mozilla :: #ateam

20 Mar 2017
02:49ekylegood day!
02:50mikelingekyle: :)
02:50ekylemikeling: glad I made someone smile. but wrong channel :)
02:52mikelingekyle: ah, it's ok. ;)
09:32whimbooTomcat|sheriffduty: good morning
09:35Tomcat|sheriffdutywhimboo: pong
09:35Tomcat|sheriffdutywhimboo: good morning
09:35whimbooTomcat|sheriffduty: do you have any idea why this is happening?
09:35bugbotBug 1285340: Firefox UI Tests, normal, nobody, NEW , TEST-UNEXPECTED-FAIL | TestDirectUpdate.test_update, TestFallbackUpdate.test_update | AssertionError: u'20160706030233' != '20160707083343' (target builids different)
09:35whimboothis looks very strange
09:36whimboomaybe lots of mahcine time is spent on that without a purpose
09:36Tomcat|sheriffdutywhimboo: seems this is something for releng
09:36whimbooso you don't know anything about it I assume
09:37whimbooI will file a Releng bug then
09:37Tomcat|sheriffdutyno not really
09:37Tomcat|sheriffdutywhimboo: or needinfo people like callek to the bug
09:37whimbooi think i will file a new bug.
09:42whimbooTomcat|sheriffduty: it's bug 1348758 now
09:42bugbotBug General Automation, major, nobody, NEW , Why do we build multiple Aurora nightly builds per day?
09:44Tomcat|sheriffdutywhimboo: i guess it has to do with taskcluster now involved
09:44Tomcat|sheriffdutyso we have taskcluster and buildbot nightlys
09:44Tomcat|sheriffdutyon trunk /aurora
09:45Tomcat|sheriffdutylinux vs osx and windows
09:46whimbooTomcat|sheriffduty: but we have three builds
09:46whimbooTomcat|sheriffduty: also I don't see buidlbot nightlies for aurora
09:47whimbooits all tc
09:48whimbook, i see it now
09:48whimboomultiple N jobs for TC
09:48whimbooeven if successful before
10:46AutomatedTester|cloudMorning all
10:46whimboogood morning!
10:46facewhimboo, good day sir!
10:47whimbooface: i was not talking to you! :)
10:47* face isn't sure how familiar you are with python execution environments
10:48whimbooface: it won't help you to distinguish between different conversations
10:48* face feels like he is missing the joke
10:48AutomatedTester|cloudFace joke
10:48faceI was thinking those minerals were extracted from the ground, but it turns out they weren't. So never mined.
10:49AutomatedTester|cloudFace haiku
10:49faceweaving my river
10:49facei refriended my fingers
10:49facebefore my clothing
11:25AutomatedTesterjgraham: regarding my sendKeys patch, looks like we need to update Firefox to match our expectation here
11:37jgrahamAutomatedTester: Probably. I think ato is back today?
11:38AutomatedTesterI will do it now
16:19janxGood morning #ateam! I have an exceptional semi-conflict with today's meeting: I can't stay longer than 30min, so if it runs over I'll have to drop out, cc jgriffin (also, welcome back!)
16:20jgriffinno worries
16:36sfinkI wonder if "31 failures per 1000 pushes" would sound more impressive than 0.031 failures/push
16:36sfinkbetter than millifailures, or failures per kilopush, at least :)
16:37jgrahamidk 31000 failures push^-1 sounds pretty good to me
16:38* sfink covers his eyes
16:39jmahersfink: heh; it it really hard to quantify the failure rate in a single number
16:55tedwe should follow netflix's lead and just change the orange factor into a thumbs up / thumbs down
17:00imaduemeI like it
17:01jgriffinmcote: ++
17:03whimboodylan: mind muting yourself?
17:05jmaherted: so always thumbs down?
17:05tedjmaher: well i guess that's the obvious pessimistic view :)
17:05globwlach: great question
17:06* ted forgot about the meeting
17:06teda meeting for a team that no longer exists
17:06tedis this now a postmortem?
17:08dylanwhimboo: can you hear me? the red light is on my headset
17:09globdylan: you're transmitting audio
17:09globdylan: make sure echo cancel and the other option are disabled in vidyo
17:10tedand i guess when your manager is leaving it's not quite the same
17:10dylanglob: why are those settings reset? they're off now
17:10ted"chaos orgchart"
17:12emorleyeugh Vidyo keeps crashing :-(
17:13* ted still needs to figure out WTF is up with vidyo on his new windows box
17:14globso, what happens with this channel, and more importantly face?
17:14globface: should we have a party?
17:14faceglob, nope
17:14janxface: yes we should
17:14* face wonders why he let wlach change the api :) I know it is better now, but frustrating
17:15tedglob: i don't see any reason to abandon the channel
17:15tedif newly-minted teams want to make their own channels as well that seems fine too
17:15tedwe've had #build for ages, i suspect a lot of the build team chatter will continue to happen there as it always has
17:15dylanI'm still in platform-op-soc too
17:15* glob nods
17:15tedregardless of org structure
17:16* janx doesn't know about the bi-weekly meeting, but definitely plans to stick around #ateam
17:16* ted ponders
17:16tedi think i only have two outdated-by-reorg t-shirts
17:16tedthe Platform Ops one and the A-Team one
17:17jmaherted: well, Platform Integrity == PI- this has potential for good channel names
17:17janxI've always had the feeling that "A-Team" was a sufficiently generic name to survive re{orgs,names,shuffles} (cf "Engineering Productivity", "Release & Productivity", etc)
17:18tedbut at one point it was an actual play on the team name :)
17:18tedAutomation & Tools
17:19jgrahamjmaher: Oh, now I suspect you of being behind the entire reorg :p
17:20tedMaher, PI
17:23janxted: the Platform Ops tshirt looks really cool, I'm a bit disappointed that I didn't join in time to get one :P
17:23dylanted: the allusion to autotools (gnu autotools) was always in my mind when I joined. :-)
17:23teddylan: hah! i don't think that was intentional
17:23tedi mean, i sure would have vetoed it ;-)
17:24dylanit would be like naming the team Dracula or typhoid...
17:25* dylan bought the gnu autotools book way back when.
17:25dylanread it cover to cover.
17:25dylanthen I installed cmake.
17:25dylanwithout ever writing a line of m4.
17:29tedi have written entirely too much m4 in my life
17:29tedi even wrote an m4 parser in python
17:29tedi will not shed a tear when we're free of autoconf
17:30dylanugh, a parser only or just an evalu-- no, I guess you can't parse it without evaluating it
17:32dylandamnit ted there goes my next weekend.
17:33dylanI'm going to try writing an m4 parser in racket.
17:33jmaherdylan: I don't understand this racket
17:37dylanit's just a scheme I've been using.
17:40jgrahamI hope that better puns weren't a goal of the reorg
17:42jmaherjgraham: we are still new to this re-org :)
17:43dylanted: woot.
17:43teddylan: IIRC the actual parsing is the easy part, it just expands macros and then starts parsing again, the hard part was supporting all the GNU m4 built-in macros
17:43teddylan: that supported just enough to evaluate
20:25KWiersobc: you want to look at autophone failures without failure messages, right?
20:25bcyep, doing that right now as a matter of fact.
20:26KWiersobc: here's one on beta:
20:26KWiersogrr... stupid long url...
20:28bcI got it
20:47bholleyKWierso: is there any way to "get previous 10" on treeheder?
20:47KWiersobholley: like, newer pushes?
20:47bholleyKWierso: yes
20:47KWiersouh... emorley ^
20:48KWiersoI'd say the famous last words "couldn't be that hard to add it if it's not there now", but I don
20:48KWierso't wanna
20:54KWiersobholley: how far back from the newest push to the repo are you? you could just drop the "tochange" parameter and load everything back to where you are.
20:54bholleyKWierso: a bit too far back for that
20:54bholleyKWierso: I did it manually
20:54KWiersobholley: I'm not seeing a bug filed for this, might be worth filing one
20:54bholleyemorley: but would be nice to have a button
20:56KWiersojmaher|afk: grr @ bug 1338320 adding a non-bugzilla recommended bug component for some paths...
20:56bugbotBug web-platform-tests, normal, jmaher, RESOLVED FIXED, annotate testing/web-platform/ with BUG_COMPONENT for tests/ directory
20:56bholleyKWierso: filed bug
20:56* bholley bug 1348973
20:56bugbotBug Treeherder, normal, nobody, NEW , Add buttons to get previous pushes in addition to the ones to get next pushes
21:04sfinkI had that in bug 1095214, which got duped to bug 1078209, which implemented a bunch of related functionality -- but it looks like it doesn't have that piece. Or it has a different thing, which is "Set as [top|bottom] of range", which seem to be broken because they leave the rev=... part in.
21:04bugbotBug Treeherder, normal, nobody, RESOLVED DUPLICATE, Allow viewing adjacent pushes in the main display
21:04bugbotBug Treeherder, normal, cdawson, RESOLVED FIXED, Unable to load prior pushes using "get next N" buttons when only one result set/revision visible
21:08KWiersosfink: that doesn't seem broken at the moment to me. clicking the get next buttons replaces the revision parameter with the tochange one
21:08KWiersowhich all appears to work, but only going down in the pushlog, not up
21:10sfinkKWierso: I guess my expectations were that you could get the previous pushes from that. So if I am on a single rev, and then select "Set as bottom of range", that I would see everything up to that rev.
21:10sfinkinstead, I get that revision, up to that revision
21:11sfinkwhich looks a lot like just that revision
21:11KWiersosfink: that's probably worth a new bug, though I'd say that's roughly the expected behavior
21:12KWiersowhere "set top/bottom" options restrict the number of pushes shown, but don't increase
21:12KWiersobut yeah, I guess both options should remove the revision param if it's set
21:13sfinkyeah. Well, I only wanted "get prev N" in the first place, so I'm happy having the new bug.
21:13sfinkI don't know if the behavior should change or not. The current behavior is conceptually simpler. Just not useful in this case.
21:13sfinkbut if there were "get prev N" buttons, nobody would hit this case anyway
21:59whimboojgraham: hi. when I want to run the webdriver tests is that the correct command?
21:59whimboomach wpt webdriver/ --webdriver-binary /usr/local/bin/geckodriver
21:59whimbooit simply fails all for me
22:00AutomatedTester"fails" ?
22:00whimbooi cannot get them to run
22:00whimbooall say ERROR or FAILED
22:01jgrahamwhimboo: From my command line: ./mach wpt testing/web-platform/tests/webdriver/newsession/ --webdriver-binary ../geckodriver/target/debug/geckodriver
22:02jgrahamwhimboo: I would expect yours to work
22:02whimboohm, i dont see newsession there
22:02whimbooin my case its
22:03jgrahamYeah, sorry newsession is something I"m working on
22:04whimbooso those are still all failing. this time with SessionNotCreatedException: Session is already started
22:05jgraham./mach wpt webdriver/ --webdriver-binary ../geckodriver/target/debug/geckodriver
22:05jgrahamRan 10 tests (1 parents, 9 subtests)
22:05jgrahamExpected results: 10
22:05jgrahamUnexpected results: 0
22:05jgrahamSo it works here
22:05jgrahamDo you have an up-to-date geckodriver?
22:05whimboojust built
22:06jgrahamHmm, not sure then. OPen a bug and add whatever logging you have, perhaps
22:06jgrahamI'm going to sleep now
22:06whimbooi will check tomorrow
21 Mar 2017
No messages
Last message: 6 days and 18 hours ago