15:01:01 <morganava> #startmeeting Tor Browser Weekly Meeting 2024-12-02 15:01:01 <MeetBot> Meeting started Mon Dec 2 15:01:01 2024 UTC. The chair is morganava. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:01 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic. 15:01:04 <morganava> welcome 15:01:10 <morganava> to december 15:01:35 <morganava> pad us usual, please update with your todos and todones, etc: https://pad.riseup.net/p/tor-tbb-keep#L254 15:06:22 <henry-x> FYI: I'm going to be away next monday and tuesday. And there's a chance I might be away at times for the rest of next week as well 15:06:34 <bellatchau> o/ 15:07:37 <bellatchau> aaaaah december, rainy days 15:07:51 <boklm> o/ 15:08:27 <morganava> henry-x: ack 15:08:29 <morganava> ok 15:08:58 <morganava> if i'm understanding calendars right, we don't have a release this week, or a rebase 15:09:26 <PieroV> No more releases until the next year 15:09:44 <PieroV> There are 6 weeks between 128.5 and 128.6 15:09:56 <morganava> amazing 15:10:00 <ma1> ESR holydays 15:10:10 <PieroV> Which is a lot, and possibly a reason for which we have one less RR version between the two ESRs 15:11:36 <morganava> my top priority for this week (apart from various grant proposals floating around) then is 14.5 stuff, specifically the rapid-release processes (code-audits, rebses, etc) and assigning issues out related to that, and following through on the plans from the release meeting last week re: labelling in milestones 15:11:51 <morganava> donuts has already started putting things in the 14.5 Milestone 15:12:02 <bellatchau> <3 15:12:32 <morganava> which you can find here -> https://gitlab.torproject.org/groups/tpo/applications/-/issues/?sort=updated_desc&state=opened&milestone_title=Tor%20Browser%2014.5&first_page_size=100 15:14:39 <morganava> it looks like we have a discussion point just added, so lets move on to that 15:14:45 <donuts> i have indeed :) 15:14:49 <morganava> specfiically: Thoughts about making tor dependencies static? 15:15:09 <PieroV> I have a couple of MRs that I kept drafted even though boklm approved them 15:15:30 <PieroV> To limit the LD_LIBRARY_PATH changes to the tor process, rather than the entire browser 15:16:09 <PieroV> I don't mind dynamic libs + LD_LIBRARY_PATH 15:16:26 <PieroV> But if we decide to go on the static libs route I'll close the MRs 15:17:30 <morganava> so basically, c-tor has various dependencies which we ship as shared libraries on some platforms (openssl, libevent, etc) 15:18:31 <morganava> which has caused some issues in the past becuse our launcher script (on Linux) tells firefox and its child processes to use our packaged shared libraries I guess under the theory that they are the most recent/up-to-date 15:19:10 <morganava> PieroV's current MR chagnes that logic to only make tor use our packaged libs rather than all of Tor Browser 15:19:16 <morganava> which woudl fix the problem 15:20:01 <morganava> however, compilng these dependencies as static libs would also solve the problem (i.e. openssl, libevent, and friends would be built directly into c-tor without *any* shared libs so tor could just run on its own) 15:20:47 <morganava> I don't have any intuition as to which approach is *better* per se, thougoh the current LD_LIBRARY_PATH patch is probably smaller/more surgical 15:21:10 <boklm> compiling as static libs sounds fine, although that might take more time to do 15:21:12 <morganava> whereas the static lib appraoch requires udpating build flags in tor's various dependencies in tor-browser-build 15:21:34 <morganava> there *is* prior art here in ricochet-build 15:21:35 <PieroV> We might have some problems if there are other libs we pull from the system 15:21:43 <PieroV> But I think we'd only load libc at that point 15:21:53 <morganava> (as this is how ricochet refresh's tor is built) 15:22:37 <boklm> ah, so that should make it easier 15:23:11 <boklm> if we already know it works fine 15:23:23 <morganava> here's the ldd output for ricochet-refresh's tor: https://share.riseup.net/#0wqtznUC3w124f-mYeUu0w 15:25:11 <morganava> hmm 15:25:18 <dan_b> oh if we're talking about fiddling with the expert bundle, can we have it ship the unversal binary for mac instead of two seperate arm64 and amd64 ones? cus I'm not sure anyones gonna want to use those. 15:25:57 <dan_b> i didn't. updated cwtch's tor, and had to rip the unified binary from TB instead of use that expert bundle 15:26:28 <morganava> I'm using those for ricochet-refresh since we ship separate x86_64 and amd64 dmgs 15:26:38 <PieroV> We're using Firefox's source code for merging the binaries 15:27:13 <PieroV> So, we have a dependency on the source code. But maybe we can add that to the expert bundle as well 15:27:35 <morganava> is that merging tooling available somewhat standalone? 15:27:41 <boklm> dan_b: how do you merge the other binaries from cwtch? 15:28:02 <PieroV> morganava: I don't know the details, maybe we can also do without that 15:28:22 <PieroV> Maybe the source/tools are used for other browser needs 15:28:38 <PieroV> And we can use something else from the toolchain we already have 15:28:57 <dan_b> boklm: i didnt have to? cwtch only ships x64 linux, x64 windows. and for android, yeah I just fetech the arm7 and arm8 libs and place them in their respective arch named jni folders 15:29:30 <dan_b> oh you mean our other deps? 15:29:49 <boklm> dan_b: I mean the other binaries included in cwtch for macos 15:31:17 <boklm> morganava: we use tor-browser.git/toolkit/mozapps/installer/unify.py 15:32:52 <boklm> but for simple binaries maybe we don't need that script 15:33:18 <morganava> yeah i would need to dig into that and see how it actually works 15:33:52 <morganava> though i suppose it lookslike either me or dan_b is going to need to do it if both of us are going to use the bundle 15:34:18 <dan_b> boklm: flutter generates a unified binary for us, but you're right, our go library we ship as two archs. it's on my todo to learn how oto combine so thanks for the tip. i only asked cus I knew from work here that we had a unified tor already 15:35:10 <PieroV> Regarding the static vs. dynamic, can we merge the MRs and think about this sometimes in the future, or should we prioritize the static thing? 15:35:50 <morganava> let's merge your MRs for now to unbreak gentoo and friends 15:36:05 <morganava> and then do the static thing sometime later in the 14.5 release cycle 15:36:15 <morganava> it feels like a nice 'intro to tor-browser-build' task 15:36:52 <morganava> so maybe a good low-priority side-project for clairehurst? 15:40:01 <clairehurst> sounds good! 15:43:26 <morganava> PieroV: re your ? in your section, please prioritize patch healththis week, and lets plan on a planning/roadmapping meeting for early next week 15:43:35 <PieroV> ack 15:43:37 <morganava> donuts, bellatchau^ 15:44:20 <bellatchau> <3 15:44:22 <morganava> i'll send out a mail w/ a pad to facilitate gathering the dev priorities for 14.5 15:45:50 <morganava> oh also, I'll be assigning out bugzilla triage issues this week for the major releases since 128 this week and working on the other rapid-release related processes 15:48:11 <morganava> any other topics to cover this morning? 15:48:17 <morganava> otherwise we can end this :) 15:48:35 <brizental> i just want to let you all know i will be working from pacific time starting next monday until the holidays 15:48:44 <morganava> !! 15:48:50 <morganava> congrats/gl ;) 15:48:57 <morganava> :) * rather no winky 15:49:03 <brizental> lol i'm not moving or anything 15:49:08 <henry-x> I have one question 15:49:24 <morganava> mmhm 15:49:28 <dan_b> ha woo 15:50:10 <henry-x> what is happening with tor-browser!684 15:50:59 * morganava re-reaidng 15:51:02 <morganava> re-reading* 15:51:57 <morganava> ahhh 15:52:13 <henry-x> it is kind of blocking mullvad-browser#312, unless we don't care about providing an obvious way to distinguish mullvad-browser nightly or alpha users 15:53:46 <morganava> how do these two issues interact with the other issue about have a formalised/supported way of detecting Tor/Mullvad Browser vs Firefox? 15:54:23 <henry-x> moreover, as-is, I think the patch would break how drop and TZP detect that the user is using Tor Browser 15:54:36 <henry-x> *drop -> securedrop 15:54:48 <morganava> right 15:55:03 <PieroV> I would be against making it too obvious though 15:56:09 <morganava> PieroV: the 'supported' detection you mean? 15:56:13 <PieroV> Yes 15:56:21 <henry-x> I don't have a strong opinion on this, since with enough work someone can detect your version by other means. I would personally just hide the `branding` assets from websites 15:56:36 <PieroV> At a certain point we were asked to create a document.isMullvadBrowser or something like this 15:56:37 <henry-x> and we can worry about the rest some other time 15:56:39 <PieroV> Thanks but not thanks 15:59:12 <morganava> so, it sounds like we should implement the allow-listed TB.css/MB.css thing for the unspoofable detection, merge something like tor-browser!684, and then drop mullvad-browser#312 entirely yeah? 15:59:34 <morganava> which would make everyone happy? 16:00:34 <morganava> well in any event we're out of time 16:00:41 <morganava> let's follow up offline 16:00:44 <morganava> #endmeeting