mozilla :: #ateam

17 Jul 2017
12:36whimboowow, our code coverage builds are ~500MB in size and this packed as bz2!
12:37whimbooekyle: ^ this is immense. I assume you are the top consumer of AWS storage
12:38ekylewhimboo: yes, it is large. I doubt I am the top consumer, maybe second.
12:40whimbooekyle: :)
12:41whimbooekyle: btw. during all hands you mentioned some improvements for content frame scripts. have those been landed already?
12:41whimboothe problem was the 0% coverage
12:42ekylewhimboo: I suspect they have, but you must to talk to marco
12:43ekylewhimboo: https://bugzilla.mozilla.org/show_bug.cgi?id=1229277
12:43bugbotBug 1229277: JavaScript Engine, normal, mcastelluccio, RESOLVED FIXED, LCOV output emits 0 coverage for some files
12:43ekylewhimboo: https://bugzilla.mozilla.org/show_bug.cgi?id=1227735
12:43bugbotBug 1227735: Code Coverage, normal, mcastelluccio, RESOLVED FIXED, Incorrect lcov output for xbl sources
12:44whimbooekyle: oh, just checked the coverage and indeed it works!
12:44whimboohttps://codecov.io/gh/marco-c/gecko-dev/src/master/testing/marionette/listener.js
12:44ekylewhimboo: excellent!
13:32globface: are you still ok?
13:32* face imagines: "jhammel's progress: arghasdfg! horrible"
13:40marcojmaher: ping
13:45jmahermarco: hi
13:58marcojmaher: on trychooser, if you select "all" platforms, linux64-ccov will be selected
13:59marcojmaher: should we make it explicit like "shell rooting analysis"
13:59marcoexplicit as in "if you click on 'All' it's not enabled, you have to explicitly select it"
13:59jmahermarco: I think we should; there is little to no chance of breaking coverage; until we have a workflow to verify patches coverage on try we would only be wasting resources
14:00marcook
14:00jmahergood find
14:20whimboowow. don't enable servo if you do a lot of code review in mozreview. it crashes like hell
14:33globwhimboo: fyi that's bug 1379617
14:33bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1379617 CSS Parsing and Computation, critical, nobody, NEW , stylo: Crash in nsIFrame::SetNextSibling
14:35whimbooglob: no
14:36whimbooat least there is bug 1381475 too now
14:36bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1381475 CSS Parsing and Computation, critical, nobody, NEW , Stylo: Crash in mozalloc_abort | abort | geckoservo::glue::Servo_ResolveStyle
14:36globwhimboo: interesting; more than one issue there then!
14:37whimbooyeah
14:38whimbootoday I feel like doing bug triage
14:38* glob passes it on to the stylo team
14:38* glob sees whimboo do the same on the bug :)
14:39whimbooi had crashes all over the place today
14:39whimboowhich causes me to flip the pref between states depending on what I'm working on
15:15whimbooato: do you know why we only allow "link text" but not general text for find_element?
16:09atowhimboo: innerText is underspecified and not interoperable, and a free-form text search of the document would not perform well.
16:09atoI voted for dropping the link text and partial link text strategies, but I was overruled.
16:22jmaheremceeaich: hi, I wanted to see if you were going to keep doing the mass closing of 21 day old intermittents, or if there is a process for this
16:30dylanjmaher: there will be some sort of automated process... that doesn't send 80k emails
16:31jmaherdylan: well, we have 6700 total intermittent-failures today, I doubt we would send >2000
16:31jmaherprobably more on tune of ~1000/week
16:32dylanthat could be as many as 20k emails
16:36emceeaichWhat do we want to do in the short term?
16:37emceeaichI could close a few at a time over the course of the week, but we'd still have the spam.
16:40whimbooato: ah, k. that makes sense
16:42* RyanVM prefers the ripping off the bandaid approach personally
16:53whimboojmaher: so i cannot work on all the major OF intermittents at once :(
16:53whimbooi'm about to get a patch reviewed which will stop the crash tests from failing
16:54whimboothen i will hop back on the navigation tests, even if we haven't gotten a reply from smaug by that time
16:54jmaherwhimboo: I know you cannot, but you are not the only person working on the marionette tests and harness
16:55jmaherI just need someone who knows the tests/harness to redirect to the right people
16:55whimboojmaher: sure but nearly everyone is out, and I'm here only the next 4 days before being on PTO too
16:56jmaherwe have 6702 current intermittent-failure bugs, right now I am only looking at 42 bugs for getting fixed- unfortunately marionette stuff shows up commonly
16:56whimbooits used more and more too
16:56whimbooi will do my best
16:56jmaherwhimboo: so we should have more people able to work on it; I assume the tests are written by others
16:56whimbooif i cannot make it until end of Thu we should disable the tests
16:57jmaheryeah, that is fine
16:57jmaherI try to treat all bugs equally :)
16:57whimboosure :)
16:57whimbooi wasn't blaming you.
16:57whimbooits just a tough schedule right now
16:58jmaheryeah, I was there last quarter on a lot of perf stuff- I don't want to add to the stress!
17:11emceeaichDylan, how much work would it be for you if we "yanked off the bandaid" like RyanVM suggests?
17:11emceeaichand just closed these
17:12dylanI have a script that can close bugs without sending emails
17:12emceeaichI have 382 intermittent bugs I can close today.
17:12dylanfor firefox os, I need to run that on the weekend
17:12dylanbut for anything short 10k bugs it should be fine
17:13dylan(for 10k bugs, it induces too much DB load for a production hours, and I have not had the desire to be around during the weekends.)
17:16dylanhttps://bugzilla.mozilla.org/show_bug.cgi?id=1381566 would probably take very little time for someone not-me to write.
17:16bugbotBug 1381566: General, normal, nobody, NEW , Add alternative query option to resolve_bugs.pl
17:18jmaherdylan: I will not be a roadblock into doing that; maybe emceeaich has other reservations given that she has a bigger picture of closing bugs
17:19emceeaichI can try doing something with that.
17:19dylanif I get a few moer things done, I can do it too
17:20dylanlet me throw a few more balls into the air before I catch this one.
17:20emceeaichI need to get into the office for weekly all hands, I'll check back in in couple.
17:20jmaher:)
17:23emceeaichhttps://twitter.com/triagegirl/status/886999271681032192
17:44whimbooato: are you still around?
17:45whimbooato: question re bug 1368434
17:45bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1368434 Marionette, normal, hskupin, ASSIGNED , Missing page load events between frame script reloaded and load listener registration
17:45whimbooato: i wonder if we should just check the readyState when starting the listener after a known reload of teh frame script
17:45whimbooif it has already the expected value we can return immediately
17:51dylanwlach: your patch is removing the max-height thing, right?
17:52wlachdylan: no, it's about changing the behaviour of the autosizer to allow vertical resizing
17:52wlachdylan: and to disable autosizing if it is ever vertically resized
17:52dylank
17:52wlachI was also going to tweak the maximum height to be a little bigger
17:52dylanwhat happens if I remove the max-height entirely?
17:53wlachbut someone else can do that too
17:53wlachdylan: it will endlessly expand
17:53dylanThat seems fine.
17:53dylanthanks!
17:53wlachdylan: so you want to take out the max-height entirely?
17:54wlachdylan: I'm not sure if that's a good idea -- a really long comment could easily be N times the size of the viewable area
17:54dylanyeah, I'll have that happen in the bug to disable it
17:55wlachdylan: oh, yeah, if it's disabled then you would probably not want to have a maximum height
17:55wlachit's only when we're autosizing that we need to be careful
17:55dylandoes autosize add any sort of class to the element it touches?
17:55wlachso many corner cases with a feature like this
17:55wlachdylan: nope
18:02wlachdylan: if you want we could make autosize do that tho
18:02wlach(just realized what your reasoning might be)
18:03wlachmy plan for fixing this issue was to hack up autosize a bunch fwiw
18:10dylanhack away, I've already forked it to our org so we're not blocked merging stuff up: https://github.com/mozilla-bteam/autosize/invitations
23:36haikWith the Mac builds now being cross-compiled, can I run a mac tool during the build from a Makefile? For example, plutil?
23:38haikprobably a better question for #build...
18 Jul 2017
No messages
   
Last message: 68 days and 5 hours ago