19:00:12 <GeKo> #startmeeting tor browser
19:00:12 <MeetBot> Meeting started Tue Feb 20 19:00:12 2018 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:12 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
19:00:16 <GeKo> hi all!
19:00:40 <GeKo> this week is the meeting everyone was longing for i guess :)
19:00:46 <mikeperry> yay GeKo is back! and no chemspills in the interim :)
19:00:49 <mcs> hi
19:00:57 <sysrqb> o/
19:01:01 <sysrqb> heh
19:01:02 <pospeselr> good morning!
19:01:02 <boklm> hi
19:01:12 <GeKo> the pad is, as usual, at https://storm.torproject.org/shared/tHoN4Ii7rLSjPE0OP4gydX4cMGadsXmRQNc-6lwru0N
19:01:20 <isabela> !
19:01:27 <GeKo> please fill in your items and start reading over the ones already written
19:01:37 <GeKo> mark those you want to discuss in bold
19:04:47 <arthuredelstein> hi everyone!
19:06:25 <GeKo> alright
19:06:30 <GeKo> let's start
19:06:36 <GeKo> it seems tjr is first today
19:06:51 <isabela> GeKo: is there a discussion part of the pad?
19:06:53 <tjr> I fill in my stuff over the course of the week and then everyone fills in below me :(
19:07:00 <GeKo> isabela: not yet :)
19:07:03 <isabela> hehe
19:07:16 <GeKo> it seems there is not much for that up yet this week
19:07:17 <tjr> So in -central we have a patch that 'fixes' the keyboard layout leak
19:07:39 <tjr> Except it turns out it doesn't work well and still leaks (Bug  https://bugzilla.mozilla.org/show_bug.cgi?id=1438795)
19:08:08 <tjr> And it turns out it breaks Copy/Paste shortcuts on what I presume are (some?) non-US-lang keyboards: https://bugzilla.mozilla.org/show_bug.cgi?id=1433592
19:08:17 <tjr> People asked that it be backed out because of that actually
19:08:39 <tjr> This seems like something we need to get fixed so ESR meets TB's acceptence criteria
19:09:06 <tjr> Except I don't know anything about this keyboard layout stuff, so I'm hoping someone can walk me through it and/or help with it.  Not during the meeting I don't think though.
19:09:53 <GeKo> i think arthur would indeed be the candidate for doing that
19:10:22 <tjr> Cool
19:10:29 <tjr> Next one: the MinGW Build on -central doesn't run.
19:10:38 <tjr> This is also important to get fixed for TB :)
19:11:05 <tjr> I looked below about the Vista debugging Richard did
19:11:17 <tjr> Am I correct in thinking it was all printf-based? No windbg or gdb-on-windows?
19:11:20 <arthuredelstein> (glad to discuss keyboard layout stuff right after this meeting)
19:11:38 <pospeselr> tjr: yeah apparently there's another flag i need to pass to my mozconfig for symbols not get stripped out
19:11:46 <GeKo> tjr: so i had the feeling dmajor helped you with that
19:11:57 <pospeselr> so i've been doing traces with env variables like a chump
19:11:59 <GeKo> where are you and what help would you need?
19:12:29 <tjr> Yea, dmajor has been really helpful in showing us _where_ the crash is, and everyone once in a while what it is. But basically I've been really lucky so far in that him, jacek, or someone else does a driveby with "Oh this is the problem"
19:12:43 <tjr> And now my luck's run out I think. So I know where it's asserting, but no idea why.
19:13:33 <GeKo> okay. did you do some debugging with gdb so far?
19:13:45 <GeKo> like using a non-debug build with symbols
19:13:54 <tjr> So I've never been able to get gdb on windows to work, so I'm curious if you can!
19:14:08 <GeKo> yes
19:14:12 <tjr> I use a mingw gdb.exe I download (not one from cygwin)
19:14:31 <tjr> But the error I get early on is https://searchfox.org/mozilla-central/source/xpcom/build/PoisonIOInterposerBase.cpp#226
19:14:31 <GeKo> there are precompiled things, yes
19:15:08 <tjr> Which I believe is entirely unrelated to anything I'm working on, and is just some debugging helper thing that assumes if you're debugging on Windows you're not doing something weird like mingw
19:15:25 <tjr> Have you hit that before? It seems old enough that it's in esr52 too
19:15:28 <pospeselr> weird, i did not hit that with mingw gdb
19:15:41 <pospeselr> everything 'worked' just didn't have symbols
19:15:54 <pospeselr> built with rbm nightly-windows-i686
19:16:00 <pospeselr> granted this is on Vista
19:16:29 <GeKo> tjr: no. so, that's with --enable-debug?
19:16:42 <tjr> Yes. You know I bet it's something that's only in debug builds.
19:16:54 <GeKo> what happens with --disable-debug?
19:17:05 <GeKo> that would be my first start to test
19:17:30 <GeKo> given that the bug you are working on is orthogonal to debug/non-debug
19:18:26 <tjr> Yea. I just set up opt non-debug builds in Tc today so it'll be easy to get builds
19:18:41 <GeKo> but, yes, i can try to help if you are still stuck
19:19:15 <tjr> Dumb question but might save me some time, pospeselr can you pastebin a printf line?  e.g. are you fprintf to a file or printf to a console, or something?
19:19:44 <pospeselr> lol yeah one sec
19:20:00 <tjr> Other than that, there is a draft help page for Mozilla's Fingerprinting Protection page, if anyone wants to read/comment on it
19:20:08 <tjr> That's it for me.
19:20:28 <pospeselr> tjr: https://pastebin.com/ZpBZhCN2 basically paste this at the top of whatever module I'm in and go
19:20:43 <GeKo> mcs: you are up
19:21:19 <mcs> I am not sure why the #19910 text is bold...
19:21:25 <GeKo> tjr: i've been using a log module recently and then doing MOZ_LOG() where needed
19:21:40 <GeKo> iirc i looked for an easy example on MDN and it worked for me
19:21:41 <mcs> So I don’t really have anything to discuss.
19:21:44 <arthuredelstein> I wanted to ask about #19910
19:22:02 <arthuredelstein> it seems this likely causes a performance regression, right?
19:22:23 <GeKo> it could be, yes
19:22:44 <arthuredelstein> So wondering if there were thoughts about what we can do. Should we be working on #5915? Or is there a way to fix the tor browser patch so it no longer has a race condition?
19:24:09 <GeKo> arthuredelstein: i think comment:22 in #19910 is the way to go
19:24:24 <GeKo> the backout won't hit stable before august
19:24:38 <GeKo> and meanwhile we should have a 0.3.4 with this stuff fixed
19:25:11 <arthuredelstein> 0.3.4 with a 5915 fix?
19:25:36 <arthuredelstein> Sounds good to me, thanks
19:25:52 <GeKo> well, that would be my plan
19:26:18 <GeKo> i think we should ask networking people about it and make sure it's on their radar
19:26:35 <mcs> Agreed. I’d feel better if an owner was assigned :)
19:26:42 <arthuredelstein> right
19:27:12 <GeKo> true, but i am not sure how the networking folks are dealing with that
19:27:16 <arthuredelstein> I do wonder if the 3875 patch might be fixable (I haven't looked carefully), because the proposed 5915 fix also has potential problems
19:27:21 <GeKo> maybe a keyword is enough?
19:27:49 <arthuredelstein> So we might want to look into that in parallel with the network team
19:27:59 <GeKo> i bet a proper fix for #3875 is pretty complicated
19:28:14 <GeKo> from the browser side that is
19:28:26 <GeKo> moreover, it's not just a thing the browser wants
19:28:35 <GeKo> but other applications as well
19:29:21 <GeKo> arthuredelstein: i guess we could ask mozilla's necko folks about what they think?
19:30:03 <arthuredelstein> we could. I actually would prefer a 5915 fix for the reason you give. But I worry it has bad UX.
19:30:18 <GeKo> how so?
19:31:10 <arthuredelstein> Well, the most naive solution is just shutting the down the socket if the server doesn't respond. But the browser reacts not by showing an error message, but seemingly waiting forever.
19:31:15 <arthuredelstein> At least that's what happened when I tried it.
19:31:53 <sysrqb> yeah,ideally the application explicitly supports sending opportunistic data and handles the failure gracefully
19:32:14 <sysrqb> we had it in the old version of torsocks, too
19:32:21 <arthuredelstein> We could serve our own error message HTML page from the tor proxy, but that requires design, translations, etc. And then we're back to a browser-specific solution again.
19:32:22 <sysrqb> but we haven't added it into the new version yet
19:32:33 <sysrqb> because it's also a pain implementing it transparently
19:32:46 <GeKo> hm, i see
19:33:10 <arthuredelstein> sysrqb: Maybe there's a way for tor to send a protocol-level error message to the browser such that it correctly display a "failed to connect" error?
19:33:26 <arthuredelstein> That would be ideal rather than duplicating the browser's error page in the tor proxy.
19:33:43 <sysrqb> arthuredelstein: maybe? that could work
19:33:55 <sysrqb> but tor won't know all the protocols it's proxying
19:34:06 <sysrqb> so that won't scale well, in general
19:34:25 <sysrqb> but i guess it could support all the official Tor Project applications?
19:34:30 <arthuredelstein> I guess I'm not sure when a socket closes if there's any way to return an out-of-band error message. But this is probably a discussion for after the meeting.
19:34:37 <sysrqb> but that could get messy...
19:34:57 <sysrqb> i think the control socket is the way to detect that
19:34:58 <GeKo> arthuredelstein: agreed
19:35:06 <sysrqb> sure
19:35:10 <tjr> Could it return an error in a message in the socks protocol. A message backwards compatible such that correctly-implementing socks protocol clients will ignore it?
19:35:45 <sysrqb> tjr: nah, everything after the handshake completes is only application protocol
19:35:50 <sysrqb> socks is done after that :/
19:35:57 <GeKo> okay, i have reviews marked bold
19:36:08 <GeKo> i need two small ones for tomorrow to get the release going
19:36:22 <GeKo> it's for #25000 and #25215
19:36:58 <mcs> Kathy and I tried to review #25000 but so far we can’t make the problem go away even when NoScript is disabled :(
19:36:59 <GeKo> boklm: and if you could look at #22612 that would be good as well
19:37:08 <boklm> GeKo: ok
19:37:21 <GeKo> mcs: do you have installed?
19:37:31 <GeKo> because that's what i needed
19:37:48 <GeKo> it seems giorgio fixed some edges cases i hit
19:38:12 <mcs> Yes but shouldn’t the test add-on work if NoScript is disabled? (I fear I don’t understand the bug although it seems somewhat simple).
19:38:48 <mcs> Anyway, Kathy and I will keep looking at #25000
19:39:02 <GeKo> aha! so what i tested is the https-everywhere preferences on about:addons
19:39:10 <GeKo> they are fixed with the patch
19:39:15 <mcs> Okay. We can test that :)
19:39:28 <GeKo> and i duped that over because i thought those were the same issues
19:39:41 <mcs> (and see if more than one problem is going on)
19:39:44 <GeKo> i guess i'll look again at the extension
19:39:45 <GeKo> thanks
19:39:51 <GeKo> sysrqb: you are up
19:40:26 <sysrqb> Okay, last week we briefly started discussing the future of Orfox
19:40:49 <sysrqb> and whether we should start releasing Orfox in parallel to Tor Browser
19:40:58 <sysrqb> while we continue working on Tor brwoser for Adnroid
19:41:07 <sysrqb> there is some risk here
19:41:15 <sysrqb> in that any bugs in Orfox become our problem
19:41:33 <sysrqb> but the current roadmap has TBA release in July/August
19:41:42 <sysrqb> which is not soon
19:42:08 <sysrqb> and the Guardian Project will be maintaining Orfox until then
19:42:19 <sysrqb> (assuming we don't take over)
19:42:25 <GeKo> how much work is involved in that?
19:42:42 <sysrqb> I don't think it is much work
19:42:53 <sysrqb> I think we can rebase the current patches onto tor-browser.git
19:43:07 <sysrqb> and we can create our own releases
19:43:16 <sysrqb> they won't be reproducible
19:43:29 <GeKo> well we need some build setup and need access to the google app store etc
19:43:33 <sysrqb> but maybe we can easily identify what causes differences and account for that
19:43:49 <sysrqb> yes, i think we can get that information form n8fr8
19:44:00 <sysrqb> but i haven't talked with him yet
19:44:08 <sysrqb> i wanted the team's opinion first
19:44:17 <GeKo> let's do that in rome and then come up with a plan?
19:44:24 <sysrqb> sure, we can do that
19:44:29 <sysrqb> that is soon anyway
19:44:35 <GeKo> yep
19:44:43 <sysrqb> cool
19:44:58 <GeKo> ok. igt0 you are up
19:45:41 <igt0> hey, so are we going to disable the sync background service in the TB mobile?
19:45:47 <GeKo> (pospeselr: i assumed i answered your question; feel free to ping me if there is still stuff needed)
19:46:33 <pospeselr> GeKo: yep I'm good, trying it out now :)
19:46:44 <GeKo> sysrqb: i forgot to ask: what's up with the orfox patch merge into tor-browser?
19:47:00 <GeKo> it seems we are close for a couple of weeks now :)
19:47:45 <sysrqb> GeKo: ah, yes, they are done, and working on ESR52
19:47:55 <GeKo> igt0: well, i guess we want to do the same as we do on desktop to start with
19:48:05 <sysrqb> GeKo: but we moved the goal post :)
19:48:23 <sysrqb> now we will probably skip ESR and go straight to nightly
19:48:50 <GeKo> ok. we had this as deliverable in our DRL proposal
19:48:52 <sysrqb> but I stopped working on that until igt0 and I finish the mobile design doc/roadmap
19:49:07 <GeKo> and i just want to make sure this is not falling through the cracks
19:49:21 <sysrqb> hmm
19:49:41 <GeKo> what would it hurt having those patches in tor-browse.git
19:49:45 <sysrqb> okay, so wehould we finish esr rebase and merge that before moving onto m-c?
19:49:57 <GeKo> especially if we are thinking to provide orfox maintenance
19:50:05 <sysrqb> i think only they're not reproducible
19:50:17 <sysrqb> i don't see much other harm
19:50:35 <GeKo> you mean they would cause reproducibility issues for desktop?
19:50:42 <sysrqb> no, only mobile
19:50:47 <GeKo> well, that's okay
19:50:57 <sysrqb> they don't touch much oustide mobile/
19:51:02 <sysrqb> *outside
19:51:07 <sysrqb> okay
19:51:18 <sysrqb> i can work with igt0 and finish that
19:51:18 <igt0> GeKo, okey, so the https://bugzilla.mozilla.org/show_bug.cgi?id=942652 is a requirement for TB mobile.
19:51:42 <GeKo> so, i think if it's not too much work, post a patch in the ticket and we just merge them to tor-browser and are done
19:51:57 <sysrqb> okay, sgtm
19:51:57 <GeKo> and can make a check behind the deliverable
19:52:08 <GeKo> and we would be prepared to take over orfox
19:52:30 <sysrqb> okay
19:54:03 <GeKo> igt0: i think i am not following. let's talk about that after the meeting?
19:54:14 <igt0> GeKo, yep
19:54:28 <GeKo> arthuredelstein: you are up
19:55:22 <arthuredelstein> I wanted to ask if anyone has any particular comments about patches remaining for uplift. See list at https://torpat.ch/short
19:55:35 <arthuredelstein> One question I have: Do we want to try to get 1382359 (our 21321) done for ESR60?
19:55:56 <mcs> There is no need to uplift #22614 (Mozilla already removed the API in Firefox 56 that we pref’d out).
19:56:19 <GeKo> arthuredelstein: yes, iam working on that one
19:56:22 <arthuredelstein> mcs: thanks, good to know. So we won't need it in TBB/ESR60 either, right?
19:56:28 <GeKo> i have a patch that is working
19:56:31 <mcs> correct
19:56:37 <arthuredelstein> GeKo: OK, sounds good.
19:56:44 <GeKo> i just need to find the time to write the additional tests
19:57:06 <GeKo> my plan is to get this finally done this week
19:57:21 <GeKo> and then rebased onto m-c
19:57:21 <arthuredelstein> Another question for mcs and brade is that in my hands the unit test for #18912 seems to be failing when I rebase it to mozilla-central.
19:57:45 <arthuredelstein> It's looking for a certain error code, but maybe that's changed. I don't know where the error code is used by us, though.
19:59:14 <arthuredelstein> We can maybe discuss that later though.
19:59:18 <GeKo> sounds good
19:59:22 <GeKo> isabela: you are up
19:59:26 <isabela> ok!
19:59:38 <isabela> really quick - i think we should start organizing our life in rome!
19:59:39 <isabela> :)
19:59:56 <isabela> specially for team meeting day, so i want to create pads for agendas etc
20:00:08 <GeKo> thanks, please do
20:00:09 <isabela> just want to let folks know and i can continue this via email too
20:00:50 <isabela> that was it :)
20:01:00 <arthuredelstein> do we have a schedule document for Rome somewhere?
20:01:07 <tjr> Does everyone know the mozilla attendee list? ethan, tim, gary, steven e., patrick mcmanus
20:01:20 <isabela> arthuredelstein: we do for the team meeting day
20:01:24 <isabela> let me get it for you
20:01:24 <GeKo> okay, regarding the group nick discussion tbb-team sounds fine to me
20:01:30 <GeKo> tjr: no, thanks
20:02:02 <GeKo> tjr: you meant tom, not tim right?
20:02:07 <GeKo> :)
20:02:20 <isabela> arthuredelstein: https://ethercalc.org/zil3cf5nm9ge
20:02:22 <tjr> No; Tim. As in I'm not going, but Tim Huang is
20:02:27 <arthuredelstein> isabela: thanks! :)
20:02:28 <isabela> i mean organize the agendas of these meetings
20:02:51 <GeKo> tjr: so what's the state of the folks from the uplift team
20:03:00 <GeKo> as i am seeing ethan on the list, too
20:03:12 <GeKo> (assuming it's the same ethan)
20:03:30 <arthuredelstein> same ethan, tim and gary :)
20:03:33 <tjr> I'm not entirely sure; but I think Ethan might be in the 'apply for a work visa' stage. I believe he's not allowed to work while apply for that.
20:03:55 <tjr> I think Ethan was going to be the guinea pig for the process, and tim and gary would follow?  Not 100% sure.
20:04:02 <tjr> But yea, all same people as Arthur said.
20:04:07 <GeKo> ok, neat
20:04:18 <GeKo> do we have anything else for today?
20:05:00 <GeKo> sounds not. thanks all for the meeting that lasted a bit longer than an hours :/ *baf*
20:05:04 <GeKo> #endmeeting