15:01:16 <richard> #startmeeting Tor Browser Weekly Meeting 2022-02-28
15:01:16 <MeetBot> Meeting started Mon Feb 28 15:01:16 2022 UTC.  The chair is richard. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:16 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:01:33 <richard> hello everyone!
15:01:40 <boklm> hi
15:01:41 <aguestuser> o/
15:01:57 <richard> it's monday, please do the usual (fill out the pad, makr points for discussion, ensure your kanban boards are up-to-date, etc)
15:01:58 <richard> https://pad.riseup.net/p/tor-tbb-keep
15:02:17 <Jeremy_Rand_Talos_> Hi!
15:08:53 <richard> ok, looks like nobody has anything in bold, but we'll give anotre minute or two for updates
15:09:51 <PieroV> I've added one now :)
15:11:20 <richard> ahem
15:11:40 <richard> ok onto discussion
15:11:59 <richard> aguestuser: di dyou want to address any of your point smissed last week?
15:12:03 <aguestuser> nope!
15:12:17 <richard> alrighty then, PieroV it's you then
15:12:20 <aguestuser> adding one about instrumentation tests...
15:12:55 <PieroV> I've seen some of you added some comments on what I wrote about the existing docs, thanks :)
15:13:16 <PieroV> Does anyone has anything to add, besides the comments on the pad?
15:13:48 * Jeremy_Rand_Talos_ checks pad again, last looked at it last night...
15:14:51 <Jeremy_Rand_Talos_> Nothing else on my end wrt the docs survey pad
15:15:09 <richard> nothing to add, though I do want to say that survey is v thorough and I approve of all the suggestions :)
15:15:40 <Jeremy_Rand_Talos_> one minor thing I probably should have mentioned,
15:16:54 <aguestuser> PieroV: it's an excellent survey and i learned about things i didn't know existed by reading it! :)
15:16:59 <Jeremy_Rand_Talos_> is that the Tor-Friendly Applications docs actually are undergoing revision right now; there's a thread on tbb-dev@ where I'm running revisions past richard
15:17:40 * Jeremy_Rand_Talos_ will update that thread later today if I don't forget
15:17:52 <richard> it's been quiet for a month :)
15:18:09 <richard> but yes we need to make sure we don't lose any in-progress doc updates as we migrate
15:18:22 <Jeremy_Rand_Talos_> richard, yeah, I typed up a long reply 2 weeks ago and forgot to hit Send :P
15:18:34 <PieroV> okay, thanks! I am going through the test issues right now, but I can also start with the docs if you prefer
15:19:05 <richard> but  I bet you def got the dopamine hit of finishing that task :p
15:19:20 <Jeremy_Rand_Talos_> lol
15:19:45 <richard> PieroV: assuming there aren't any blockers on tor-browser#40773 and tor-browser#40774 I'd say do the docs migrations, then test fixes
15:20:19 <richard> (I'm planning to get these into alpha/nightly this week after I do the esr rebase today)
15:20:26 <PieroV> There are still a few comments of mine that haven't been answered, yet
15:20:43 <PieroV> On tor-browser!254, I mean
15:21:22 <richard> ok assuming they are dev (not design) related we can follow up on this offline
15:21:28 <PieroV> And I'll have to open a MR for strings (I was waiting for them to be definitive on tor-browser)
15:22:25 <PieroV> it's dev stuff, but minor
15:22:59 <PieroV> There's still a problem with sizing, IIRC
15:23:15 <richard> hmm ok good to know
15:23:18 <richard> alright beyond that
15:23:49 <PieroV> right, the cards are not done yet
15:24:06 <PieroV> we never discussed the carousel thing
15:24:33 <richard> iirc there were some design q's around how to best do that right?
15:24:40 <PieroV> (since now we don't have any way to change bridge lines, the carousel could be a nice way to disable some of them)
15:24:50 <richard> given that certain configs send over 10 or so PTs to tor
15:24:56 <PieroV> yeah, exactly
15:25:33 <richard> ok, I think for alpha a sub-optimal UX for the preferences bridge display is acceptable given all the other new features
15:25:40 <richard> unless you object donuts
15:26:01 <donuts> it's been on my plate to look into, but I've been stretched pretty thin recently
15:26:10 <richard> i know you are being hammered by UX tasks left right and center
15:26:17 <donuts> what am I objecting to? pushing to alpha?
15:26:46 <PieroV> pushing to alpha with bridge cards not yet complete
15:26:47 <richard> going with the current bridge carousel implementation in alpha despite it not being what we eventually want to ship in 11.5
15:27:07 <PieroV> I haven't implemented the carousel, yet
15:27:19 <donuts> oh well I'm halfway through a design review of the last build, gimme a few more days to complete that first if that's okay
15:27:20 <PieroV> it was just an idea donuts had
15:27:34 <donuts> yep, I can provide some design guidance there too
15:27:39 <richard> mk
15:27:45 <donuts> thanks both :)
15:28:09 <richard> so a heads up for timelines, I'm planning on rebsaing for alpha next Monday, so if we can get sign-off on the design sooner rather than later that would make my life a bit easier
15:28:20 <richard> on the design build*
15:28:29 <donuts> it's gonna be fixes rather than signoff
15:28:29 <richard> design review** etc etc
15:28:39 <donuts> can we hold off until a later alpha?
15:28:52 <PieroV> I think we can
15:28:56 <richard> that works for me
15:28:59 <PieroV> I'd have to rebase the MR anyway
15:29:03 <donuts> awesome, ta :)
15:29:09 <donuts> I'll start posting them asap
15:29:16 <PieroV> thanks
15:29:31 <PieroV> we could drop a commit from the alpha
15:29:34 <richard> (so that'd be in April)
15:29:45 <PieroV> The one that implements a pair of unit tests
15:30:01 <PieroV> one is already covered in tbb-testsuite, for the other one I opened a MR
15:30:40 <PieroV> The commit to drop would be Bug 12620: TorBrowser regression tests
15:30:53 <PieroV> https://gitlab.torproject.org/tpo/applications/tor-browser/-/commit/ff1d9e7b2e22ae7dc7e4a998db655de2b05aa567
15:31:37 <richard> ok neat
15:31:51 <richard> be sure to note that in the MR if you haven't already
15:32:24 <richard> alright the only other thing on my end to mention
15:33:19 <richard> i'll be implementing boklm's proposed work-flow around tracking backport and fixed issues for releases going forward so hopefully this will make our lives easier
15:33:40 <richard> and of course, I'm now without my chonky dev PC atm, so I'll need a pair of builders for 11.07 on Friday
15:33:58 <aguestuser> "chonky"! ;)
15:34:05 <richard> he's a v beefy boi
15:34:19 <richard> but my laptop is not up to the task :p
15:34:23 <PieroV> I can build it
15:34:33 <boklm> I can build too
15:34:37 <richard> oh and I forgot to mention
15:35:15 <richard> sysrqb, boklm: my yubikeys arrived, so perhaps this week we can walk though the release signing process
15:35:46 <richard> anyway I'll ping you offline after this meeting
15:35:55 <richard> ok, aguestuser you're next in the discussion q
15:36:13 <aguestuser> i tried to queue up context in the pad
15:36:17 <aguestuser> (line 67)
15:36:47 <sysrqb> richard: that was fast
15:36:48 <aguestuser> it's regarding review of tor-browser-bundle-testsuite!17
15:37:00 <sysrqb> richard: wfm
15:37:40 <aguestuser> upshot is that the script for downloading and runing the tests could use a small few tweaks (mainly to use new versions of things -- including onion v3 addresses)
15:38:09 <aguestuser> but the real blocker seems to be the underlying tests -- i would expect them to always fail
15:38:14 <PieroV> why do we need to use an onion service for downloading?
15:38:14 <aguestuser> (all of them)
15:38:46 <aguestuser> PieroV: it's a good question! but i guess if it's true that the tests are totally borked, my question is whether it is worth paying any attention to the tooling at all until they are fixed
15:38:53 <aguestuser> and fixing them might take a minute
15:38:57 <PieroV> in desktop we don't test bootstrap from Tor Browser
15:39:16 <richard> my preference is to only have tests running automatically once they are passing
15:39:35 <richard> otherwise it's more noise in the inbox to ignore/incorrectly ignore
15:39:36 <PieroV> We bootstrap it before launching it, and pass the details through env variables
15:40:09 <aguestuser> richard: same here. so if we do that, should we split out the scrip that runs the desktop tests so we can get that into CI?
15:40:26 <sysrqb> i suspect the onion service is irrelevant now and it was due to our Nightly builds being served from an onion service (until ~1 year ago)
15:40:33 <aguestuser> PieroV: i'd love to look at that! we'd still need to inject some logic to bypass the homescreen
15:41:02 <aguestuser> and i need to try to understand how we want to do something similar to / different from moz's MockWebBrowser
15:41:21 <aguestuser> (i'd expect we need somethign vastly different but have no understanding of if we have any prior art for that)
15:41:22 <richard> aguestuser: seems reasonable to me, what's the alternative?
15:42:14 <aguestuser> richard: alternative is merging with failing android tests. seems non-controversial we don't want to do that!
15:42:33 <richard> ah I see
15:42:36 <aguestuser> but.. checking in b/c PieroV and i are "reviewing" an MR (from about a year ago) that i am proposing we completely reject and restructure
15:42:45 <PieroV> +1
15:42:46 <aguestuser> into 2 different cards with different scopes
15:42:58 <PieroV> I was about to ask if we really need the new desktop script
15:43:17 <PieroV> (the one you'd split out from the current MR)
15:43:19 <aguestuser> PieroV: say more... -- i had ignored that part of the MR...
15:43:47 <aguestuser> PieroV: if the desktop script adds nothing useful we could just close the MR and reference it from a new issue with the scope of first fixing the tests
15:44:07 <PieroV> What was the MR number again?
15:44:20 <PieroV> tor-browser-bundle-testsuite!14 right?
15:44:34 <aguestuser> yes!
15:44:38 <aguestuser> (sorry, i put !17)
15:45:39 <PieroV> I think actually boklm can give more insightful comments about that script
15:45:41 <boklm> yes, I think we don't need the desktop script as we already have some code for downloading the latest nightly (which we can reuse in the android case too)
15:46:55 <aguestuser> when picking up the new issue (to actually get android tests working), i could use some input from sysrqb on any prior art for (1) creating (or faking) tor connection (so that we can proceed to main browser screens), (2) performing (or mocking) http calls (so that we can assert useful things in screens)
15:47:11 <aguestuser> doesn't need to happen in this meeting!
15:47:25 <aguestuser> or prior art from richard and PieroV about how we do this on dekstop!
15:47:44 <aguestuser> doesn't need to be in this meeting -- this meeting just seems like a great time to get people's attention! :)
15:47:53 <PieroV> on desktop we use this: https://firefox-source-docs.mozilla.org/python/marionette_driver.html
15:48:00 <PieroV> but it's deprecated
15:48:29 <PieroV> so I don't know if it can help for Android
15:48:39 <aguestuser> PieroV: there's a whole "robot" framework for driving the UI
15:48:50 <aguestuser> (they don't seem to be able to keep metaphors straight across platofrms?)
15:49:00 <aguestuser> and it's pretty similar to stuff i've used for android UI tests before
15:49:16 <aguestuser> mainly curious how we go about faking/stubbing/mocking http calls and/or tor circuits
15:49:19 <aguestuser> (if at all)
15:49:52 <aguestuser> (moz seems to feed a whole bunch of canned http responses back to the UI, and i want to develop an opinion on how/if we'd do similar)
15:49:54 <PieroV> we use that driver to load real pages
15:50:18 <PieroV> but I think we also start a HTTP server to load local .html files
15:50:29 <aguestuser> kk.
15:50:39 <PieroV> I'm not sure though
15:50:42 <aguestuser> if i can get away with not mocking/stubbing too much to get started that would be my pref
15:50:53 <aguestuser> the env var injection you mentioned above... where does that live?
15:51:11 <PieroV> TBBTestSuite/TestSuite/BrowserBundleTests.pm
15:51:22 <PieroV> there's a function called pre_tests
15:51:30 <sysrqb> (marionette is responsbile for running a local web server)
15:51:58 <aguestuser> sysrqb: aha! so different responsibility from robots which just drive ui
15:52:26 <Jeremy_Rand_Talos_> PieroV, are you referring to the TOR_SOCKS_HOST etc vars, or is this something different?
15:52:29 <aguestuser> PieroV: i will look at `pre_tests`
15:53:05 <PieroV> yes, exactly that ones
15:53:21 <aguestuser> sysrqb: also curious -- before we introduced the tor connect screen -- did these tests pass? or was this the first effort to get passing tests (and then they got obsoleted by introduction of tor connect screen?)
15:54:16 <Jeremy_Rand_Talos_> PieroV, ok cool.  This ties in with the Tor-Friendly Apps docs then, because that doc is where some of those vars are documented (though not as completely as I'd like)
15:55:17 <aguestuser> regardless, i think PieroV and i have a good outcome as for what to do with tor-browser-bundle-testsuite!14. thanks all! :)
15:55:50 <richard> alright great
15:56:03 <sysrqb> aguestuser: there is a gradle build parameter we use in Gitlab CI that bypasses the connect screen
15:56:03 <richard> any last minute topics in our last 4 minutes?
15:56:22 <PieroV> richard: do you want to take torbutton!72 for review if I continue the MR?
15:56:22 <Jeremy_Rand_Talos_> one minor thing
15:56:44 <PieroV> Its author told me they can't continue working on it for the Ucraine war :(
15:56:49 <Jeremy_Rand_Talos_> boklm, is the ARM MR still blocked on me at all?
15:57:00 <richard> PieroV: yeah that works for me
15:57:06 <richard> yeah i saw
15:58:17 <richard> yeah I noticed lenovo is releasing an ARM-based thinkpad relatively soon
15:58:26 <PieroV> (also another q: in the calendar we have the release meeting also for today, are we going to have it?)
15:58:58 <richard> yes I believe so
15:59:05 <PieroV> okay, thanks
15:59:05 <donuts> I think the frequency is supposed to be every two weeks
15:59:26 <richard> alright with that i'm a call it
15:59:33 <richard> have a good day/evening everyone
15:59:33 <sysrqb> o/
15:59:35 <richard> #endmeeting