15:05:29 <richard> #startmeeting Tor Browser Weekly Meeting 2024-06-24
15:05:29 <MeetBot> Meeting started Mon Jun 24 15:05:29 2024 UTC.  The chair is richard. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:05:29 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:05:51 <richard> our pad per usual -> https://pad.riseup.net/p/tor-tbb-keep
15:06:14 <Jeremy_Rand_36C3[m]> Hi!
15:06:16 <richard> update with your todos and todones and any discussion points please and thank you :)
15:06:43 <manuel[m]12> o/
15:08:45 <bellatchau> o/
15:10:45 <richard> ok
15:10:57 <richard> 13.5 went out last week!
15:10:59 <richard> \o/
15:11:14 <richard> nothing catastrophic came in over the weekend, though i haven't checekd my email yet this morning
15:11:20 <richard> so perhaps there are surprises lurking
15:11:46 <richard> the difference in Android package size is about 5mb which is a cool thing we probably should have maybe called out in the blog post but oh well
15:12:22 <richard> the post got more comments than usual, mostly positive apart from the old OS support being dropped in 14.0
15:12:22 <PieroV> 5MB more or less?
15:12:30 <richard> oh 5MB less
15:12:37 <richard> from our re-packaging efforts in the spring
15:12:44 <ma1> very cool indeed
15:13:05 <richard> PieroV: I did see your gitlab issue re fenix size increasing in 128 so looks like we have some fun challenges coming this summer
15:13:33 <Jeremy_Rand_36C3[m]> richard: that kinda ties into my discussion point I guess
15:14:05 <richard> but overall it looks like a  pretty smooth release and I'm not anticipating any need for an emergency update
15:14:50 <richard> my only announcement for this week is to please fill in the crab.fit i sent out to the team RE the esr 128 issue audit process meeting
15:15:07 <PieroV> richard: well, since the Play Store numbers for x86 are very low and x86 on Android on real hardware is more or less dead we could keep building but stop providing it through the Play Store
15:15:48 <richard> PieroV: honestly probably not the worst idea long term, but I'd rather keep it around for the one person using an x86 tablet somewhere :p
15:15:59 <richard> but i won't feel too bad dropping it if it's the only one over-budget
15:16:41 <Jeremy_Rand_36C3[m]> As much as I hate x86 as an architecture, I do like having diversity in arches
15:17:08 <PieroV> They could download them from our site/Telegram bot
15:17:20 <richard> if anyone needs guidance on what to work on this week please ping me after this meeting
15:17:23 <PieroV> (and F-droid maybe?)
15:17:23 <ma1> are we still planning for F-Droid? I believe I've seen some discussion about removing "anti-feature" recently.
15:17:59 <richard> i think gitlab user Mynecol (iirc) has been working on changes required off and on this release
15:18:29 <richard> not sure what remaining work there is there, but vaguely recall them being blocked on something 115-specific
15:19:22 <PieroV> No
15:19:28 <PieroV> It's problems with apksigcopier
15:19:49 <PieroV> Even though we have a way to strip the QA signature and sign the unsigned APKs
15:20:06 <PieroV> (up until a few months ago we were signing APKs without stripping the QA signature, but the tools somehow worked anyway)
15:20:40 <PieroV> But this was a problem because we aligned the zip after the QA signature and apksigcopier was not happy with that
15:21:31 <PieroV> So, there was a single final missing step, but we might have to contact some expert of Android signing
15:21:41 <richard> ah ok, maybe it was just that we weren't going to priortise fixing it for 115
15:22:08 <PieroV> Then F-droid updated their Debian version, maybe this was also a problem
15:22:18 <PieroV> But I'm also updating our build containers to the current stable
15:22:31 <richard> fun :3
15:22:47 <richard> Do you all have any discussion points, need help with/are blocked on anything?
15:22:56 <PieroV> I think Jeremy_Rand_36C3[m] has a discussion point
15:23:00 * Jeremy_Rand_36C3[m] has a discussion point
15:23:06 <PieroV> And after that I'd like to update you on my current 128 status
15:23:10 <Jeremy_Rand_36C3[m]> Pasting it:
15:23:13 <Jeremy_Rand_36C3[m]> "What's the status on pluggable transport compression (e.g. GoBusybox)? Is it currently high priority? Would it be helpful if I spent some time on it? If so, deadline? Is there an open GitLab issue I should be keeping an eye on?"
15:23:31 <Jeremy_Rand_36C3[m]> (I've been out of the loop since Lisbon since I was still traveling at other conferences)
15:23:43 <Jeremy_Rand_36C3[m]> So I wasn't sure what follow up was done after that session
15:23:45 <PieroV> Jeremy_Rand_36C3[m]: we haven't worked on that
15:23:56 <PieroV> Because we gave the highest priority to releasing 13.5
15:23:59 <PieroV> And then to 128
15:24:03 <Jeremy_Rand_36C3[m]> Alright
15:24:16 <PieroV> I think it won't be needed for armv7+aarch64
15:24:17 <Jeremy_Rand_36C3[m]> It sounded like 128 was going to inflate the size quite a bit?
15:24:28 <PieroV> We should still have budget for those platforms
15:24:50 <PieroV> But we'll definitely need to do something for x86/x86_64
15:24:59 <Jeremy_Rand_36C3[m]> ok
15:25:09 <PieroV> I hacked together an unpatched Fenix build + tor + PTs
15:25:24 <PieroV> And I hit ~120MB after repacking with maximum compression
15:25:26 <Jeremy_Rand_36C3[m]> is there a GitLab issue open for this category of experimenting?
15:25:43 <PieroV> We have a meta, let me find it
15:26:15 <PieroV> https://gitlab.torproject.org/tpo/applications/tor-browser/-/issues/42607
15:26:37 <Jeremy_Rand_36C3[m]> If I can persuade one of my funders to cover me to try this, would that be of benefit?
15:26:58 <PieroV> That we can keep shipping Android x86 on Play Store
15:27:11 <PieroV> Err, sorry, I read what would be the benefit
15:28:12 <Jeremy_Rand_36C3[m]> Alright so it sounds like there would be some benefit?
15:28:27 <PieroV> So, we haven't applied our patches yet
15:28:40 <PieroV> They do remove some assets iirc
15:28:57 <PieroV> But shaving out 16MB compressed seems kinda hard
15:29:00 <richard> jeremy: we have a meta issue here tracking the various apk size reduction issues -> https://gitlab.torproject.org/tpo/applications/tor-browser/-/issues/42607
15:29:14 <richard> including the things we want to try from the session in lisbon
15:29:38 <Jeremy_Rand_36C3[m]> PieroV: yeah it sounds like we'd want to do a bunch of things to get the size down if possible
15:29:58 <Jeremy_Rand_36C3[m]> Since I doubt a single tweak is going to shrink it that much
15:30:05 <Jeremy_Rand_36C3[m]> But yeah every bit helps
15:30:16 <PieroV> Jeremy_Rand_36C3[m]: very appreciated
15:30:37 <PieroV> A trick would be to use a better JS minifier than the one Moz has
15:31:23 <Jeremy_Rand_36C3[m]> PieroV / richard : alright I'll talk to the funder in question and see if he can cover some of my time on this. Whether I'll succeed is of course uncertain, but yeah I'll keep you posted and I'll read through that meta issue
15:32:22 <PieroV> Thanks! And thanks again for the talk about this you did in Lisbon
15:32:37 <Jeremy_Rand_36C3[m]> PieroV: no problem, I really enjoyed that session
15:32:47 <Jeremy_Rand_36C3[m]> Glad you guys enjoyed it too :)
15:33:45 <Jeremy_Rand_36C3[m]> Alright that's all for me
15:33:53 <richard> \o/
15:34:06 <richard> anything else on the agenda for today?
15:34:23 <PieroV> Yes
15:34:38 <PieroV> I'd like to give a status update about 128
15:34:48 <richard> yes please!
15:34:53 <PieroV> As you know, there's the MR with the rebase up for review
15:35:13 <PieroV> I managed to get builds from that for all platforms, including Android
15:35:20 <PieroV> (even though only GeckoView is patched)
15:35:53 <PieroV> A good news is that the local build steps haven't changed that much
15:36:38 <Jeremy_Rand_36C3[m]> PieroV: you had something you wanted to cover after me?
15:36:50 <PieroV> And you can use the same gradle incantation you used for the firefox-android.git repository. However, you have to specify a correct mozconfig (e.g., with the `MOZCONFIG` environment variables, but maybe it's still possible to do it also in local.properties)
15:37:06 <PieroV> Jeremy_Rand_36C3[m]: yep, on it :)
15:37:28 <PieroV> Also, the desktop part for tor-browser-build is ready for review as well
15:37:48 <PieroV> However we should keep talking about the gnullvm patch
15:38:03 <PieroV> We didn't continue after last week
15:38:26 <richard> this the windows Rust toolchain issue?
15:38:35 <PieroV> And I've been working on the Android part. Offline builds have become even harder, but still possible I think/hope.
15:39:04 <PieroV> richard: yes, it's the Rust/Windows toolchain issue
15:40:20 <PieroV> Currently I'm fighting with Gradle build scripts. I have to patch a file to prevent it from downloading a zip and use the file we build locally instead. I might need some help if I can't really get that working
15:41:05 <PieroV> Hopefully, once I'm done with this, I should be able to push Android to the current MR we have for tor-browser-build
15:42:30 <PieroV> I think once we merge everything, we should wait a few nightlies to test the updater
15:42:41 <richard> what's the issue with the grade scripts? is it like a baked in thing to the build system or more of a edit this shell script to cp instead of wget type problem?
15:42:57 <PieroV> (maybe I could try to produce some fake nightlies after I'm done with Android)
15:43:15 <PieroV> Then only if the updater is okay push an alpha
15:43:46 <PieroV> At this point, I think we're going to release 14.0a1 by ~10 July (or later)
15:44:00 <PieroV> richard: Gradle is the build system for Android
15:44:08 <richard> right
15:44:14 <PieroV> Up until a few months ago it used to check if the nimbus-fml binary existed
15:44:37 <PieroV> Now it doesn't check anymore, and instead downloads the artifact from Moz immediately
15:44:49 <dan_b> lol. but we can just patch that behaviour back in then?
15:45:09 <PieroV> dan_b: there was a big refactor around there
15:45:20 <dan_b> hahaha of course there was
15:45:33 <dan_b> clairehurst, how's your gradle-fu?
15:45:42 <PieroV> This is the regressor: https://github.com/mozilla/application-services/commit/9197c0bdce6b3be5c35feab0b01f8bffbb4fa6be
15:45:42 <dan_b> mine is... poor
15:46:00 <dan_b> pierov: 0_o
15:46:12 <richard> that's a bit of a patch
15:46:25 <dan_b> ooooh
15:46:25 <richard> is it time to purge application-services? :D
15:46:31 <dan_b> so this is hte year I learn groovy?
15:46:55 <PieroV> richard: it'd be at least another 4.5MB compressed removed
15:47:07 <dan_b> oooooh actually yeaaah... we dont... use any of that do we
15:47:08 <dan_b> 😄
15:47:15 <dan_b> (just reading the README
15:47:23 <PieroV> I think it'll be easier to remove the `fetch` part of this file: https://github.com/mozilla/application-services/blob/main/tools/nimbus-gradle-plugin/src/main/groovy/org/mozilla/appservices/tooling/nimbus/NimbusGradlePlugin.groovy
15:47:54 <PieroV> And somehow call the equivalent of cp. We could use some environment variable to tell Gradle where to find our build of nimbus-fml
15:50:11 <PieroV> And that's where I've arrived with my work on 128
15:51:34 <richard> well, looking forward to taking it for a spin in the coming weeks :D
15:52:07 <richard> alright, unless there's anything else, let's end this
15:52:20 <richard> have a good week folks \o/
15:52:21 <ma1> ty, o/
15:52:28 <richard> #endmeeting