[05:25] bregma: added https://bugs.launchpad.net/unity-scope-audacious/+bug/1229028 which blocks automatic testing of unity stack [05:25] Ubuntu bug 1229028 in Unity Guayadeque Scope "Search tests fail for scopes guayadeque and audacious" [Critical,New] === jamesh_ is now known as jamesh\ === jamesh\ is now known as jamesh [06:59] * tsdgeos waves hello [07:34] pstolowski: hi there, did you get that problem we had before I went into holidays fixed? [07:35] tsdgeos: hi! nope, it's waiting for you ;) [07:35] damn! should have gotten a longer holiday :D [07:35] j/k [07:36] ok, will get into it after reading the ton of email [07:39] bah, we do something really wrong with power management on the phones [07:39] left the house with the Nexus4 turned off and full of baterry [07:39] just to come back and it even doesn't react to charging [07:47] oh my, and we shelved qt 5.1 [07:47] sigh [07:57] larsu: ping [07:57] tsdgeos: good morning [07:57] larsu: morning [07:58] larsu: the mouse wheel on the volume thing still has some issues here, you aware of it? [07:58] no, what are those issues? [07:58] * larsu never uses that feature… [07:59] middle mouse scroll once, then left click on the indicator [07:59] takes like 4 seconds to draw [07:59] woah [07:59] not for me [08:00] this is on saucy I presume? [08:00] yep [08:00] is it only after freshly logging in or does it happen all the time? [08:01] all the time [08:01] i guess it may also be the reason that mouse wheeling doesn't seem as it would work [08:01] since it only changes volume like once each 4 seconds [08:01] i can't really mousewhell and notice an immediate change in volume [08:01] ya I definitely don't have this [08:01] :-/ [08:02] can you execute `dbus-monitor destination=com.canonical.indicator.sound` while you do this [08:02] oh lol [08:02] do the messages appear after the 4 seconds or immediately when you scroll? [08:03] they appear during 4 seconds [08:03] kind of :D [08:03] something fights between 0.392637 and 0.392624 [08:03] http://paste.ubuntu.com/6144572/ [08:05] mind = blown [08:05] * larsu wonders what's going on there [08:07] tsdgeos: I assume `dbus-monitor sender=com.canonical.indicator.sound` is similar (with the difference that instead of SetState calls you see Changed signals) [08:08] yep [08:08] larsu: some float vs double or something? [08:08] might be [08:08] but it shouldn't even update the state in response to an update of the state [08:09] I'll look into it, thanks for bringing it up [08:09] if you want to make a bug, that would be superb ;) [08:09] (but not strictly necessary) [08:10] larsu: against indicator sound? [08:10] tsdgeos: yes please [08:22] larsu: https://bugs.launchpad.net/indicator-sound/+bug/1229076 [08:22] Ubuntu bug 1229076 in The Sound Menu "Mouse wheeling in Saucy makes the indicator unresponsive" [Undecided,New] [08:24] tsdgeos: thanks! I'll try to look at it today or tomorrow [08:25] thanks! [09:01] paulliu, let me know if you need a hand with the music filter grid [09:08] Saviq: ok. [09:18] paulliu, should be as simple as fixing the "clicked" signal signature? [09:30] Saviq: yeah.. I'm fixing it and testing it right now. [09:31] paulliu, cool, thanks - Thomas is asking for an ETA in #ferrets [09:40] Saviq: can you have a quick look and tell me if you're happy with the direction it goes before I clean it up and add tests? https://code.launchpad.net/~mzanetti/unity8/switching-previews/+merge/186991 [09:45] mzanetti, looking good, but it's limited to a single category - can we make it not? [09:45] Saviq: uh... is that really wanted? [09:45] Saviq: I guess it should be possible [09:45] mzanetti, yes, you're supposed to be able to just get through the whole scope [09:46] mzanetti, what's the solution? same as for the calendar? [09:46] Saviq: right now I just put the category's model to the ListView, that's it [09:47] for including all... I'd need to try a bit [09:47] mzanetti, yeah, I think we need to go with the "3-item ListView" that you add/remove items to [09:47] /from [09:47] mhm... [09:47] mzanetti, there's one more thing... [09:48] Saviq: should the contents still only be loaded once you actually switch to a preview or should it be cached for all 3? [09:48] mzanetti, on demand [09:48] mzanetti, the list view in the back needs to scroll to show you the currently previewed item [09:48] ok. then the solution for cancelling proposed by pstolowski still works... he proposed in auto cancelling a old request when a new one comes in [09:49] yeah, that's fine [09:49] Saviq: scrolling with the current OpenEffect is an issue tho [09:49] mzanetti, yeah, you need to enable: true it for the time [09:49] ah right... that might work [09:49] ok. will check it out [09:49] thanks [09:50] mzanetti, looking good, though, Oren will be happy :D [09:50] Saviq: one more thing: the design shows a little arrow pointing to the current item [09:50] any idea how to get that into the openeffect? [09:50] mzanetti, let's leave that for later, I'd say [09:50] ack [09:51] mzanetti, same with highlighting the previewed item [09:51] mzanetti, that's sugar, where the actual feature is a big usability improvement [09:51] heh... missed that in the design so far [09:51] I fully agree === davmor2_ is now known as davmor2 [10:17] mzanetti: ping [10:17] pstolowski: hey man [10:18] mzanetti: hi! you don't have the nullptr check branch for Friday's crash, do you? [10:19] pstolowski: no, not really... as I said: when I wanted to actually implement it I realized that I'd really need to be able to reproduce in order to find out what else breaks if I take the DeeModel's sourceModel away [10:20] we might as well end up with a crash in some other line of code [10:20] mzanetti: yeah... so do you mind if I take it? [10:20] pstolowski: please do :) [10:20] mzanetti: not that I know precisely what to do about it :P [10:21] pstolowski: well, I'd say it'd be a good idea investigating into that if that returns the nullptr [10:22] seems like it should return an empty model instead of null [10:30] mzanetti: Saviq: what's the phablet-flash command one has to run nowadays? [10:30] phablet-flash cdimage-touch --pending ? [10:30] tsdgeos, actually no [10:31] tsdgeos: depends if you want the read-only image or not [10:31] phablet-flash ubuntu-system --channel devel-proposed [10:31] Saviq: ta [10:31] devel-proposed is a bit broken today [10:31] tsdgeos, and then `adb touch /userdata/.writable_image; adb reboot` [10:31] so might not be the best one to flash [10:31] Saviq: is apt-get able to install unity build deps on the ro image nowadays? [10:32] might be better using phablet-flash ubuntu-system --channel saucy [10:32] popey, well, for development... it's probably good to get the latest, even if *a bit* broken? [10:32] popey, how broken, btw? [10:32] sure [10:32] download-manager broken [10:32] so you cant download click packages [10:32] popey, k, we'll keep in mind, thanks [10:32] mzanetti, not without rw-ing the image [10:33] Saviq: last time I tried I rw-ed the image and it still bailed out [10:33] mzanetti, /me tries [10:33] Saviq: I think the place where dpkg stores its db is not large enough or something like that happened [10:34] mzanetti, oh [10:34] stupid jetlag [10:34] hi guys! [10:35] greyback: hey man [10:35] mzanetti: hey, I need to catch up with you on the devdays stuff [10:36] greyback: yep [10:36] wanna do that now? [10:37] popey, btw, "rm: cannot remove '/etc/init/ssh.override': Device or resource busy" any idea? [10:38] why you doing that out of interest? [10:39] popey, well, we probably shouldn't indeed [10:40] popey, i.e. "initctl start ssh" is enough (we'll fix in our scripts) [10:40] popey, but it's weird anyway [10:40] and that's after making the fs rw? [10:40] (and rebooting) [10:45] Saviq: hmm... I'd say run_on_device -s should still enable ssh permanently [10:45] pstolowski: i commented in https://code.launchpad.net/~aacid/unity8/lvwph_update_section_header/+merge/183457 [10:45] pstolowski: isn't the testcase enough to convince you the fix is good? [10:46] popey, yes [10:46] mzanetti, I'd say it shouldn't - as long as it will start it with every run... [10:47] Saviq: hmm... ok... I've been abusing run_on_device -s to make the device usable for my general purpose (which includes enabling ssh) [10:47] mzanetti, exactly ;) [10:48] mzanetti, btw, you can edit /userdata/system-data/etc... instead of making it rw [10:48] mzanetti, and that's possibly the correct way [10:48] Saviq: for enabling ssh? [10:48] mzanetti, yeah [10:48] Saviq: no... you need to delete that .override file [10:49] Saviq: and the bind-mounted stuff doesn't allow that [10:49] mzanetti, `echo "auto" > /userdata/system-data/etc/init/ssh.override` [10:49] mzanetti, that should work [10:49] Saviq: doesn't do for me [10:49] Saviq: as long as that file is here (even if empty) ssh doesn't autostart here [10:49] not sure why, tho [10:49] tsdgeos: as I said, I'm ok to approve if you think it still makes sense [10:49] pstolowski: i do [10:50] popey, actually that might be the reason for the error - the file is also in /userdata/... (I assume overlaid?) [10:50] tsdgeos: ok [10:51] mzanetti, anyway, run_on_device seems to work here [10:51] we *do* need a -c option to at least remove unity8-build-deps from the device... [10:52] Saviq: yep. it does. except that it fails to install build-essentials. haven't looked into that yet [10:52] mzanetti, installed fine here, it's building for me [10:52] Saviq: since a week or so I need to manually apt-get build-dep unity8. otherwise cmake wouldn't be installed after a fresh flash... [10:52] mzanetti, yes - that's it [10:53] mzanetti, drop the unity8-build-deps .deb [10:53] from where? [10:53] mzanetti, it only installs it when it's older than debian/control [10:53] mzanetti, from ~phablet/ [10:53] mzanetti, /me wanted to save generating / installing it every time [10:53] hmm... I'm pretty sure I had this also after flashing with -b [10:54] mzanetti, /me doubts it [10:54] but obviously that doesn't work out well [10:54] not 100% positive on it... [11:15] mzanetti, noticed one more thing you should tackle while "crossing" the category boundary [11:16] mzanetti, it shouldn't be possible to go "out" of the filtered model [11:16] Saviq: yeah... I'm kinda struggling with that currently [11:16] Saviq: as in the ScopeView I don't even have access to all those models [11:17] mzanetti, we might need to expose them [11:18] tsdgeos, here's a bug "spawned" out of the other one https://bugs.launchpad.net/unity8/+bug/1229144 [11:18] Ubuntu bug 1229144 in Unity 8 "Missing scope headers" [High,Triaged] [11:20] Saviq: i see [11:21] Saviq: what I find a bit weird is that apparently model.results in the delegate gives a *model. but calling data(index, "results") gives a string [11:21] mzanetti, is data() invokable? [11:21] yeah [11:22] mzanetti, did you try get() instead? [11:22] mhr3: there is no get() in QSortFilterProxyModelQml [11:23] and I'm afraid there can't be [11:23] mzanetti, why are you getting sortfiltermodel? [11:23] is that the count>0 wrapper thing? [11:23] mhr3: scopeview.categories [11:24] mzanetti, that returns DeeListModel [11:24] s [11:24] print("...........................finding model", categoryView.model) [11:24] ...........................finding model QSortFilterProxyModelQML(0x1db4470) [11:24] ...........................finding model QSortFilterProxyModelQML(0x1db4470) [11:24] damn copy/paste [11:27] mzanetti, scopeView.categories is SortFilterModel scopeView.scope.categories should give you access to the DeeListModels [11:30] mhr3: ah... that gives me a object with "count" 0 [11:32] mzanetti, what exactly are you doing? [11:32] mhr3: in onPressAndHold I get the model and the index of the pressed item [11:33] mhr3: but the thing is, the Preview must be able to walk over all categories [11:33] mhr3: so I need to find out which model this is (if there are any other models before and after this) [11:33] you mean rows in the model [11:35] mhr3: if you look from the ScopeListView POV yes, if you look at it from the FilterGrid POV not, in that case its really other models before/after this model [11:35] but in the onPressAndHold handler I've only got data from the FilterGrid POV [11:37] mzanetti, you have the delegateItem.model, isn't that enough? [11:38] mhr3: no... that only points to the model for this category [11:38] but I need all categories [11:39] mzanetti, wasn't the previewing supposed to work only within a category? [11:39] mhr3: not any more since Saviq is back :P [11:39] mhr3: no... I think design wants it to include the whole scope [11:39] mhr3, is that the case for unity7? [11:40] mhr3, mzanetti ok, looks like we need clarification from design [11:40] Saviq, yes [11:40] * Saviq always assumed it's "global" [11:40] Saviq, moreover it'd be terribly complex, imagine you cross to a completely different renderer [11:40] from a code POV I think including all categories seems like a major headache right now. [11:40] indeed [11:41] mhr3, different renderer doesn't matter - we're not "replacing" the model for the current renderer [11:41] from a usability POV... not sure... I personally think crossing categories might be confusing... but not entirely sure [11:41] mhr3, even if you mean different *category* renderer, doesn't really pose much more headaches [11:42] * Saviq asks Oren [11:42] Saviq, yes i meant category renderer [11:42] ack [11:42] going from previewing grid to carousel [11:43] i can see the transition to make it nice really complex [11:43] mhr3, doesn't feel more complex than enabling preview in the carousel itself [11:43] hmm, perhaps [11:44] mhr3, the view behind the preview would just scroll, and then we'd need to scroll the carousel to show the entry in question [11:44] anyway, crossing categories seems silly :P [11:44] maybe, /me assumed it'd work :) [11:44] immagine you swipe through the list of available apps and suddenly you end up in dash plugins. [11:44] especially since it's not like you can expect all the previews in a category to be of the same type [11:44] exactly, yeah [11:45] bbiab === greyback is now known as greyback|food [12:06] Saviq: already got an answer from Oren? [12:09] mzanetti, no [12:09] mzanetti, he's in a meeting [12:10] ok [12:10] mzanetti, btw, as part of it, you could improve the situation on bug #1224555 [12:10] bug 1224555 in Unity 8 "[Dash] Previews do not match visual designs" [High,Triaged] https://launchpad.net/bugs/1224555 [12:11] mzanetti, re: header, at least [12:11] Saviq: afaik paulliu is working on that [12:11] paulliu: are you? [12:13] mzanetti: yeah.. [12:13] Saviq: ^ [12:13] paulliu, can you please update the above bug? [12:14] assign yourself and such [12:14] sure. [12:19] mzanetti, mhr3 Oren confirmed - stop on category boundary [12:19] Saviq: nice [12:19] good, that should actually be doable :) [12:20] Saviq: so I'll clean this mess up and propose for merging [12:20] mhr3: yeah, its actually done already [12:37] ok, our autolanding doesn't work [12:37] fginther, help ;( [12:37] fginther, http://s-jenkins:8080/job/unity8-autolanding/ [12:38] fginther, the three landings of unity8 today failed in the same exact way - gallery apps tests failed on phones [12:40] Saviq: hey, with today's trunk I'm getting qml errors about 'Hud' type; what do I miss? [12:40] pstolowski, what error exactly? [12:40] right... so under otto it fails 'cause we're not killing notify-osd [12:40] mzanetti, remember ↑? [12:40] Saviq: yeah... [12:40] mzanetti, what did we decide re: notify-osd vs. our own notification interface? [12:41] Saviq: export `dbus-launch` [12:41] mzanetti, and why didn't we do that yet? [12:41] Saviq: actually I had hoped to do this in the autopilot stuff itself. but that doesn't work [12:41] mzanetti, thing is, I'm afraid if the whole shell is run under a "new" dbus [12:41] Saviq: because the pythong notify libs are being loaded before the ctor is called. so they load the env var before we have a chance to change it [12:42] * mzanetti really wonders why he can't type python without a g at the end [12:42] mzanetti, that might be even worse - whatever is run in the session won't be available [12:42] Saviq: ah, nvm, it's actually Nick's filter-selector branch, trunk runs fine [12:42] pstolowski, for that you probably need a newer UITK [12:42] pstolowski, or even Nic's branch, for that matter [12:42] Saviq: yeah [12:42] pstolowski, it's linked on Nic's MR [12:43] Saviq: what do you mean with that? what would be running what's not available? [12:43] mzanetti, if stuff isn't dbus-activated, but upstart-managed === greyback|food is now known as greyback [12:43] mzanetti, if we run the whole suite under a new dbus [12:43] mzanetti, we won't get those [12:43] yeah... might cause issues. I agree [12:44] we really should have an "empty" session [12:44] or better: it might hide issues we have [12:44] without unity7 [12:44] +1 [12:44] that's what we did in the vms so far [12:44] I had unity7 shut down and only started a plain X [12:44] fginther, didrocks, can we have otto run unity8 tests without unity7? [12:45] we can work around issues - like we did until now - but their amount will grow and cause us carrying cruft :/ [12:47] pete-woods, https://code.launchpad.net/~pete-woods/libusermetrics/color-themes-gconf/+merge/186770/comments/426459 [12:47] pete-woods, let's continue the discussion here, if we need to, latency on MRs is too big ;) [12:48] Saviq: relocatable schema sounds good to me [12:48] Saviq: is this supported by stuff like dconf-editor? [12:48] pete-woods, sure [12:49] pete-woods, a relocatable schema is just one that doesn't have a default path associated with it [12:49] pete-woods, so you need to explicitly point it at some place - and there can be multiple of those [12:49] Saviq: okay, then we just have a fixed schema in some place with a string list of the theme "names"? [12:50] pete-woods, shouldn't even be needed [12:50] pete-woods, as long as we put all of them under a known path [12:50] Saviq: how do I iterate over them? [12:50] so all of the objects under /somewhere/themes/ would be those [12:50] pete-woods, you just ls /somewhere/themes/ and assume all of theme are that [12:51] Saviq: is there some "children of node X" type method in the Qt QPI? [12:51] pete-woods, of sorts [12:51] pete-woods, dunno, larsu, around? === _salem is now known as salem_ [12:53] Saviq: yep === mzanetti is now known as mzanetti|food [12:53] larsu, so, we need multiple sets of values for "infographic palettes" [12:53] larsu, so that would be palette1/foreground palette1/background etc. [12:54] larsu, would relocatable dconf schemas be a good way to store those in dconf? [12:54] larsu, question no. 2 - to iterate over those, should they just all be stored under a common path, so, say /com/canonical/infographic/palettes/{palette1,palette2} etc.? [12:55] and if so, do we need a *list* of those paths stored somewhere, or would iterating over /com/canonical/infographic/palettes/'s children be enough [12:56] Saviq: definitely a question for desrt, not me [12:56] let's move this to #ubuntu-desktop :) [12:56] larsu, and, would we have APIs for that in the Qt gsettings APIs? [12:56] larsu, k [12:56] Saviq: we don't have APIs like that in gsettings-qt, but I can certainly add it [12:57] larsu, k, let's see what desrt says [12:59] pete-woods, join #ubuntu-desktop please? [13:02] Saviq: do you have some kind of repro steps for https://bugs.launchpad.net/unity8/+bug/1225391 ? [13:02] Ubuntu bug 1225391 in Unity 8 "Installed Applications is blank after doing a search" [High,Triaged] [13:02] i can't repro it here [13:03] tsdgeos, well, yeah, search, clear, search, clear, search, clear... [13:03] * Saviq tries [13:04] tsdgeos, right, can't repro [13:04] mhr3, can you ↑? [13:04] * Saviq flashes maguro === iahmad is now known as iahmad|afk [13:05] errrr [13:05] how am i supposed to stop a search? [13:06] i.e. hide the search bar [13:07] hmmm [13:08] we aren't resizing the shell when the OSK shows so i can't see some stuff without hiding the OSK [13:08] :/ [13:08] tsdgeos, it's "persistent" [13:09] tsdgeos, so if it happens, you can scroll down to see [13:09] tsdgeos, bug #1213034 [13:09] bug 1213034 in Unity 8 "Can't dismiss keyboard by tapping outside of search entry" [High,Fix committed] https://launchpad.net/bugs/1213034 [13:09] huh? how is that fix committed? [13:10] exactly my question :D [13:11] not being able to dismiss it is a pain [13:11] since i can't see everything easily [13:12] tsdgeos, yeah, it's a regression [13:14] and that's where we need an autopilot test :-) [13:14] * tsdgeos hides [13:15] tsdgeos, not ap, qml test is enough [13:16] tsdgeos, and completely +1 [13:16] Saviq: autopilot to test the OSK is not really there, no? [13:16] tsdgeos, that shouldn't be a test in unity8, though [13:16] tsdgeos, it should be tested with the SDK - we should just make sure our text field is unfocused [13:16] true [13:17] i'm lost now [13:18] fginther, didrocks, please help ;( [13:18] did https://code.launchpad.net/~aacid/unity8/lvwph_update_section_header/+merge/183457 failed to merge because some tests in the gallery fail? [13:18] tsdgeos, right [13:18] why? [13:18] Saviq: I don't know how we can do it (and in hangouts about Mir TBH) [13:18] tsdgeos, that's something you missed - we're running *some* tests across the board [13:18] tsdgeos, so as to test if something doesn't break something else [13:19] Saviq, sorry, I'm trying to catch up now. [13:19] tsdgeos, so, gallery with unity8, some other apps with UITK etc. [13:19] so we never get anything merged because all our tests are unstable :D [13:19] fginther, I can give you a summary [13:19] tsdgeos, or theirs ;) [13:19] fginther, our tests assume we're running in a clean session (no unity7 running) [13:20] fginther, and while we could think of working around stuff [13:20] so i just keep approving until magic happens? [13:20] tsdgeos, nah, right now we have a different issue [13:20] ook [13:20] tsdgeos, 'cause we've moved from VM'ed tests to real hardware ones [13:20] tsdgeos, which means we're under unity7, and that causes issues we need to tackle first [13:20] i see [13:20] Saviq, ah, right [13:21] fginther, we'll grow the number of workarounds unnecessarily, IMO [13:22] to test in a non-real environment anyway, where unity8 runs in a unity7 session [13:22] Saviq, so we essentially need to kill unity7 first, right? [13:22] fginther, not even that [13:22] fginther, ideally we shouldn't start it at all [13:22] fginther, i.e. start a unity8 session instead [13:23] Saviq, got it, the VM setup is coming back to me now [13:23] fginther, otherwise we might end up hiding real issues [13:23] fginther, case at hand - notify-osd is prestarted with unity7 [13:24] fginther, and while we could kill it so that unity8 takes the notifications over - doesn't sound like a correct solution [13:24] tsdgeos, it's really hard to repro, just keep searching it will happen at some point [13:24] fginther, we did think of running the tests under a separate dbus session, but again - that might cause some other things to fail, 'cause we *do* expect some other things to be pre-started [13:25] or at least might [13:25] tsdgeos, i just had it happen in the video scope [13:25] Saviq, I think I know how to fix this, but it might take most of the day. Do we need to fall back to the VMs to keep the testing moving? [13:25] fginther, we can't merge anything currently, 'cause our mediumtests are failing [13:25] tsdgeos, it seems to be more common when you expand a category [13:25] fginther, if we can go back to VMs for the time being, that could be a solution for us indeed [13:26] Saviq, ack, will set that up first [13:26] mhr3, ah! I think you hit the jack pot [13:26] mhr3, tsdgeos: expand "More suggestions", search, reset search → "More suggestions" empty [13:26] greyback, Hey, so we had a conference call where the focused observers came up. [13:27] greyback, It was mentioned that the app guys might want to have separate wakeup and focus events going to the shell. [13:27] tedg: yes sorry I missed it, I was traveling [13:27] greyback, That way it'd be "wake up" -> "send url" -> "focus" [13:27] greyback, Yeah, no problem. [13:27] tedg: I don't follow. Can you give me an example? [13:28] greyback, We had Saviq track your movements... we knew exactly where you were. [13:28] tsdgeos, uh oh, just managed to break the LVWPH again... [13:28] greyback, So on the case of an app that is in the background and it gets send a new URL to open. [13:28] * greyback looks around nervously [13:28] Saviq: wops [13:28] tsdgeos, expand "Installed", tap "SEARCH" [13:28] tsdgeos, breakage [13:28] greyback, So then if it was paused, it would have to be unpaused before getitng the URL [13:29] greyback, And it'd be better if it could get the URL before being shown to the user. [13:29] tsdgeos, but yeah, if an expanded category disappears [13:29] tsdgeos, it doesn't come back [13:29] mhr3, can you confirm ↑? [13:29]