[06:57] <zbenjamin> Saviq: i see my branch has been merged... is it already on the image?
[06:57] <Saviq> zbenjamin, http://people.canonical.com/~ogra/touch-image-stats/211.changes
[06:58] <Saviq> zbenjamin, not yet apparently
[06:58] <Saviq> zbenjamin, the next one
[06:58] <zbenjamin> Saviq: :/
[06:58] <Saviq> zbenjamin, ah no
[06:58] <Saviq> zbenjamin, yeah, it's there in 211
[06:58] <Saviq> zbenjamin, sorry, eyes didn't wake up yet :)
[06:58] <zbenjamin> Saviq: but not on the RTM image i guess
[06:58] <Saviq> zbenjamin, no, it's in a silo for that
[06:58] <Saviq> zbenjamin, awaiting QA sign-off
[06:59] <Saviq> zbenjamin, http://people.canonical.com/~platform/citrain_dashboard/#?distro=ubuntu-rtm&q=landing-001
[06:59] <zbenjamin> Saviq: ok because thats what i use , on the devel-proposed image the click packagage signing checks are enabled and make it impossible to install click packages from the SDK
[06:59] <Saviq> zbenjamin, I believe that was pulled back again
[07:00] <zbenjamin> Saviq: ok then i'll check that later
[07:00] <Saviq> zbenjamin, if you get me a .click package I can test out for you, just flashing 211
[07:06] <Cimi> morning
[07:08] <zbenjamin> Saviq: sent you a mail
[07:08] <Saviq> zbenjamin, tx, will check in a mo
[07:11] <tsdgeos> Saviq: ping
[07:12] <tsdgeos> i didn't realize your seeLessStuckBottom did so many things different to mine
[07:12] <tsdgeos> in particular you changed expandedCategoryId to expandedCategoryItem which is wrong
[07:13] <Saviq> tsdgeos, was not me, was Daniel actually
[07:13] <Saviq> tsdgeos, http://bazaar.launchpad.net/~unity-team/unity8/seeLessStuckBottom/revision/1128
[07:14] <tsdgeos> good
[07:14] <tsdgeos> code is mor declarative
[07:14] <tsdgeos> and doesn't work :)
[07:16] <tsdgeos> shall i simply revert the whole change?
[07:17] <Saviq> tsdgeos, what doesn't work there? I didn't notice any wrong behavior when testing?
[07:17] <tsdgeos> Saviq: well, storing the pointer is wrong
[07:18] <tsdgeos> you can't compare pointer delegates
[07:18] <tsdgeos> because the next delegate represent the same category will obviously have a different pointer
[07:18] <Saviq> tsdgeos, right, so what's broken is expansion across searches?
[07:18] <tsdgeos> that's why we store the category id and not the pointer
[07:18] <tsdgeos> Saviq: and expansion across scrolling
[07:19] <tsdgeos> expand, scroll, scroll back, broken
[07:19] <Saviq> tsdgeos, right
[07:19] <tsdgeos> that might get fixed once/if we merge the branch i did to keep all the categories created
[07:20] <Saviq> yeah, but not across searches
[07:20]  * Saviq was never fond of the persistence of the expansion TBH
[07:20] <Saviq> across searches that is
[07:21] <Saviq> zbenjamin, status in bug #1361349 will tell you what's in RTM btw
[07:21] <Saviq> zbenjamin, hmm and I didn't get no email?
[07:21] <Saviq> ah
[07:22] <Saviq> junk
[07:22] <zbenjamin> Saviq: so you consider my click package junk? ;)
[07:22] <Saviq> zbenjamin, thunderbird does
[07:23] <tsdgeos> Saviq: so where do you want that reversion? in a new separate branch or is part of two_see_more_bugfix ?
[07:24] <Saviq> zbenjamin, yeah, pkcon install-local installed it fine
[07:24] <Saviq> tsdgeos, the same, just name it "three_see_more_bugfix" ;)
[07:25] <tsdgeos> ok
[07:33] <zbenjamin> Saviq: awesome thx for testign
[08:05] <Cimi> tsdgeos, isn't like the indentation incorrect here? https://code.launchpad.net/~aacid/unity8/dashOverviewFromTempScopePreview/+merge/232378
[08:05] <tsdgeos> Cimi: no
[08:05] <tsdgeos> unless you mean that the whole list is
[08:05] <tsdgeos> then yes
[08:06] <Cimi> tsdgeos, yes
[08:07] <Saviq> tsdgeos, btw, sorry for missing the expansion behaviour, I didn't look too closely as Daniel top-acked it already
[08:07] <Saviq> tsdgeos, sounds like we need a test, but let me try and verify that that's still the behaviour we want (persistency across searches that is)
[08:08] <Saviq> tsdgeos, btw, your popover test fix didn't help apparently :/
[08:08] <tsdgeos> yeah i saw
[08:08] <tsdgeos> can try making that last test a trycompare
[08:08] <tsdgeos> will do in a bit
[08:15] <Saviq> tsdgeos, so you basically use valgrind to slow stuff up during a test run locally?
[08:15] <tsdgeos> that's what i did yesterdya yeah
[08:15] <Saviq> nice ;)
[08:15] <tsdgeos> sometimes it is unworkable
[08:16] <tsdgeos> since it takes too much
[08:16] <tsdgeos> and then tryCompare fails
[08:16] <tsdgeos> because takes more than 5 sec
[08:17] <tsdgeos> you know what?
[08:17] <tsdgeos> the test fails all the time here now :D
[08:26] <tsdgeos> Saviq: and the test is righfully changing
[08:27] <Saviq> tsdgeos, oh interesting
[08:27] <tsdgeos> something's bringing back the focus to the text field after clicking in a popover result here in tryPageHeader
[08:28] <tsdgeos> and from what i can see it isn't us
[08:31] <tsdgeos> now what made this break, i do not know
[08:33] <tsdgeos> the pluin is from 26th
[08:33] <tsdgeos> pluin -> SDK
[08:33] <tsdgeos> so may well be that
[08:40] <tsdgeos> Saviq: yeah SDK
[08:41] <Saviq> tsdgeos, yay :
[08:41] <tsdgeos> they made their popup destruction restore focus
[08:41] <tsdgeos> which i can see how it makes sense
[08:41] <tsdgeos> but broke us
[08:43] <tsdgeos> i think i can workaround, let me see
[08:43] <Saviq> tsdgeos, since when do bindings with functions introspect the function body ?¿
[08:43] <tsdgeos> they don't afaik
[08:44] <tsdgeos> it may just happen that you get the binding updated for some other reason
[08:44] <Saviq> tsdgeos, http://pastebin.ubuntu.com/8167099/
[08:44] <Saviq> tsdgeos, that's due to find*Child
[08:44] <tsdgeos> :S
[08:44] <Saviq> or actually findInvisibleChild
[08:44] <tsdgeos> weird
[08:44] <Saviq> because findChild looks at children
[08:45] <tsdgeos> Saviq: no, can't workaround it
[08:46] <Saviq> tsdgeos, I think I saw the bad behaviour too
[08:46] <Saviq> but can't pinpoint when
[08:46] <tsdgeos> Saviq: shall i do a quick MR against SDK adding a "propery bool restoreFocus: true" to Popup?
[08:47] <Saviq> tsdgeos, hmm, why would it restore focus?
[08:48] <MacSlow> Saviq, just pushed the fix for the "white-bar on top of sim-unlock snap-decision" to my branch... I guess it's too late for the silo, right?
[08:48] <Saviq> MacSlow, it landed already, yeah ;)
[08:48] <tsdgeos> Saviq: because is the new default, i jus want to expose the property so we can set it to false
[08:48] <Saviq> MacSlow, but that's fine, we'll get another one
[08:49] <MacSlow> Saviq, I could swear I had that fix pushed before... I don't remember seeing it when I did local tests... no idea how that slipped
[08:49] <Saviq> tsdgeos, looks like https://bugs.launchpad.net/ubuntu-ux/+bug/1353966 ?
[08:50] <Saviq> hmm
[08:50] <tsdgeos> Saviq: no?
[08:50] <Saviq> tsdgeos, no ;)
[08:50] <MacSlow> Saviq, I'll to a "paranoid" check of all of my other branches... I don't want to be surprised like that again :)
[08:50] <Saviq> aargh devel branches ffs
[08:50] <tsdgeos> Saviq: https://bugs.launchpad.net/bugs/1234982 is the bug it tries to fix
[08:51] <tsdgeos> don't see how it helps
[08:51] <tsdgeos> but it's the bug number attached to the change
[08:52] <Saviq> tsdgeos, http://bazaar.launchpad.net/~ubuntu-sdk-team/ubuntu-ui-toolkit/staging/revision/1211 then
[08:52] <tsdgeos> Saviq: yes, that one
[08:52] <tsdgeos> see the change in popupbase
[08:53] <Saviq> tsdgeos, mhm, talk to zsombi please
[08:54] <Saviq> tsdgeos, or well, if we create the popup before we focus (but we do create it on focus don't we?)
[08:54] <Saviq> oh yay, got a popup in the middle of the screen now
[08:54] <Saviq> because it restored focus... uugh
[08:55] <tsdgeos> Saviq: yes, we create it on focus, only way to know we have to create it
[08:55] <Saviq> tsdgeos, mhm, I'm not totally sure I agree with that behaviour
[08:56] <Saviq> but maybe I don't know everything
[09:42] <tsdgeos> Saviq: our qmluitests take 2h in CI ?¿
[09:43] <Saviq> tsdgeos, yup, an hour to install deps / build + 40 sthg to run the tests :/
[09:43] <tsdgeos> :/
[09:43] <Saviq> tsdgeos, I have plans to improve that, but time...
[09:43] <tsdgeos> sure :D
[09:55] <Saviq> Wellark, huh, is it expected that one of the SIMs goes "no signal" when the other is on call?
[09:57]  * Saviq gets No signal all the time on one of the SIMs, need to swap them...
[10:07] <Saviq> hmm so the no signal is a lie
[10:09] <Cimi> Saviq, looks like we might have indeed issues with the carousel and card size
[10:09] <Cimi> Saviq, i was trying with scope tool
[10:09] <Cimi> Saviq, do you have a category handy I can override?
[10:09] <Cimi> I don't have enough music to show me carousel
[10:10] <Saviq> Cimi, just get more music?
[10:10] <Saviq> Cimi, or override any random category?
[10:11] <Cimi> Saviq, yeah but some others are broken too
[10:11] <Cimi> Saviq, the carousel seems to clip
[10:11] <Cimi> so I was wondering what we want to support
[10:11] <Saviq> Cimi, yes, because it only supports overlay
[10:11] <Saviq> Cimi, that's fixed in the scope already
[10:11] <Saviq> Cimi, don't look at carousel, that is fine
[10:12] <Cimi> Saviq, I tried this in grooveshark http://paste.ubuntu.com/8167638/
[10:12] <Saviq> Cimi, look at when you tap on a result and open
[10:12] <Saviq> Cimi, carousel is fine, leave it
[10:12] <Cimi> ok
[10:12] <Cimi> this is broken btw
[10:12] <Cimi> shows text clip
[10:12] <Saviq> yes, because it's not overlaid
[10:12] <Saviq> when it should be
[10:12] <Saviq> carousel assumes there's nothing below the card (per design)
[10:13] <Saviq> Cimi, and that's fix committed to the scope already
[10:13] <Saviq> Cimi, what is not fine is the single large card when you open an artist search
[10:15] <Cimi> Saviq, how do I open an artist search?
[10:15] <Saviq> Cimi, just tap on an artist in music (the tool does not support canned queries, so in the real thing)
[10:17] <Cimi> I don't have artists??
[10:17] <Cimi> I have few coldplay albums
[10:17] <Cimi> I tap it opens albums
[10:17] <Cimi> oh maybe I need two artists
[10:17] <tsdgeos> easy joke incoming "Coldplay don't produce art thus no artist"
[10:20] <Saviq> ;)
[10:24] <Wellark> Saviq: yes
[10:24] <tsdgeos> j/k it's like when Skrillex added this sample at the end of one of his songs https://www.youtube.com/watch?v=Gs3ocG5yW88
[10:24] <Wellark> but the signal strength should return
[10:24] <Wellark> when you end up the call
[10:24] <Wellark> Saviq: you can still check with awe and abeator
[10:25] <Saviq> Wellark, yeah weird thing is stuff works fine, but the indicator just goes to no signal for no reason
[10:25] <Wellark> Saviq: it goes to no signal if ofono tells it to go no signal
[10:25] <Wellark> Saviq: so A) there really is no signal
[10:25] <Saviq> Wellark, sure, I understand, I didn't say that's your fault
[10:25] <Wellark> B) ofono is not updating the signals properly :)
[10:26] <Saviq> Wellark, just wanted to know your experience
[10:26] <Wellark> Saviq: your experience matches mine
[10:26] <Wellark> except that the signal should recover once you hang up the phone
[10:26] <Saviq> sure it did
[10:26] <Saviq> but then it was switching back'n'forth after that
[10:26] <Saviq> even though everything seemingly worked
[10:27]  * Saviq wonders if the names should be mapped to SIM serial numbers...
[10:27] <Wellark> Saviq: the back and forth is another bug
[10:27] <Wellark> which will get fixed on the next ofono planging
[10:27] <Saviq> Wellark, oh ok, yeah, my 2nd SIM does that, regardless which SIM I put in
[10:28] <Saviq> I mean I swapped them and it's still the second one that jumps about
[10:29] <tsdgeos> Saviq: we're hitting a few asserts on tests when running on a debug qt (i.e. one that has asserts enabled)
[10:29] <tsdgeos> :(
[10:29] <tsdgeos> QFATAL : qmltestrunner::Dash::test_temp_scope_dash_overview_all_search_temp_scope_favorite_from_all() ASSERT: "e->isAccepted()" in file items/qquickitem.cpp, line 4490
[10:30] <Saviq> tsdgeos, oops
[10:30] <Saviq> huh, interesting
[10:30] <tsdgeos> which is ultra weird
[10:30] <tsdgeos> since it's a keyevent
[10:30] <tsdgeos> and i don't see us having key events on that test
[10:30] <Saviq> it seems the second SIM slot is slow for data...
[10:30] <tsdgeos> Saviq: yes
[10:30] <tsdgeos> Saviq: i think someone mentioned 3g is only available on first
[10:31] <tsdgeos> rings a bell
[10:31] <Saviq> well then the indicator is wrong
[10:31] <Saviq> says H still
[10:35] <tsdgeos> ask around, my bell may be broken :D
[10:44]  * Saviq sends stupid photos around, wonder when the sensor will break
[10:50] <Saviq> 006087
[10:51]  * Saviq managed to remember a 16-random-char password for U1 SSO... #brainisstupid
[10:51] <Cimi> Saviq, I put another artist, no luck
[10:51] <Saviq> Cimi, search for it?
[10:52] <Saviq> Cimi, what do you see in the Music scope?
[10:52] <Cimi> albums
[10:53] <Saviq> Cimi, that's Music or My Music?
[10:54] <Cimi> music
[10:54] <Saviq> Cimi, where are those albums from?
[10:54] <Cimi> local
[10:54] <Cimi> here
[10:54] <Saviq> Cimi, and it says "My music" above them? and there's Grooveshark results below?
[10:55] <Cimi> yes
[10:55] <Saviq> Cimi, I only get artists there, not albums...
[10:56] <Saviq> Cimi, do you have tags in your music at all?
[10:56] <Cimi> Saviq, yes I have artist
[10:57] <Cimi> albums, genres, tracks
[10:57] <Cimi> inside my music navigation
[10:57] <Saviq> Cimi, I've artists, albums, genres, tracks...
[10:58] <Saviq> Cimi, so your music selection must really not sit together with the media scanner ;P
[11:02] <Cimi> Saviq, so can you tell me what I can put in scope-tool?
[11:02] <Cimi> pls?
[11:02] <Saviq> Cimi, I don't think you'll get the exact result with the scope tool as you can't override the result data
[11:02] <Saviq> Cimi, in card tool
[11:02] <Cimi> I'll unassign myself for now then
[11:03] <Cimi> I copied 3GB of music
[11:03] <Cimi> still no artist
[11:04] <Saviq> Cimi, http://pastebin.ubuntu.com/8168021/
[11:04] <Cimi> trying thaty
[11:04] <Saviq> Cimi, and see that implicitHeight is wrong
[11:04] <Cimi> tryCard?
[11:06] <Cimi> ah basically we want to not allocate height for summary here>?
[11:10] <tsdgeos> Saviq: http://paste.ubuntu.com/8168055/
[11:13] <Saviq> Cimi, implicitHeight should always be the real, minimum, height of the card (no forced height)
[11:13] <Saviq> Cimi, because it's used in vertical journals where there's no "allocation" happening
[11:14] <Cimi> Saviq, it is weird because we have   text: cardData && cardData["summary"] || ""; \n\  height: text ? implicitHeight : 0; \n\
[11:14] <Cimi> Saviq, maybe we need text != "" ?
[11:14] <Cimi> on height?
[11:16] <Saviq> Cimi, well, text [11:16] <Saviq> Cimi, that seems to be the problem
[11:18] <larsu> starting unity8/trunk gives me a black window and some warnings about a lightdm mock module not being found
[11:21] <larsu> does anyone have an idea what could be going on?
[11:22] <facundobatista> Holas
[11:26] <Saviq> larsu, how are you starting it?
[11:27] <Saviq> larsu, ./build.sh and ./run.sh are best set up to take care of it now
[11:27] <larsu> Saviq: running the binary directly (but setting QML2_IMPORT_PATH to what run.sh sets it to)
[11:27] <larsu> run.sh doesn't work for me, because I don't have unity8 installed on my system
[11:28] <Saviq> larsu, you need LD path too
[11:28] <Saviq> larsu, just copy data/unity8.conf to ~/.config/upstart
[11:28] <larsu> hm? Doesn't qml load .so files directly?
[11:28] <larsu> oh, interesting
[11:28] <larsu> that makes it work, thanks
[11:29] <Saviq> larsu, if you care about the dash, copy unity8-dash.conf, too
[11:29] <larsu> oooh, colored indicators
[11:29] <Saviq> larsu, ;)
[11:30] <Saviq> but note that by default the dash that launches with ./run.sh is the system-wide one (so it will fail in your case), we need to fix the run script to support the split-dash situation
[11:30] <Saviq> larsu, means you got the mock indicator backend
[11:30] <larsu> dash works for me
[11:30] <larsu> seems to be a mock as well
[11:30] <Saviq> larsu, just an image ;)
[11:31] <Saviq> see the SURFACE part? ;)
[11:31] <Saviq> larsu, yeah, we can't do window management under X11 ya know
[11:31] <larsu> ya, makes sense
[11:32] <larsu> I'm just trying to test whether doing s/StatusIcon/Icon works
[11:33] <Saviq> larsu, not sure why you'd get mock indicators, though :/
[11:34] <larsu> Saviq: a stray QML2_IMPORT_PATH. Now I don't get any indicators and the system's dash
[11:35] <Saviq> unless you copied unity-dash.conf to upstart and you don't have unity8 installed, there can be no dash as upstart doesn't know the job...
[11:35] <Saviq> unity8-dash
[11:35] <Saviq> and even if you did copy it
[11:35] <Saviq> the job doesn't know how to run the non-installed dash
[11:35] <larsu> I copied it
[11:35] <Saviq> so should fail
[11:35] <larsu> I have a dash though
[11:35] <Saviq> larsu, inside unity8?
[11:36] <larsu> ya...
[11:36] <Saviq> larsu, that's just an image
[11:36] <Saviq> mzanetti, gonna kill ya ↑
[11:36] <Saviq> we need to replace that image :P
[11:36] <mzanetti> ?
[11:37] <mzanetti> nah...
[11:37] <mzanetti> I like it :)
[11:37] <Saviq> mzanetti, I'm gonna defer people to you every time someone mentions a dash inside unity8 under x11 :P
[11:37] <mzanetti> we might stamp a red "Facsimile" on it though
[11:37] <larsu> argh, now it stopped working again
[11:37] <Saviq> mzanetti, we have a yellow SURFACE stamped on it now
[11:37] <Saviq> aanyway
[11:38]  * Saviq goes to try out BT in the car + /food
[11:38] <larsu> "unity8 start/running"  - but no window
[11:38] <mzanetti> Saviq: can you pair something now?
[11:38] <larsu> Saviq: bon appetit
[11:38] <Saviq> larsu, .cache/upstart/unity8.log
[11:38] <mzanetti> Saviq: I still can't pair anything BT
[11:38] <Saviq> mzanetti, it's supposed to work, will find out soon (don't have anything BT at home other than laptops and phone)
[11:38] <Saviq> s
[11:39] <mzanetti> that would be awesome
[11:39]  * Saviq wonders how will google sheets behave with no internets
[11:39] <mzanetti> Saviq: they don't
[11:39] <larsu> oh neat: *** Error in `/home/lars/src/unity8/trunk/builddir/src/unity8': corrupted double-linked list: 0x0000000000ef3a10 ***
[11:39] <mzanetti> just don't
[11:39] <mzanetti> hmm
[11:39] <Saviq> mzanetti, I have the sheet opened, didn't complain in plane mode yet ;)
[11:39] <larsu> also, a binding loop
[11:40] <mzanetti> larsu: what are you trying to do?
[11:40] <larsu> mzanetti: run unity from trunk (just shelved my changes)
[11:40] <mzanetti> larsu: shouldn't crash then :/
[11:40] <mzanetti> larsu: you using ./run.sh?
[11:40] <larsu> yep
[11:41] <mzanetti> hmm... well, sometimes I have such crashes when it doesn't rebuild everything. eg. after shelving
[11:41] <mzanetti> larsu: try a ./build -c
[11:41] <mzanetti> if its still crashing then, we need to debug
[11:41] <larsu> k
[11:44] <larsu> mzanetti: same crash
[11:44] <mzanetti> meh
[11:44] <larsu> I wonder why this happens now - it worked 10 minutes ago
[11:44] <mzanetti> yeah... weird
[11:44] <larsu> there's a bunch of image provider warnings as well
[11:45] <mzanetti> they shouldn't be fatal though
[11:45] <larsu> for the launcher icons
[11:45] <larsu> no, they aren't. The fatal one is the currupted double linked list error
[11:45] <larsu> works in gdb of course :(
[11:46] <mzanetti> ...
[11:46]  * mzanetti tries current trunk
[11:46] <mzanetti> working fine here
[11:47] <mzanetti> hmm...that's not good
[11:47] <larsu> might be related to copying unity8-dash.conf to ~/.config/upstart?
[11:47]  * larsu tries without
[11:48] <larsu> indeed
[11:48] <larsu> I still get the dash with my currently running apps though
[11:49] <mzanetti> larsu: I can't repro even with copying the file over
[11:49] <mzanetti> what's the currently running apps?
[11:49] <larsu> it's a heisenbug. I'm getting it again now
[11:49] <larsu> mzanetti: gnome-terminal, firefox, thunderbird
[11:50] <mzanetti> larsu: how are they related to unity8?
[11:50] <larsu> I don't know?
[11:59] <larsu> mzanetti: got it in gdb now: http://paste.debian.net/plain/117983
[12:02] <mzanetti> hmmm plugins/Unity/Indicators/rootactionstate.cpp:56
[12:02]  * mzanetti looks
[12:03] <larsu> looks like a noreml connect call to me - maybe @menu is dangling?
[12:03] <mzanetti> yeah...
[12:03] <mzanetti> m_menu is set from the outside using setMenu()
[12:03] <mzanetti> maybe the outside deletes it at some point
[12:03] <larsu> right
[12:03] <mzanetti> dednick: hey, can you maybe have look? ^
[12:04] <mzanetti> dednick: first guess would be m_menu is parent-less and QML engine garbage collects it
[12:05] <mzanetti> well, assuming the pointer ends up in QML at some point
[12:05] <mzanetti> yeah. its a Q_PROPERTY(). so most likely that's it
[12:06] <mzanetti> dednick: thanks for the trace, seems helpful
[12:06] <mzanetti> err larsu ^
[12:06] <dednick> mzanetti: hm. i remember something similar with the actions i think
[12:08] <larsu> mzanetti, dednick: manually setting a parent in that function doesn't resolve it though
[12:09] <mzanetti> I think the ownership is determined when it first reaches the qml context
[12:09] <larsu> and it can't be changed?
[12:09] <mzanetti> and as setMenu is called *from* QML that ship has sailed at that point
[12:09] <mzanetti> yes it can be changed
[12:09] <larsu> but qml doesn't notice?
[12:10] <mzanetti> yes, it does, but doing that here would probably be too late already (as the connect below crashes already)
[12:10] <mzanetti> and even if it would work, it would be a bad hack
[12:10] <mzanetti> but if you want to try, its something like setOwnership in QQmlContext iirc
[12:11] <larsu> yeah it's a hack - I'm just trying to suport your hypothesis :)
[12:11] <mzanetti> larsu: http://qt-project.org/doc/qt-5/qqmlengine.html#setObjectOwnership
[12:11] <larsu> hm, I don't have acces to the qmlengine in here though
[12:12] <dednick> huh. looks weird to me. if the menu was destroyed, why is it being set?
[12:13] <larsu> dednick: mzanetti's idea is that it is set and GCed right after
[12:13] <dednick> right after what?
[12:13] <larsu> setting it
[12:13] <dednick> but it's crashing in the set.
[12:14] <mzanetti> yeah... which doesn't line up I just realize
[12:14] <larsu> ah right. Is the GC running in the same thread?
[12:14] <mzanetti> hmm... actually
[12:14] <dednick> #11 0x00007fffd0b3087e in RootActionState::setMenu (this=0xcab270, menu=0xcfd530)
[12:14] <mzanetti> no it's not
[12:14] <larsu> so it might stil be that
[12:14] <larsu> but judging by how often I see it, I doubt it
[12:14] <larsu> it would need to be gced at exactly the right time every time
[12:14] <mzanetti> larsu: it doesn't always crash for you, right?
[12:15] <larsu> mzanetti: no, but more than half the time
[12:15] <mzanetti> can you try qDebug() << menu->parent()
[12:15] <larsu> in gdb a bit less often
[12:15] <mzanetti> would be interesting what it says in the non-crashing case
[12:15] <Saviq> mzanetti, hah! it did work (sheets with no wifi)
[12:15] <larsu> sure
[12:15] <Saviq> mzanetti, it only saves the changes after you press "ready"
[12:16] <mzanetti> Saviq: last time I tried it it broke the whole doc
[12:16] <Saviq> mzanetti, so I was able to do the changes and it synced them up
[12:16] <Saviq> mzanetti, wfm
[12:16] <Saviq> mzanetti, but, BT, did not, my car kit requires a reverse pairing (it's the car kit that initiates and says what pairing code to use)
[12:16] <Saviq> mzanetti, so it found the phone and as soon as it tried to connect it failed
[12:17] <Saviq> hah, note != comment
[12:18] <dednick> larsu: has this only been recent?
[12:18] <larsu> mzanetti: gcc complains  that unitymenumodel only has a parent(QModelIndex &) function
[12:18] <larsu> dednick: haven't run unity in a while
[12:18] <dednick> this must have to do with lp:~nick-dedekind/unity8/lp1328646
[12:18] <larsu> dednick: just tried it and hit this error
[12:18] <mzanetti> now that's strange
[12:19] <dednick> messed around with the object ownership of models
[12:19] <larsu> is qasbatrctitemmodel not a qobject?
[12:20] <mzanetti> it is
[12:20] <mzanetti> and it should have a QObject* parent() method
[12:21] <dednick> i wonder if something is trying to access it's parentItem, which it doesnt habve
[12:21] <dednick> but is accessed through the qml parent property
[12:22] <dednick> larsu: is it latest image?
[12:22] <larsu> dednick: it's unity8 trunk on my laptop
[12:22] <dednick> larsu: ah. running desktop shell in window?
[12:23] <dednick> ./run i mean
[12:23] <dednick> ./run.sh
[12:24] <larsu> mzanetti: http://paste.debian.net/plain/117987
[12:24] <larsu> dednick: yes
[12:25] <mzanetti> larsu: you said it inherits QAbstractItemModel?
[12:26] <mzanetti> larsu: then this might be the issue actually... I wonder how it compiles at all then
[12:26] <larsu> mzanetti: I think it does, yes. Why=
[12:26] <larsu> ?
[12:27] <mzanetti> not sure how you even can mask the parent() method of a QObject
[12:27] <mzanetti> http://qt-project.org/doc/qt-5/qabstractitemmodel-members.html
[12:28] <larsu> ya... setParent() works fine
[12:28]  * larsu loves c++
[12:29] <larsu> mzanetti: casting it to QObject makes it compile. parent() returns NULL
[12:31] <dednick> hm. weird that it's not happening to me.
[12:32] <larsu> ya, I get this fairly consistently now
[12:32] <dednick> larsu: can you try put a tracepoint in plugins/Unity/Indicators/unitymenumodelcache.cpp ln:41  (should be a connection to the QObject::destroy of the model)
[12:33] <larsu> sure
[12:34] <dednick> larsu: it's the qml trying to access the parent i think
[12:34] <dednick> mzanetti: ^
[12:34] <dednick> maybe
[12:34] <dednick> not sure why though
[12:35] <mzanetti> dednick: ?
[12:35] <mzanetti> yeah, QML is accessing the parent for sure
[12:36] <dednick> i meant about compaining about parent not existing
[12:36] <mzanetti> yeah, that's really weird
[12:36] <dednick> i'm pretty sure nothing explicitly tries to set the parent of the model.
[12:37] <dednick> oh. hm. maybe it's because of the javascript ownership
[12:37] <dednick> hm.
[12:37] <dednick> i have no idea
[12:37] <mzanetti> dednick: is it explicitly set to javascript ownership?
[12:37] <dednick> no. but it's a invokable function to get the model
[12:38] <larsu> dednick: it never arrives at that line
[12:38] <mzanetti> yeah... if it has no parent, i.e. parent() == nullptr, then it'll be javascript ownership
[12:38] <mzanetti> and then qml might GC it at any point
[12:38] <mzanetti> and its not safe to pass it in setModel() back to c++
[12:38] <dednick> larsu: huh. so that means the menu isn't being destroyed :/
[12:40] <dednick> mzanetti: but the gc happens on main thread? wont occur at the point inside setMenu right?
[12:40] <mzanetti> dednick: I think gc also happens in other threads... that behavior changed a bit with Qt 5.2 and Qt 5.3.
[12:40] <mzanetti> which is why we didn't have any of those crashes before, and since Qt 5.3 I fixed the parenting in like 10 places already
[12:41] <larsu> be back in ~20 mins
[12:42] <dednick> sounds dodgey to me
[12:45] <dednick> mzanetti: so. the only way gc can take place on the model is after we've released the model property from:
[12:45] <dednick> http://bazaar.launchpad.net/~unity-team/unity8/trunk/view/head:/qml/Panel/Indicators/CachedUnityMenuModel.qml#L32
[12:45] <dednick> right?
[12:46] <mzanetti> not sure about that...
[12:46] <mzanetti> dednick: no... I think if the using QML context is destroyed it would clean up the model property too
[12:46] <dednick> mzanetti: althought how does the referencing work? if there are multiple properties using the return from a c++ function, does it need to wait for all of them?
[12:46] <dednick> to go out of context.
[12:47] <mzanetti> hmm... I think so yes
[12:48] <dednick> if it's javascript owned that is. so if I point 2 of these "CachedUnityMenuModels" at the same menu model in the cache, they both need to get cleaned up?
[12:48] <mzanetti> I think yes
[12:48] <dednick> for the model to get deleted
[12:48] <dednick> hm.
[12:48] <mzanetti> but... not that I would ever have tried it
[12:48] <dednick> because that would explain things.
[12:48] <mzanetti> right
[12:48] <dednick> actually, there is a unity-test for that
[12:48] <dednick> unit-test
[12:49]  * mzanetti can't type unit without the y and more either
[12:49] <dednick> hehe
[12:49] <dednick> qmltest atually.
[12:49] <dednick> but that's relying on gc actually cooperating with the gc() function
[12:49] <dednick> which it doesnt really
[12:51] <dednick> anyway. i don't understand why this is happening since the model isn't being deleted.
[12:52] <mzanetti> dednick: I don't know the code around it and how it works... so I'm not that much of a help
[12:52] <dednick> otherwise we should hit: http://bazaar.launchpad.net/~unity-team/unity8/trunk/view/head:/plugins/Unity/Indicators/unitymenumodelcache.cpp#L41
[12:52] <dednick> mzanetti: i'm not supposed to be doing anything other than prompts at the moment. otherwise the gunn will be upset
[12:53] <mzanetti> dednick: can't you just pass (this) as argument in line 39 here?
[12:53] <mzanetti> dednick: then its owned by c++ and all is well
[12:53] <mzanetti> haha
[12:53] <dednick> mzanetti: lol. then Saviq isn't happy
[12:53] <mzanetti> why not?
[12:53] <dednick> because he wanted qml ownership!
[12:54] <mzanetti> ah
[12:54] <mzanetti> then make him fix the crash
[12:54] <Saviq> dednick, no, I didn't want refcounting in QML ;)
[12:54] <mzanetti> :)
[12:54] <dednick> Saviq: well this is what you get for that!
[12:54] <Saviq> dednick, didn't we really end up saying that these models will basically never get destroyed? (provided that they can switch profiles)?
[12:55] <dednick> Saviq: yeah. not sure why they are.
[12:55] <mzanetti> well, if they're never destroyed, one more reason to keep c++ ownership.. imho
[12:55] <dednick> mzanetti: well, they "can" be
[12:56] <dednick> mzanetti: just that they aren't under normal ops
[12:56] <mzanetti> apparently :)
[12:56] <dednick> if they aren't being used anymore
[12:56] <Saviq> *gasp* we don't have pastebinit in rtm?
[12:56] <mzanetti> ok well... I'm sure you guys figure it... I really lack knowledge how this whole menumodel construct works
[12:57] <dednick> mzanetti: it's not a menumodel problem :) it's purly qml isuee. so you're perfect match
[12:57] <dednick> :)
[12:57] <mzanetti> no its not...
[12:58] <mzanetti> if its purely a QML issue I'd say either don't pass the object back to c++ (in setModel()) or make it c++ owned
[12:58] <mzanetti> but both doesn't seem to be options because of the way the menumodel works
[12:58] <dednick> mzanetti: well, i don't pass it back via setModel. i use qml property
[12:58] <dednick> :)
[12:58] <mzanetti> dude
[12:59] <dednick> which is the same as any other qml object. they're all c++ backed
[12:59] <mzanetti> you don't want me to explain what Q_PROPERTY() expands to, domyou?
[12:59] <mzanetti> :)
[13:00] <dednick> but yeah. creating in c++ then passing back and forth probably not best. blame Saviq
[13:00] <dednick> i do!
[13:01] <tsdgeos> any idea where this is coming from?
[13:01] <tsdgeos> Failed to get image from provider: image://scaling/0.666667//home/tsdgeos_work/phablet/unity8/croppedImageMinimumSourceSize/tests/qmltests/Dash/artwork/avatar@12.png
[13:01] <tsdgeos> specially the scaling part
[13:02] <dednick> Saviq: otherwise i'm going to c++'ise CachedUnityMenuModel and put ref counting in there.
[13:02] <dednick> or shared pointers or whatever
[13:03] <Saviq> dednick, or let's just not cache them
[13:03] <dednick> Saviq: and have 3 models for each indicator
[13:03] <Saviq> dednick, you can share those without having the factory
[13:03] <Saviq> tsdgeos, UITK
[13:03] <dednick> Saviq: yeah but it would be realy nasty
[13:04] <Saviq> dednick, if you can get it to work proper without refcounting in QML, fine
[13:04] <Saviq> dednick, using shptr or whatever
[13:04] <mzanetti> meh for shptr :D
[13:04] <dednick> mzanetti: c++ ownership!
[13:05]  * larsu apologizes for the discussion he started ;)
[13:05] <mzanetti> larsu: no... its good you found the issue!
[13:06] <dednick> larsu: i don't really see how this is happening in the first place. should only occur if an indicator is being removed.
[13:07] <Saviq> greyback__, http://paste.ubuntu.com/8168840/, thread 25, does that look like a lock to you? (doesn't to me)
[13:07]  * Saviq blames scopes
[13:07] <dednick> larsu: bug#?
[13:07] <larsu> dednick: I on't see any indicators
[13:07] <larsu> dednick: no bug
[13:07] <larsu> I just found it
[13:08] <tsdgeos> Saviq: grrrr
[13:08] <dednick> larsu: does it happen pretty much as soon as the app window opens?
[13:08] <dednick> larsu: can you log one? i can't sort it right away. busy with other things
[13:08] <Saviq> tsdgeos, what? me?
[13:08] <tsdgeos> Saviq: no the uitk :D
[13:09] <tsdgeos> why is it changing source of my images?
[13:09] <Saviq> tsdgeos, anyway, the avatar shouldn't ever be @foo
[13:09] <Saviq> tsdgeos, because it's doing it for the @foo mechanism
[13:09] <larsu> dednick: it happens at different times. Sometimes before the window opens, sometimes after I close it
[13:09] <larsu> dednick: yes, I'll file a bug
[13:09] <dednick> larsu: but you have no indicators?
[13:09] <dednick> larsu: that's weird
[13:10] <larsu> yes
[13:10] <larsu> very weird indeed
[13:10] <dednick> might explain things
[13:10] <Saviq> larsu, didn't you have the coloured indicators?
[13:10] <dednick> or why i can't reproduce
[13:10] <Saviq> or is that gone now
[13:10] <larsu> Saviq: those went away after I removed the mock import pats
[13:10] <larsu> *paths
[13:10] <Saviq> mhm
[13:11] <dednick> coloured indicators? hang on. is this in testcase?
[13:12] <larsu> dednick: no, it's when you run unity8 in fake mode (./run.sh -f)
[13:13] <dednick> larsu: with no mock paths?
[13:13] <larsu> no, it adds the mock paths in that case
[13:14] <dednick> larsu: does bug also happen when running non fake mode?
[13:15] <larsu> dednick: the other way around: it _only_ happens in non-fake moe
[13:16] <larsu> *mode
[13:16] <dednick> larsu: ah. ok
[13:16] <dednick> thought that sounded weird
[13:17] <larsu> sorry if I wasn't clear
[13:22] <Saviq> facundobatista, can we report bugs against the remote scopes somewhere (like wrong header colours and such)?
[13:23] <facundobatista> Saviq, https://bugs.launchpad.net/ubuntu-rest-scopes/+filebug
[13:23] <Saviq> facundobatista, thanks
[13:23] <facundobatista> Saviq, 90% is faster just tell me (if they are small changes), but as you wish :)
[13:24] <Saviq> facundobatista, it was davmor2 finding it difficult to read in some scopes (7dig, grooveshark)
[13:24] <Saviq> facundobatista, foreground color should be white
[13:24] <Saviq> according to https://docs.google.com/a/canonical.com/file/d/0B8I8ZVKH-8SsS01CcE93TFNUSGc/edit
[13:24] <facundobatista> Saviq, it's a matter of response time... for example you tell me to change a couple of colors, I'll probably propose a branch in the next hours... but I don't know when I'll read the mails to find out a new bug was proposed, etc
[13:24] <Saviq> or white-ish
[13:25] <facundobatista> Saviq, you're saying, for example, the background color for the 7digital header?
[13:26] <Saviq> facundobatista, no, foreground color
[13:26] <Saviq> facundobatista, icons
[13:26] <Saviq> facundobatista, back and favorite and search
[13:26] <Saviq> facundobatista, so page-header.foreground-color or some such
[13:26] <facundobatista> oh, wait
[13:26] <facundobatista> Saviq, I can control that?
[13:26] <davmor2> facundobatista: Currently any header that is dark on the scopes makes it really hard to see the dark grey settings and favourite icons to the right of the header
[13:26] <Saviq> facundobatista, of course :)
[13:26] <facundobatista> Saviq, I got confused, sorry, yesterday I read this
[13:26] <facundobatista> PageHeader.ForegroundColor = default header text color (defaults to the overall foreground color)
[13:27] <facundobatista> and I thought "ah, the text, not the icons"
[13:27] <Saviq> facundobatista, yeah, it should say "also applies to icons...."
[13:27] <Saviq> facundobatista, and it shouldn't say 'default'...
[13:27] <zbenjamin> Saviq: should the url dispatcher bring the scope up on the screen?
[13:27] <facundobatista> davmor2, Saviq, Daniela mentioned changes to those colors, I thought I couldn't control them
[13:27] <Saviq> zbenjamin, in the dash, yes
[13:27] <zbenjamin> Saviq: because on the most recent devel-proposed the screen just flickers and nothing else :/
[13:28] <Saviq> zbenjamin, restart url-dispatcher
[13:28] <Saviq> zbenjamin, ah oh
[13:28] <Saviq> zbenjamin, let me try
[13:29] <Saviq> zbenjamin, `url-dispatcher scope://clickscope`
[13:29] <davmor2> facundobatista: So we all on the same page now though right?
[13:29] <Saviq> zbenjamin, works fine here?
[13:29] <zbenjamin> Saviq: the clickscope is the store?
[13:29] <Saviq> zbenjamin, apps
[13:29] <zbenjamin> Saviq: try any other please
[13:29] <zbenjamin> url-dispatcher scope://com.ubuntu.scopes.youtube
[13:29] <Saviq> url-dispatcher scope://com.canonical.scopes.clickstore
[13:29] <zbenjamin> maybe i do it wrong
[13:30] <facundobatista> davmor2, maybe, Daniela mentioned changes for 7digital and TWC, but not for Grooveshark
[13:30] <zbenjamin> ooo that works
[13:30] <Saviq> zbenjamin, youtube is wrong
[13:30] <Cimi> Saviq, is not implicitHeight of the card being calculated incorrectly, is the cardHeight in cardTool that forces an height that is uncorrect
[13:30] <Saviq> Cimi, that's not used in vertical journal
[13:30] <zbenjamin> Saviq: how to find out the correct url?
[13:31] <Saviq> zbenjamin, favorite it, `gsettings get com.canonical.Unity.Dash favorite-scopes`
[13:31] <Saviq> zbenjamin, there's probably a *real* way somewhere, but that works ;D
[13:31] <Cimi> Saviq, implicitHeight is correct
[13:31] <Cimi> Saviq, I just tested
[13:32] <Cimi> Saviq, but cardHeight is fixed regardless of the summary size
[13:32] <larsu> dednick: https://bugs.launchpad.net/unity8/+bug/1362609 - can't assign you for some reason, but I've subscribed you
[13:32] <zbenjamin> Saviq: how do i favorite it?
[13:32] <Saviq> Cimi, should not be in vertical journal
[13:32] <Saviq> zbenjamin, go to overview
[13:32] <Saviq> zbenjamin, press on the star icon
[13:32] <zbenjamin> aaa
[13:32] <dednick> larsu: ta
[13:33] <Saviq> zbenjamin, _youtube
[13:33] <Saviq> zbenjamin, you need the hook name too
[13:33] <dednick> larsu: can you add a u8 startup log to that as well?
[13:33] <zbenjamin> Saviq: ok i have something to try now ! thanks :)
[13:33] <Saviq> url-dispatcher scope://com.ubuntu.scopes.youtube_youtube
[13:33] <Saviq> zbenjamin, coolz
[13:34] <Saviq> Cimi, CardVerticalJournal does not force the card height on its delegates
[13:34] <larsu> dednick: sure
[13:35] <Cimi> Saviq, fixes to 18.5 gu, no~?
[13:35] <Saviq> Cimi, no
[13:35] <Cimi> yeah no
[13:36] <Cimi> I was testing normal ones anyway
[13:36] <Saviq> Cimi, and I tracked the wrong expanded height of a vjournal down to implicitHeight being incorrect (because it was the same as the whole journal's height - just one item)
[13:36] <Saviq> Cimi, and it was cut off
[13:36] <greyback__> Saviq: hmm, thread25 (renderer thread) is sitting, waiting to be told to render. (processEventsAndWaitForMore)
[13:37] <Saviq> Cimi, I can measure for you
[13:37] <Saviq> greyback__, yeah, what I thought
[13:37] <greyback__> Saviq: usaully means the GUI thread blocked
[13:39] <Cimi> Saviq, I am seeing mistakes in testCard
[13:39] <greyback__> Saviq: scopes use zeromq?
[13:39] <Cimi> Saviq, just clear out the summary text and the height is fixed
[13:39] <Cimi> implicitHeight changes
[13:40] <Saviq> Cimi, #Cimi?
[13:40] <Cimi> Saviq, I am there
[13:43] <Saviq> Cimi, don't use me for debugging then please
[13:43] <Cimi> Saviq, i debugged and I discovered height is fixed
[13:43] <Cimi> from cardTool
[13:43] <Cimi> and I told you :)
[13:43] <Saviq> Cimi, in the test, maybe, that doesn't mean that's the case in CardVerticalJournal
[13:44] <Cimi> Saviq, would be good to see then
[13:44] <Cimi> Saviq, this cardVerticalJournal
[13:45] <Cimi> I cannot enable it apparently
[13:46] <Cimi> Saviq, testing vertical-journal in scope tool works fine here
[13:47] <Cimi> Saviq, maybe it's a particular combination of components for the artits?
[13:49] <Cimi> Saviq, I think is that one for scope tool http://paste.ubuntu.com/8169147/
[13:49] <Cimi> nope
[13:49] <Saviq> Cimi, yeah, as long as it has only one result
[13:50] <Saviq> pstolowski, any idea why Cimi wouldn't get artists in the music aggregator?
[13:50] <tedg> Saviq, Do you have any idea where all these bad AppIDs could be coming from? https://errors.ubuntu.com/bucket/?id=ubuntu-app-launch-invalid-appid
[13:50] <tedg> Saviq, It's kinda reaching epidemic proportions.
[13:51] <Saviq> tedg, what am I looking at in those reports?
[13:51] <Cimi> Saviq, ok reproduced
[13:51] <Cimi> Saviq, with scope tool
[13:51] <Saviq> tedg, like can I see what appid was passed?
[13:51] <tedg> Saviq, Mostly the numbers, the data is stripped (different patch), but UAL is throwing an error when an invalid AppID is given.
[13:52] <pstolowski> Cimi do you have local music files? are they correctly tagged?
[13:52] <Cimi> Saviq, because the bands I am listening too don't deserve to be called artists
[13:52] <tedg> Saviq, It's *there* but Whoopsie strips it right now :-/
[13:52] <Saviq> tedg, a) someone has a stale entry in the launcher
[13:52] <Cimi> pstolowski, sure
[13:52] <Saviq> tedg, b) click scope didn't notice a upgrade / removal
[13:53] <Saviq> tedg, ... and didn't tell the dash to refresh
[13:53] <tedg> Saviq, Hmm, okay. It's just happening 500 times a week on phones right now.
[13:53] <tedg> So it must be in the testing somewhere.
[13:53] <Saviq> tedg, ah yeah, testing could makes sense too
[13:53] <Saviq> tedg, but it would help a *lot* to see the actual appid used
[13:54] <Saviq> tedg, I'll have a look in my /var/crash when I run a unity8 suite next
[13:54] <tedg> Saviq, Yeah, working on that patch. Tradeoffs in DB space, etc.
[13:56] <pstolowski> Cimi, and what exectly do you see when going to My Music -> artists department? empty page?
[14:02] <MacSlow> tsdgeos, do you happen to know why test_popover of PageHeaderLabel fails -> https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-utopic/1158/testReport/%28root%29/qmltestrunner/PageHeaderLabelTest__test_popover
[14:02] <tsdgeos> MacSlow: sdk changes
[14:02] <tsdgeos> MacSlow: https://code.launchpad.net/~aacid/unity8/pageHeaderFailingTest/+merge/232528 should fix it
[14:03] <tsdgeos> but on first run CI i had a dozen unexpected autopilot errors
[14:03] <tsdgeos> i triggered CI again since they seem unrelated
[14:03] <tsdgeos> and may be just a CI hicukup
[14:03] <tsdgeos> +spelling
[14:04] <MacSlow> tsdgeos, guess it just typical CI-routine :)
[14:04] <Cimi> pstolowski, no artist department
[14:05] <MacSlow> tsdgeos, getting stuff through CI makes one feel like Don Quixote :)
[14:05] <tsdgeos> old and crazy? :D
[14:05] <MacSlow> tsdgeos, hehe... not like that :)
[14:07] <MacSlow> tsdgeos, Sisyphus would have been a better choice of analogy I guess
[14:07] <pstolowski> Cimi, no artists department in "My Music" scope? (don't confuse it with "Music" which is the aggrgeating scope)
[14:10] <Cimi> MacSlow, http://www.engadget.com/2014/08/28/leap-motion-s-next-sensor-is-designed-specifically-for-virtual-r/?ncid=rss_truncated
[14:11] <MacSlow> Cimi, very cool idea... the leap-motion is indeed next on my list of geek-toys to get hold of
[14:12] <Saviq> wha!
[14:12] <Saviq> this video is not available in your country ffs
[14:13] <Cimi> Saviq, here too, didn't watch it
[14:14] <Cimi> it's bs
[14:14] <Cimi> tried with a proxy, is not even available in US
[14:14] <tsdgeos> mterry: ping
[14:14] <MacSlow> Cimi, Saviq: tor-browser to the rescue maybe
[14:15] <mterry> tsdgeos, hello!
[14:15] <Saviq> MacSlow, yeah, and then I got police on my back eh? ;)
[14:15] <tsdgeos> mterry: i'm testing the silo with https://code.launchpad.net/~mterry/unity8/interactive-while-locked/+merge/231253
[14:16] <tsdgeos> mterry: is it correct that if locked, i can launch the phone, but then from the phone i don't get the launcher?
[14:16] <mterry> tsdgeos, yes
[14:17] <tsdgeos> that's weird
[14:17] <tsdgeos> but ok
[14:17] <mterry> tsdgeos, yeah...
[14:18] <tsdgeos> mterry: where do i edit that in system settings?
[14:18] <mterry> tsdgeos, the idea was for (a) security reasons, reducing the surface of attack -- because that emergency phone mode is actually in your session, if we screw up and let you actually switch to another app, that's bad.  We have code to prevent that, but just for safety, we hid the launcher and also (b) if you are actually making an emergency call, we didn't want any extra UI to confuse or accidentally interact with
[14:18] <mterry> tsdgeos, I had to rebuild USS this morning after a release -- confirm that you have USS from the silo
[14:19] <mterry> tsdgeos, but you edit it in the security panel
[14:19] <mterry> tsdgeos, under the lock phone setting
[14:19] <tsdgeos>  *** 0.3+14.10.20140828-0ubuntu1 0
[14:19] <tsdgeos>         500 http://ppa.launchpad.net/ci-train-ppa-service/landing-009/ubuntu/ utopic/main armhf Packages
[14:19] <tsdgeos> ah
[14:19] <mterry> tsdgeos, awesome
[14:19] <tsdgeos> see it
[14:20] <tsdgeos> now that you mention it, makes sense :D
[14:21] <dandrader> kgunn_, could you please lift your "needs fixing" on https://code.launchpad.net/~nick-dedekind/unity8/prompts-in-prompts/+merge/232362 . it merges fine now. you could replace it with a "abstain" for instance
[14:22] <tsdgeos> mterry: so what things can i open from the locked launcher?
[14:22] <tsdgeos> phone nly?
[14:22] <mterry> tsdgeos, everything else will open, but you have to unlock first
[14:22] <kgunn_> dandrader: yep, sorry bout that,slipped my mind
[14:22] <mterry> tsdgeos, phone is only one that you can actually interact with without unlocking
[14:22] <tsdgeos> mterry: and then what's the point?
[14:23] <mterry> tsdgeos, slightly faster?
[14:23] <tsdgeos> faster what?
[14:23] <mterry> tsdgeos, I think the indicator switch is more interesting
[14:23] <dandrader> kgunn_, np. I know it's easy to forget follow up on such small things
[14:23] <tsdgeos> mterry: i still have to type my passcode, can't see how it's faster
[14:24] <mterry> tsdgeos, er, not faster, but more natural how about.  like the flow is "click, unlock" rather than "unlock, click"  -- you lead with the action you want
[14:24] <mterry> tsdgeos, but honestly it's not a big deal
[14:24] <tsdgeos> ok
[14:24] <tsdgeos> let's not argue about what is more natural
[14:24] <mterry> tsdgeos, I think the reason you would disable it is if you had sensitive icons on it
[14:24] <tsdgeos> i like my natural blue
[14:24] <tsdgeos> :D
[14:24] <mterry> tsdgeos, it's enabled by default
[14:24] <mterry> tsdgeos, I think it's just a privacy toggle to hide the apps you have
[14:26] <tsdgeos> ok
[14:26] <dandrader> dednick, wow, "make trySessionContainer" is awesome now!
[14:47] <dednick> dandrader: :) what you asked for!
[14:47] <Cimi> Saviq, #Cimi
[14:48] <mzanetti> is there party at #Cimi's?
[14:49] <dednick> dandrader: can you read and let me know what you think: http://paste.ubuntu.com/8169568/
[14:49] <dednick> dandrader: i've come up with a solution, but it needs some more eyes.
[14:53] <dandrader> jesus, that's a long conversation
[14:58] <Cimi> tsdgeos, is there a container of cards that clips?
[14:59] <tsdgeos> Cimi: not sure what you mean
[14:59] <Cimi> tsdgeos, cards in vertical journal seems clipped
[14:59] <Cimi> when you only have 1 card
[14:59] <tsdgeos> cards in vertical journal are clipped at XY gu
[14:59] <tsdgeos> XY = don't remember, 20 i think
[14:59] <tsdgeos> not clipped
[15:00] <tsdgeos> but you have to press see more
[15:00] <tsdgeos> so yes, clipped :D
[15:02] <Cimi> tsdgeos, with 1 only?
[15:03] <Cimi> tsdgeos, would you mind having a look at https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1362160 ?
[15:03] <tsdgeos> Cimi: it does not matter
[15:03] <tsdgeos> height is height
[15:03] <tsdgeos> an be 1000 items or 1
[15:03] <Cimi> tsdgeos, it's true that missing summary is complicating things, but the bottom edge is still clipped
[15:03] <tsdgeos> it clips at 20
[15:04] <Cimi> so here is different
[15:05] <tsdgeos> is that a VJ?
[15:05] <tsdgeos> why would anyone use a VJ there?
[15:06] <tsdgeos> if that's a VJ
[15:06] <tsdgeos> i don't see where is the problem
[15:06] <tsdgeos> besides it looking ugly as hell
[15:06] <tsdgeos> it's cut at 35gu how VJ is supposed to do
[15:07] <tsdgeos> no?
[15:07] <Cimi> well
[15:07] <Cimi> I never seen this
[15:08] <tsdgeos> this = what?
[15:09] <Cimi> tsdgeos, that bug
[15:09] <tsdgeos> i've never seen it either
[15:09] <Cimi> tsdgeos, I don't have artist department
[15:09] <dandrader> dednick, (about that conversation) that's the reason we have MirSurfaceItem::release(). It's so that unity8 can decide when to really destroy a mir surface. So qtmir tells unity8 "hey, that surface/app/session/whatever is no more". then unity8 animates things away. once unity8 animations are done. it finally calls surface.release() so that qtmir can finally destroy the mir surface
[15:10] <dandrader> dednick, when a client dies, its surface is not automatically destroyed
[15:10] <tsdgeos> Saviq: why do you say the implicit height is wrong at https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1362160 ?
[15:10] <dandrader> dednick, as the compositor (unity8 through qtmir) still holds a reference to it
[15:11] <dandrader> dednick, only once the compositor releases its reference to a mir surface is that it gets destroyed
[15:11] <Saviq> @unity, while I have most of you here: http://pastebin.ubuntu.com/8169708/ TAGS!
[15:11] <dednick> dandrader: ya. so my solution is just to delay the release for 100ms, and if it starts removing the session then cancels the release (and gets released with the session).
[15:11] <tsdgeos> Saviq: so you've decided to kill 0.1.16=
[15:11] <mzanetti> phew
[15:11] <tsdgeos> that's news
[15:11] <tsdgeos> i thought you liked it
[15:11] <Saviq> tsdgeos, ;D
[15:11] <mzanetti> :D
[15:11] <dandrader> dednick, more exactly, every application (mir client) has a a ring buffer of surfaces. the compositor always points/holds one of them and the client another
[15:12] <dandrader> dednick, to one of the surfaces in the ring buffer I mean
[15:12] <tsdgeos> Saviq: not killing  7.85+14.10.20140428.2-0ubuntu1 ?
[15:12] <dandrader> dednick, so while the client is rendering onto a surface, the compositor is displaying another one.
[15:12] <Saviq> tsdgeos, it's there
[15:13] <mterry> Saviq, fixed
[15:13] <tsdgeos> Saviq: ah tight
[15:13] <tsdgeos> right
[15:13] <dednick> dandrader: yep. i understand it's not "really" destroyed. we can hold on to it. but under normal circumstances we dont want to (just closing an app) . it's only if we're going to animate the prompt transition that we want to hang on.
[15:14] <dandrader> dednick, we hang on to it when switching between app surface and its screenshot for instance, when a suspended app is killed due to OOM conditions
[15:15] <dednick> dandrader: ya. but not when an app closes.
[15:15] <dednick> dandrader: which is my prpblem
[15:15] <dednick> since the surface gets destroyed first.
[15:15] <dandrader> dednick, what do you mean by "when an app closes"?
[15:15] <dandrader> dednick, swiping it up in the spread?
[15:15] <Saviq> dandrader, no, Qt.quit()
[15:16] <Saviq> dandrader, that's when the job's gone from ual too
[15:16] <dednick> dandrader: in my case it's when the prompt process closes.
[15:16] <dandrader> Saviq, dednick ahhh... an application closing itself
[15:16] <dednick> the surface gets destroyed first
[15:16] <dednick> and then we want to animate the transition
[15:16] <dednick> but the surface is already gonners
[15:17] <dednick> so we need to hang on to it for awhile until we know that the session hasn't been closed so that we can do the transition with the surface still existing
[15:19] <Saviq> dandrader, see, https://launchpad.net/ubuntu-rtm/+source/unity8/8.00+14.10.20140827.2-0ubuntu1 just landed in rtm and I pushed it to lp:unity8/rtm-14.09 for you https://code.launchpad.net/~unity-team/unity8/rtm-14.09 ;)
[15:25] <cwayne> Saviq: \o/
[15:33] <tsdgeos> Saviq: ok, all my extra tags should be gone
[15:33] <Saviq> cwayne, what up? rtm landing? :)
[15:33] <cwayne> Saviq: yep :)
[15:34] <Saviq> cwayne, it has davmor2's stamp on it
[15:34] <Saviq> with barely a complaint!
[15:34] <davmor2> Saviq: no I dare not stamp on it, it'll break
[15:34] <Saviq> davmor2, IT WILL NOT!
[15:35]  * davmor2 hands the pieces to Saviq told you
[15:38] <mzanetti> Saviq: do we already use dbus-testrunner or similar somewhere in unity=
[15:39] <mzanetti> ?
[15:39] <Saviq> mzanetti, grep doesn't work? :D
[15:39] <Saviq> mzanetti, dbus-test-runner
[15:39] <mzanetti> oh... it does work with the correct input...
[15:39] <mzanetti> thanks
[15:48]  * Saviq goes for another round of BT testing then
[15:54] <Saviq> FAIL
[15:55] <mzanetti> lol
[15:55] <mzanetti> Saviq: same issue still?
[15:56] <Saviq> mzanetti, yeah, I don't think we support reverse pairing yet
[15:56] <dandrader> dednick, "so my solution is just to delay the release for 100ms" <- is that in qtmir code?
[15:56] <mzanetti> Saviq: need a bug for that
[15:56] <dednick> dandrader: not yet
[15:56] <mzanetti> Saviq: btw there is no such thing as reverse pairing in the spec :)
[15:56] <Saviq> mzanetti, that's what I'm afraid of
[15:57] <mzanetti> Saviq: well... its an incoming pairing... and has to work just like outgoing pairing
[15:57] <dednick> dandrader: and it's going to have to be about 1s actually. seems to take a bit of time between destroying surface & closing session. going to put in a fadeout while it's deciding.
[15:58] <mzanetti> Saviq: but I'm afraid we don't have a bt service... do we?
[15:58] <mzanetti> oh right... the indicator
[15:58] <dandrader> dednick, not sure I got it. if mir client (a prompt session) just destroys its surface and hang on to the session you do one thing. but if it also closes the session you do another?
[15:59] <Saviq> mzanetti, yeah, we need a snap decision from the bt indicator I think
[15:59] <mzanetti> Saviq: yeah, all pairing needs to be handled by that, even for outgoing
[15:59] <mzanetti> also incoming obex push requests
[16:00] <Saviq> mzanetti, yeah probably, not sure how it works now, but it feels like settings are quite involved
[16:00] <mzanetti> Saviq: and I think without that pairing-agent not even incoming connections (without pairing) would work when system settings is not opened and in foreground :/
[16:00] <mzanetti> as bluez always asks back iirc
[16:01] <Saviq> mzanetti, well it should work for already paired devices
[16:01] <Saviq> mzanetti, but yeah, we need the indicator to handle this
[16:01] <mzanetti> Saviq: it depends on the trust level (by specification)
[16:01] <mzanetti> Saviq: you should be able to mark a device as trusted
[16:02] <Saviq> MacSlow, friendly reminder, strip tags on lp:~macslow/unity8/fix-1348092
[16:02] <dednick> dandrader: erm. trying to work out the details still. basically always hang on to the surface for awhile after it's destroyed. But if we close the session, it hangs onto it until the session is destroyed.
[16:02] <mzanetti> and if that's not the case we'd need to open a snap decision which asks for ok
[16:02] <Saviq> dednick, you got 0.1.16 in prompts-in-prompts too
[16:02] <dednick> Saviq: damnit. i removed it yesterday
[16:02] <MacSlow> Saviq, doh... ok
[16:02] <dednick> Saviq: keeps going back into trunk!
[16:03] <dednick> Saviq: removed.
[16:04] <Saviq> dednick, remember to remove locally too
[16:04] <dednick> Saviq: yup
[16:05] <dandrader> dednick, I still don't fully understand the problem (might only get it when I see the code and try out for real), but as a general rule, IMHO, qtmir should only destroys a surface once unity8 okays it. so that we can ensure that unity8 animates what it has to animate. so there should be no need for those timers/delays....
[16:06] <dandrader> dednick, so maybe the same would apply for sessions.....
[16:06] <dednick> dandrader: if you mir destroys an application surface on close, when is it ok for u8 to destroy it?
[16:08] <dandrader> dednick, qtmir would somehow tell that the surface was closed. unity8 responds to it by animating it away and once it's done it callls surface.release().
[16:10] <dednick> dandrader: right. but when a session closes, it's the session that needs to animate away
[16:11] <dandrader> dednick, so we would need a similar qtmir-unity8 "protocol"  for sessions...
[16:11] <dandrader> or something in that spirit
[16:12] <dednick> dandrader: but when a client closes, the session is destroyed some time after the surfaces (about 200ms or so). so there's nothing to animate, since the surfaces are already gone.
[16:12] <tsdgeos> Saviq: CI passed on https://code.launchpad.net/~aacid/unity8/pageHeaderFailingTest/+merge/232528
[16:12] <tsdgeos> \o/
[16:13] <dednick> dandrader: the qtmir-unity8 protocol for sessions is what my branch does
[16:13] <dandrader> dednick, the surface is only gone is unity8 already finished animating it (called surface.release)
[16:13] <dandrader> is unity8/if unity8
[16:14] <dednick> dandrader: yes. so i end up swiping down an empty session container :)
[16:16] <dandrader> dednick, I don't see DarkenAndFadeInAnimation being used for anything. Is it a leftover? something for the future?
[16:17] <dednick> dandrader: yeah, it's just left over. where supposed to be different types of animations for different surface types
[16:17] <dednick> but nor supported yet.
[16:17] <dednick> dandrader: i'll remove
[16:23] <Saviq> tsdgeos, awesomes
[16:28] <MacSlow> Saviq, cleaned lp:~macslow/unity8/fix-1348092 only 8.0-tags remain
[16:43]  * mzanetti listens to music using his bluetooth headphones and an ubuntu phone and loves it!
[16:43]  * popey phones mzanetti to break that experience
[16:44] <mzanetti> popey: if that works through the headphones too its the "Bye bye Nokia N950"
[16:45] <popey> try it ㋛
[16:45] <popey> known broken for months
[16:45] <mzanetti> I only manage today to pair something for the first time (using the UI only)
[16:46] <mterry> Does anyone have time for reviewing https://code.launchpad.net/~mterry/unity8/unlock-via-dbus/+merge/232428 ?
[16:46] <mterry> It's blocking some adb work
[16:46] <mterry> @unity ^
[16:46] <dandrader> mterry, I don't
[16:50] <tsdgeos> mterry: why remove lockscreen.hide(); ?
[16:51] <mterry> tsdgeos, Oh I had a comment in the merge but it got lost by later commits: "Just a quick comment on this change -- I realized it was an unnecessary call so I took it out.  greeter.login() does the hiding."
[16:51] <mterry> tsdgeos, just unrelated cleanup really
[16:54] <tsdgeos> mterry: don't understand this
[16:54] <tsdgeos> UNLOCK_OUTPUT=$(adb shell "if [ \"\$(id -u)\" = 0 ]; then sudo -u phablet -i $GDBUS_CMD; else $GDBUS_CMD; fi" 2>&1)
[16:54] <tsdgeos> why do i need sudo if i'm root already=
[16:54] <tsdgeos> ?
[16:54] <Saviq> tsdgeos, -u phablet
[16:55] <tsdgeos> doh
[16:55] <mterry> tsdgeos, yeah, that bit was there to future-proof when adb changes which user it is
[16:56] <tsdgeos> looks good to me then
[17:00] <dednick> tsdgeos: mzanetti: SurfaceContainer.qml:47:19: Unable to assign bool to [unknown property type]
[17:00] <dednick> http://paste.ubuntu.com/8170572/
[17:00] <dednick> any idea? it's happening totally randomly inside a qmltest
[17:00] <dednick> every 10 or so runs it happens
[17:12] <dandrader> dednick, is that trunk?
[17:13] <dandrader> ah, the pastebin
[17:13] <dednick> dandrader: yeah
[17:13] <dednick> dandrader: oh. no, my branch
[17:14] <dandrader> dednick, prompt-in-prompt or something else. that looks like something else
[17:16] <dandrader> yeah, I've no idea
[17:16] <dandrader> (that qml warning)
[17:16] <dednick> dandrader: it's prompt in prompt
[17:16] <dednick> the new one that i just pushed
[17:17] <dandrader> oh, time to pull again
[17:19] <dandrader> dednick, was already writing in my review that SurfaceContainer was quite empty now, not warranting the existence of a separate component (thus to be merged with SessionContainer)
[17:19] <dednick> dandrader: heh. indeed. although it is used for opacity in the app window
[17:20] <dandrader> ?
[17:20] <dednick> switching between showing surface and showing splash
[17:20] <dednick> dandrader: or screenshot
[17:20] <dandrader> ah, I keep forgetting that AppWindow extends SessionContainer
[17:24] <dandrader> dednick, I wonder if we could merge SessionContainer with AppWindow. might make it easier understand and maintain the interaction between the logic in those two files
[17:25] <dandrader> just an idea
[17:25] <dednick> dandrader: ? can't really. A SessionContainer has child SessionContainers which don't use things like splashscreens
[17:26] <dandrader> dednick, or make AppWindow have a SessionContainer inside it instead. composition instead of inheritance
[17:26] <dandrader> dednick, right
[17:27] <dednick> dandrader: i thought about using composition, but thought it would introduce another level of items which takes more time to render.
[17:28] <dandrader> dednick,not if that new level doesn't render anything itself but just positions its children
[17:28] <dednick> dandrader: as in Item { Item {} } takes longer to render than Item {}
[17:28] <dednick> greyback_: ^ ? didn't you do some perf monitoring on this?
[17:29] <dednick> when i originally did prompt sessions i had a few levels which was slowing things down.
[17:29] <greyback_> dednick: Items don't actually draw anything, they add a couple of extra nodes to the scenegraph though
[17:30] <greyback_> so it's slightly slower to render the tree, but the difference would be tiny
[17:30] <dandrader> hmm
[17:31] <greyback_> Items can clip though, clipping often has a cost
[17:32] <dandrader> oh yeah, clipping should be avoided as much as possible
[17:32] <dandrader> but are disabled by default
[17:32] <dednick> dandrader: well i dont mind. i can use composition if you really want it.
[17:33] <dandrader> dednick, I'm still reviewing. so just an idea at this point
[17:33] <dandrader> dednick, btw, are you changing that mp much still?
[17:33] <dednick> dandrader: na, i think i'm done now
[17:34] <dednick> greyback_: finished the qtmir one as well. :)
[17:34] <dednick> greyback_: MR is up for grabs :)
[17:34] <dednick> you know you want to.
[17:34] <dednick> only 4k lines!
[17:34] <greyback_> I really don't
[17:35] <dednick> hehe
[17:35] <dednick> greyback_: quite a bit of it is from the other MP
[17:36] <greyback_> dednick: the fun part is figuring out what is and what isn't :)
[17:36] <greyback_> dednick: commit message & checklist when you're ready
[17:36] <dednick> 1k lines of tests. although i just realised they arent valid anymore
[17:37] <dednick> damn. i think i commented them out
[17:37] <dednick> probably not good
[17:42] <greyback_> jesus dude you're a hero
[17:42] <greyback_> dednick: ^
[17:46] <dednick> greyback_: ? what, for disabling tests? :)
[17:46] <greyback_> dednick: heh, no, just the hugeness of this MR
[17:47] <dednick> one of my smaller ones i know
[17:47] <dednick> i think my record still stands at 12k
[17:48] <greyback_> that's your strategy, is it? shock & awe - nobody dares critique it!
[17:48] <greyback_> note - tests are disabled in tests/tests/pro still
[17:52] <cwayne> did the scope:// url stuff land in url-dispatcher?
[17:53] <cwayne> and also, can i use that to open a scope to a specific department?
[17:54] <Cimi> Saviq, you here?
[18:00] <greyback_> dednick: high-level question - does your code support the case where a trust helper appears before the applicaiton does?
[18:00] <greyback_> +surface
[18:01] <dednick> greyback_: um. should do. i might not have tested that eventuality though
[18:01] <greyback_> dednick: ok. Was just curious
[18:01] <dednick> greyback_: oh. before the application does? no
[18:01] <dednick> greyback_: before the app surface, yes.
[18:02] <greyback_> dednick: before app surface is what I wanted to know. Thanks
[18:02] <dednick> greyback_: pl
[18:02] <dednick> ok )
[18:02] <dednick> bah!
[18:02] <greyback_> dednick: can you share tools and instructions on making a nested trust session?
[18:03] <dednick> greyback_: sure
[18:05] <mterry> kgunn_, any problems with landing silo 009?  It has unity8 while you have a version of unity8 in silo 002
[18:07] <dednick> greyback_: added to mp
[18:07] <greyback_> dednick: ta
[18:25] <mhall119> Saviq: can a scope's Art and Icon fields use relative filenames?
[18:26] <mhall119> it doesn't seem to find mine
[18:26] <kgunn_> mterry: go for it....if we have to rebuild we'll rebuild
[18:28] <cwayne> mhall119: they should
[18:29] <Saviq> mhall119, if they're in click, yes
[18:38] <mhall119> cwayne: Saviq: are there any restrictions or requirements for those files?
[18:41] <Saviq> mhall119, they should be square, similar to those https://drive.google.com/drive/#folders/0B8I8ZVKH-8SsT1V6MmVGM01uRE0
[18:41] <Saviq> mhall119, because they will be cropped to the aspect ratio of the dash
[18:43] <Saviq> mhall119, in reality only one of them will be used soon
[18:47] <mhall119> Saviq: cwayne: I think my problem was that the dash scopes manager doesn't update when I install a click package update, but it does on restart
[18:47] <mhall119> is there a way to force it to refresh without restarting  my emulator?
[18:49] <cwayne> none that i know of
[18:54] <mhall119> pkill -f unity8-dash worked :)
[19:02] <mhall119> Saviq: are we going to be able to use an image for the background, rather than just a color?
[19:25] <Saviq> mhall119, no
[19:38] <mhall119> :(
[20:27] <mhall119> Saviq: thanks for all your help today, I was able to make quite a bit of progress
[20:33] <Saviq> mhall119, sure
[20:33] <Saviq> mhall119, on backgrounds for scopes, there's a slew of issues that would arise of that
[20:34] <Saviq> mhall119, different aspect ratios, readability etc
[20:54] <mhall119> Saviq: but we have background images on the default ones and the dash management screen, so it's a problem that'll need solving anyway
[20:55] <mhall119> and readability is technically already a possible issue, it's just up to the scope developer to not be dumb
[20:55] <mhall119> I understand that there are issues though, and until they're worked out I can understand not supporting it
[20:56] <mhall119> still, it would be nice to have
[20:56] <mhall119> both for my own scopes, and being able to change the Apps scope background again :)
[20:56] <Saviq> mhall119, sure, that's ultimately a design decision on how this should work, and it was explicit that they did not want image backgrounds on scopes (the background we see now is "see through")
[20:56]  * mhall119 misses the purple salad
[20:56] <mhall119> "see through"
[20:59] <Saviq> mhall119, that basically means it's static behind the dash, not at the layer of scopes, where their background is