14:58:57 <richard> #startmeeting Tor Browser Weekly 2023-09-11
14:58:57 <MeetBot> Meeting started Mon Sep 11 14:58:57 2023 UTC.  The chair is richard. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:58:57 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:59:07 <donuts> o/
14:59:12 <ma1> o/
14:59:18 <jagtalon> o/
14:59:21 <dan_b> o/
14:59:23 <boklm> o/
14:59:23 <Jeremy_Rand_36C3[m]> Hi!
14:59:27 <clairehurst> o/
14:59:30 <PieroV> o/
14:59:35 <richard> meeting pad as usual: https://pad.riseup.net/p/tor-tbb-keep
14:59:48 <thorin> /o
15:00:29 <ma1> is it just me or gitlab is extremely slow
15:00:35 <richard> We can likely keep it short for today unless other folks have discussion points
15:00:41 <richard> gitlab *is* slow today per chat in #tor-dev
15:00:47 <anarcat> ma1: it's not just you
15:00:49 <richard> and also my own personal experience :3
15:01:06 <richard> anarcat: do you have an alert for 'gitlab'? :3
15:01:14 <anarcat> i do not
15:01:25 <anarcat> i have human alerting systems, pretty reliable
15:01:47 <ma1> (heard me yelling)
15:01:50 <richard> :3
15:01:51 <richard> ok
15:02:09 <richard> anyway, the big thing for this week is likely the 13.0a3 release
15:02:19 <ma1> 13.0a4?
15:02:21 <PieroV> 13.0a4
15:02:23 <richard> iirc we were waiting for tor-browser!721 and various android things
15:02:26 <richard> yes 13.0a4
15:03:19 <dan_b> android things: I think I got to all of clairehurst's comments last week so hopefully that'll be good to go early this week
15:03:22 <richard> i've already done the release prep pending tags so we're good there, short of changelog updates
15:03:46 <richard> boklm, dan_b: have either of you looked into the remaining nightly build failures?
15:04:00 <PieroV> richard: I think it's the Snowflake patch thingie
15:04:04 <PieroV> I have a MR for it
15:04:10 <richard> ack thx
15:04:21 <richard> donuts: macOS nightlies seemed to have built oveer the weekend if you were waiting on that btw
15:04:47 <PieroV> It's the one that completely removes parameters from Snowflake. I got an informal approval in #tor-anticensorship, but not a formal approval yet
15:04:47 <boklm> it's the tor-onion-proxy-library failure
15:04:54 <donuts> richard: thanks, pierov sent me a couple of builds on thursday but I'll install the latest and greatest
15:04:59 <boklm> (which is the snowflake thing)
15:05:03 <richard> ah ok
15:05:12 <clairehurst> dan_b: I think so too
15:05:19 * Jeremy_Rand_36C3[m] does have a brief thing to bring up today, see pad
15:05:36 <ma1> I'm linking to 13.04 relprep also cypherpunks1's tor-browser!727 and tor-browser!729 which I verified & merged in the weekend
15:05:45 <richard> very nice
15:06:01 <richard> someone over the weekend also submited an MR to fix the asserts in the debug build too \o/
15:06:26 <PieroV> richard: I'd like to comment a little bit on that
15:06:45 <PieroV> Not in this meeting, in the issue itself
15:07:17 <richard> wfm, i figured you'd have thoughts
15:07:20 <richard> :)
15:07:23 <Jeremy_Rand_36C3[m]> lol at the GitLab perf issues causing the IRC bot delay
15:07:38 <PieroV> I wouldn't merge it already. It's very interesting and happy someone investigated, but moving to 001-/000- seems a workaround to heal the sympthoms rather than the cause
15:09:11 <richard> I'm inclined to agree but from what i remember the 'right' fix would be rather complicated
15:09:46 <richard> anway I'm happy to punt it since it's not user-facing
15:10:07 <ma1> I also hope to fix tor-browser#42084 from hackerone before 13.0a4, since it's user-facing.
15:10:21 <ma1> (i.e. tonight or tomorrow morning, right?)
15:10:43 <richard> we wouldn't tart building until tomorrow at the earliest
15:11:06 <PieroV> There's also the Snowflake problem (the other one)
15:11:26 <PieroV> So, we have to see if the new tag comes out, or if we have to switch to the current HEAD of main before it's actually tagged
15:11:47 <PieroV> Because Snowflake on Android is broken on alpha (API level things)
15:13:14 <richard> for for everyone, as usual please prioritise ~Next and ~"13.0 Stable" labled issues
15:13:40 <richard> I'll see about updating the release calendar today to match reality
15:13:51 <Jeremy_Rand_36C3[m]> (should I go next or are other things still on the agenda for the meeting before I talk?)
15:13:54 <richard> Jeremy: I'll hand it over to you
15:14:00 <Jeremy_Rand_36C3[m]> ok
15:14:31 <Jeremy_Rand_36C3[m]> so, as requested, I asked the Whonix Team to test latest Tor Browser Alpha to see if the control port / Tor daemon refactors broke anything for them
15:14:53 <Jeremy_Rand_36C3[m]> They tested it out, and they didn't find anything obviously wrong, but they did see something weird that they wanted me to flag for you
15:15:13 <Jeremy_Rand_36C3[m]> Specifically, when they click the New Identity button, they no longer observe a NEWNYM signal getting sent to the control port
15:15:20 <Jeremy_Rand_36C3[m]> But, the exit IP does still change
15:15:40 <Jeremy_Rand_36C3[m]> Their assumption is that this is intentional, and that using NEWNYM was removed in favor of just cycling the SOCKS5 auth strings
15:15:58 <Jeremy_Rand_36C3[m]> But they wanted me to check with you on whether that is in fact what's going on, or if this is an unintentional behavioral change
15:16:03 <PieroV> It should not be, we still have the code to send it
15:16:12 <Jeremy_Rand_36C3[m]> Bug is filed at tor-browser#42085
15:16:32 <PieroV> Do they use transproxy, right?
15:16:35 <Jeremy_Rand_36C3[m]> PieroV: ah, interesting. Then I'm glad they did the testing.
15:17:05 <Jeremy_Rand_36C3[m]> PieroV: no, they don't use transproxy mode, they just use env vars to skip Tor launch and set a custom SOCKS and Control port
15:17:19 <Jeremy_Rand_36C3[m]> I think they use Unix sockets for both SOCKS and control port but I'd have to double check
15:18:00 <Jeremy_Rand_36C3[m]> (There is an optional feature in Whonix to switch to transproxy mode but that's not what they were testing with)
15:18:45 <PieroV> I can't load the issue (GitLab not working well)
15:19:12 <PieroV> But we have a logger for the new identity, you can raise its level with the dev pref browser.tordomainisolator.loglevel
15:19:21 <Jeremy_Rand_36C3[m]> PieroV: the GitLab issue mostly just says what I typed here, but also has a link to the Onion-Grater log file that Patrick saved in case it helps you
15:20:05 <Jeremy_Rand_36C3[m]> PieroV: ok, so should I ask them to re-run the test with that pref maxed out so that we can see what might be going wrong?
15:20:07 <PieroV> It might even be some race condition, I had one at a certain point
15:20:27 <PieroV> Yes please, but maybe we should wait tor-browser!721
15:21:27 <Jeremy_Rand_36C3[m]> PieroV: what's the relationship between 721 and this bug? (Sorry, I haven't been keeping track of which MR's are for what parts of the refactors)
15:21:37 <PieroV> This MR changes a lot of stuff
15:21:41 <PieroV> It's the actual refactor
15:21:45 <richard> (so after 13.0a4)
15:22:08 <PieroV> richard: we'd need to do a QA session before releasing 13.0a4
15:22:19 <richard> yes
15:22:20 <PieroV> So, after we've built, but before we actually release
15:22:34 <PieroV> Unless you want to send 13.0a4 on fast lane, and merge 721 in 13.0a5
15:22:38 <Jeremy_Rand_36C3[m]> OK. FWIW I think the Whonix guys would probably not mind running the test twice, if it has a chance of turning up more useful debug logs. So maybe I should ask them to test again now with the log level maxed out, and then again after that MR is included in Alpha?
15:22:56 <PieroV> We're at a week from Moz's tag. At this point a delay in 13.0 sounds very likely
15:23:01 <richard> do you think buliing 13.0a4 tomorrow afternoon/evening is a reasonable expectation?
15:23:17 <PieroV> We could do a 13.0a5 next Monday/Tuesday, after we receive the tags
15:24:04 <PieroV> Jeremy_Rand_36C3[m]: if they don't mind yes please, it'll be great to catch other regressions !721 might add
15:24:26 <Jeremy_Rand_36C3[m]> PieroV: got it, I'll relay the request to them
15:24:31 <PieroV> (of course, they're not intentional, but it's really a big change)
15:24:36 <PieroV> Thanks!
15:25:04 <Jeremy_Rand_36C3[m]> (and I'll thank them for pointing out that behavior even though they thought it was intentional -- good communication is important)
15:25:22 <richard> yes please :)
15:25:31 <PieroV> (it's the final piece of the big work I've done during all the Summer)
15:26:34 <thorin> note: still waiting uBO's update in mozaddons - https://github.com/gorhill/uBlock/releases/tag/1.52.0 - Put uBO's icon in nav bar by default
15:26:38 <Jeremy_Rand_36C3[m]> PieroV: dumb question: where do the logs get written when that pref is maxed out?
15:26:45 <Jeremy_Rand_36C3[m]> Is it written to stderr, or somewhere else?
15:26:48 <PieroV> To the JS console
15:26:52 <PieroV> Ctrl+Shift+J
15:26:53 <Jeremy_Rand_36C3[m]> Ah ok
15:26:58 <Jeremy_Rand_36C3[m]> great thanks
15:27:15 <PieroV> They're also written to stdout, but I don't remember if nightlies to that. Dev builds surely do
15:28:32 <Jeremy_Rand_36C3[m]> anyway that's all I wanted to bring up, so whoever else has things to say can take the floor :)
15:29:34 <richard> ok, I don't have anything else for you all today
15:29:42 <ma1> Are backports expected during the weekend?
15:29:54 <PieroV> ma1: not this one, the next one
15:30:06 <richard> i mean generally nothing's expected during the weekend ;)
15:30:19 <ma1> good, otherwise I'd have to do it in Tor's stall at the Global Gathering :D
15:30:35 <richard> could be a fun exhibit for the community
15:30:39 <PieroV> Moz's release is scheduled for September 26
15:30:40 <ma1> ikr
15:30:49 <Jeremy_Rand_36C3[m]> PieroV: oh, forgot to ask, what's the maximum log level for that pref?
15:30:56 <PieroV> Jeremy_Rand_36C3[m]: all
15:31:01 <richard> watch as our security engineer backports patches.. in front of a live studio audience!
15:31:02 <Jeremy_Rand_36C3[m]> ok thanks
15:31:28 <ma1> lol
15:32:04 <richard> ok have a good week everyone
15:32:09 <ma1> thanks, you too
15:32:10 <richard> see you all on IRC o/
15:32:15 <Jeremy_Rand_36C3[m]> thanks! Get some sleep PieroV :)
15:32:22 <PieroV> Thanks! o/
15:32:24 <richard> :)
15:32:26 <ma1> o/
15:32:30 <richard> #endmeeting