Hello!
Below are the notes from our latest Tor Browser meeting. As usual the IRC log can be found at meetbot.debian.org:
http://meetbot.debian.net/tor-meeting/2018/tor-meeting.2018-07-23-17.59.log....
The notes from our pad are:
Tor Browser Meeting Notes
Monday July 23, 2018 Discussion: -I [tjr] read through the Hacking document. Had a few observations: "For historical reasons, Tor Browser tickets are spread across several Trac components: "Tor bundles/installation", "TorBrowserButton", "Firefox Patch Issues", and "Tor Launcher". We are considering consolidating many of these components and switching to keywords instead, but that hasn't happened yet." I don't think this is still true? https://trac.torproject.org/projects/tor/wiki/doc/TorBrowser/Hacking#Howweus...
Maybe we could move Orfox/Orbot to another page? [GeKo: I'll open a ticket for restructuring the Hacking document]
when a new nightly build is available on https://people.torproject.org/~linus/builds/. <- I don't think this is active anymore?
[boklm: ah yes, this part of the page (about QA and Testing) needs some updates][GeKo: I'll open a ticket for this item]
-How do we want to use Github (if at all)? [GeKo: I'll think a bit more about it and then wil write a mail later this week as reply to Isa getting the discussion started on the mailing list taking into account what we discussed during the meeting] -Moving away from storm to some etherpad for meeting notes? [GeKo: We'll decide that next week when Arthur is back] - Sandboxing meeting: what should we prepare? Any agenda? (Did I miss it?)
mcs and brade: Last week: - Finished reviewing undocumented bugs since FF52esr for #22074. - Debugged and created a workaround for #26514 (intermittent updater failures on Win64 (Error 19)). - Tried to test Snowflake on macOS 10.9 for #26251 (Adapt macOS snowflake compilation). - Got sidelined due to #26876 (tor.real fails to start on macOS 10.9). - Helped with triage of incoming tickets. - Reviewed a couple of patches: - #26603 (remove obsolete HTTP pipelining prefs) - #26353 (First request goes over catch-all circuit) This week: - Prepare for and attend sandboxing futures meeting. - Start on #25695: Activity 5.1: Redesign Tor Browser homepage ("about:tor") - Review our notes from #22074 (undocumented bugs since FF52esr) and file additional tickets if necessary. - Get back to #26381 (about:tor page does not load on first start in localized Windows bundle) - After the Tor Browser nightly builds include the network team's fix for #26876, do some testing for #26251 (Adapt macOS snowflake compilation to new toolchain).
pospeselr: Moving house tomorrow and Tuesday so won't be in the meeting Won't have internet installed until *sometime* on Wednesday Last week: - Portland - #26540 patch works now (pdfjs range-based request first party isolation) - failed to reproduce windows DNS leak we got email'd about - fixed up #26456 patch and ran automated try server tests to verify changes don't break anything (haven't had a chance to look at results yet) This Week: - moving
- #21785 (storage api)
- #tbbfingerprinting
boklm: Note: currently traveling to PETS so I will be missing second half of the meeting (after 18:30 UTC) Last week: - vacation This week: - finish dealing with backlog - attending PETS - can help with building a new alpha if we want to do one this week [GeKo: Not this week, but hopefully next week] - fill binutils ticket
tjr
- More #26476 work. I have reduced even more differences between the two toolchains, still crashes.
- Confirmed that the TC build runs though: https://treeherder.mozilla.org/#/jobs?repo=try&revision=9a1a8cc2c8224e38...
- Confusingly though, if I remove --disable-maintenance-service I get a NSS build error.... If nothing else gives me a result I guess I'll dig into that more
- I'm getting pushback on landing required NSS patches into -esr60... so I am delaying pushing back on that until I have more ducks in a row regarding tests running [GeKo: What are we talking about here? And what would be the impact if we don't get those patches landed? I wonder if we could cope with that and maybe it would be more useful to focus on getting nightlies with mingw-w64 running again?] [tjr: The impact would be that we can't get the gcc or clang MinGW builds running in ESR branch. I have been focusing on that rather than mingw-clang on -central.] [GeKo: Well, RyanVM told me a while ago, that tests for mingw-w64 are running (to catch regressions due to security bugfixes etc. so, I was under the impression that there are things running on ESR60 for us, even it is not exactly what we ship, hence my question][tjr: esr60 is running the x86 build only. I forgot about that. These NSS patches are holding up the x64 build and tests (assuming I can the damn tests fixed for whatever else is breaking them).] [GeKo: Okay, the plan here is to let the 32bit esr60 on Mozilla's infrastructure and the manual 64bit testing before release be enough and focus on mozilla-central again instead]
- But that is going slowly/hard to allocate time for
- This is holding up landing the build jobs for x64 or mingw-clang
- I don't think the mingw-clang build (with stylo) is as stable as Jacek thinks it is [GeKo: Why not? Any bug reports/pointers?][tjr: I don't know anything more than "I tried to run it and it didn't always start for me, or stay running. Haven't investigated, or compared with/without stylo or asked Jacek.]
- A few other outstanding issues for MinGW:
- jemalloc on mingw-gcc (This is a shame, because it means Tor Browser will be lacking exploit mitigations)
- mingw-clang sandbox
sysrqb: Last week: Received good feedback on TBA/Orfox rebased patches (26401) and worked on revisions Attended HOPE Spoke at HOPE Contributed to joint statement against white nationalists/fascists at HOPE and in our spaces Talked with some people about sandboxing Tor Browser This week: Finish 26401 Read Isa's sponsor 8 mail and respond Sandboxing meeting tomorrow at 15:00 UTC Review igt0's torbutton modification for mobile
sisbell: Last Week: Android toolchain + licenses working in RBM Worked on getting arm target working on Rust (nearly complete) This week: Complete arm target on rust Working browser on Android Create branch for public review Cleanup build changes
GeKo: Last week: -coped with my backlog -bunch of reviews (most importantly, and most time-consuming the one for #26401; then #26590, #26795, #26477, #26569, #26216, #9145, #25485) -worked further on #26475 (there might be light on the end of the tunnel due to alex's last comment given that I can already feel glandium's "this-drives-me-crazy" mentioned in https://github.com/rust-lang/rust/issues/52044. Thus the bug has to be the same. :) ) -helped with Sponsor8 reporting (sysrqb, igt0: please add items I forgot, see tbb-dev mail) [sysrqb: yes!] -worked a bit on #26628 (backporting an Origin Attributes bug that could lead to browser crashes) -looked a bit a proxy bypasses on mobile and resumed network code review (#22176) -looked over remaining UX issues for both mobile and desktop alphas This week: -finally tracking down #26475 and hopefully finding a fix for it -more reviews -more network code review (#22176) -I need someone looking at #26874 (pospeselr can you do that?) and I need someone working on the onboarding part (#25695) (mcs/brade could you do that); sysrqb, igt0: who has #25696 on his plate? [sysrqb: I can put it on my plate, unless igt0 wants it] [igt0: either way is fine] -PSA: I have to be AFK from 8/2-8/12, alas (this is one of the things I could not postpone as the planning for it where already done before Mozilla moved the ESR release, I am sorry for that)
igt0: Last week: - Tried to make the preferences.xul work on mobile, started to migrate the code to xhtml (#26884) - Investigated how the onboarding works on android and reviewed #25696 (Design of alpha onboarding for Tor Browser for Android) - Investigated about MAR on android. (Firefox team never used it) This Week: - Finish #26884 - Review again #26401
sukhe: Last Week: - I continue with #26476 (crash on Windows) with tjr, #12968 (HEASLR). - I have given up on up #26251 (Snowflake) again. I am going to wait for the mingw-w64/clang builds perhaps. - Worked on #25485; waiting for more discussion to finalize This Week: - Continue the above tickets unless something comes up.
Georg