17:32:08 #startmeeting tor browser 6/24/2019 17:32:08 Meeting started Mon Jun 24 17:32:08 2019 UTC. The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:32:08 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:32:24 sysrqb2: :( 17:32:33 hi 17:32:37 hello 17:32:42 i am optimistic :) 17:32:44 hi 17:32:49 but it is taking longer than i hoped 17:32:58 the backups are promising 17:33:03 I had to go through laptop recovery a few weeks ago, took several days 17:33:10 https://storm.torproject.org/shared/tHoN4Ii7rLSjPE0OP4gydX4cMGadsXmRQNc-6lwru0N is teh pad 17:33:12 And then hardware failure, 2 weeks 17:33:24 please add your items and mark things bold you want to talk about 17:35:39 okay, let's go 17:36:06 it seems i am the only one with something bolded, so i'll be first 17:36:30 i thought about giving a summary about the all hands weeks from a browser perspective so that we are roughly on the same page at least 17:36:37 hi 17:36:42 sorry, got distracted, I'm here 17:37:12 hmm, it seems we have two "Week of June 24, 2019" on the pad 17:37:40 weird 17:37:47 looks like it got copied down 17:38:18 fixed 17:38:19 okay 17:38:35 i sent a mail about the fennex/fenix transition to eveyone with my sessions notes 17:38:52 pili: i just fwded that one to you as i forgot to add you in the first place 17:39:03 that was one of the bigger sessions 17:39:32 we set down with tjr and the uplift team and went over the fingerprinting bugs and prioitized them + thought about potential solutions 17:39:47 it's not clear yet how much time the uplift team will have to work on those 17:40:02 (I added one minor bold item for after, btw) 17:40:09 but if there is time we have a plan and priorities which is good 17:40:37 it realized that not all of our fingerprinting bugs are on file in bugzilla (in particular the recent locale related ones i filed) 17:40:48 so, i guess we should fix that at some point 17:41:10 but, anyways, it should not affect the priorities and thus is not so urgent 17:42:00 the other big meeting was about the future of first-party isolation (fpi) and enhanced tracking protection (etp) which made me a bit nervous 17:42:28 because i did not have the feeling that mozilla thought fpi was worthwhile and focused on etp instead 17:43:04 i was very pleased to see that i was wrong and work is under way inside of mozilla to use (dynamic) fpi together with etp 17:43:51 i promised to help with testing and filing bugs where we know about fpi breaks 17:44:10 so that we can test whether the dynamic version helps 17:44:27 dynamic = relaxing the requirements a bit for those things that would break with fpi 17:45:16 the interface for etp in firefox is currently confusing as that option is under the privacy options but is talking about performance as well 17:45:34 and it's not exactly clear what is disabled where in the particular options 17:45:45 What mechanism is used to decide to relax the requirements? Is that based on a list of domains or ? 17:46:02 however, that got me thinking and i feel if we communicate that properly then using etp for _performance_ reasons is fine with me 17:46:42 mcs: not a list, i think some heuristics. however, i'd need to look up exactly how this is supposed to work 17:47:12 what is the timeline for enable it in tor browser? 17:47:14 GeKo: OK; thanks. Using heuristics sounds promising 17:47:29 it sounded similar to Webkit/Safari's ITP, relying on user "interaction" with a frame from a 3rd party 17:47:34 but it wasn't clear 17:47:40 anyway, i filed #30939 for the general idea 17:47:41 i'm not sure they know the heuristic yet 17:47:46 input is welcome 17:47:51 antonela: there is no timeline yet 17:47:56 we may want to figure out how two shields will co-live at the same chrome 17:48:03 but i think starting to work on that this year is not unreasonable 17:48:12 i also linked some ux/ui docs regarding that feature 17:48:35 i then had a chat with luke who is responsible for wasm 17:48:36 cool, we can talk about it during the dev meeting 17:48:51 and i think we can enable wasm in tor browser 9 17:48:58 and bind it to the sec settings 17:49:08 and we found a way to allow extensions to use wasm 17:49:19 which "just" needs a patch :) 17:50:07 sysrqb2 showed some first work on using selenium for tor browser performance measurements which looks promising 17:50:24 as a result i filed #30392 17:50:30 err 17:50:32 #30932 17:50:46 i think that's all from me for that 17:50:53 questions? amendments? 17:51:02 sysrqb2: tjr: ^ 17:51:15 neat. thanks GeKo 17:51:23 i don't think i got the fennec/fenix email, could i? 17:51:56 oh, indeed, sorry 17:52:03 let me fwd it to you as well 17:52:10 thank you 17:53:42 pili: you are next 17:53:45 hi 17:54:08 I just wanted to check that everything is on track for the fundraising banner to be released with the next "planned" release :) 17:54:15 I believe antonela provided some assets today 17:54:34 yes, we should figure out who will implement the changes 17:54:46 yes, i did 17:54:52 i guess we should do that this week 17:55:03 to have time for review and translations 17:55:04 and emmapeel "did something" :) so that the copy will be in the translation memory 17:55:25 let me dig up what she did so we're all on the same page 17:56:34 oh yeah, she added it to the tweet file 17:56:43 so I'm hoping it's being translated... :) 17:56:57 what is a tweet file? 17:57:47 I guess it's a file they have with a list of tweets to be translated? 17:57:58 but apparently this allows the string to be added to the translation memory 17:58:03 well, i don't know :) 17:58:34 me neither ;) 17:58:35 emmapeel is afk until 6th July 17:58:53 okay 17:59:04 but apparently this should allow it to be translated and picked up while she is away 17:59:09 acat: can you put adding the banner on your list for this week? 17:59:16 ready for the time of the release (fingers crossed... :/) 17:59:30 i'd do it otherwise but i really want to get the esr68 things reviewed 17:59:45 so we have a first pass done of a lot of the parts 18:00:16 GeKo: sure 18:00:25 great 18:00:42 i guess we can reuse a lot of the logic for our yearly donation banners 18:00:59 although this one is not as sophisticated it seems 18:01:26 so, maybe look at the revert of the last donation banner changes earlier this year in torbutton 18:01:40 and get some inspiration 18:01:57 There is a “Tor Project tweets” resource on Transifex. 18:02:01 boklm: so what do we plan to do with #28672? 18:02:17 should we just land what we have so we can think about nightly builds? 18:02:31 or do you want to track down the repro issues first? 18:02:43 mcs: that sounds like the one ;) 18:04:16 ah, I missed the updates on #28672. I should look at the repro issues this week. 18:04:34 okay, please do 18:05:01 regarding toolchain issues: i think a good that would be figuring out how we can produce 32bit mar-tools 18:05:06 that's important 18:05:17 and i am not sure how best we can do that 18:05:26 ideally, we'd use the build process we have right now 18:05:27 do we have a ticket for this? 18:05:52 in which cases do we need 32bit mar-tools? 18:06:32 not a separate one. i collect all the linux issues in #30321 right now 18:06:37 see last commit 18:06:54 without 32bit mar-tools there is no release signing at the moment 18:07:02 ah ok 18:07:05 hence the importance 18:07:14 I can look at this 18:07:22 thanks 18:07:31 feel free to file a new ticket for that if you want 18:07:43 ok 18:08:00 up to now it was so hackish that i was not sure whether to file a ticket or just fix it while doing the transition work 18:08:33 okay, other status updates, comments, questions? 18:10:15 boklm: linux_esr68_v7 is the latest tor-browser-build branch i have 18:10:27 let's move on to discussion then 18:10:28 ok 18:10:31 i have two items 18:11:35 apparently users are starting to use tor browser on a prerelease macOS 10.15 version 18:11:38 which is great 18:11:55 the drawback is that this is currently busted due to signing changes 18:12:24 #30126 is the current bug on our side 18:12:30 there are many pieces to it 18:12:58 part if it might be a firefox bug (https://bugzilla.mozilla.org/show_bug.cgi?id=1556733) 18:13:25 but i fear the new scheme will be enforced in the version that will be releases later this year 18:13:57 and if so, it seems we need to upgrade our macos version on the signing machine or need a newer one 18:14:08 i think at least 10.13.5 18:14:41 and i am not sure whether it would be working at all as it seems we need to be able to reach a timestamping server while signing 18:14:59 which is currently not possible in our setup as we locked it down that way 18:15:05 so, we need a plan 18:15:38 currently the one i have contains 3 pieces 18:15:49 1) we need to investigate the exact requirements 18:16:06 2) we need to think about getting those working in our signing setup 18:16:32 3) we need to think about a fallback solution in case we don't get it working in time 18:17:07 3) could involve having someone just a properly set up macOS system to be able to sign bundles in the mean time 18:17:40 comments? suggestions? 18:17:53 mcs: brade: my plan was to put you on 1) 18:18:00 I don’t know a lot about this, but it seems like notarization is something you do after the “traditional” gatekeeper signing. 18:18:12 as i assume you need to investigate that stuff for your other work anyway 18:18:25 do we know when this macOS version will be published as stable? 18:18:38 probably end of september 18:18:47 GeKo: Sure, we can take a look. 18:18:51 so we have like three months during esr transition 18:18:58 I think you can disable timestamp server option 18:19:03 mcs: yes, but not as a separate thing 18:19:12 But that may have changes since I last checked 18:19:28 dunno, might be a first stopgap if possible 18:20:42 I assume the signing machine is not connected to the network, but notarization requires sending bits to Apple. 18:21:02 (but I am just starting to learn about this too) 18:21:05 yes 18:21:10 i started with https://developer.apple.com/documentation/security/notarizing_your_app_before_distribution 18:21:40 https://bugzilla.mozilla.org/show_bug.cgi?id=1556733 leads to all the other bugs in bugzilla 18:22:04 concerned with the new signing scheme and might provide further reading material 18:22:11 GeKo: Kathy and I saw those bugs but haven’t made time to dig in. 18:22:20 no worries 18:22:30 but we should start with that plan now-ish 18:22:53 to have some time to think about workaround if needed 18:23:06 *workarounds 18:23:17 okay, that was the first platform related item 18:23:22 the other is concerned about windows 18:23:27 Do we need this for Tor Browser 8.x or “just” the esr68-based one? 18:23:32 Tming is close.... 18:23:54 Timing 18:24:01 i suspect we need it for the last esr60 version, too 18:24:10 GeKo: OK; thanks. 18:24:29 but i suspect as well that we'd get all the necessary things from mozilla as they'd need it for esr60, too 18:24:41 but i might be wrong on both accounts 18:25:13 so windows 18:25:44 we get lately more and more bug reports that tor browser on windows is not working in some capacity 18:26:02 like bookmarks don't get saved and it does not start properly 18:26:40 the interesting part is that a lot of those problems can be solved if moving the bundle to somewhere else 18:26:45 this is on windows 10 fwiw 18:27:08 i think that happens because ms is starting to lock down folders for read-only access 18:27:13 for security reasons 18:27:18 and the desktop is one of them 18:27:44 relevant ticket #s? 18:27:49 i forgot about where i read that but i think teor posted once a link to that feature in a comment to one of our bugs 18:28:22 pospeselr: i can look but there is no ticket with "locking down desktop breaks on win 10" yet 18:28:34 some of those issues were resolved on irc, too 18:28:41 so, the questions is waht we do about that 18:29:08 an obvious answer would be "follow the macos model" 18:29:28 another option would be looking for a different canonical location to put our bundle 18:29:57 what is the macOS model in this context? 18:30:10 and why do we have the desktop as the default install location to begin with? 18:30:16 to have it still self-contained (which we would lose if we go the macOS model where we have the binaries in /Applications and the profile dir in the usual profile location) 18:30:37 pospeselr: we have it as we wanted to give users a) a self-contained bundle 18:30:38 ah I see 18:30:45 and b) at a place where they can easily find it 18:31:16 otherwise they install it and later on need to search their c:\ to figure out where it actually is 18:32:29 well the canonical place to put it would be %ProgramFiles% (but that does require a separate place for the profile) 18:32:43 #18367 is the ticket for moving our data out of the application directory on Windows 18:33:06 yes 18:33:07 yeah sure 18:33:36 It would be good to understand why only some users encounter this issue. 18:33:46 Maybe some AV or Windows Defender setting? 18:34:06 also, %ProgramFiles% requires admin privileges, right? 18:34:17 to write to o 18:34:30 ("installing")..yeah, k 18:34:31 doesn't chrome actaully install itself to the profile somewhere? 18:34:36 iirc? 18:34:52 it's been awhile since I used windows beyond testing tb 18:35:09 seems like something that needs more investigation 18:35:17 mcs: good point this stuff could also interfere 18:35:41 so, i guess i should start with filing a ticket just for that issue 18:36:02 it would be a good idea to outline what properties we want to have with respect to tor browser and the installer 18:36:04 and we can collect all the info there and start then with finding a startegy forward 18:36:13 pospeselr: yes, agreed 18:37:10 I found this, which looks interesting (but might be unrelated): https://www.windowscentral.com/how-enable-controlled-folder-access-windows-10-fall-creators-update 18:37:13 ie, right now you can 'install' tor browser on a pc you don't have admin privileges to, but that has downsides (ie the weirdness w/ respect to read-only desktop) 18:37:36 (Windows Defender has a “Controlled folder access” option) 18:37:52 yeah, that might totally be it 18:37:55 yeah that looks suspect 18:38:43 okay, i'll try to get all the info we have into one place 18:38:50 and we can start from there 18:38:52 thanks 18:38:57 anything else for today? 18:39:14 Not from me 18:39:27 nope 18:40:52 nope2 18:41:03 nope3 18:41:10 I'm good 18:41:17 okay, i am calling it then (sorry it took a bit longer this time) *baf* 18:41:20 #endmeeting