15:00:21 #startmeeting Tor Browser Weekly Meeting 2024-09-09 15:00:21 Meeting started Mon Sep 9 15:00:21 2024 UTC. The chair is morganava. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:21 Useful Commands: #action #agreed #help #info #idea #link #topic. 15:00:23 blesssed date it would seem 15:00:25 o/ 15:00:27 o/ 15:00:37 the pad per usual -> https://pad.riseup.net/p/tor-tbb-keep 15:00:49 o/ 15:00:50 o/ 15:00:52 o/ 15:00:53 14.0a4 went out last week (thx dan_b for building) 15:01:26 with 14.0a5 planned for this week 15:02:11 also, I will have unreliable availability this week on Wed and Thurs, so if we have meetings or 1:1s on those days we'll need to either cancel or reschedule 15:02:26 o/ 15:02:26 Hi! 15:02:37 gosh so many people 15:03:20 PieroV: re your point, as far as i know, all we maybe know is that mozilla is planning ot announce their 13.5 plans 'soon' 15:03:44 morganava: it's out on whattrainisit 15:03:45 can you give us a staus update re the the updater channel splitting patches and next steps on that? 15:03:50 it's in the calendar 15:03:54 https://whattrainisitnow.com/release/?version=esr 15:03:55 ooh based 15:04:32 Also, I had an announcement first 15:04:44 I'm AFK next week (I'm at the reproducible build summit) 15:04:55 ooh have fun! 15:05:03 * boklm will be there too 15:05:05 And will be AFK also on Mon 23 and Tue 24 (taking a day off in Germany since I'm there and then traveling on May 24) 15:05:25 I'll have my laptop with me, so I might be able to do emergency work 15:05:49 Thanks, dan_b :) 15:06:02 :D 15:06:42 Regarding the update channel: I created a 115.15.0esr-13.5-2 branch with the updater patches 15:06:49 weird so 13.5 support is officially ending before 140 starts 15:06:51 alright 15:07:03 It also has ma1's security backports, so we should be able to use it for a build already, we just need to tag it 15:07:12 morganava: "We will re-evaluate this decision in early 2025 and announce any updates on ESR 115's end-of-life then." 15:07:44 is it just for windows, or ar ethey also backporting macOS? 15:07:46 In the red banner at the top of the page 15:07:49 security fixes 15:08:09 macOS, according to the same banner 15:08:20 oh i was on the wrong page 15:08:22 sorry sorry 15:08:45 well alright then, looks like we've some certainty now 15:09:00 Yep, but we have a hard deadline anyway 15:09:07 November 1st or 3rd, I don't remember 15:09:30 well yes, for android 15:09:34 Yep 15:09:41 115.17 is scheduled for October 29 15:09:56 The next release is scheduled for November 26 15:10:07 But I really hope we don't go that late 15:10:24 As I'd prefer 14.5 to arrive early 15:11:06 Anyway, 13.5a10 should be ready for the tor-browser side, but I haven't done anything on tor-browser-build 15:11:12 I was waiting for the official confirmation 15:11:52 Do we need a proper alpha channel, or do we want only one for testing the updater in the 115 series? 15:12:18 In the latter case we don't even need another maint branch, but we could temporarily set 13.5a10 on rbm.conf... 15:13:43 i guess the q is how confident are in in the changes 15:13:58 I hope a lot 15:14:04 my understanding is you've been able to exercise the relevant scenarios in the -2 branch (on Windows at least?) 15:14:05 As they are in 14.0a4 15:14:32 Yes, I think I still have the relevant configuration in tb-build-03 15:14:32 ah rightt 15:14:58 You can start from this one: https://tb-build-03.torproject.org/~pierov/torbrowser/nightly/tbb-nightly.2024.08.01/ 15:15:12 And see everything works as expected 15:15:31 not sure we're can get more confident than that then 15:15:48 I posted the details in some comments in a GitLab MR or issue, let me see 15:16:15 https://gitlab.torproject.org/tpo/applications/tor-browser/-/merge_requests/1138#note_3068348 15:18:17 /o 15:21:15 alright i'll have a look after this meeting 15:21:38 I don't think we have anything imposing a timing on us 15:21:47 given the esr extension, we can include these patches in a regularly scheduled 13.5.4 15:21:53 But I'd say the sooner the better 15:22:03 morganava: so, are you proposing we skip 13.5a10? 15:22:06 or maybe the week before regularly scheduled so it doesn't get awkard for non-legacy users 15:22:30 PieroV: is there a good reason to ship a 13.5.10? 15:22:33 PieroV: is there a good reason to ship a 13.5a10? 15:22:43 morganava: have a watershed also on alpha 15:23:26 erm 15:23:39 So that we don't need to think whether there are differences between alpha and stable users (even though some users will be excluded from this watershed) 15:23:47 so this would affecct those users on legacy paltforms running 13.5a9 15:23:53 And also for maximum peace of mind 15:24:02 Yes 15:24:07 who didn't get the 14.0a1 upgrade 15:24:38 but we have no plans of maintaining a 13.5 alpha series do we? 15:24:55 Indeed. An alternative if we want to be aligned is to make the last 14.0a a watershed 15:25:20 No, I don't see a reason to do alphas, since they're going to be only Moz updates 15:25:26 mmhm 15:25:52 sorry why does the last 14.0aX need to be a watershed? 15:26:05 i thought we just need a watershed 13.5.X 15:26:26 So that we know that all the alpha users got the updater patches and we can do also a couple of other changes that were planned with the first watershed 15:26:46 E.g., dropping a fontconfig thing in start-tor-browser 15:27:06 And simplifying the migration procedures 15:28:04 Also, it'd be great to backport the search engine update for the same purpose (drop the Twitter, YouTube, and Yahoo search engine extensions before the watershed) 15:28:05 ohh ok 15:28:33 so to ensur esmoetime in the future when we drop win 10 support, some poor 14.0a4 users doesn't try to upgrade to unsupported 16.0a1 15:28:46 right ok and the other things too 15:29:13 There's also the hash change 15:29:14 What does watershed mean in this context? 15:29:29 clairehurst: a watershed version is an update all TB users upgrade through 15:29:30 clairehurst: it means that you force users to go through a mid version instead of a direct update 15:30:00 eg if X is a watershed, all users on versions older than X will first upgrade to X then to current 15:30:09 E.g., users from 13.0.10 will go through 13.5.4 before updating to 14.0 15:30:22 Ah cool thanks! 15:30:35 However, we will have to keep hashes for a while for nightly builds 15:30:53 Because wathersheds on nightly builds would be annoying 15:31:06 (or we just break very old builds at a certain point...) 15:33:18 ok so tldr, we plan to release 13.5a10 as a watershed update, and 14.0aX also as a watershed update, and finally the last 13.5.X as a watershed update 15:33:54 and of course our release calendar is again completely out of date 15:34:10 Yes, that sounds like a plan to me 15:34:34 well at least we have a bit of breathing room with the extended ESR schedule 15:34:37 Maybe instead of keeping the update data forever in cdn.tpo, at a certain point we can modify the update responses to point to archive.tpo 15:36:14 good idea 15:37:56 morganava: bw, it'd be interesting to see how many users are pinging from 13.5a10 15:38:06 * btw 15:38:23 it would be 15:38:26 I think we can't right now, but maybe it's a query that can be added to metrics.tpo 15:38:36 we still have a copy of 11.5.8 on cdn.tpo and could use archive.tpo URLs instead 15:38:37 hiro opened an issue for me about that a while ago 15:38:44 boklm: yep, good idea 15:38:52 mmhm 15:38:54 Is 11.5.8 one that has all the various languages? 15:39:06 I think yes 15:39:34 that's from awhile ago wow 15:39:40 morganava do we need to update the warning message that is currently showing on windows <= 8.1? 15:40:15 morganava: yep, so we don't have to keep it on cdn.tpo, but we don't have to delete the update responses either, as they're very light 15:40:44 AFAIK changing the URLs in the update responses doesn't involve signing again or something like that, but just running the right incantations of sed : 15:40:46 :) 15:40:55 henry-x that's a good question 15:41:44 PieroV: yes, we only need to update the xml files in tor-browser-update-responses.git/update_pre12.0/release 15:41:48 I suppose we could soften the messaging, let them know' ok you're on a supported release for now but the future is scary plz upgrade' 15:42:06 or just remove the banner altogether 15:42:40 henry-x can you open an issue and cc myself and donuts? 15:42:41 +1 for a softer message still encouraging to upgrade or switch OS 15:42:50 ok 15:43:01 "hey plz thank Moz for this" :) 15:43:34 i like to believe our whining helped a little :3 15:43:45 but perhaps not lol 15:44:34 ok any other topics of announcements for today's meeting? 15:44:38 Is there any question about support for old mac os being extended? 15:44:53 old macOS is also being extended 15:44:54 It is, but I think we're not 15:45:25 well hmm 15:46:08 ok i'm of the opinion that if we're maintaining a 13.5 legacy channel, we may as well do macOS too 15:46:30 it's barely additional work 15:46:39 It's twice the work though :P 15:46:43 4 builds instead of 2 15:46:52 computer work and signer work vOv 15:47:07 i mean 15:47:08 Computer costs power, it isn't green :P 15:47:13 ok objectively there's hardly any users 15:47:36 Also, we could probably build only x86_64 for macOS 15:47:38 on the other hand good-will? or at least no bad-pr from those 5 users lol 15:47:55 I don't believe M1 is supported on anything older than 11 or 10.15 maximum 15:48:04 (nor M2 and followers) 15:48:20 Well, Apple users are used to be left unsupported 15:48:49 how annoying is it going to be to remove the fat-dmg pipeline from the build? 15:48:59 well actually testbuilds are only single-arch aren't they? 15:49:00 It's supported on testbuilds 15:49:02 ack 15:49:12 But it's more a boklm question 15:49:36 i think i'd rate that as nice to have for an initial 13.5.X but not a blocker 15:50:13 yes, we have makefile targets for single arch builds, but we need to check if it's still working fine as I think we're not using them very often 15:51:56 in theary using the browser-macos-x86_64 target instead of browser-macos should be enough 15:52:32 well, i know we have at least 1 users, since a legacy mac user emailed me over the weekend asking about legacy support (but that was before mozilla finalised their plan) 15:53:27 do we know how many legacy macos users mozilla have? 15:53:41 according to the graphs 15:53:43 very very few 15:53:55 I know NoScript has many users on Firefox 78 because of MacOS :/ 15:54:32 (or, rather, vocal users) 15:54:32 esh 15:55:06 basically my take is if there's no extra dev work involved lets support macOS, otherwise lets' forget about it and move on with our lives 15:55:21 i'm fine spending time on biulds and the minimially extra time to sign+notarise etc 15:56:14 ok we've 4 minutes 15:56:36 if there's ntohing else, then we can call it 15:57:13 I'm good 15:57:26 Nothing from me 15:57:42 nothing from me 15:58:20 ok, if you are somehow looking for tasks, here's our relevant query for 14.5 -> https://gitlab.torproject.org/groups/tpo/applications/-/boards?not[label_name][]=14.5%20stable&label_name[]=14.0%20stable 15:58:30 have a good week everyone o/ 15:58:32 #endmeeting