the ticket, can you look at it today?
18:28:15 <sisbell> sysrqb: I have a new version of tor-android-service with the latest tor
18:28:18 <GeKo> which means a patch for review up by wednesday
18:28:33 <sysrqb> sisbell: okay
18:28:37 <sisbell> So if you want to take it, that's fine. I also have implementation
18:29:12 <sisbell> I'll take a look at the ticket
18:29:59 <sysrqb> sisbell: I'd prefer you take it, but i'm worried the commit you mentoined on that ticket doesn't have all the necessary pieces
18:30:41 <sysrqb> so if you can take a look at it and get us the updated branches, then we can review them
18:31:09 <sisbell> sure, I missed the mozconfig commit, I'll add that
18:31:34 <sysrqb> if you don't think you can get this done by wednesday, then i can take it
18:31:48 <sysrqb> (before wednesday)
18:32:05 <GeKo> there is #31564 as well
18:32:11 <GeKo> which we should get done by wed
18:33:13 <sisbell> That issue has the tricky part about the JDK but I think it is doable by Wed
18:33:36 <sysrqb> do you think you can do both, or only one in that timeframe?
18:33:52 <GeKo> sysrqb: let's talk about that one after the meeting
18:33:58 <GeKo> err sisbell:
18:34:59 <sisbell> i can have x86 issues done by today
18:35:52 <sisbell> If there are no hidden issues after the JDK change in issu #31564, I can have it done by Wed. ass well
18:36:36 <sysrqb> okay, we can discus this some more after the meeting
18:36:37 <sisbell> I'll know more by this evening and then we can go from there
18:36:42 <_hc> FYI, my tor-android fork is building for 32-bit and 64-bit
18:37:03 <sysrqb> _hc: excellent, thanks
18:37:22 <sysrqb> the only two discussion points this week are from me
18:37:25 <_hc> with openssl 1.1.1d and tor
18:37:44 <sysrqb> 1) just keep in mind the release (likely at the end of this week) - so get in those patches early
18:38:12 <sysrqb> 2) i wanted to ask if anyone has an opinion about enabling fuzzyfox after the esr transition
18:38:22 <sysrqb> but we can discuss that at another meteting
18:38:25 <sysrqb> *meeting
18:38:48 <sysrqb> and i think that's it
18:38:57 <sysrqb> any last comments/questions/concerns?
18:39:50 <sysrqb> okay, thanks everyone, sorry this meeting ran a little over time.
18:39:53 <sysrqb> have a good week!
18:39:58 <antonela> thanks!
18:40:00 <sysrqb> #endmeeting