18:01:24 <sysrqb> #startmeeting Tor Browser Meeting 8 September 2020
18:01:24 <MeetBot> Meeting started Tue Sep  8 18:01:24 2020 UTC.  The chair is sysrqb. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:24 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:01:51 <sysrqb> pad: https://pad.riseup.net/p/tor-tbb-2020-keep
18:05:23 <sysrqb> mikeperry: ahf: acat: are you updating the pad?
18:05:40 <gaba> hi
18:05:46 <GeKo> o/
18:08:08 <mikeperry> yah done
18:08:14 <acat> sysrqb: i'm done too
18:08:39 <ahf> yep
18:08:45 <acat> although i will probably have time for more todo items than the ones i picked
18:09:01 * sysrqb refreshes
18:10:00 <acat> are https://gitlab.torproject.org/tpo/applications/tor-browser/-/issues?milestone_title=Tor+Browser%3A+10.0 the ones with highest priority?
18:10:10 <sysrqb> ahf: did you update the pad?
18:10:21 <acat> well, that's only tb, sorry
18:10:24 <ahf> i think so?
18:10:26 <sysrqb> mikeperry: i assume your dates are wrong?
18:10:43 <ahf> isn't it tor-tbb-2020-keep ?
18:10:50 <sysrqb> ahf: ah, i see now
18:10:53 <ahf> ya, the one mike is updating in
18:10:57 <sysrqb> yeah, sorry. maybe just slow updating
18:11:00 <mikeperry> ah dag fixed
18:11:02 <GeKo> acat: we could debate whether #40117 should be on this list
18:11:11 <GeKo> but i put it on for now
18:11:26 <GeKo> as i stumbled over that while reviewing the 79-81 bugs
18:11:28 <sysrqb> tor-browser##40117
18:11:31 <sysrqb> tor-browser#40117
18:11:33 <GeKo> but it's affecting desktop, too
18:11:36 <GeKo> nice try :)
18:11:45 <sysrqb> tpo/applications/tor-browser#40117
18:11:47 <sysrqb> :)
18:11:53 <sysrqb> no torbot here
18:11:53 <GeKo> yay!
18:11:55 <GeKo> thanks
18:12:06 <ahf> oh, i guess we could get it in here too
18:12:35 <GeKo> tpo/applications/tor-browser#40093 seems to be annoying, too
18:12:46 <GeKo> but i don't have so much insight on the fenix side
18:12:56 <sysrqb> o/
18:13:32 <sysrqb> i think  is related to noscript click-to-play
18:13:53 <sysrqb> *#40093
18:14:10 <sysrqb> but i didn't investigate more than that
18:14:15 <GeKo> you mean we see that on stable, too?
18:14:19 <GeKo> that would be surprising
18:14:29 <GeKo> or is that the mix of the click-to-play and esr78?
18:14:38 <sysrqb> the latter
18:14:51 <GeKo> okay.
18:15:33 <sysrqb> but, while it is a bad usability issue for medium security level, i think we have higher priority tickets right now
18:15:57 <sysrqb> so i'm okay with delaying it until after 10.0
18:16:27 <acat> ok, i had put that one #40093, i can change it for sth else if needed
18:16:46 <acat> *had put it in my list
18:16:48 <sysrqb> i haven't looked at mike's network audit yet
18:16:58 <GeKo> acat: i think the fenix patch for loading the extensions is important
18:17:19 <sysrqb> but patching any of tose is a higher priority
18:17:24 <GeKo> right now, that piece is missing
18:17:25 <sysrqb> and, yes, loading extensions
18:18:07 <acat> but there are in-review MR for the extensions, right?
18:18:50 <GeKo> yup
18:19:09 <acat> well, except noscript in tor-browser-build, not sure if you had picked that one?
18:19:10 <GeKo> so, i guess going over the open items mike found during the proxy audit is it then
18:19:27 <GeKo> let's see
18:19:39 <acat> ok
18:20:03 <mikeperry> yah so some of these probably need some live app testing?
18:20:16 <mikeperry> like I am not sure how to get the system webview, if that is even possible
18:20:20 <mikeperry> but there's def code for it
18:20:41 <mikeperry> and the external app/intent stuff
18:21:37 <GeKo> acat: i think the noscript and https-e part for tor-browser-build landed already on master
18:21:38 <GeKo> 07dbd7701e89ae984bb852d4b5113707066b3df3
18:22:00 <GeKo> let me know if that's not enough, but i thought it would be
18:22:51 <sysrqb> acat: if you can create child tickets for issues we should further investigate/patch based on mikes audit
18:22:56 <sysrqb> that will be a big help
18:23:08 <acat> sysrqb: sure
18:23:11 <sysrqb> if the extension loading piece is pending review
18:23:13 <acat> it's mostly for android, i assume
18:23:18 <sysrqb> yes
18:23:22 <sysrqb> for android
18:24:03 <GeKo> i am looking at the LOAD_FLAGS_BYPASS_PROXY part
18:24:16 <GeKo> but i think that's not really meant to bypass the network level proxy
18:24:23 <GeKo> just the proxy cache
18:24:39 <mikeperry> ah, phew
18:24:46 <GeKo> but i did not file tickets for the other parts yet
18:24:52 <mikeperry> I was gonna feel real stupid if I missed that one for all these years ;)
18:24:58 <GeKo> :)
18:25:00 <mikeperry> it's in old gecko-dev code
18:25:10 <ahf> hehe
18:25:20 <GeKo> yeah, i looked briefly at it and will look closer tomorrow/on thursday
18:25:37 <GeKo> here is what mdn says:
18:25:39 <GeKo> This flag specifies that any intermediate proxy caches should be
18:25:39 <GeKo> bypassed (That is, that the content should be loaded from the origin
18:25:39 <GeKo> server).
18:25:45 <mikeperry> I saw th flag by chance looking at other stuff
18:25:59 <GeKo> so, my panic mode went away a bit
18:26:05 <mikeperry> which makes sense -- it probably doesn't actually change the proxy usage as you said, which is why I didn't note it before
18:26:12 <GeKo> yah
18:26:21 <GeKo> but let's confirm once and for all ;)
18:26:49 <GeKo> sysrqb: so, one thing that is bothering me is rbm
18:26:56 <GeKo> that is rbm#40004
18:27:02 <GeKo> uh
18:27:05 <ahf> oh wow
18:27:06 <sysrqb> woah. colors
18:27:09 <GeKo> nice, color-mode
18:27:14 <ahf> what the hell happened there
18:27:19 <sysrqb> lol
18:27:20 * GeKo bows infront of torbot
18:27:30 <sysrqb> heh
18:27:54 <sysrqb> yeah. didn't happen in #tor-dev
18:28:01 <sysrqb> in any case....
18:28:26 <GeKo> i think we get releases out without fixing that issue
18:28:32 * sysrqb opens ticket
18:28:34 <GeKo> or investigating, but...
18:29:07 <GeKo> it's kind of ridiculous to wait like 2 hours for rbm before a 10minute build is done
18:29:26 <gaba> torbot is evolving...
18:29:42 <GeKo> let alone that this hits us badly when we want to get chemspills out fast
18:30:09 <GeKo> but i suspect optimizing the dependency checking algorithm won't be a trivial thing to do
18:30:28 <GeKo> aka it's likely a huge timesink
18:30:42 <sysrqb> right
18:30:47 <acat> i could try taking a look
18:30:58 <GeKo> oh, and essentially only boklm knows that part well and it's in perl
18:31:20 <GeKo> acat: yeah, feel free
18:31:26 <sysrqb> yeah. i did look at sme of that code previously
18:31:39 <sysrqb> but it's perl and it goes deep in recursion
18:31:47 <ahf> so some graph resolution makes it take 2h to compute?
18:31:56 <GeKo> yes
18:32:00 <ahf> oh wow ok
18:32:13 <sysrqb> it's basically a depth-first search on a dependency tree
18:32:23 <GeKo> + a couple of hours more at other stages
18:32:37 <GeKo> for all the other mobile archs
18:32:55 <GeKo> so we easily have like 10hours+ rbm involved when building  mobile
18:33:20 <acat> yeah, it's painful
18:33:23 <GeKo> which is insane
18:33:48 <ahf> wild
18:33:59 <acat> i spent a whole weekend building because i had to do some changes, and rebuild a few times
18:34:03 <GeKo> acat: so, yeah, if you feel like it go for it, i think it's important enough
18:34:27 <GeKo> if we can shave off a few hours that's already a huge win
18:34:38 <sysrqb> acat: and please don't hesitate asking any questions if you get stuck or confused
18:34:51 <GeKo> i started to look into that over the weekend but i am happy to hand that task over :)
18:34:53 <acat> sure
18:34:54 <sysrqb> maybe GeKo or I or (or boklm, if we're lucky) can help
18:35:14 <acat> thanks
18:35:23 <GeKo> boklm already had some ideas on the ticket we could play with
18:35:58 <GeKo> but it felt more like workarounds to me
18:36:09 <GeKo> but maybe that could be a first step, dunno
18:36:31 <acat> ok
18:38:03 <sysrqb> okay. should we think about an alpha this week?
18:38:31 <sysrqb> even if it is only desktop
18:38:44 <GeKo> if we get some mobile nightly builds done before we could include the mobile part :)
18:38:57 <GeKo> yes, we should
18:38:58 <sysrqb> i would love inclduing that part
18:39:11 <GeKo> we need to test the https-e inclusion
18:39:17 <sysrqb> yep
18:39:26 <GeKo> i think before that should go to alpha
18:39:29 <acat> and noscript
18:39:45 <acat> but yeah :)
18:39:45 <ahf> i think the glob fix can go in tomorrow. david and i talked a bit about it today, and i'm just going to give it a test run and see that it all is OK in tor land if that is what is holding back the nightly there
18:39:50 <GeKo> we should have it in a nightly first, so we can test the update scenarios
18:39:55 <GeKo> *scenario
18:40:05 <GeKo> ahf: great
18:40:21 <sysrqb> ahf: awesome
18:40:38 <sysrqb> ahf:  do you have CI for android yet?
18:40:39 <GeKo> acat: i am not concerned about noscript for tor-browser#40091 :)
18:40:49 <ahf> for tor? nope
18:40:54 <sysrqb> ah, okay
18:41:08 <ahf> unless it happened while i was on vacation - i haven't checked that
18:41:10 <ahf> but i don't think so
18:41:20 <GeKo> sysrqb: let's aim for building on friday and releasing on mon/tue?
18:41:31 <sysrqb> ahf: yeah, okay, np. we can talk about that later
18:41:38 <GeKo> if we feel mobile is ready then let's include it
18:41:44 <sysrqb> GeKo: yes, that is the timing I had in mind, too
18:41:59 <sysrqb> yep
18:42:35 <sysrqb> i *think* mobile can be ready
18:42:50 <sysrqb> but i always hit roadblocks
18:42:57 <sysrqb> so it's difficult to know for sure
18:43:23 <GeKo> assume there are a bunch of unknown unknowns still :)
18:43:35 <sysrqb> yep
18:44:00 <acat> yeah, it has been the same for me recently with the extension bundling :)
18:44:14 <GeKo> nice work, though :)
18:44:33 <acat> after the third approach tried :)
18:44:44 <sysrqb> :)
18:44:45 <sysrqb> indeed
18:45:42 <sysrqb> okay, anything else for this meeting?
18:45:47 <sysrqb> if not, then i'll get back to work :)
18:45:56 * ahf good
18:45:57 <GeKo> i am fine
18:46:20 <acat> i'm good too
18:46:21 <sysrqb> okay, cool. thanks everyone, have a nice week
18:46:29 <sysrqb> #endmeeting