15:06:02 <PieroV> #startmeeting Tor Browser Weekly Meeting 2022-04-19
15:06:02 <MeetBot> Meeting started Tue Apr 19 15:06:02 2022 UTC.  The chair is PieroV. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:06:02 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:06:09 <PieroV> Here we are!
15:06:37 <PieroV> I think we've all updated our pad and we can start with the discussions
15:07:02 <PieroV> I see only my points
15:07:22 <PieroV> The first one is about Yawning Angel's goutls fork
15:07:36 <PieroV> It's been deleted for some reason (somebody should contact them to ask why)
15:08:16 <PieroV> Now builds fail because fetch is stuck at git authentication
15:08:29 <Jeremy_Rand_Talos__> Does Yawning still work at Tor?  It's been a long time since I saw him around here but I'm not sure I've been watching the right places
15:08:35 <boklm> I can send Yawning an email to ask about that (unless one of you wants to do it)
15:08:51 <PieroV> We still have the code (in git_clones), and we could push to our GitLab
15:08:58 <boklm> Jeremy_Rand_Talos__: no
15:09:03 <PieroV> Jeremy_Rand_Talos__: it's a core contributor, but not a TPI employee
15:09:19 <Jeremy_Rand_Talos__> I see, thanks
15:09:38 <sysrqb> i believe this is a question for the anti-censorship team, as well
15:09:40 <PieroV> boklm: if you already have their contact/have already contacted them in the past it would be great, thanks
15:09:52 <PieroV> sysrqb: yeah, I agree
15:09:53 <sysrqb> this is a dependency of one of their projects (kinda)
15:10:15 <PieroV> and IIRC they wanted to mirror obfs4 as well
15:10:38 <PieroV> So I think that we could ask them whether they want to mirror goutls in tpo/anticensorship as well
15:10:57 <sysrqb> yes, that seems like a good plan
15:11:05 <PieroV> If needed, I can push with the history, since I already had the copy cached in git_clones
15:12:08 <PieroV> If nobody wants to add anything to this topic, I'd go ahead with Android
15:12:53 <boklm> I think we can have a mirror in someone's personal repo to unblock nightly builds, then later create it in tpo/anticensorship depending on what Yawning answers
15:13:46 <PieroV> sounds good to me
15:14:32 <PieroV> (we can/have to publish the code anyway, as it's GPLv3, but asking is more polite :))
15:15:14 <Jeremy_Rand_Talos__> I mean, if it were me, I'd be grateful for someone telling me my server was busted
15:16:31 <PieroV> Jeremy_Rand_Talos__: it's the only Yawning's repo that isn't available anymore, and it's hosted on gitlab.com
15:17:14 <Jeremy_Rand_Talos__> Weird.  :/
15:18:16 <PieroV> Anyway, I'd go with Android problems and updates
15:18:56 <sysrqb> +1
15:19:03 <PieroV> I was finally able to (incrementally) build outside tor-browser-build
15:19:19 <sysrqb> woo!
15:19:29 <PieroV> Mozilla devs use a special update channel that they call default
15:19:51 <PieroV> By switching to it we can use Mozilla's "pipeline" successfully \o/
15:20:14 <sysrqb> awesome
15:20:17 <PieroV> (sorta, I've managed to do so for a pair of days and now I have some problems with colors)
15:20:38 <sysrqb> heh
15:20:39 <PieroV> I literally mean colors: AAPT: error: resource color/photonDarkGrey90A60 (aka org.torproject.torbrowser_debug:color/photonDarkGrey90A60) not found.
15:21:35 <boklm> is the "update channel" used on android builds too?
15:22:15 <PieroV> I've also found the problem with extensions: it was a JS minifier that misbehaved with comments (it's a known bug, and the JS minifier has been enabled by default only recently, that's why we've never had this problem before)
15:22:39 <PieroV> boklm: we use the beta channel, but I think it's used for build dependencies, rather than for real updates
15:23:09 <boklm> ah, ok
15:23:58 <sysrqb> it may still control some features in Gecko (for geckoview) within nightly/early_beta/late_beta preprocessor conditionals, as well
15:24:19 <PieroV> sysrqb: that's an interesting point
15:24:32 <PieroV> a bad news is that we have one or two more crashes
15:25:14 <PieroV> I could find one in my local builds, but it happens on debug code
15:25:39 <PieroV> It doesn't happen always, and it's related to some remote-profile stuff that we should disable
15:26:12 <PieroV> The second crash, however, happens only in the build created with tor-browser-build
15:26:31 <PieroV> And I haven't figured out where it happens
15:26:45 <sysrqb> are both at startup?
15:27:08 <PieroV> No! The profile is at startup
15:27:27 <PieroV> But the other crash happens after one minute, or some seconds after the bootstrap
15:27:51 <PieroV> But like the crash we had with v96, you need to clear the storage to reproduce it
15:27:56 <sysrqb> this second crash comes from libxul, too?
15:28:19 <PieroV> Yes, and it has glean in logcat's rough stack trace
15:28:27 <sysrqb> mmmm
15:28:37 <PieroV> But I have a good news as well, hopefully
15:29:12 <PieroV> I've found the --disable-install-strip option, that keeps all the debug symbols also in the apk (or so it should)
15:29:30 <sysrqb> oh, nice
15:29:40 <PieroV> I'm doing a nightly build with it, and see if we get a better stack trace
15:30:21 <PieroV> In the meantime, I think I've also resolved the signing problem, by using apksign from the SDK, instead of the Debian's one
15:31:20 <boklm> nice
15:31:47 <Jeremy_Rand_Talos__> (Should a bug maybe be filed with Debian about that...?)
15:31:57 <PieroV> So, we're proceeding, but until we solve the crash we won't be able to release
15:32:30 <PieroV> Jeremy_Rand_Talos__: not really, it's just an old version. I don't think they'd fix it (unstable already has the tool from SDK 31)
15:33:04 <Jeremy_Rand_Talos__> PieroV, ok, makes sense.  Maybe they'd put it in backports, but probably not enough users for them to bother
15:33:23 <PieroV> yep...
15:34:12 <PieroV> I think that most people use the official one compiled by Google anyway along with Android Studio
15:34:23 <GeKo> PieroV: if you could take notes in your debugging process and add them somewhere that would be awesome
15:34:42 <GeKo> --disable-install-strip was a known option and it's sad you needed to re-find it
15:34:59 <GeKo> maybe we can take the opportunity to avoid that yet another time...
15:35:01 <PieroV> GeKo: yep, I'm thinking on how structuring a series of pages on building and debugging for the wiki
15:35:24 <GeKo> nice
15:35:40 <PieroV> I've added some comments on the issue that have a lot of information, and I'd like to transform them to pages, to avoid having their content lost
15:36:29 <PieroV> I have a last point on the Android updates (at least while I keep investigating on the crashes)
15:36:38 <PieroV> Right now, I'm working on geckoview v99, but Mozilla released also v100 and v101
15:37:06 <sysrqb> indeed
15:37:13 <PieroV> what should we do?
15:37:58 <PieroV> (considering that the current stable has startup crashes as well)
15:38:17 <sysrqb> my recommendation is finding a balance between debugging 99 and moving forward onto 100
15:38:30 <sysrqb> (and soon 101)
15:39:21 <sysrqb> we know falling behind on patch rebasing, toolchain updates, and other dependencies only makes the transition more difficult in the future
15:39:43 <sysrqb> but, you have a "working" build that you can debug now
15:40:14 <PieroV> hi richard!
15:40:20 * Jeremy_Rand_Talos__ waves hello to richard
15:41:09 <richard> o/
15:41:12 <PieroV> we're discussing on how to proceed with Android and we've quickly discussed about nightly builds failing because of a goutls fork possibly turned into private now
15:41:31 <richard> yeah sorry moved the meeting and then didn't update my calendar
15:41:36 <richard> so gj me
15:42:43 <PieroV> sysrqb was suggesting me to find a balance between debugging v99 (it has yet another crash, on the apk produced with tor-browser-build but not on my "local" build) and start with the v100 rebase
15:42:58 <PieroV> and I think it makes sense
15:43:10 <richard> yeah that makes sense
15:43:53 <richard> and good catch on noticing yawning's repo disappearing, I was going to bring that up too :3
15:43:54 <PieroV> the crash like the v96, so it happens at the first launch (after a while), but then it works okay, so we could even try to run some QA steps on it
15:44:41 <PieroV> yeah, I've noticed a pair of days ago, but I saw that we also have an issue on GitLab
15:44:56 <richard> what do you mean? regrading qa steps?
15:45:11 <PieroV> No, sorry, it was re the Yawning's repo
15:45:35 <richard> no no, I mean what qa steps did you have in mind
15:46:11 <PieroV> the usual stuff, trying the bootstrap, bridges, and some other steps from this document:
15:46:24 <PieroV> https://gitlab.torproject.org/tpo/applications/tor-browser/-/wikis/Quality-Assurance
15:46:36 <PieroV> but any suggestion is welcome :)
15:46:48 <richard> ooooh got you
15:47:23 <PieroV> the rationale is that we know that we have the crash, but at least we could catch any additional problem in this way
15:47:43 <richard> yes good plan
15:49:03 <richard> boklm: if you're around on irc later, i'd love to chat about the signing process :D
15:49:19 <boklm> richard: yes, we can chat after the meeting
15:49:31 <richard> at the very least, I can share the scripts we use for signing ricochet-refresh with you
15:49:36 <richard> a bit of a simpler problem, but still
15:49:45 <richard> but yeah build signing
15:49:46 <richard> wow
15:50:18 <richard> ok I don't have anything else for this meeting
15:50:24 <boklm> richard: by the way I opened tor-browser-build!437 with some improvements to the signing scripts
15:50:25 <richard> seems like y'all have things taken care of
15:50:40 <richard> love it, I will take a look asap
15:51:07 <PieroV> okay, do anyone else have something to bring up?
15:51:39 <richard> nope
15:51:52 <boklm> nope
15:52:29 <PieroV> 'kay, before ending the meeting let me remind you of the release meeting in 2 hours :)
15:52:40 <richard> also not on my calendar
15:52:42 <richard> thx PieroV
15:52:49 <PieroV> :)
15:53:01 <PieroV> Thank you all!
15:53:03 <PieroV> #endmeeting