18:02:06 <richard> #startmeeting Tor Browser Release Meeting 2022-10-31
18:02:06 <MeetBot> Meeting started Mon Oct 31 18:02:06 2022 UTC.  The chair is richard. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:02:06 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:02:11 <richard> bam
18:02:13 <richard> welcome everybody
18:02:20 <richard> pad: https://pad.riseup.net/p/tor-browser-release-meeting-keep
18:02:21 <PieroV> wb
18:04:58 <richard> ok ok
18:05:00 <richard> discussions
18:05:39 <richard> 11.5.7 should just be a targeted release to enable seamless locale to multi-locale transition w/o having to go through english first
18:05:52 <richard> so once that patch is in i'll start release preps
18:06:20 <PieroV> I've tested the sticky prefs locally to see how they work
18:06:46 <PieroV> They should do what we want, but we need to force a pref write
18:07:04 <PieroV> So, we need to modify projects/tor-browser/build in maint-11.5 to make the lang pref sticky
18:07:28 <richard> so what's the high level idea here, we backup the user's locale in 11.5.7 so it can be restored in 12.0
18:07:29 <PieroV> Then we need to modify tor-browser to write the intl.locale.requested at least once
18:07:53 <PieroV> richard: yes. Sticky prefs are written to prefs.js also when they match the default value
18:08:08 <PieroV> Normal prefs don't, and are just removed from that file when they match the default value
18:08:50 <richard> ok
18:09:22 <richard> sounds like we have a plan there
18:09:34 <PieroV> Yes, but we need a tor-browser-build issue + MR before the rel prep
18:09:55 <richard> yes all the usual
18:10:17 <PieroV> No, I mean for the tor-browser/build changes
18:10:45 <PieroV> (as the projects/tor-browser/build script)
18:11:31 <richard> ah right
18:11:42 <richard> the locales are injected in tor-browser-build right?
18:11:48 <PieroV> richard: yes
18:11:53 <richard> in stable
18:11:56 <PieroV> It appends to 000-tor-browser.js
18:12:01 <PieroV> Also in main
18:12:13 <PieroV> (if multi-lingual is disabled)
18:12:38 <PieroV> We still have the possibility to enable/disable multi-lingual builds
18:12:39 <richard> sounds like we have code to purge in main then :D
18:12:56 <richard> unless there's a good reason to keep them around
18:12:57 <PieroV> 12.5 to be on the safe side?
18:13:05 <richard> oh yes for sure
18:13:10 <richard> after 12.0
18:13:16 <PieroV> wfm
18:15:14 <richard> is there going to be any manual migration logic needed in the 12.0 series to migrate the pref, or does this 'just work' already using mozilla stikc pref machinery?
18:15:47 <PieroV> richard: I expect sticky prefs will work
18:16:01 <PieroV> They are made to keep prefs when switching from a channel to another one, that is our case
18:16:17 <richard> ok great
18:16:17 <PieroV> https://udn.realityripple.com/docs/Mozilla/Preferences/A_brief_guide_to_Mozilla_preferences#Sticky_Preferences
18:16:24 <richard> onto 12.0a5 then
18:16:38 <PieroV> wait
18:16:45 <PieroV> The second point was different from this
18:16:55 <richard> oh ok
18:17:22 <PieroV> It was: during today's meeting, we discussed about people updating from 11.5.6 (or earlier) directly to 12.0
18:17:44 <PieroV> I.e., skipping a pair of months of updates or mroe
18:18:37 <PieroV> We might be able to do something for them, but it would involve creating incrementals for localized channels even though we use multi-lingual packages
18:19:12 <richard> ah so bringing back the hack i whipped up over the weekend
18:19:20 <PieroV> Not really
18:19:32 <PieroV> We should customize the 000-tor-browser.js for each locale, I fear
18:19:55 <PieroV> So we should build the localized packages only to create the mar
18:20:38 <PieroV> + complications to switch from 12.0 to 12.0.1, if we stop doing that at 12.0.1
18:21:18 <richard> hmm
18:21:27 <PieroV> personally, I'm not sure it's worth the effort
18:21:40 <richard> and there's the question of how long do we maintain that
18:21:57 <PieroV> yes
18:22:08 <richard> iirc from one of the gitlab threads
18:22:23 <richard> boklm mentioned that it upgrade to English, becaues his system language is english
18:22:46 <PieroV> richard: yes
18:22:52 <richard> so am i right in understanding that currently people upgrade from older versoins will upgrade into their system language?
18:22:52 <PieroV> In tor-browser#17400
18:23:12 <PieroV> richard: you are right
18:23:50 <richard> honestly that seems fine to me
18:24:41 <richard> ok
18:24:44 <PieroV> okay
18:24:52 <richard> can we detect if a user has upgrade from a stable version from *before* the lang pref was saved?
18:25:06 <richard> like can we tell we're in the fallback case
18:25:09 <PieroV> richard: maybe we can
18:25:18 <PieroV> But we wanted to show a message in 12.0 in any case
18:25:20 <PieroV> Also to new users
18:25:36 <PieroV> See tor-browser#41378
18:25:43 <richard> (and if so, do we want to provide some UX syaing hey change your preferred lagnauge in about:preferences#whatever
18:25:55 <PieroV> If I understood correctly, we want to show that bar at the first start
18:26:04 <PieroV> https://gitlab.torproject.org/tpo/applications/tor-browser/-/issues/41378#note_2845190
18:26:28 <richard> oh perfect
18:26:58 <richard> yeah i vote we just have that UX if he pref isn't set (either by old -> new migration, or first ever launch)
18:27:06 <richard> and just avoid all of that complicated build stuff
18:27:18 <PieroV> yes, we can check if a pref isn't default
18:27:38 <PieroV> There's a method of Services.prefs
18:27:57 <PieroV> Or even easier: we can check if intl.locales.requested is empty
18:28:02 <PieroV> (which means use OS language)
18:30:51 <richard> ok sounds lik ethere is a path forward there
18:31:10 <PieroV> +1, we can proceed with the 3rd point for me :)
18:31:32 <richard> let's just make sure henry-x is at least looped in on the UX for an accessibility review
18:32:03 <richard> if they don't just end up implementing the feature entirely
18:32:10 <richard> ok so for 12.0a5
18:32:20 <richard> my take is that we should release as soon as we have macOS universal bundles
18:32:47 <PieroV> okay, I think its main purpose is to test the updater
18:32:53 <richard> that is the last remaining major feature for 12.0
18:33:10 <richard> yes
18:33:10 <PieroV> richard: could you please do a last pass through the board?
18:33:24 <richard> yeah i can do that
18:33:29 <PieroV> So that we have only possible blockers for 12.0a5 in ~Next?
18:33:33 <richard> i did a pass friday or thursday
18:33:45 <richard> ok good plan
18:33:52 <PieroV> yes, but we have 29 issues in ~Next
18:34:00 <PieroV> Like "Use locked prefs"
18:34:05 <richard> anything that's not a real blocker can be downgraded to ~Backlog
18:34:09 <PieroV> I think that could be even future
18:34:19 <richard> hm true
18:34:46 <richard> tor-browser#40899
18:35:32 <PieroV> the short story of that issue is that we disable already a lot of stuff
18:35:42 <PieroV> But I think we should remove them from about:preferences, too
18:35:54 <richard> yes
18:36:18 <richard> we should probably have a single about:preferences purge commit
18:36:22 <richard> seems easy enough to bunch all together
18:36:46 <richard> stick it after the base browser preferences commit?
18:37:09 <richard> or maybe we en dup needing two purges, one for base and one for tor-browser
18:37:18 <PieroV> yeah, works for me
18:37:26 <PieroV> There are a couple of interesting things to disable
18:37:33 <PieroV> 1. notifications
18:37:34 <richard> regardless, not a blocker for 12.0 but a nice to have
18:37:49 <PieroV> 2. credit cards and addresses
18:38:15 <PieroV> The two latest items
18:38:30 <PieroV> of the first section
18:39:58 <richard> let me review this entire ticket offline and respond there
18:40:07 <PieroV> wfm
18:40:07 <richard> overall this looks like a lot of good stuff
18:40:31 <richard> I will say I am actually excited for mullvad-browser to be my daily driver
18:40:52 <PieroV> :)
18:41:54 <richard> ok and finally multi-lingual buils
18:41:57 <richard> are in fact
18:41:58 <richard> excellent
18:42:07 <PieroV> \o/
18:42:23 <PieroV> Let's hope users are excited too
18:42:32 <richard> well yeah hopefully
18:42:44 <PieroV> From a build point of view I hoped for more
18:43:05 <PieroV> It still took a while to build the various browser projects
18:43:11 <richard> we can probably expect like, an hour turnaround from unsigned -> published in 12.0
18:43:20 <PieroV> But I am very happy about signing timings
18:43:28 <richard> that is true
18:43:42 <PieroV> Also, browser has become a parallelizable project
18:43:51 <PieroV> Since it uses single threaded compression
18:44:05 <richard> build performance in general may be something we can work on in the future
18:44:30 <PieroV> richard: what was previous time?
18:45:14 <richard> end to end signing -> publishing was around 12 hours depending
18:46:16 <richard> a bit less if one has a persistent internet connection and are paying attention
18:46:47 <PieroV> So, very nice improvement
18:47:02 <richard> yeah
18:47:02 <richard> stellar
18:47:39 <PieroV> I don't have anything else for this meeting
18:47:45 <richard> alright sam
18:47:47 <richard> same*
18:47:49 <PieroV> (but can we make it 1 hour in the calendar? :))
18:47:59 <PieroV> It's still 30 mins
18:48:03 <richard> oh
18:48:06 <richard> yeah can do
18:48:39 <richard> 3endmeeting
18:48:42 <richard> #endmeeting