18:00:27 <GeKo> #startmeeting tor browser
18:00:27 <MeetBot> Meeting started Mon Aug 28 18:00:27 2017 UTC.  The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:00:27 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:00:30 <GeKo> hi!
18:00:52 <GeKo> and welcome to another round of tor browser reporting
18:01:14 <GeKo> mcs and brade are not here today
18:01:51 <GeKo> boklm is probably still on vacation
18:01:52 <arthuredelstein> hi everyone
18:01:59 <GeKo> ah arthur is here, hi!
18:02:11 <GeKo> arthuredelstein: do you want to go first today?
18:02:15 <msvb-lab> Hello folks.
18:02:39 <GeKo> hi
18:02:46 <arthuredelstein> Sure!
18:02:48 <GeKo> if not i can do
18:02:48 * tjr waves
18:02:52 <GeKo> o/
18:02:59 <GeKo> go ahead then
18:03:10 <arthuredelstein> This week I revised my patch for #22989
18:03:36 <arthuredelstein> I also wrote a new patch for #18101 that I am in the process of posting
18:04:12 <arthuredelstein> That latter patch will probably need more testing on legacy systems however.
18:04:40 <arthuredelstein> I also worked a bit on ubsan patches
18:05:08 <arthuredelstein> And I started looking into STACK
18:05:21 <arthuredelstein> Finally I reviewed a number of Mozilla uplift patches
18:06:17 <arthuredelstein> This week I will try to work on STACK and possibly ubsan. And I want to see what other fingerprinting and/or regression tickets we have
18:06:21 <arthuredelstein> That's it for me
18:06:42 <GeKo> how is your coming week supposed to look like?
18:08:17 <GeKo> arthuredelstein: ^
18:08:44 <arthuredelstein> sorry, not sure I understand your question
18:09:15 <GeKo> ah, sorry i did not read your last message
18:09:45 <GeKo> i think finishing up #22343 might be a worthwhile thing?
18:10:13 <arthuredelstein> no problem. Ah yes, definitely. I forgot to mention that one but I want to finish it as well!
18:10:22 <GeKo> cool
18:10:47 <GeKo> oh, could i ask you to review #23258 quickly today?
18:10:58 <GeKo> i'd like to get a new stable release out to fix that
18:11:32 <GeKo> it should be easy to test and it's basically just some whitelisting we did not do yet
18:11:35 <GeKo> arthuredelstein: ^
18:11:39 <arthuredelstein> yes, will do
18:11:44 <GeKo> thx
18:11:49 <GeKo> okay, i can go i guess
18:12:46 <GeKo> last week reviewed a bunch of tickets, #18101, #23240, #13398, #22989 and #20375
18:13:02 <GeKo> i worked on #23258 and #22692
18:13:17 <GeKo> and am currently testing the latter without selfrando enabled
18:13:47 <GeKo> my plan is to get it included into the next alpha if possible to have it availabe in the stable series as fast as possible
18:14:04 <GeKo> i spent some time on #16010 amd working further on the design document update
18:14:21 <tjr> Can I ask where you are with the sandbox?
18:14:34 <GeKo> i started to review the tor-browser-build work boklm has been working over the past weeks and months
18:15:46 <GeKo> tjr: yes, i did not make much progress alas, as i keep being distracted by other stuff. the code compiles and shipping level 0 works (which does not do much)
18:16:05 <GeKo> the error seems to me not related to the sandbox itself
18:16:14 <GeKo> (the one i get on start-up)
18:16:21 <tjr> And you're just commenting out the __try's?
18:16:27 <GeKo> yes
18:16:46 <GeKo> i have a proper debug build now and plan to spend quite some time on it this week
18:16:59 <tjr> Okay, cool.
18:17:15 <GeKo> i hope i can have some substantial progress this week
18:17:37 <GeKo> my medium term goal is shipping the sandbox enabled in the next alpha (end of sep)
18:17:42 <arthuredelstein> I guess without the __try's it should just crash if there is an error, right?
18:17:51 <GeKo> yes
18:17:58 <arthuredelstein> That seems like an improvement :)
18:18:08 <GeKo> heh
18:18:42 <GeKo> alright, this week i plan to do serious work on #16010
18:18:57 <GeKo> and hope to get a release out
18:19:18 <GeKo> then i hope we can switch to rbm officially, the next alpha is planned to get build with it
18:19:38 <GeKo> and i aim at making further progress with the design doc update
18:19:43 <GeKo> that's it for me.
18:19:51 <tjr> I can go.
18:19:57 <GeKo> please do
18:20:05 <tjr> I worked pretty much exclusively on MinGW all last week, and made a lot of progress
18:20:26 <tjr> I got a green build on TaskCluster, which involved solving several new automation-related bugs in our build system
18:20:46 <tjr> gcc for Windows exposes a ton of new warnings (which we treat as errors in automation)
18:21:01 <tjr> I started fixing them, but eventually just turned off warnings-as-errors and hope/expect to be able to land it that way
18:21:22 <tjr> I filed a few dozen warning bugs, intend to file a new dozen more, and fixed probably 75% of the ones I did file
18:21:32 <GeKo> wow
18:21:52 <tjr> I got a review on half of the most important patch to get it into TaskCluster and it wasn't that bad. I'm fixing the stuff today
18:22:04 <tjr> The outstanding things are:
18:22:35 <tjr> stylo doesn't work with mingw. Fixing it is going to be a big big thing, I think. https://bugzilla.mozilla.org/show_bug.cgi?id=1390583
18:22:42 <tjr> the sandbox
18:23:11 <tjr> enable-webrtc and the crashreporter have a bunch of small bugs I've backburnered while i work on the 'main thing' but I'll clean those up in short order
18:24:18 <tjr> And then I expect I'll have to do some debugging to understand why the browser (and maybe the installer) don't run (I think the browser at least crashes a lot) and if that's related to my hack-and-slash of skia ( https://bugzilla.mozilla.org/show_bug.cgi?id=1364560 ) or what.
18:25:16 <tjr> I mentioned in the thread about the mozilla rust/servo people coming to the dev meeting that I'd suggest cornering them to talk about the stylo build issues
18:25:20 <GeKo> tjr: re the stylo thing. i saw your mail and, yes, talking to the mozilla rust folks about a good plan is on ma to do list
18:25:26 <GeKo> ha
18:25:26 <tjr> ahha! perfect
18:25:32 <GeKo> :)
18:25:42 <tjr> That's it for me
18:26:56 <GeKo> tjr: fwiw: my sekrit plan is to modernize our linux/window toolchain in the next couple of months using a newer binutils/gcc (gcc 6.x)
18:27:12 <tjr> Oh!
18:27:15 <msvb-lab> sekrit?
18:27:28 <GeKo> i guess having this ready when we switch to esr59 does not seem so unreasonable
18:27:47 <tjr> So it's super easy for me to try out other versions of gcc now that it's in TC (at the expense of a slightly slower compilaiton)
18:27:49 <GeKo> yeah, there is no bug for it yet :)
18:28:12 <GeKo> yes
18:28:13 <tjr> I threw in 6.4.0 just to see and hit a bug, and moved on
18:28:25 <tjr> And 5.4.0 works now after I fixed an include problem
18:28:33 <arthuredelstein> Mozilla's still on gcc 4.9, right?
18:28:38 <tjr> Yes
18:29:09 <tjr> I saw the progress happening on Win64, I suspect I will have a lot of stuff to do for that as well, but I'm basically ignoring it until I get the 32 bit into TC...
18:29:30 <GeKo> that's perfectly fine
18:29:38 <tjr> Oh and one more thing! https://bugzilla.mozilla.org/show_bug.cgi?id=1392844
18:29:48 <GeKo> we are supposed to have alphas with 64bit windows later this year
18:29:48 <tjr> It's the 'Make sure Stylo respects our prefs' bug
18:30:03 <tjr> I don't know for certain what else to say in there, they think they got everything
18:31:08 <GeKo> good question in that bug
18:31:57 <GeKo> i am not sure but we already have a bunch of fingerprinting stuff upstreamed with tests
18:31:59 <arthuredelstein> tjr: So I guess my question is if they are running our 418986 tests with stylo activated
18:32:10 <GeKo> yes
18:32:21 <arthuredelstein> that's where the CSS-related tests are
18:32:54 <arthuredelstein> I think a goal of Stylo is that it should pass all existing tests, but I'm not sure if that's already in place.
18:33:18 <tjr> Okay I can put that in the bug
18:33:37 <arthuredelstein> thanks
18:34:01 <GeKo> who else is here for a status update?
18:34:12 <arthuredelstein> I guess once stylo is on by default then it will automatically be undergoing the tests on the try servers
18:35:06 <GeKo> okay, discussion time
18:35:18 <GeKo> do we have anything for that today?
18:35:22 <kingu> I am hoping to see more of the languages popular enough to see substantial work done make it into official packaging #20628 #22710 #17400
18:36:45 <GeKo> kingu: the hard one is #17400
18:37:10 <GeKo> and it is essentially the only blocker we need to figure out
18:38:30 <arthuredelstein> So is the main question related to the build process?
18:38:45 <arthuredelstein> How many bundles do we want to build, store, etc.?
18:39:05 <GeKo> that's one aspect.
18:39:25 <kingu> If so, then arguably it the smaller issue. Tor attracts heaps of translators, and it does very little in the way of comfort to work on something for years, only to see it narrowly missing the point.
18:39:30 <GeKo> but the more problematic thing is how to present the locale selection dialog to the user
18:39:55 <GeKo> in a way that it is understandable and that the user is not confused
18:39:56 <arthuredelstein> It seems like there are two separate possible tasks.
18:39:59 <kingu> I would say all the languages that are allowed onto the translation platform, that people put effort into.
18:40:38 <arthuredelstein> One is to implement a multi-lingual browser. A second is simply to increase the number of single-language bundles we build. We could work on the latter without having to implement the former.
18:41:21 <arthuredelstein> My question is, how many single-language bundles can we afford to offer? I'm not running builds so I don't have a feeling for that.
18:42:04 <kingu> The way I see it, increase community and adoption by a good amount, or start out later but being smooth about it.
18:42:36 <GeKo> we don't have the infrastructure to support localized builds like mozilla does
18:42:57 <kingu> I don't see how it is costly other than for "wasted" bandwidth once bundled into one
18:43:12 <kingu> In terms of what?
18:43:17 <GeKo> so the current plan is to ship the localized builds as we do right now
18:43:33 <GeKo> and bundle the other locales into one bundle
18:43:51 <GeKo> kingu: bundling all locales into one bundle is not optimal for a lot of users
18:44:08 <GeKo> the optimal thing would be having localized bundles for every language we support
18:44:32 <GeKo> but that's not feasible if we want to increase our reach substantially (which we want)
18:44:32 <arthuredelstein> GeKo: Would it be feasible to increase the number of localized bundles at all, or are we at the limit of our capacity?
18:44:50 <GeKo> not sure.
18:45:29 <kingu> What does it take, and what is there a lack of?
18:45:47 <GeKo> we could think about that one harder while trying to make progress on the other part which we need anyway
18:46:00 <arthuredelstein> Yes
18:46:16 <arthuredelstein> I guess the importance of languages follows a pareto distribution or something similar
18:46:23 <arthuredelstein> So there are diminishing returns for single-language builds
18:46:26 <GeKo> yep
18:46:30 <kingu> In my head the "Boy howdy, now you can get the Tor-browser in language x" is an article ready to see publishing
18:47:20 <GeKo> yeah. i think we should get the discussion  going again with the ux team and isabela and try to update #17400
18:47:20 <arthuredelstein> But there might be value in adding some localized bundles for a few more languages.
18:47:22 <GeKo> accordingly
18:47:31 <GeKo> indeed
18:47:33 <arthuredelstein> Does Firefox already have some kind of language choice prompt?
18:47:38 <GeKo> no
18:47:50 <GeKo> they ship localized builds for every supported locale
18:47:53 <GeKo> for all platforms
18:48:15 <msvb-lab> ...and the installer starts in those languages too.
18:48:22 <GeKo> yes
18:48:55 <GeKo> okay. i'll try to poke linda and see how we can get that moved forward (again)
18:49:14 <arthuredelstein> It's something I would be interested in working on if we can agree on a plan
18:49:30 <GeKo> arthuredelstein: what about a `tbb-backport` keyword to have this more transparent and not just in my head?
18:49:40 <GeKo> arthuredelstein: sounds good, i'll keep that in minde
18:49:41 <kingu> Small languages often have big impact, it seems distasteful to diminish the merits of some of them in a computer science context
18:49:43 <GeKo> *mind
18:50:22 <arthuredelstein> GeKo: I like that idea
18:51:02 <arthuredelstein> kingu: I didn't intend to diminish importance of small languages at all. Thanks for bringing up this important issue
18:51:26 <arthuredelstein> I hope we can get as many languages covered as possible.
18:51:26 <GeKo> okay. after thinking about it for a while i am a fan as well. i'll go over the tickets that could be in this group and tag them accordingly
18:52:14 <kingu> :)))
18:52:30 <arthuredelstein> kingu: And I feel the pain of translators whose work hasn't been deployed yet
18:53:06 <msvb-lab> How can we find them when searching trac, like the tag will be tbb-locale?
18:53:30 <msvb-lab> GeKo: ^
18:53:31 <msvb-lab> Or do you have another tag in mind?
18:53:43 <GeKo> what do you mean with "them"?
18:54:01 <GeKo> the tickets i was talking about?
18:54:08 <msvb-lab> The tickets that you want to triage and sort under the locale umbrella.
18:54:17 <GeKo> ah, no
18:54:26 <msvb-lab> I misunderstood?
18:54:37 <msvb-lab> I'm a fan of this effort too.
18:54:41 <GeKo> that are only those that could be candidates for a backport into the stable series
18:54:55 <GeKo> which will be markes with `tbb-backport`
18:55:01 <GeKo> *marked
18:55:02 <msvb-lab> Okay, I understand. Thanks.
18:55:14 <GeKo> the locale related things don't have a keyword (yet)
18:55:23 <arthuredelstein> I also like the idea of a tbb-locale tag
18:55:37 <msvb-lab> arthuredelstein: THere already is one ;)
18:56:07 <GeKo> maybe, dunno yet
18:56:09 <msvb-lab> ...just not being used for much. Probably some tickets need to be triaged if we want to emphasize localization.
18:56:18 <arthuredelstein> Hm, I don't see one
18:56:21 <GeKo> anything lse for today? we are running out of time...
18:56:25 <GeKo> *else
18:57:14 <msvb-lab> arthuredelstein: Ups you're right. The one I was looking at is tbb-fingerprinting-local.
18:57:20 <msvb-lab> Never mind, end the meeting...
18:57:34 <msvb-lab> tbb-fingerprinting-locale
18:58:13 <GeKo> okay. calling it then. thanks everyone *baf*
18:58:16 <GeKo> #endmeeting