14:59:38 #startmeeting Tor Browser Weekly Meeting 2022-10-24 14:59:38 Meeting started Mon Oct 24 14:59:38 2022 UTC. The chair is richard. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:59:38 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:59:39 hello 14:59:41 Hi! 14:59:51 meeting pad as always: https://pad.riseup.net/p/tor-tbb-keep 14:59:55 * Jeremy_Rand_36C3[m] is surprised that I remembered to fill out the pad ahead of time 15:00:10 * Jeremy_Rand_36C3[m] hopes I've turned over a new leaf and this will be a trend 15:01:15 so I don' thave any discussion points this week 15:01:41 in terms of release schedule, 11.5.5 should be signed today 15:02:03 \o/ 15:02:32 boklm: I haven't caught up on email yet, so if you've signed the 11.5.5 apks thanks, if not can do that after the meeting? 15:02:45 richard: yes, I signed them 15:02:58 then thanks it is :) 15:03:00 * dan_b woke up late but is now caught up writing 15:03:34 richard: https://gitlab.torproject.org/tpo/applications/tor-browser-build/-/issues/40628#note_2846845 15:04:00 and I'm hoping to start 12.0a4 release prep once I've finished signing 11.5.5 15:04:21 so if there are any more things to get into the alpha get those MRs out 15:04:26 ok beyond that do we have any bolded items? 15:04:31 * richard skims 15:04:36 richard: I think some issues might be moved from ~Next to ~Backlog, if ~Next is still for 12.0a4 15:04:50 you said there was a build error with goservice in the nightlies yeah? 15:04:51 I'd like to get my final MR for multilang in it 15:05:03 (which happens to be a bolded item, too) 15:05:11 makes sense to me 15:05:13 dan_b: uh oh, goservice is one of my things, yes? 15:05:28 dan_b: yes i think so 15:05:29 dan_b: yes. Do you receive emails about nightly builds? 15:05:29 yes, I opened a ticket for the goservice error 15:05:37 https://tb-build-05.torproject.org/~richard/logs/goservice-linux-x86_64.log 15:06:01 tor-browser-build#40659 15:06:04 (do dan_b, ma1 and henry-x know about nightly builds and their emails?) 15:06:18 boklm: can you tag me in the ticket? I'm guessing it's simple to fix, but if you guys can't get it dealt with easily, I can try to investigate it. 15:06:20 PieroV: hmmm... don't think so? 15:06:35 it might be a good idea to add them to the email 15:06:42 dan_b: we have nightly builds (they take more than one day atm, so they're not really nighly builds :P) 15:06:47 I don't get any nightly build emails 15:06:58 henry-x: if you haven't started already, I'd suggest going with qemu and virt-manager over virtual box for setting up VMs 15:07:09 if only because setting up macOS vms is v easy with qemu 15:07:19 When they finish, they send an email, that tells which platforms worked and which ones failed 15:07:24 and in my experience the two don't play well together 15:08:08 +1 for Qemu, VirtualBox's bios needed to be compiled with proprietary tools a while ago, and it was moved to contrib in Debian, I don't know if other distributions did something, too 15:08:16 And Qemu is more performant 15:08:32 Jeremy_Rand_36C3[m]: ok, done 15:08:47 (I only brought up virtual box because it's what I initially used back in the day and discovered it didn't play well with others) 15:08:48 boklm: I don't recognize the `gspt` dependency that it says is missing, did that maybe get added because some other dependency got bumped? 15:08:49 For Windows you can use official MS images, too. They don't ship the Qemu image, but VMWare's works fine also with Qemu (and it's very compressed, too) 15:09:03 I was going to use GNOME boxes, which apparently uses QEMU 15:09:06 Jeremy_Rand_36C3[m]: I think it's related with the go version update 15:09:11 (I mean, maybe that dependency was always there and I've forgotten about it) 15:09:29 dan_b: I would hold off on the go1.4 and other go stuff (beyond the build failure) until after 12.0 15:10:11 boklm: yeah could be, though it's odd that a Go stdlib update would cause some random github.com repo to get pulled in. 15:10:26 Jeremy_Rand_36C3[m]: we worked on go recently 15:10:29 in favor of other open FF102-esr labeled issues 15:11:12 so go got upgraded to 1.19 and i updated goxsys to HEAD of a few days ago as well 15:11:18 that was it I believe on my end 15:12:25 ok I can def wait on further go work! 15:12:28 boklm: if adding that dependency as a new project doesn't solve it, a fallback fix would maybe be to disable cgo for all of the Namecoin Go projects. They should build fine without cgo, and the only thing they use cgo for is privilege dropping, which is not a thing in Tor Browser's usage. 15:14:52 boklm: can you prioritize signing a macos aarch64 testbuild tomorrow? 15:15:07 richard: ok, I can do that 15:17:32 ok, beyond that I think everything looks good here 15:17:39 I have a bolded item 15:17:39 if there's nothing else from anyone else here I'm happy to adjourn 15:17:47 For the language order 15:18:00 I will start an aarch64 testbuild now to have it ready for signing tomorrow 15:18:12 I have a MR that sets it in code order, but it hasn't been reviewed, and merged 15:18:17 boklm: ack 15:18:18 ah, did emmapeel say anything about that? 15:18:47 PieroV: i'm happy to merge it as is and open a ticket later for localization review 15:18:52 re: lang order, the current code lgtm, but I was waiting for the final "political" decision. 15:18:59 some light reading on the topic doesn't suggest a 'right' way to do it 15:19:05 ma1: okay 15:19:22 I think emmapeel said to follow the site? 15:19:34 And the site is ordered by language code, from what I can see 15:19:46 So like the current patch 15:19:48 Except for English 15:19:55 Which is set as first 15:19:58 the only alternative I would consider is initially having the top N commonly used locales at the top (similar to how se prioritize certain countries in connection assist) 15:20:25 but... we're about to lose metrics on that once we only have multi-locale bundles 15:20:55 The ordering has to kind of work with any language packs that the user may have installed as well 15:20:56 okay, so let's stop the multi-locale thing :P 15:21:03 Having English as first lets you switch to English if you're lost and go from there to pick yours. 15:21:09 so currently it'd be like, English, Russian, German, and Farsi 15:21:31 I wouldn't put english first. Just use the code ordering. 15:21:49 I could understand having English as first for the reason ma1 says, but don't like it that much 15:22:02 But I really dislike richard's alternative 15:22:30 I don't think the list is long enough to warrant pinning popular languages 15:22:41 the use case is also pretty different for connection assist 15:22:53 we're really just pinning the countries we have settings for there 15:23:00 re language packs, we don't have a list of languages we support, at the moment 15:23:27 We could hack a preference that is filled by tor-browser-build to avoid having multiple lang lists, and follow it 15:23:47 I agree with the above fwiw 15:23:56 what do you mean by "multiple lang lists"? 15:24:13 (and why don't the locale lists in rbm.conf count?) 15:24:29 So, that languages we officially support are always first, even when users had old profiles with additional langpacks (downloads are disabled now) or install them manually 15:24:50 henry-x | what do you mean by "multiple lang lists"? <-- currently we have rbm.conf, which is the authoritative one 15:25:18 But we have to update translations on torbutton and version them 15:25:30 oh, so you don't want to re-specifiy the same list in different places? 15:25:41 not if we can help it :D 15:25:42 Exactly, because keeping things in sync is difficult 15:26:16 Once we complete tor-browser#40924, we will not need to import translations in that way anymore 15:26:30 But rather update the translation repo commit hash in tor-browser-build 15:26:38 And we'll finally use only the list of rbm.conf 15:26:56 (but at the moment we don't have a way to tell Firefox the list of rbm.conf, afaik) 15:28:16 donuts: do you know if we have a design ticket for the website updates needed 15:28:32 for Alpha, yes 15:28:33 since we no longer need the language list in alpha? 15:28:35 not for stable yet 15:29:02 https://gitlab.torproject.org/tpo/web/lego/-/issues/53 15:29:47 please feel free to input there, and we'll make the changes happen 15:30:25 * donuts adds ~For Applications team 15:31:12 henry-x: can you take a look at our Windows installer and see if it needs to be tweaked/upated as well? 15:32:04 I'll open a ticket for that 15:32:17 I've never worked with anything on Windows. I would need some direction 15:33:39 follow the friendly paperclip across the grassy field under a clear blue sky, there you will find a man named william gates 15:33:45 he is the wizard of windows 15:33:49 lol 15:33:54 haha 15:34:30 p much I don't quite recall what our windows installer UX is like and if there's anything relevant to locales in there so we should make sure we don't need to update our nsis configuration 15:35:04 maybe long-term we can pre-set the user's chosen locale from the nsis installer itself 15:35:05 I'm almost certain I recall something locale-related in the NSIS script, from when I had to mess with it half a year ago 15:35:09 Nobody really knows anything about the Windows installer... And if they do, they wish they don't have to deal with it :P 15:35:17 i vaguely remember that being a thing in windows land 15:35:34 It prompts for the locale to use for the installation 15:35:42 PieroV: honestly I don't find NSIS that painful 15:35:53 But I haven't modified the NSIS script to somehow inject it in the preferences 15:36:00 Everything in the macOS ecosystem makes me about 10x as stressed as the NSIS stuff 15:36:34 Jeremy_Rand_36C3[m]: do you know that it has assembly code in it, that compiles only with MASM and GCC? 15:36:41 Call me biased, but that was enough for me :P 15:36:42 our windows installer NSIS script is in tor-browser-build/projects/browser/windows-installer/torbrowser.nsi 15:37:22 PieroV: compared to literally every other Windows installer system, which can't cross-compile without Wine, I'll take whatever NSIS has 15:38:05 Maybe I'm just accustomed to Windows things doing insane stuff 15:38:23 Yeah, I don't know about the others 15:38:51 But it's one of the two projects preventing us to get rid of GCC on Windows, at the moment. The other one is Rust 15:39:15 PieroV: yeah, don't ever try to use InnoSetup or PyInstaller if you think NSIS is bad :P 15:40:14 clearly we should just package it as an appx once win7 support is dropped from firefox 15:40:18 * richard has said too much 15:40:40 hahaha 15:40:49 richard: is that the same thing as MSIX or is there some other "X" thing in Windows now? 15:41:08 Jeremy_Rand_36C3[m]: I think it's the format of Windows Store? 15:41:14 vOv i'm afraid all of my windows knowledge is 5 years out of date at this point 15:41:30 PieroV: I'm almost certain MSIX is a Windows Store thing too 15:41:43 You can distribute MSIX installers outside of the Store too though 15:41:44 Okay, then you probably know more than me :D 15:42:02 yeah it was originally for their windows 8 UDP/UWP apps, but they extended it to native win32 for use in their store at some point in the past N years 15:42:42 but anywya 15:42:43 * ma1 has heard too much 15:42:47 Anyway I briefly investigated MSIX for Namecoin and gave up when I discovered that MSIX only recently added support for services, and they don't let you use sandboxed service accounts, for Satan knows what reason 15:43:10 i think we can worry about app packaging and distribution in general for 12.5 15:43:20 beyond 'just get it working' we're at now 15:43:35 any other bold points I've missed? 15:43:38 Last thing maybe from me for 12.0a4 15:43:44 mk 15:43:44 We should really call for test test and test 15:43:54 For tor-launcher and multi-lingual 15:44:04 yess, I can ping comms 15:44:10 Thanks <3 15:44:35 yeah agreed :) 15:44:42 A dedicated forum post would be great as well 15:44:50 maybe community can help with that 15:45:03 12.0a4's changelog is going to be *MASSIVE* 15:45:04 I'll bring up at the UX team meeting tomorrow 15:45:19 did we have any automated tests that covered some of the tor-launcher stuff? 15:45:26 yeah poor flutter launched with win desktop support packaging for UWP and MS had like already announced UWP deprecation 15:45:41 they've since updated to MSIX which i assume is some new x version of .msi ? 15:45:43 henry-x: I think we used the old launcher for the few tests we had 15:45:48 for Cwtch we also use NSIS 15:45:49 is our ETA for release still the 28th? 15:45:55 henry-x: we *used* to in tor-browser-bundle-testsuite 15:46:29 dan_b: remind me to chat with you sometime about Cwtch and Flutter, I'd like to run some brainstorms past you... 15:46:38 any time! 15:46:49 grab me in IRC 15:46:56 or i m up for voice on bbb 15:47:46 ok i'm going to call it then 15:47:50 have a good week everyone! 15:47:53 dan_b: yeah sure, will reach out once the Outreachy excitement has calmed down a bit :) 15:47:57 Thanks richard ! 15:47:57 Thanks, good week! 15:48:02 #endmeeting