mozilla :: #testpilot

10 Aug 2017
02:09fwiwCaspy7: nice
11:43Caspy7heh, yet another Mozilla dev suggesting they'd like Container Windows
11:43Caspy7with another replying to agree of the benefit
11:44Caspy7not a hard sell, just another datapoint
11:44Caspy7at the very least may be a good initial webextension to build
12:02fzzzyyeah i think container windows would make more sense for me too
12:04fzzzywe get some test pilot exceptions in sentry from a user that is running firefox 11 :o
13:42fwiwis there any interesting usage for the container tab outside of the default one?
14:41fzzzyfwiw: logging into different accounts on the same service in different windows
16:38JSON_voorheesThis thread is awesome
16:38_6a68Caspy7: sure, but what happens when you tear a tab off of container window A and move it into container window B? there are a ton of edge cases between tabs/windows
16:41Caspy7_6a68: do the same thing we do with Private windows :)
17:04fzzzy_6a68: i think it would have to just be disallowed
17:34_6a68I dunno, I think we have to build on the user's mental model of how tabs/windows work, that's the incredible challenge of UI engineering :-P
18:07b4handfzzzy: lorchard is there something needed to be done to sign the testpilot addon? It won't install on circleci, I am getting a geckodriver error that says it is not signed
18:08lorchardOh crap, yeah, Firefox just changed the signature requirements and we don't sign at all on CircleCI anymore
18:08lorchardnot sure what to do about that though :/
18:09lorchardMight need to defer testing addon installation to figure out how to make that work
18:09b4handIt would need to be signed when it is built correct?
18:10lorchardWell, we only sign on production release now
18:10fzzzywe used to sign it on circle but stopped a few days ago
18:10lorchardCan be built unsigned and installed via about:debugging manually
18:10fzzzymaybe all those prefs related to allowing unsigned addons would work?
18:10lorchardNot sure how that works in automated testing
18:11chuckDo we have any idea what is causing the rash of uninstallable add-on and inaccessible experiment detail page reports we've been getting?
18:11chuck#2755, #2763, #2749, #2746, etc
18:12lorchardNo idea yet, but been looking into the errors in 2746 yeah
18:13chuckI've tried to repro in Windows 10 and on macOS on each release channel and haven't seen any problems at all.
18:13lorchardThat is the problem, yes
18:14chuckLet me know if I can help winnow it down, stuff I'm working on is pretty low-priority atm.
18:14lorchardBeen trying to repro, looking at sentry reports, no leads so far
18:14lorchardNo idea what to say to help other than to poke at things
18:15chuckOn it!
18:15lorchardTried reproducing on OS X, Windows, and Linux on nightly, dev edition, and release and can't reproduce myself
18:16b4handfzzzy: I have those prefs set...
18:16lorchardand the stack traces in sentry reports don't really point at anything in our code, which makes me think it's something async somewhere that lost association with the place where the problem actually happens
18:17fzzzyb4hand: ok well i guess not then
18:17fzzzyone person in the first bug report said it went away when they used a new profile. maybe they would be willing to donate the broken profile for testing.
18:20fzzzyI really don't understand the tracebacks where require is the top function in the stack
18:22lorchardYeah, I don't really understand what's happening in those stack traces
18:26fzzzywell, time to switch off after a few years of no problems. it's been consuming >100% cpu regularly for a week or so
18:29lorchardYeah, I switched to Atom along with switching to Windows, but I switched back to Vim yesterday
18:32bwintonI moved to VSCode. It's been decent.
18:36fzzzyI'm going to try again. I used it for at least a decade
18:48fwiwJSON_voorhees: >_< that thread... I don&#39;t know if they&#39;re trolling or not
18:52fzzzyi don&#39;t think so
18:55b4handfzzzy: lorchard apparently there prefs only work on nightly/dev and on circle we install release/beta....
18:55* b4hand feels stupid
18:56b4handMaybe want to add that to the docs
18:57JSON_voorheesfwiw: lol i know, i think it&#39;s just an older guy. I like what he is saying though
18:57SeburoHiya. Is there any news on the GNU/Linux Nightly issue? (Not chasing....honest)
18:57JSON_voorheesfwiw: just not really understanding the current landscape
18:59JSON_voorheesHe doesn&#39;t understand that it&#39;s out of our hands to support xp
19:00fwiwXP is too successful I think
19:01fwiwit makes people reluctant to part with them :D
19:03Caspy7_6a68: I did test and we disallow tab dragging from private windows to normal
19:04Caspy7so it create mental model that &quot;this thing is isolated&quot;
19:04Caspy7cannot interact outside its box
19:06_6a68oh yeah, that&#39;s a good point
19:12fzzzySeburo: it is a platform issue. is it still not fixed? we should escalate the platform issue, if we can figure out exactly which one it is
19:12Seburofzzzy: Ok. Let me try here (and now) and I will let you know.
19:14JSON_voorheesfwiw: os makers have learned from that, it shows by the quality of desktops os&#39; now of days
19:15Seburofzzzy: I have installed the main TP add-on, but when I try to install an experiment, I cannot open the experiment page.
19:17fzzzySeburo: does that happen with a fresh profile? we&#39;ve been getting a lot of reports of that but can&#39;t repro (save the failing profile to help debug)
19:19Seburofzzzy: That is on an existing profile. I cannot try with a fresh profile (in the middle of a SUMO day right now), but will do in the next few days and will let you know.
19:20fzzzySeburo: do you see an error in the console?
19:20fzzzymaybe this error we can&#39;t repro is on linux
19:21SeburoUsing webconsole from the dev tools, I see:Unhandled promise rejection
19:21SeburoTypeError: is null
19:21SeburoStack trace:
19:21Seburo[object Object]
19:21Seburohold on, there is more....
19:28fzzzythat helps thanks
19:31Seburofzzzy: I am also seeing &quot;Content Security Policy: The page&#39;s settings blocked the loading of a resource..&quot; followed by website addresses.
19:34fzzzyok, that is something different. i know what the e is null error is now!! thanks!!
19:34fzzzyi&#39;ll open a pr later tonight
19:36Seburofzzzy: I hope that helps. All the website addresses look like links to fonts.
22:29Caspy7I&#39;m going to bed soon (I think) so not real awake, also not real familiar with it myself, but someone may want to see about answering this one
11 Aug 2017
No messages
Last message: 11 days and 7 hours ago