18:58:48 <sysrqb> #startmeeting Tor Browser meeting 22 February 20201
18:58:48 <MeetBot> Meeting started Mon Feb 22 18:58:48 2021 UTC.  The chair is sysrqb. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:58:48 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:58:59 <Jeremy_Rand_Talos> Hello!
18:59:01 <sysrqb> err. year = 2021
18:59:07 <GeKo> hi!
18:59:11 <sysrqb> hello!
18:59:16 <sysrqb> Pad: https://pad.riseup.net/p/tor-tbb-keep
18:59:38 <pospeselr> o/
18:59:39 <sysrqb> I'd like to welcome back pospeselr ! *cough*
18:59:42 <sysrqb> o/
18:59:51 <pospeselr> oh hey fancy meeting you folks here
18:59:54 <boklm> hi!
18:59:56 <GeKo> p/
18:59:59 <Jeremy_Rand_Talos> Welcome back pospeselr!
19:00:01 <GeKo> o/
19:00:03 <GeKo> actually
19:00:17 <GeKo> and a special hi! to boklm too from me :)
19:00:23 <sysrqb> :)
19:00:24 <boklm> :)
19:00:31 <pospeselr> hi everyone
19:00:58 <acat> o/
19:02:17 <sysrqb> Jeremy_Rand_Talos: I hope you're warmer this week than last week
19:02:44 <acat> wb pospeselr!
19:02:48 <Jeremy_Rand_Talos> sysrqb, yes, heating just came back this morning.  :)
19:02:58 <pospeselr> :)
19:03:23 <sysrqb> that's good news, at least
19:03:40 <Jeremy_Rand_Talos> indeed, thanks :)
19:06:29 <sysrqb> alrighty.
19:07:18 <sysrqb> This is release week. 10.0.12 and 10.5a11 should be published ~tomorrow
19:07:48 <sysrqb> 10.0.12 is signed and ready to go. Signing 10.5a11 is in progress
19:09:08 <sysrqb> We may have an emergency release later this week or next week, when we have a patch for fixing the recent issues on Arch/Fedora/OpenSuse/etc.
19:09:17 <sysrqb> that will be a Linux-only release
19:10:07 <boklm> a firefox issue?
19:10:15 <sysrqb> Please upadte your boards, if needed: https://gitlab.torproject.org/groups/tpo/applications/-/boards
19:10:25 <sysrqb> boklm: no, this is only affecting Tor Browser 10.0
19:10:41 <sysrqb> it seems related to a change within glibc 2.33
19:10:58 <boklm> ok
19:11:06 <sysrqb> https://gitlab.torproject.org/tpo/applications/tor-browser/-/issues/40328
19:11:33 <GeKo> sysrqb: it then seems we should disentangle all the bugs you duped over tho that one
19:11:45 <sysrqb> but I duped tor-browser#40350 and tor-browser#40335 into that ticket
19:11:50 <GeKo> because not all users had a newer glibc i think
19:11:58 <sysrqb> GeKo: i thought they did
19:12:23 <sysrqb> but the tab crash seems related to freetype
19:12:28 <sysrqb> while arch didn't update its freeetype versoin
19:12:45 <GeKo> yep
19:13:33 <sysrqb> i suspect they are all related to an incompatibility in the sandbox
19:14:24 <sysrqb> but I am planning on debugging tor-browser#40328 and then finding if the fix for that also fixes the crashes on the other distros
19:14:45 <sysrqb> and then reopen one of the duped tickets if it doesn't solve the issue
19:15:43 <GeKo> sounds good.
19:16:11 <sysrqb> i can re-open one of the freetype-specific crash issues for collecting more info there
19:16:26 <sysrqb> but i think we have enough data for identifying the problem
19:17:59 <sysrqb> acat did you successfully deply the android instrumented test last week?
19:18:09 <sysrqb> *deploy
19:18:30 <acat> no, i think we'll need a machine that is kvm capable
19:18:36 <sysrqb> hrm
19:18:41 <sysrqb> nested kvm.
19:18:56 <sysrqb> okay
19:19:10 <acat> the latest arm-based images are for android 25
19:19:22 <acat> and i couldn't make my scripts work with those
19:19:46 <sysrqb> yeah, using an arm-bsaed emulator would be sadness
19:19:59 <acat> so i just made a MR with the scripts that work in my local setup, and i hope it's easy to make them work in a kvm-capable machine
19:20:12 <sysrqb> okay
19:20:18 <sysrqb> thanks
19:21:14 <sysrqb> that's tor-browser-bundle-testsuite!14
19:21:22 <acat> yes
19:21:48 <GeKo> acat: hey, regarding tests. did you move the test suite run away from your user?
19:22:18 <GeKo> to a role account, so we can debug and run things in a group?
19:22:54 <acat> GeKo: i will do next, i was planning to update the repo to use the latest scripts in that MR in the machine
19:22:59 <acat> with the tb-tester user
19:23:37 <GeKo> just before you are leaving would be fine :)
19:24:18 <acat> sure
19:25:17 <sysrqb> acat: did you try running a x86 emulator on the machine, even without kvm support?
19:25:26 <sysrqb> (maybe that fails, i don't remember)
19:25:37 <acat> yes, it fails
19:25:40 <sysrqb> okay
19:25:58 <acat> i did not try to run an old x86 image actually, that might work
19:26:27 <acat> but i think it will probably be still too slow
19:26:38 <sysrqb> yeah
19:26:49 <boklm> should I be added to the tb-tester ldap group?
19:26:59 <sysrqb> boklm: yeah, I can add you
19:27:06 <sysrqb> (or, request you be added)
19:27:09 <boklm> thanks
19:28:52 <sysrqb> This week we re-start the process of rebasing onto Mozilla beta
19:29:03 <sysrqb> I'll open the usual tracking tickets for that
19:29:43 <sysrqb> boklm: GeKo usually works on moving tor-browser-build onto the newer toolchain
19:29:55 <sysrqb> but we're hoping you can start taking on that responsibility
19:30:05 <boklm> ok
19:30:38 <GeKo> i can help you with that explaining where to look and what worked so far
19:30:46 <sysrqb> i think he'll be happy to help you with the tricks he's learned
19:30:49 <GeKo> and where we still have open issues in the process
19:30:57 <GeKo> yeah
19:31:23 <pospeselr> what all is being upgraded?
19:31:31 <sysrqb> "Everything"
19:31:44 <boklm> is the rebase for both desktop and android?
19:31:58 <pospeselr> oh fun
19:32:16 <sysrqb> pospeselr: it varies with every version, but it could be any part of the toolchain that mozilla update
19:32:32 <sysrqb> boklm: the priority is getting android builds
19:32:41 <GeKo> boklm: the toolchain update is mostly for android
19:32:53 <GeKo> but i have a branch for linux, too
19:33:01 <sysrqb> ^
19:33:10 <GeKo> which i keep updated for the time in the future where we are at the point to actually merge it
19:33:12 <sysrqb> I hope we don't hit any blockers for moving Linux
19:33:28 <GeKo> and, more importantly, to have something for the test suite to run
19:33:29 <pospeselr> got it
19:34:50 <sysrqb> I'll open new issues for this works and assign them
19:35:01 <boklm> so android nightly builds are based on beta, and desktop still on esr but we plan to move them to beta too at some point?
19:35:17 <sysrqb> yes, that's correct
19:35:30 <sysrqb> hopefully that happens before we need to transition to the next esr, too
19:35:53 <sysrqb> but I still like the idea of having a single stable tor-browser git brnach, and a single alpa
19:36:00 <sysrqb> *alpha
19:36:19 <sysrqb> and not maintaining four branches on two very different versions of firefox
19:36:31 <sysrqb> (increasingly different)
19:38:09 <sysrqb> okay, GeKo, i think you added the discussion topic on moving the nightly builds?
19:38:50 <GeKo> yes
19:39:00 <GeKo> i think we should start moving the nightly builds over
19:39:08 <GeKo> at the begin of this cycle
19:39:12 <GeKo> if there are no blockers left
19:39:14 <sysrqb> wfm
19:39:29 <GeKo> which i think we don't have?
19:39:34 <GeKo> is that right, boklm?
19:39:43 <boklm> yes, I think everything is ready
19:39:47 <GeKo> great
19:39:54 <sysrqb> i saw boklm's nightly build had some issue with building win64 yesterday (?), while the new serer did not
19:40:03 <sysrqb> maybe that was a temporary issue
19:40:18 <sysrqb> but seeing the new server succeed was nice
19:40:22 <GeKo> so one thing to consider is that we likely need the old build system and the new one in parallel for the time being
19:40:27 <boklm> yes, it looked like a temporary issue but I did not investigate much
19:40:41 <sysrqb> boklm: no problem considering we have the new one
19:40:43 <sysrqb> :)
19:40:47 <GeKo> i am not sure if everyone who is building releases has the resources for that
19:41:04 <GeKo> we have the sunet machine which can still do stable builds
19:41:19 <GeKo> but won't do nightlies and alphas based on the new system
19:41:40 <GeKo> and we have the tpo machine(s) which can only do the new thing
19:41:52 <GeKo> so that could be one pair for release building
19:42:02 <GeKo> i am not sure what else we have
19:42:55 <sysrqb> ah, i see. yes
19:43:23 <sysrqb> boklm said he could begin building again, too
19:43:26 * boklm has a local machine that can do both the new and old system
19:43:30 <GeKo> we *could* think trying to use the new system for stable, too, at some point
19:43:48 <GeKo> but i'd like to have at least some alpha cycles first
19:44:02 <sysrqb> in theory, boklm and I could build stable, if needed
19:44:05 <GeKo> i guess ideally we just use the new one for 10.5 onward for desktop
19:44:15 <sysrqb> yep
19:44:17 <GeKo> well, i can build, too
19:44:25 <GeKo> but i only have access to the tpo machines
19:44:38 <sysrqb> and sunet?
19:44:42 <GeKo> that is my dev machine i had can't do full builds anymore
19:44:43 <boklm> does the stable still need wheezy?
19:44:51 <GeKo> yes sunet
19:44:57 <GeKo> and the new tpo ones
19:45:07 <GeKo> boklm: yes
19:45:14 <GeKo> for centos 6
19:45:29 <boklm> ah, I think the new system doesn't work with wheezy
19:45:34 <GeKo> yep
19:45:49 <GeKo> we'd need sacrifice centos 6 here, too
19:45:52 <boklm> (with wheezy containers)
19:46:05 <GeKo> which would be a bit shitty in the midst of a series
19:46:21 <GeKo> in particular as we said 10.5 will be the first one without centos 6 support
19:46:23 <sysrqb> that would be a large change for a stable series
19:46:32 <GeKo> yeah
19:46:34 <sysrqb> so i would definitely prefer waiting until 10.5 stable
19:47:19 <GeKo> okay
19:47:35 <GeKo> so, i'll review the remaining patches for the switch and merge them this week
19:47:44 <GeKo> and then send an announcement mail to tbb-dev
19:47:48 <GeKo> and we are done :)
19:47:51 <GeKo> sounds good?
19:47:54 <boklm> sounds good
19:47:58 <sysrqb> sounds good to me
19:48:08 <GeKo> great
19:49:09 <sysrqb> really good work on getting this working
19:49:15 <sysrqb> congrats aboklm and GeKo
19:49:16 <GeKo> sysrqb: i might be able to do the alpha publishing and hopefully merge the publishing instructions, too
19:49:24 <sysrqb> err, boklm
19:49:30 <GeKo> yeah, it's been boklm's baby
19:49:31 <boklm> thanks
19:49:33 <GeKo> nice work!
19:50:12 <sysrqb> GeKo: sure, sounds good to me
19:50:14 <sysrqb> thanks
19:50:19 <GeKo> sysrqb: so, if you give me signed bundles i can plan to pick that up
19:50:25 <sysrqb> cool
19:51:55 <sysrqb> Jeremy_Rand_Talos: did you have any documentation about what changes you needed to make for working on cirrus CI?
19:53:29 <Jeremy_Rand_Talos> sysrqb, most of it is pretty non-invasive.  Added a few lines to the container start/finish commands in rbm.conf to support saving/restoring container filesystems, and added a few lines to some toolchain build scripts to support resuming from a halfway completed build
19:53:53 <sysrqb> neat
19:54:11 <Jeremy_Rand_Talos> sysrqb, I will try to write up more detailed docs, this literally first worked Sunday morning
19:54:25 <sysrqb> ok, no worries or rush
19:54:30 <sysrqb> i was just curious
19:54:33 <sysrqb> congrats on getting it working
19:55:07 <SomeHacker> > 22 February 20201
19:55:11 <SomeHacker> weird, fix the topic
19:55:19 <Jeremy_Rand_Talos> sysrqb, thanks :)  Note that Namecoin doesn't use Rust, so I didn't patch the Rust build script, but the same thing I did for clang should work in Rust
19:55:26 <boklm> Jeremy_Rand_Talos: is that without using CRIU?
19:55:36 <Jeremy_Rand_Talos> boklm, right, it's not using CRIU
19:56:33 <Jeremy_Rand_Talos> boklm, it's a really dumb implementation of checkpoints that abuses the fact that you can send SIGINT to GNU Make, run make again, and it'll resume where it left off
19:57:08 * sysrqb TIL
19:57:20 <sysrqb> Lastly, huge thanks to acat for all of their work in tor browser
19:57:51 <acat> thanks, it was a pleasure :)
19:57:52 <sysrqb> I have we see you around here after this week
19:57:58 <sysrqb> *I hope
19:58:06 <Jeremy_Rand_Talos> +1
19:58:10 <GeKo> +1
19:58:12 <pospeselr> +2
19:58:12 <boklm> +1
19:58:46 <acat> sure, i hope i can somehow keep contributing  :)
19:59:42 <sysrqb> Good luck with your next position :)
19:59:54 <acat> thanks!
20:00:02 <sysrqb> okay, I think that wraps up this weekly meeting
20:00:04 <sysrqb> thanks everyon!
20:00:07 <sysrqb> +e
20:00:10 <sysrqb> #endmeeting