[07:48] <pitti> Good morning
[07:56] <jibel> Good morning
[08:16] <jibel> DanChapman, good morning, I continue my investigation on the libautopilot-gtk case
[08:18] <jibel> DanChapman, I found that each time ubiquity/autopilot crashes there is also an ATK critical message in the logs
[08:18] <DanChapman> jibel, good morning o/ great :-)
[08:18] <jibel> DanChapman, it happens for every flavor with or without U1
[08:19] <jibel> DanChapman, so we can eliminate U1 from the list of suspects and I think the problem comes from the webkit view
[08:19] <jibel> here are the logs with the crashes I found http://people.canonical.com/~j-lallement/ubicrashes.log
[08:20] <jibel> on atk and the crash, I don't know if one is the consequence of the other or if they are 2 symptoms of the same problem
[08:22] <DanChapman> yes i mentioned that to pitti a while back, but i thought it should carry on running unless you expliciltly set Gtk to fail on critical
[08:22] <jibel> xnox, in ubiquity is there a way to no load the webkit view in the slideshow at run time?
[08:23] <jibel> I didn't find any from the code but maybe missed something
[08:23] <DanChapman> jibel It doesn't check the webkitview but the scrolled window
[08:23] <jibel> DanChapman, yes but I think the message ATK critical comes from webkit
[08:23] <DanChapman> I changed that a while back as the webkitview properties were tempremental at the time. I could try sweitching it back
[08:24] <DanChapman> ahhh i see
[08:24] <DanChapman> sorry. still early
[08:30] <jibel> DanChapman, that could mean the parent doesn't exists, which could also leads to the crash in autopilot if it refers to a non-existent object
[08:34] <jibel> DanChapman, and another symptom when there is a crash, the slideshow also have big fat scrollbars
[08:44] <jibel> DanChapman, are you doing anything with webkitwindow other than checking if it is visible? I'd like to remove it
[08:44] <jibel> in progress_test_page() that is
[08:45] <jibel> *progress_page_tests
[08:45] <DanChapman> jibel, ahh! that darn scrollbar! jibel I was only using it to assert we moved to the next page since we lost the page title as it's in the webview
[08:46] <DanChapman> I think as long as we can  be sure that the progressbar is progressing ok then that's enough :-)
[08:49] <jibel> DanChapman, k, I'll do a few run without it, it looks like a race where the gtkscrolledwindow really exists after the webview is created
[08:49] <jibel> hence the randomness
[08:54] <DanChapman> jibel, that makes sense now.
[09:30] <xnox> jibel: if the slideshow package is uninstalled, the webview will not be started.
[09:30] <xnox> jibel: however, if it crashes it's a bad sign =(
[09:34] <jibel> xnox, I'll remove the code that tests if the webview is visible first. Ubiquity or the slideshow don't crash, only autopilot does :)
[09:35] <jibel> and the visible side of the bug are the fat scrollbars
[09:44] <DanChapman> jibel, it would appear the webkit_scrolled_window is visible/exists from the start. so we could actually do assertTrue(scrolledwindow.visible) at any point/page in the test. So could it be more of a bug that the webkitview root object is unable to get parent rather than it being non-existent
[09:46] <DanChapman> as it's a pointless assertion it should be removed anyway
[10:18] <DanChapman> jibel this passes http://paste.ubuntu.com/6998946/ asserting the scrolled window during the welcome_page_tests.
[10:46] <DanChapman> jibel, I just did a trial removing ubiquity-slideshow-ubuntu which the webview didn't load and I no longer get the atk critical warning.
[11:06] <davmor2> Morning all
[11:06] <jibel> Morning davmor2
[11:07] <jibel> DanChapman, sure but I don't want to disable the slideshow completely. We had bug in the past where loading it crashed ubiquity, so I just disabled the check and if the slideshow makes the installer crash, it will be detected anyway
[11:13] <DanChapman> jibel cool... makes sense :-) Fingers crossed it resolves the issue
[11:25] <DanChapman> jibel wow it borked on the progressbar this time :-(
[11:26] <jibel> DanChapman, I'll remove the slideshow to validate it is really a problem with webkit
[11:32] <jibel> DanChapman, do you know what is the failure on lubuntu. It says '/home not found'
[11:32] <jibel> ?
[11:32] <DanChapman> jibel, for custom install?
[11:34] <DanChapman> it's just when creating partitions it's taking too long to update the UI i'll change that to use wait_select_single
[11:34] <DanChapman> should resolve it
[11:35] <jibel> DanChapman, k, I'm starting a new run without the slideshow, if it is still crashing, then ... I don't know
[11:36] <jibel> DanChapman, I'll try to produce a test case with just the transition to the slideshow
[12:09] <jibel> woohoo, 3 runs and no crash of ubiquity without the slideshow!
[12:09]  * jibel restarts $WORLD
[12:13] <DanChapman> jibel, yes it looks like it  has done the trick 'so far'  ;-p I'm going to have to re-write a fair bit of the custom install test rather than sleep longer to fix the partition fails
[14:15] <rvr> elopio: I'm here :)
[14:21] <elopio> rvr: hah, right.
[15:04] <elopio> jfunk: we had the first meeting about landing qt5.2, with Pat coordinating it.
[15:04] <elopio> thanks for pushing on that.
[15:05] <jfunk> yw
[15:05] <jfunk> it seems you're not the only one who wants this out the door
[15:05] <jfunk> Chris Lee is working on the AP job
[15:06] <jfunk> not sure if you're aware
[15:09] <elopio> jfunk: yes, we can use it now.
[16:39] <elopio> balloons: ping. I'm looking at the weather app failures.
[16:39] <elopio> I can't reproduce the ones jenkins is showing, but I get a loading popup that doesn't disappear.
[16:39] <balloons> elopio, hey, did you see martin's comments? we're working on landing a branch
[16:40] <balloons> elopio, https://code.launchpad.net/~martin-borho/ubuntu-weather-app/anchors-refactoring/+merge/207793
[16:40] <balloons> elopio, I would test and work on that branch.. I don't really care about the dashboard results
[16:40] <elopio> balloons: ok, I can agree with that :)
[16:41] <elopio> balloons: the recommended way to run the tests now is with click buddy provision and all the nice things you mentioned on your last blog post, right?
[16:42] <balloons> elopio, yes for sure.. it's pretty slick
[20:04] <elopio> I'll take a break now. bbl.
[20:04] <elopio> ubuntu-qa, remember the lightning talks today.
[20:08] <robotfuel> elopio: that's not on my calender, where are they?
[20:10] <elopio> robotfuel: http://ubuntuonair.com/calendar/
[20:16] <Letozaf_> balloons, hi
[20:16] <balloons> elopio, ^^
[20:16] <balloons> Letozaf_, hello :-) I think reminders might launch for you now
[20:17] <Letozaf_> balloons, unfortunatly no :(
[20:17] <Letozaf_> balloons, file:///home/letozaf/autopilot-tests/new-reminders-app-tests/src/app/qml/reminders.qml:262: Unable to assign null to QQuickAnchorLine
[20:17] <Letozaf_> balloons, when I launch it with autopilot launch
[20:18] <Letozaf_> balloons, I do not know qml well enough but looks like the null is wrong here: anchors { right: null; left: parent.left; }
[20:18] <balloons> Letozaf_, did you rebase your branch to trunk?
[20:19] <Letozaf_> balloons, yes
[20:20] <Letozaf_> balloons, mmm ...
[20:20] <balloons> Letozaf_, if elopio doesn't otherwise come around, he was hoping to fix us the rssreader tests with you; go through and refactor them. I think he was thinking next week
[20:21] <Letozaf_> balloons, ok, but what about reminders app ?
[20:22] <balloons> I'm looking now :-) I agree it still doesn't launch, sigh
[20:22] <balloons> however, the error you spotted isn't the reason; you can see it launching reminders; aka the installed version
[20:23] <Letozaf_> balloons, yes the installed version works fine
[20:24] <balloons> Letozaf_, also, I tried from trunk and autopilot launch doesn't come up. My assumption actually is you need to pass the plugin to it
[20:24] <balloons> that's the issue. it's not a straightforward launch
[20:24]  * balloons thinks
[20:39] <Letozaf_> balloons, I am looking at rssreader tests, the ones that aren't skipped are working on the desktop, so you probably want me to fix the skipped ones right ?
[20:40] <Letozaf_> baba
[20:41] <balloons> Letozaf_, yea, the skipped tests should certainly be looked at trying to get them re-enabled. But also, the tests are a bit hard to maintain and write and rather large. The goal is to try and break the existing tests down to be easier to read and write as well
[20:41] <Letozaf_> balloons, ah ok
[21:04] <Letozaf_> balloons, I have some questions on rssreader app tests, do I have to ask e1opio or can I ask you ?
[21:21] <elopio> hey Letozaf_, I'm making some changes on the rss tests, so you can ask me.
[21:22] <elopio> I was hoping you could give me a hand :)
[21:22] <Letozaf_> elopio, hello
[21:22] <Letozaf_> elopio, sure
[21:22] <Letozaf_> elopio, let me know what you want me to do
[21:23] <elopio> Letozaf_: I'm making a first branch, so I would like you to review it. I hope it will be ready tomorrow.
[21:23] <Letozaf_> elopio, balloons told me you want to refactor them
[21:23] <Letozaf_> elopio, fine, I will be around tomorrow same time as today
[21:27] <elopio> Letozaf_: ok, thanks.
[21:27] <Letozaf_> elopio, yw