=== chihchun_afk is now known as chihchun [05:24] Good morning everyone === OutOfControl is now known as benonsoftware [06:38] Good Morning [06:38] morning DanChapman [06:43] hey DanChapman, hey elfy [06:43] hi pitti - and congrats on the Tech Board thing [06:43] hey elfy; thanks (have been doing TB for quite a while :) ) [06:44] yep - I know that too :) [06:44] I might hide away - but after 7 years of it I sort of have a good idea of what's going on lol [06:45] though I did venture out of the shadow of the forums to the community council [06:49] Hey elfy and pitti o/ [06:53] pitti while i remember bug 1261685 I managed to break it down into a small test case which I attached to a comment do you need a new bug for it as that one is fix released?, I didn't have very much success in tracing it any further [06:53] bug 1261685 in autopilot-gtk (Ubuntu) "libautopilotgtk.so crashed with SIGSEGV in GtkNode::MatchStringProperty (this=0xa8b6c80, name=..., value=...) at /build/buildd/autopilot-gtk-1.4+14.04.20131128.1/lib/GtkNode.cpp:271" [High,Fix released] https://launchpad.net/bugs/1261685 [06:54] DanChapman: indeed, thanks for working on a reproducer; I'm catching up with bug mail, I just opened that one :) [06:54] DanChapman: new bug report would be nice [06:54] DanChapman: I'll try your reproducer [06:55] pitti, excellent thanks, i'll get a new bug report done for it === psivaa_ is now known as psivaa === vrruiz_ is now known as rvr [09:43] Happy new year to everyone! [09:51] DanChapman: I indeed get a crash with just http://paste.ubuntu.com/6708207/, always at iteration 57 [09:51] DanChapman: but it's not quite exactly the same crash as in bug 1261685 as that part of the code was changed [09:51] bug 1261685 in autopilot-gtk (Ubuntu) "libautopilotgtk.so crashed with SIGSEGV in GtkNode::MatchStringProperty (this=0xa8b6c80, name=..., value=...) at /build/buildd/autopilot-gtk-1.4+14.04.20131128.1/lib/GtkNode.cpp:271" [High,Fix released] https://launchpad.net/bugs/1261685 [09:57] pitti, thats strange how it's always iteration 57. Your right, sorry about that, I will update the bug report [09:58] DanChapman: thanks for the new bug report, I updated/assigned it [09:58] DanChapman: I'll figure out how to debug it from here, many thanks for the reproducer! [10:06] pitti no problem :-) [10:17] DanChapman: meh, I can't actually get it to crash (no core file, nothing in dmesg, etc.); tough.. [10:17] I clearly get the graying out/org.freedesktop.DBus.Error.NoReply though [10:27] DanChapman: are you sure it crashes with the reproducer? i. e. do you get an apport .crash file or something? if so, would you mind putting that somewhere? [10:31] DanChapman: err, WTF -- with teh g_warning -> g_debug fix for bug 1250452 it suddenly works [10:31] bug 1250452 in autopilot-gtk (Ubuntu) "Constant warning output running Gtk tests with autopilot 1.4" [Low,In progress] https://launchpad.net/bugs/1250452 [10:33] pitti, yeah I hit a brick wall trying to find it. It's a really tricky one, I couldn't get a crash file either from it [10:33] pitti lol really??? o-O [10:33] DanChapman: running with your script now [10:34] it's like the spewage overflows some of autopilot's stdout/err capturing buffers, and that might need some time to catch up during which the d-bus time out happens [10:38] DanChapman: so you didn't actually see it crash, but you saw ubiuity (or whatever) gray out and hang, and the DBus.NoReply error? [10:38] $ (D=`pwd`; cd tests/autopilot; GTK_PATH=$D/lib autopilot run -v tests.test_force_segfault) [10:38] Ran 4 tests in 186.526s [10:38] OK [10:38] DanChapman: ^ that's your script [10:38] (with trunk) [10:42] DanChapman: bug updated, would appreciate a confirmation/difference to what you see === chihchun is now known as chihchun_afk [10:44] pitti I haven't got it to crash locally but yes ubiquity was doing the same with greying out in exactly the same way polling on the progressbar then getting the dbus, I pushed up patch to the tests to see on jenkins with the 0.3 second sleep, it's been somewhat better but the odd test 1 or 2 a day are still failing with it but in general it helped [10:45] pitti sure i'll give it a run now :-) [10:45] DanChapman: want an amd64 libautopilot.so to test? (with the g_debug change) [10:45] or a .deb? [10:46] pitti, a .so would be handy :-) [10:47] DanChapman: http://people.canonical.com/~pitti/tmp/libautopilot.so [10:47] DanChapman: or http://people.canonical.com/~pitti/tmp/libautopilot-gtk_1.4+14.04.20131218-0ubuntu1_amd64.deb if that's more convenient [10:47] (just a straight bzr bd -- -b from trunk) [10:47] thanks :-) [10:51] DanChapman: I committed the new test to a branch; I'll wait for your testing results for submitting, though [11:01] Morning all [11:16] pitti all seems to be good from this end. +1 from me :-) [11:16] DanChapman: nice! [11:20] DanChapman: great, I'll land the new test case and then ask for getting trunk into trusty [11:20] DanChapman: many thanks for confirming! [11:21] pitti, No problem, thanks for the efficient fix :-) [12:05] jibel: does run-adt-test currently work for you? first time I try it since the holidays (with -sUl), it hangs here [12:08] it waits for /dev/vdc (the pristine VM image) which apparently never comes up === rbasak_ is now known as rbasak [12:21] pitti, it is not the first thing I tried this morning, but I'll check. [12:35] pitti, (qemu) pci_add: command not found [12:35] (qemu) pci_add auto storage file=/dev/shm/adt/trusty-amd64-tabix-20140107_133008.FoLnAX.img,if=virtio,index=2,readonly [12:35] unknown command: 'pci_add' [12:36] ^ [12:48] pitti, in 1.7.0+dfsg-2ubuntu5 * Drop d/p/fix-pci-add: upstream does not intend for pci_add to be [12:49] supported any longer. [12:49] replaced by device_add [12:49] https://bugzilla.redhat.com/show_bug.cgi?id=602590 [12:49] bugzilla.redhat.com bug 602590 in qemu-kvm "Disable pci_add, pci_del, drive_add" [Medium,Closed: currentrelease] [12:55] jibel: ah, many thanks; I'll adjust [13:11] jibel: meh; there's very little documentation about that, will take me a while; if this blocks the CI ADT tests, I can also just comment it out for now [13:12] pitti, I cannot find much documentation either and syntax is non-obvious. I don't think it block adt tests, the change is only in trusty. [13:13] jibel: ah, in CI we run this from precise/raring, right? [13:13] pitti, yes, precise and saucy [13:15] and errors are not better than documentation: Can't hot-add drive to type 7 [13:41] pitti, I can hot plug an scsi disk and get a /dev/sda but no luck with virtio [13:41] jibel: I pinged hallyn in #u-devel about it, maybe he knows [13:42] pitti, good, thanks [14:00] pitti, just FTR, here are the commands to hotplug an scsi disk. but if I replace if=none with virtio it fails with "can't hot-add drive to type 7" [14:00] I hope there is something simpler [14:02] pitti, with the link http://paste.ubuntu.com/6709139/ :) [14:02] jibel: merci [14:11] pitti: "Drop --override-control option". Hey, I used that! [14:16] rbasak: oha; OOI, what's the use case? [14:17] rbasak: we can certainly bring it back, but I didn't know anyone who used it, and it seems rather strange (and adds unnecessary complexity) [14:17] Right now, I wanted to enable some tests that were disabled in the archive due to breaks-testbed not implemented. I think I have used it in the past for other things, though. [14:17] I'm trying to work around, but I can't seem to get --unbuilt-tree to work. [14:18] rbasak: how does it fail? (it even has test coverage now) [14:19] cp: cannot open ‘/tmp/adt-downtmp/ubtree0-ubtree/./debian/changelog’ for reading: Permission denied [14:19] I'm not sure if that's an issue with adt-virt-lxc or something I'm doing wrong. [14:19] I'm still looking at it. [14:28] pitti: I think it's a regression, though haven't spent time trying to confirm it.+ [14:28] pitti: adt-virt-lxc defaults to suggested-user=ubuntu, and you have to say --gain-root=sudo to adt-run, so that it does "debian/rules binary" with sudo. [14:29] pitti: adt-run picks up on that and does, effectively, --user=ubuntu as default in that case. [14:30] rbasak: right; that's similar how schroot behaves now (since the last release) [14:30] pitti: http://paste.ubuntu.com/6709295/ [14:30] pitti: I think it then can't copy from $origpwd because that isn't readable by the ubuntu user. [14:31] rbasak: ah, I only tried it with schroot, where the suggested user is the same as the calling one (which is different with lxc) [14:32] pitti: btw, I also needed a workaround for adt-virt-lxc and the latest lxc in trusty. The wait implementation inside adt-virt-lxc assumes that it can see inside the guest tree from the host as a normal user, but latest lxc sets /var/lib/lxc o-a. So I set it to 755 as a quick workaround. [14:35] rbasak: indeed, "lxc-ls" is now also broken due to that [14:35] was that a deliberate change, or a bug? [14:35] I'm not sure, but I presume "sudo lxc-ls" is required. [14:35] it wasn't until a few weeks ago [14:36] OK, so I mean is *now* required :) [14:36] did someone complain that they have super-s3kr1t container names which are never running? (because you can see their names in ps when they are..) [14:36] stgraber: ^ [14:39] pitti: it's a deliberate security change [14:40] without that, any user would be able to execute a setuid binary from any of the container's rootfs or cache [14:40] stgraber: how so? [14:40] stgraber: I understand making /var/lib/lxc/ 0700 [14:40] but /var/lib/lxc itself? [14:42] we indeed could have done it on /var/cache/lxc and /var/lib/lxc/*, changing LXC so that containers dirs are always created 0700 instead of respecting the umask, but /var/lib/lxc + /var/cache/lxc was just easier [14:42] the obvious downside is that you can't list containers as non-root now [14:42] but at the same time, lxc-ls as non-root won't be listing /var/lib/lxc in 1.0 anyway [14:42] (as we have unprivileged containers and lxc-ls lists ~/.local/share/lxc/ now) [15:40] DanChapman,balloons, Hello Good Eve!!! [15:40] good eve senan [15:46] senan hey there, sorry i missed your messages over the holidays [15:51] DanChapman, Its ok dan [15:57] DanChapman, can you check the issue I mentioned [16:11] senan, I have looked into testing the file chooser dialog in general, what you need to do is select_single('GtkTreeView', BuilderName=u'') this gives you the 'Places' treeview then you need to select_single the GtkTreeViewAccessible with the same globalRect as your current selected treeview, from the treeview accessible object you will be able to select any of the treeview items with .select_single('GtkCellTextAccessible', accessible_name=' [16:11] item label value') [16:13] senan, do you need an example? [16:16] DanChapman, Yes [16:31] DanChapman, Did you mean the browse_files_tree_view ? [16:39] DanChapman, Can you please give me an example [16:39] senan just sorting it now :-) [16:40] ok dan [16:42] mhall119: balloons: can you please approve me to https://launchpad.net/~ubuntu-touch-coreapps-test-writers/+members I'm helping out on enabling all core apps tests pass in the emulator? [16:42] into that is. [16:42] xnox, sure thing [16:43] done [16:44] senan http://paste.ubuntu.com/6709963/ use this all you have to do is pass the treeview object and the treeitem label name [16:45] senan you want the other treeview and not browse_files_tree_view. The one with no name [16:46] DanChapman, under PlacesSidebar right ? [16:51] balloons: thanks!~ === micahg_ is now known as micahg [16:56] DanChapman, Is it possible to directly access GtkTextCellAccessible ? [16:58] senan it is but you have to be sure its coming from the treeview you want, hence you need to match the GtkTreeViewAccessible.globalRect with the treeview then select it from there to be sure your in the correct treeview [16:59] DanChapman, So first I need to get the TreeView object under PlacesSidebar and pass it to the function you just created [17:08] senan yes so with the treeview object do self.get_tree_item(treeview, 'Computer') and you will get the 'Computer' tree item which you can then click [17:16] xnox, I need to try out your autopilot testing with the emulator :-) [17:16] balloons: well instructions on the bug reports and on ubuntu-phone mailing list [17:16] balloons, howdy can you just confirm a bug for me if you right click the text area of gnome-calculator is the menu over sized? [17:17] DanChapman, yes it is.. wild [17:17] DanChapman, if you right click the numberpad first, then the textarea, it works fine [17:19] balloons, thanks. Your right it does!, it seems it's only when the text cursor is visible [17:25] DanChapman, It worked!!! Thanks [17:25] senan YAY!! your welcome :-) [17:28] DanChapman, One more issue...when launching Scan Remote Folder [17:29] DanChapman, its launching NautilusConnectServer Dialog [17:29] DanChapman, I believe its a separate process === davmor2_ is now known as davmor2 === balloons_ is now known as balloons [17:38] DanChapman, NautilusConnectServerDialog, I am not able to catch it === bfiller is now known as bfiller_afk [17:43] Good Night All!!! [17:46] senan i wouldn't worry about scanning remote repositories as it's generally not recommended to depend on external connections [17:46] good night senan === stgraber_ is now known as stgraber === Claudinux is now known as Guest33318 === balloons is now known as Guest99854 === balloons_ is now known as balloons === salem_ is now known as _salem === bfiller_afk is now known as bfiller === Ursinha-afk is now known as Ursinha === Ursinha is now known as Guest50378 === Ursinha-afk is now known as Ursinha === manjo` is now known as manjo === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === _salem is now known as salem_ === salem_ is now known as _salem