mozilla :: #ateam

20 Apr 2017
05:59Silne30maja_zf|afk: thanks.
06:46jgrahamAutomatedTester: Will review today
12:48atojgraham: Mind having a look at https://github.com/mozilla/webdriver-rust/pull/90?
12:48jgrahamato: Already am
12:50atoThanks!
12:56jgrahamato: Done
12:58atoThanks again!
12:58atoAutomatedTester: Are we ready for a release then?
13:03jgrahamAutomatedTester: Please do a final run of the selenium tests to check that everything is OK
13:07atoIve bumped the relevant dependencies on geckodriver master now and updated the changelog.
13:14atoAutomatedTester, jgraham: This looks wrong: https://github.com/SeleniumHQ/selenium/blob/master/java/client/src/org/openqa/selenium/remote/RemoteWebElement.java#L96
13:14atoThe Selenium Python client, on the other hand, is correct.
13:15atoNevermind, this is where the JSON body actually gets constructed: https://github.com/SeleniumHQ/selenium/blob/master/java/client/src/org/openqa/selenium/remote/http/W3CHttpCommandCodec.java#L242
13:15atoThat looks OK.
13:16atoI think.
13:16atoNot familiar with this new Scala-esque Java 8 style.
13:17atoUh, theres lots of other suspicious things in this file though.
13:27atojgraham: I guess this is a good sign running geckodriver master against wdspec in m-c: https://gist.github.com/andreastt/ed5f7a734fea50b5fb5981bdf70648ef
13:27atomaja_zf|afk: Also, its fascinating watching the drag and drop tests (=
13:37* armenzg tries to work; ends up profiling Nightly
14:39marcojmaher|afk: ping
15:24jmahermarco: pong
15:34marcojmaher: I wanted to ask you about https://bugzilla.mozilla.org/show_bug.cgi?id=1314305
15:34bugbotBug 1314305: Code Coverage, normal, nobody, UNCONFIRMED , Enable e10s mode for mochitest plain on the linux64-ccov build.
15:34whimbooato: hi. i wonder if someone could make me a contributor for geckodriver
15:34whimbooi basically have no rights
15:35jmahermarco: yeah, I think I commented in the bug; but we have concerns that e10s will not emit coverage data for all processes and or confuse our tooling
15:35marcojmaher: yeah, I remember; I just wanted to try and see what happened
15:36atowhimboo: I cant, no. Youll have to ask AutomatedTester.
15:37whimbooi see
15:37marcojmaher: I think we just need to make sure that all processes exit cleanly
15:37atowhimboo: But you should have it (-:
15:38atoI mean, get.
15:38whimbooI will send him an email
15:47whimbooato: mind adding the following link to https://github.com/mozilla/geckodriver/issues/93? https://www.hskupin.info/2017/01/23/using-selenium-and-webdriver-to-interact-with-insecure-ssl-pages-in-firefox/
15:49whimbooi dont hvae permission for that
15:52jmahermarco: ok, hopefully this is not a big issue!
16:10atowhimboo: Done.
16:11atowhimboo: Can you look at https://reviewboard.mozilla.org/r/129960/ again?
16:40dklhttps://www.quora.com/What-do-Mozilla-employees-think-of-Chrome
16:48whimbooato: will do soon
16:49whimbooato: just noticed a problem with your interaction.flushEventLoop
16:49whimboowhich breaks our click + page load logic
16:49whimbooi think we have to remove it
16:52atoI cant really comment on that. d-:
16:52atoAnyway, Im going home soon.
17:16mrbkapDoes anybody know why I didn't get Linux builds for https://treeherder.mozilla.org/#/jobs?repo=try&revision=6874b25be46a97e7ce49017be4028b21426e3a97 ?
17:19marcochmanchester: ping
17:19jmahermrbkap: let me look
17:19jmahermrbkap: linux32
17:19jmaherI see linux64 runs
17:20mrbkapjmaher: Yeah, it ran, but I didn't get builds in the builds directory?
17:20jmahermrbkap: the build for linux64 opt is inside the tc() cluster
17:20mrbkapjmaher: is that expected?
17:20jmahermrbkap: https://treeherder.mozilla.org/#/jobs?repo=try&revision=6874b25be46a97e7ce49017be4028b21426e3a97&selectedJob=92812821 (you can see the build in the job details pane)
17:20jmahermrbkap: not really, I believe that this is a recent change that dustin did
17:20jmahernot really sure
17:21jmaherhmm
17:21jmaherpossibly related to https://bugzilla.mozilla.org/show_bug.cgi?id=1357465
17:21bugbotBug 1357465: Task Configuration, normal, dustin, RESOLVED FIXED, attributes.html says test_platform is the full test platform, but it only contains up to the `/`
17:22jmaherprobably not related to that
17:22jmahermrbkap: maybe that is normal now
17:25mrbkapjmaher: so if I want to produce archived linux builds in the future I should pass "linux" and not "linux64"?
17:25jmahermrbkap: linux == linux32; linux64 == linux64
17:25jmaherdoes that help?
17:26mrbkapjmaher: Not exactly. I was more wondering why the Linux build didn't show up at https://archive.mozilla.org/pub/firefox/try-builds/mrbkap@mozilla.com-6874b25be46a97e7ce49017be4028b21426e3a97/
17:26marcojmaher: can we make the linux64-ccov a debug build?
17:26mrbkapjmaher: Which I assume is related.
17:26marcojmaher: do I just need to add "ac_add_options --enable-debug" to "browser/config/mozconfigs/linux64/code-coverage"?
17:27jmahermarco: possibly, you might need a few other options- I would look at what we do for linux64 opt vs debug; if it is just --enable-debug, then yeah
17:28marcojmaher: ok, thanks!
17:28marcojmaher: enable-debug will also fix the e10s problem
17:28jmahermarco: prepare for increasing the chunks by at least 50% if not double
17:28jmahermarco: will you compare debug coverage vs opt?
17:28jmahermarco: why is that?
17:28jmaherwhy is it that e10s will be solved?
17:28marcojmaher: in opt builds, we are using "_exit" to terminate the content processes
17:29marco"_exit" doesn't update the coverage counters
17:29marcoin debug builds, the content process exits by returning from main
17:30marcowhich, instead, should update the coverage counters
17:30jmaheroh!!
17:30jmahermarco: good call
17:30jmahermarco: ideally we will just run 1 build config
17:31jmaherso maybe debug* is the way to go
17:32marcoI'll file a bug with the details
17:33jmahermarco: ++
17:34jmahermrbkap: I will follow up, I don't know if we post to that dir anymore
17:34mrbkapjmaher: ok, thanks a lot.
17:36jmahermrbkap: so linux* builds are done in taskcluster, win/osx are done in buildbot; this is why the linux related builds are not in archive.mozilla.org
17:38jmahermrbkap: the build is here: https://queue.taskcluster.net/v1/task/fJ0Kw1K0Q6ywG-gg3KhUFQ/runs/0/artifacts/public/build/target.tar.bz2; not sure if there is a simple way to find that; do you have a tool that looks in archive?
17:42chmanchestermarco: pong
17:42marcochmanchester: read above re e10s and ccov
17:43chmanchestermarco: that's good news it sounds like!
17:43marcoyep :)
17:43marcoI'm doing a try build now
17:44chmanchestergreat
17:44marcowith debug enabled
17:44marcoand I'll inspect the results to see if they make sense
17:45jmaherekyle: ^
17:52mrbkapjmaher: sorry, just saw your question. I don't have a tool, but I was trying to give a link to QA to test. In the future I'll point them at target.tar.bz2
17:54jmahermrbkap: ok
20:14gpsTIL about betamax, a Python library for recording HTTP requests for subsequent usage in tests https://betamax.readthedocs.io/en/latest/index.html
20:15gpsthis seems much better than mocking
20:16gpssince you can run tests in "live" mode to record requests. if an upstream API changes, you can just re-run tests in live mode to get new data. no manually mucking with mock data
20:26AutomatedTestergps: that looks cool
20:44tedgps: interesting!
20:53Silne30AutomatedTester: Hey, man. Your A-Team bootcamp doc is pretty awesome. Has it helped with community engagement for you?
20:54AutomatedTesterSilne30: its mostly created by wlach iirc (thanks though)
20:54AutomatedTesterI think it has helped a little
20:54wlachyeah it didn't help as much as I hoped it would
20:54wlachalso it's pretty out of date at this point
20:59wlachI think if I had to do it again, I would make it less specific to ateam (which no longer exists) and try to make a bunch of specific guides about how to contribute to particular areas
21:01maja_zf|afkwhimboo: AutomatedTester: re https://bugzilla.mozilla.org/show_bug.cgi?id=1358020 -- halp, could you follow-up on this if you have time today? i wonder if beta builds have different behaviour wrt to keyboard shortcuts?
21:01bugbotBug 1358020: Marionette, normal, nobody, NEW , Permaorange in webdriver/actions/special_keys.py when Gecko 55 merges to beta 2017-06-12
21:07whimboomaja_zf|afk not today
21:07whimbooi'm about to head to bed
21:07whimbooalso the broken updates have higher prio for me right now
21:07whimbooi'm sure this can wait
21:08whimboowe have plenty of time until June
21:08maja_zf|afkwhimboo: gah, they made it sound so urgent :\ --- okay, i'll take care of it when i'm back then
21:09maja_zf|afkhave a good night!
21:09whimboomaja_zf|afk: its just annoying to have so many failing tests
21:09whimboothey want green test results :)
21:09whimboomaja_zf|afk: enjoy your remaining day
21:20Silne30maja_zf|afk: Still around?
21:20Silne30Trying to get this mach command up and running.
21:21Silne30Using the firefox UI tests as somewhat of a template.
21:21AutomatedTesterSilne30: you can ask in #build
21:23Silne30AutomatedTester: Thanks.
21 Apr 2017
No messages
   
Last message: 6 days and 18 hours ago