MirvTrevinho: morning! I just filed https://bugs.launchpad.net/unity/+bug/1211174 which is blocking releasing at the moment, can you make any sense of it?06:36
ubot5Launchpad bug 1211174 in Unity "unity.tests.test_home_lens failing" [Critical,New]06:36
Mirvandyrock doesn't seem to be here yet06:36
Mirvnot sure yet if it's something from the unity stack or if something from the platform stack (where the failing test so far was run) could cause it06:39
jameshhi sil210007:02
mzanettigood morning!07:04
mzanettiSaviq: hi07:06
Saviqmzanetti, mornin'07:06
mzanettiSaviq: any updates I should have?07:06
Saviqmzanetti, nothing major, only thing I think is that Rick has set our priorities for August (see ubuntu-engineering@)07:07
Saviqmzanetti, so we'll need to potentially reshuffle the work and help wherever needed to reach that07:07
* mzanetti reads mails07:08
mzanettiSaviq: progress on application management?07:08
Saviqmzanetti, there is07:08
mzanettiare we using mir on the phone yet?07:09
Saviqmzanetti, not by default, no07:09
Saviqmzanetti, we're hoping for the end of the week07:09
mzanettiI've seen dednick workarounded the CPU battery draining issue07:10
mzanettimakes me happy :)07:10
sil2100jamesh: hi! lucene++ is waiting for sponsoring, I think someone should pick it up soon07:12
sil2100jamesh: I'll poke the patch pilot as well07:12
Saviqmzanetti, yeah, unfortunately a temporary and short-lived solution07:12
mzanettiSaviq: who's the man for the network indicator service? isn't that larsu?07:12
mzanettimaybe we should subscribe him to the bug too07:12
Saviqmzanetti, it's not a service issue07:13
Saviqmzanetti, it's a generic glib vs. Qt one07:13
mzanettiwell, it seems to crash07:13
jameshsil2100: okay, thanks07:14
didrockssil2100: I think jamesh has other requests, I asked him to wait for you being back :)07:15
jameshthe lucene++ one is the main one, so I can clear the merge request queue07:15
Saviqmzanetti, ah, that's the "wrong" indicator backend anyway07:16
Saviqmzanetti, it's chewie-client and not indicator-network that's used, still07:16
sil2100jamesh: \o/ what's the other request? I'm back from holidays now so ready for action07:16
sil2100jamesh: btw. did you send me an e-mail with those media-scanner branches?07:17
jameshsil2100: I may need a little more help once I've got everything merged and building in Jenkins07:17
sil2100jamesh: or is it just one branch right now?07:17
mzanettiSaviq: any pointers where to start fixing this one? https://bugs.launchpad.net/ubuntu-ux/+bug/121020007:17
ubot5Launchpad bug 1210200 in Ubuntu UX "[osk] OSK overlaps launcher when on an app and then navgiate to launcher" [High,New]07:17
Saviqmzanetti, with Mir07:18
Saviqmzanetti, shell is always just a single surface currently, so there's no way to composite them in the right  order yet07:18
Saviqmzanetti, although reaching for the launcher should dismiss the keyboard anyway07:19
jameshsil2100: https://code.launchpad.net/hollywood/+activereviews has all the pending work.  The boost-fixes branch is probably the easiest one to do a test build against, if that's what you're after.07:19
mzanettiSaviq: should it?... not sure about that07:19
mzanettiSaviq: I wouldn't say so07:19
jameshsil2100: they all extend the changes in the saucy-fixes branch07:19
Saviqmzanetti, add ubuntu-ux to the bug, then07:19
Saviqmzanetti, I think that's what I heard - and it kinda makes sense07:19
Saviqmzanetti, if you're going for the launcher, most of the time you'll focus another app, so OSK is out of context anyway07:20
dednicklarsu: ping07:45
mzanettiMacSlow: veebers: my X crashed...07:46
mzanettianyways, I think we were done, right?07:46
MacSlowmzanetti, yup07:46
veebersmzanetti: d'oh :-\ You didn't miss much, yeah was jsut the wrapup07:46
larsudednick: morning07:47
dednicklarsu: good morning!07:52
dednicklarsu: i've proposed a couple of branches to qmenu/unitymenumodel on friday. Has to do with some leaking qobjects caused by glib callbacks issueing calls directly to qt.07:54
larsudednick: awesome, I'll have a look in a bit. I fixed the action handling and added the parameter07:54
larsudednick: ooh, but I didn't push it yet. Will do now07:55
dednicklarsu: cool. thanks, i'll take a look at that and update unity8 to handle.07:55
larsudednick: only use model.action.activate in menu items with a "x-canonical-type". For the normal ones, just call model.activate(). That will handle the parameter for you if the menu item has a target set07:56
dednicklarsu: i'd much prefer if we can always using the action.07:58
larsudednick: why?08:01
dednicklarsu: we don't pass the root model around in qml, just the row data. so we don't have access to the model.activate function from outside the list08:01
larsumodel.activate is accessible in every delegate, no?08:02
larsuthere shouldn't be any reason why the menu item components need access to the root model08:03
dednicklarsu: larsu because then they need to know about their own index within the model and such.08:04
dednicklarsu: actually, i may change how they call activate08:05
larsudednick: I don't understand. Are you doing something substantailly different from examples/unitymenumodel.qml?08:05
dednickmight be better to signal value changes back to the delegate and let it call activate.08:05
dednicklarsu: it's a hell of a lot bigger. has to be much more modular.08:05
larsudednick: of course. I mean structurally. The components inside the delegates can always access all roles of its row, can't they?08:06
larsuand they should always have the ListView attached property08:07
dednicklarsu: the delegates can, but the delegates arent the menu items.08:07
dednicklarsu: nevermind. i'm going to change it.08:07
dednicki think it needs to change anyway.08:08
larsuhm, okay :)08:08
dednicklarsu: actually had a q about thte x-canonical-type. there are some menus (eg mute) which don't set the type [to switch] it but have a bool state parameter. Should i be parsing the type in this case to know it's checkable?08:19
dednick*parsing the state type08:20
dednicklarsu: unity7 is just a checkable menu type, but in unity8 i'm guessing we want a switch.08:20
larsudednick: ah, right. I should be exporting that through the role, you shouldn't need to parse the state08:25
dednicklarsu: ok, so it'll always be in x-canonical-type?08:26
larsudednick: thinking about that right now08:26
larsudednick: it's not really in the implementation, because x-canonical-type is something we added on top of gmenumodel08:27
larsuwhere checks and radios are in gmenumodel itself08:27
larsubut this will lead to two type systems for you...08:27
larsukinda weird08:27
dednicklarsu: hm. the radio is an intersting one.08:28
larsudednick: is it easier for you if I consolidate everything into x-canonical-type?08:28
larsudednick: usually, checks/radios are handled by the same widget as normal menu items08:28
larsui.e., we just draw a check or a radio thing in the margin if gtkmenutracker tells us to08:28
dednicklarsu: i so have a solution in place for the parsing type.08:29
larsuparsing type?08:29
dednicklarsu: just checking if it's a boolean. if it is, pop a switch in the menu08:30
larsudednick: don't do that, switches and checks are separate things08:30
dednicklarsu: er. how so.08:31
larsudednick: I'm sure mpt will gladly answer that one ;)08:31
dednicklarsu: i'm pretty sure there are no checks in unity808:32
larsuwhat do you mean? In the toolkit?08:32
dednickin the design08:33
larsudednick: https://wiki.ubuntu.com/Sound#Phone08:33
dednicklarsu: ahh. hm. ok08:34
dednicklarsu: how do i know it's a check then ?08:34
larsudednick: not at all right now, I'm trying to make up my mind about a good solution08:35
larsudednick: we have two options:08:35
larsu(1) consolidate with x-canonical-type08:35
larsu(2) make a separate thing, as gtkmenutracker does it08:35
dednicklarsu: role in unitymenumodel08:35
larsuif you have separate widgets for check and radio menu items, (1) will be easier08:36
larsuotherwise, (2) will be easier (because all you need is a "Check { visible: model.is_check }" or some such)08:36
dednicklarsu: yes, but i'm guessing we wont be able to render a radio in that case. there's no "group" role is there?08:37
dednickdifferent'y from a check i mean08:37
larsuit would be is_check and is_radio08:37
larsuor a string property that is set to "check" or "radio"08:37
larsuwhatever you prefer :)08:38
dednicklarsu: sep roles probably best.08:38
dednicknow i need to check if we actually do have a checkbox in sdk...08:39
larsudednick: okay, so (2) with "is_check" and "is_radio". Coming up.08:39
dednicklarsu: ta.08:39
larsuoh, isCheck and isRadio, of course08:40
* larsu always forgets about the camels08:40
dednicklarsu: i removed model.activate when i changed to action.activate. need to put it back?08:43
larsudednick: I already did08:44
dednicklarsu: ah. i was just looking at r10008:44
larsudednick: done08:55
dednicklarsu: that was quick. thanks.08:56
CimiSaviq, ping :)08:58
CimiSaviq, can you throw me all your ideas about why a signal might not be received externally by what is using a component?08:59
tsdgeosit's not well connected? the signal is not emitted?09:03
SaviqCimi, can we see some code?09:05
CimiSaviq, http://paste.ubuntu.com/5976602/09:05
Cimibasically I added a swapped signal09:06
Cimiand in my test case is not sent through :-\09:06
Ciminot always I mean...09:06
Cimilet me push the code09:06
CimiSaviq, lp:~unity-team/unity/background-ugly09:07
SaviqCimi, here's my output when setting my background to an empty string and back to the right path http://pastebin.ubuntu.com/5976613/09:09
CimiSaviq, you see it's missing one09:09
SaviqCimi, it's missing the "swapped" print, that it?09:10
CimiSaviq, yeah09:10
tsdgeosthe crossfadeimage thing (i.e. receiver) is nto there?09:11
Cimitsdgeos, I have one onSwapped handler in Shell.qml09:13
Cimitsdgeos, it's not receiving one09:13
tsdgeosah sorry, got confused by the diff then09:13
Cimiwhich is problem number 109:13
Cimithen I have problem number 2 which is understanding why the swapImage() function is not called when crossfading :)09:14
CimiI think it might be because of the animation09:14
Cimibut that's a different thing ;)09:15
SaviqCimi, http://paste.ubuntu.com/5976628/ seems to print "received swapped" every time09:16
CimiSaviq, what did you change?09:17
SaviqCimi, it's because you have wo handlers09:17
CimiSaviq, even with one didn't change...09:17
SaviqCimi, now I added onSwapped: in CrossFadeImage09:18
SaviqCimi, and indeed I'm not getting the "outside" one09:18
Saviqtsdgeos, when you have Foo.qml: Item { onSomething: }; and then Foo { onSomething: }, would you expect both to be called or just the overridden one?09:19
tsdgeosSaviq: i think both are called09:19
tsdgeosbut can't be sure tbh09:19
Saviqtsdgeos, yeah, in a simple example both are09:22
SaviqCimi, I think your problem is that you're emitting the signal twice in a very small time window09:24
SaviqCimi, I "reliably" see that the internal onSwapped is called twice09:24
CimiSaviq, that's why I was trying to put wait(2000)09:24
SaviqCimi, when the external onSwapped is only used once09:24
SaviqCimi, it does feel like a bug, but I can't find an easy way to reproduce09:26
CimiSaviq, it does feel like a bug to me too09:26
CimiSaviq, spent hours on friday09:26
SaviqCimi, either way, do you need the "internal" onSwapped?09:26
CimiSaviq, I put it to test if it was printing correctly09:27
SaviqCimi, and why did you go against the Image { } that was just testing whether the image is ok and to check its aspect ratio09:27
CimiSaviq, mzanetti removed it09:28
CimiSaviq, because this simplifies...09:28
SaviqCimi, yeah, but that means that you would see a flicker in the crossfade, no?09:29
SaviqCimi, or will the crossfade not happen if the "new" image is wrong?09:29
CimiSaviq, I wanted to know when crossfadeimage correctly loaded and shown a new image09:29
Cimiif two images are correct, onStatusChanged doesn't change (always "Image.Ready" I think)09:30
CimiSaviq, onSourceChanged is not enough to use, because it's too early to get the right status09:30
Cimiso was thinking of emitting a signal when the image is correctly loaded, that's why the "Swapped"09:30
Cimiwhich I could rename to loaded or so..09:31
Cimibut if it's not reliable I'm doomed09:31
SaviqCimi, sure, I understand, I just want you to make sure that when you set an invalid background, it goes directly to the default one and not to black and then to the default one09:32
CimiSaviq, ahh I see09:33
CimiSaviq, this is a good reason to keep the image loader09:33
CimiSaviq, crossfadeImage should handle that though09:33
CimiSaviq, how about adding an image loader in crossfadeimage?09:40
CimiSaviq, loader/tester09:40
SaviqCimi, it should already "have" one, i.e. the "next" image - you start loading it and report status updates from that image09:41
SaviqCimi, so when Image.Error is spat out of it, you know the source that's set is wrong09:42
CimiSaviq, yes but I want to know the last image that was swapped correctly...09:44
SaviqCimi, why?09:44
CimiSaviq, so I can store it09:44
CimiSaviq, if it fails, it reverts09:44
SaviqCimi, you don't want to store the "last correct value", though, do you?09:45
SaviqCimi, we want to fall back to the default09:45
CimiSaviq, you asked to fallback to the last sane image09:45
SaviqCimi, not for the default?09:45
Cimilike I have default, I set red, then blue, then wrong09:45
CimiI want to go back to blue09:46
SaviqCimi, that could be handled in CFI internally09:46
SaviqCimi, but anyway I don't think we should just "ignore" the change09:46
SaviqCimi, but instead go to the default09:46
SaviqCimi, if I said otherwise, sorry, I'm an a$$09:46
Cimino problem09:47
CimiI discovered the craziness of qml and its signals :O09:47
larsudednick: is the qevents branch a temporary workaround for the qt bug?09:52
dednicklarsu: no, dont think qt will do anything about it.10:03
dednicklarsu: that's the impression i got from them anyway10:03
larsudednick: bummer. I assume we might hit this in other projects as well.10:04
dednicklarsu: indeed.10:04
larsudednick: anyway, I'll merge it, we need to get this rolling ;)10:04
SaviqCimi, yeah, I think that's the correct thing to do10:10
CimiSaviq, you mean I should revert to default without overriding the gsettings key?10:10
SaviqCimi, yes10:10
SaviqCimi, don't touch gsettings, but display the default10:10
CimiSaviq, so gsettings will keep the broken image?10:11
SaviqCimi, yes10:11
SaviqCimi, that's the correct thing to do10:11
SaviqCimi, we shouldn't be overwriting someone's settings just because our image failed to load10:11
SaviqCimi, we don't know the reason why it failed to load10:12
SaviqCimi, e.g. it might be unavailable right now, but on reboot will be there, for example10:12
tsdgeosmzanetti: Saviq: so what we do with the ofono branch?10:14
Saviqtsdgeos, let me ping jounih for feedback on it10:14
tsdgeosSaviq: btw weird thing about the qdbusxml2cpp, we (as with my kde hat) run it on compile time all the time afaik10:15
tsdgeosbut not going to block on that10:15
tsdgeosmaybe we need to change the copyright notice though? or?10:16
Saviqtsdgeos, yeah, was surprised to read that, but it does actually make sense10:16
Saviqtsdgeos, same for qmltypes - we should only generate them on change, not build-time10:16
Saviq(or do it build-time and compare to see if there's a difference and block on that, actually)10:17
Saviqdifference between the checked-in one and the generated one, that is10:17
mzanettiSaviq: tsdgeos: for the ofono branch we have 2 main questions:10:22
mzanettia) should it be part of unity8 or not (I'd say yes)10:23
mzanettib) some open design figures where I would not block on as it will be refined once designers actually get it on their phones.10:23
tsdgeosmzanetti: a) looks good to me, but i agree with some of the comments in which i should be able to use my phone even without entering the sim and then i should have some kind of indicator or way to enter it once "running"10:26
tsdgeosthat's how most of the phones i've had work10:26
mzanettitsdgeos: yeah... you can cancel the PIN entry already now10:27
mzanettitsdgeos: the only thing we need to figure is how to enter it at some later point in case it was cancelled. But we have a meeting scheduled for tomorrow for that10:27
tsdgeosah good stuff10:28
larsudednick: there's a FIXME before you send the datachangeevent, but it seems to work fine. Is that just a left over?10:28
dednicklarsu: yeah. i think i was just tagging code that needed the events put in10:29
dednicklarsu: i'll remove quick.10:29
larsudednick: don't worry about it, I'm removing it with the merge10:30
dednicklarsu: ok. thanks10:30
larsudednick: your branches are merged. Let me just add a "isToggled" role so that you can find ääout whether to show a check by doing "visible: isChecked && isToggled"10:31
larsuoh, "isCheck" of course10:31
dednicklarsu: toggle will be stored in state parameter no?10:32
larsudednick: only for checks. For radios, you'd need to compare to "target", which "isToggled" does for you10:33
tsdgeosSaviq: greyback: so how do we fix https://launchpadlibrarian.net/147378097/buildlog_ubuntu-saucy-amd64.platform-api_1%3A0.18.3%2B13.10.20130807-0%2B201308121010~113%2B90~saucy1_FAILEDTOBUILD.txt.gz ?10:33
dednicklarsu: ah. i see. ok10:33
Saviqgreyback, any idea why https://code.launchpad.net/~phablet-team/+archive/mir/+sourcepub/3419610/+listing-archive-extra failed, then? and if it's the correct branch anyway?10:33
Saviqgreyback, it's built from lp:platform-api via https://code.launchpad.net/~phablet-team/+recipe/platform-api-mir-daily10:33
larsudednick: done. It's ready to merge from my side.10:34
greybacktsdgeos: I'll fix the FTBFS, I understand the Mir change that caused it10:34
greybackSaviq: ^^10:34
tsdgeosawesome :-)10:34
dednicklarsu: cool. thanks for that. i'll check it all out asap10:34
Saviqgreyback, cool10:35
larsudednick: okay. I'll be gone for the next hour or so. Lunchtime :)10:35
Saviqgreyback, let me know, I'll handle the recipes and the rebuilds10:35
greybackSaviq: ack10:35
tsdgeosgreyback: also was talking about getting a nice autolander and CI for unity-mir, what do you think? is it too soon or can we do that already?10:36
greybacktsdgeos: I was waiting for sil2100 's go-ahead, he had to block on that for some reason.10:39
greybacksil2100: can we get auto-landing on unity-mir, pretty please?10:40
tsdgeoshe ran away :D10:44
tsdgeossil2100: did you get the "unity-mir autolanding" msg?10:57
greyback_sil2100: it was "can we get auto-landing on unity-mir, pretty please?"10:57
mzanettiSaviq: there are rumors that we want to move away from the ./build script. Is that true?10:58
sil2100greyback_: !10:59
sil2100That's music to my ears10:59
sil2100I missed it, but now I hear it10:59
sil2100greyback_: does it build with everything that's in saucy now?11:00
greyback_sil2100: ah yes, that was why we've to wait. Unfortunately not yet, ricmm has 2 MRs up to fix that11:01
Saviqmzanetti, we already can11:01
Saviqmzanetti, a simple cmake; make11:01
Saviqmzanetti, is all we need11:01
Saviqmzanetti, ./build is just a helper11:01
mzanettiSaviq: I know... I used that a lot11:01
mzanettiSaviq: still it does some other things too, doesn't it?11:02
Saviqmzanetti, it installs build and runtime deps11:02
Saviqmzanetti, but that can be done manually "the usual way"11:02
Saviqmzanetti, that's ut11:02
mzanettiyeah... just wondering as Wellark told me it would be obsolete and will be removed "soon"11:02
sil2100greyback_: since auto-landing requires that we are able to build it without any additional PPAs - if that's possible right now, I can add it tot the config ASAP11:02
Saviqmzanetti, that's probably an incorrect statement11:02
Saviqmzanetti, I don't see a reason to remove it, really11:03
Saviqmzanetti, other than maintainership11:03
greyback_sil2100: not possible yet. When it is, you'll be first to know11:03
sil2100greyback_: awesome, waiting for the signal then ;)11:05
SaviqWellark, I don't think we'll be removing the build script any time soon, it's just a helper script11:05
SaviqWellark, it's not needed any more, but speeds up things11:05
Saviqwe might remove the mention of it from CODING11:05
mzanettiSaviq: is there a solution (e.g. from unity7) how to generate the initial list of apps in the launcher or do we need to figure something ourselves=11:05
Saviqmzanetti, gsettings11:06
Saviqmzanetti, com.canonical.Unity.Launcher favorites11:06
Saviqmzanetti, we should have an .override for it on the phone, but that's about it11:06
mzanettiSaviq: not talking about where to store them. but rather where to get a default list in case the setting is empty11:06
Saviqmzanetti, from gsettings' default11:07
Saviqmzanetti, overridden by the .override file that I don't know what we'll ship it with yet11:07
Saviqseb128, larsu, any updates on the "how do we handle settings different between form factors" topic?11:07
Saviqi.e. default launcher favourites, default background and such?11:08
Saviqand whether changing a setting "propagates" to different ffactors?11:09
seb128Saviq, I've not seen any recent discussion about that, I guess it would make more sense to have the feature build into the stack (e.g having db profiles) rather than having different keys (eg Launcher.<form>) and have the code bind to the right one11:09
mzanettiSaviq: hmm... are QSettings not to be used?11:10
Saviqmzanetti, aren't QSettings being deprecated?11:11
mzanettiSaviq: that would be news to me11:11
tsdgeosSaviq: they were, but they didn't write any replacement, so they undeprecated it :D11:11
mzanettioh boy... and I know the reason for all this :D11:12
Saviqmzanetti, and anyway - we use gsettings in Ubuntu extensively, so gsettings-qt is the first place to look at11:12
Saviqmzanetti, like we have for the background, for example11:12
mzanettiSaviq: not saying I would know better, but does it really make sense to switch away from everything G* while still keeping the settings?11:13
Saviqmzanetti, about that, we had to revert the background stuff - it broke autopilot, 'cause the default background wasn't there (no idea how that got through the upstream merger)11:13
Saviqmzanetti, we're not "switching away from everything G*"11:13
Saviqmzanetti, not where we're using it extensively on the desktop (like for gsettings)11:14
Saviqmzanetti, and where there isn't a real alternative11:14
mzanettihmm... I don't know enough about GSettings to judge that11:14
mzanettijust feels weird to completely write Qt apps but then use gsettings instead of QSettings. Thinking about portability of apps for example11:15
mzanettinot a real issue with unity, I agree11:15
WellarkSaviq: ok, I will make sure unity8 launcher uses gsettings. I will check out gsettings-qt11:15
WellarkSaviq: is it in main/universe?11:16
mzanettiWellark: we already use it in unity iirc11:16
dednickanyone know if it's possible to block signals in qml?11:16
mzanettidednick: define "block" signals11:16
greyback_tsdgeos: Saviq: https://code.launchpad.net/~gerboland/unity-mir/fix-FTBFS-mir-951/+merge/17966511:16
dednickmzanetti: change in property value causing signal. equiv of QObject::blockSignals11:17
WellarkSaviq: do you have any other technical requirements for the launcher backend?11:17
Wellark14:06 < Saviq> mzanetti, com.canonical.Unity.Launcher favorites11:17
WellarkSaviq: so, reuse the schema from unity7?11:18
SaviqWellark, yes11:18
Wellarkand paths11:18
SaviqWellark, wherever it makes sense, we should use the same11:18
WellarkSaviq: do you have a spec for me or is it the old "dig the source, luke"11:18
SaviqWellark, I have nothing, can't say there is nothing, though11:18
mzanettidednick: what's the issue with QObject::blockSignals11:18
dednickmzanetti: ? i'm in qml...11:19
tsdgeosgreyback_: shall i "auto-merge" ?11:19
Saviqdidrocks, are the unity gsettings schemas documented somewhere?11:19
mzanettidednick: ah... qml. lemme check11:19
greyback_tsdgeos: please11:19
SaviqWellark, they are effectively documented in the schema itself11:20
Saviqtsdgeos, greyback_I can trigger platform-api build?11:20
SaviqWellark, http://bazaar.launchpad.net/~unity-team/unity/trunk/view/head:/com.canonical.Unity.gschema.xml11:21
mzanettidednick: can't find anything integrated in QQuickItem. However, I guess you can move the onSignal: methods into a Connections {} item and set that to enabled/disabled11:21
=== MacSlow is now known as MacSlow|lunch
didrocksSaviq: I don't think they are TBH :/11:22
dednickmzanetti: ah. ok, thanks.11:22
Saviqdidrocks, the .xml is probably enough, really11:22
WellarkSaviq: ok, thanks!11:22
Saviqgreyback_, hmm, the FTBFS was on lp:platform-api, though11:22
mzanettidednick: if there isn't a enabled property there, you can set/unset the target property11:22
WellarkSaviq: seems the schema is installed by libunity-core-6.0-711:22
greyback_Saviq: odd, I just checked here and it was fine11:22
SaviqWellark, yes11:22
Saviqgreyback_, https://code.launchpad.net/~phablet-team/+archive/mir/+sourcepub/3419610/+listing-archive-extra11:23
Wellarkso, either, we split out the schemas to independent package, or unity8 will depend on unity-core or we do the middle ground: on systems where libunity-core is installed we use that and on systems where there is no unity7 we provide additional package to install the schema11:24
SaviqWellark, it already depends on unity-core11:25
WellarkSaviq: or will unity8 have it's own unity-core11:25
WellarkSaviq: unity-core does not bring in any unity7 binaries?11:25
SaviqWellark, no, it's just a utility library for "unity shells"11:25
SaviqWellark, whether it's unity7, unity-2d or unity811:26
SaviqWellark, but there are plans to split the schemas out anyway, I believe11:26
WellarkSaviq: ok, sweet. I was worried if we run into trouble with pure unity8 systems or unity7/unity8 hybrids11:27
Wellarkbut it's taken care off11:27
seb128mzanetti, we use gsettings for system settings, not app settings11:36
seb128mzanetti, and we are talking about writting a gsettings backend for qgsettings, maybe11:36
mzanettiseb128: ah ok11:36
seb128mzanetti, gsettings has some advantage, those keys are shared being component, not a db specific to an app11:36
seb128mzanetti, they also need to have system/vendor override and lockdown support11:37
seb128mzanetti, gsettings gives us all that today11:37
Wellarkseb128: sweet :)11:37
mzanettiah, I see. thanks for clarifying11:37
Wellarkseb128, Saviq: how complete is gsettings-qt?11:37
Wellarkdoes it give access to everything we need?11:37
SaviqWellark, complete enough, with Qt and QML bindings11:37
seb128Wellark, define "complete", it's working and being used in system settings and phone-app11:38
seb128Wellark, it has c++ and qml apis11:38
Wellarkcomplete as API parity with the GObject API11:38
seb128not yet11:38
seb128but you don't need most of things11:38
SaviqWellark, yeah, so in launcher-backend you'd probably use the c++ api11:39
WellarkSaviq: yeah, AFAIK all I need is a way to write a list of values and read them back11:40
SaviqWellark, yup, that's what it'll give you11:40
Wellarkand get notification if somebody else messes with the list (webapps adding launcher entries, etc)11:40
Wellarkgsettings supports multiple writers, right?11:41
seb128you have read/write/notification on changes11:41
larsuSaviq, mzanetti: last I heard was that "someone" wanted to write a gsettings backend for qsettings11:55
larsufor apps11:55
larsuunity and other desktop components need more flexibility than qsettings, so they use gsettings directly (through a qt wrapper)11:56
=== tvoss is now known as tvoss|quick_lunc
=== alan_g is now known as alan_g|lunch
=== _salem is now known as salem_
WellarkSaviq: what package is providing the gicon qml loader?12:20
SaviqWellark, ui toolkit plugin12:20
larsuWellark: but but but don't use it.12:22
Wellarklarsu: what should be used instead?12:30
=== MacSlow|lunch is now known as MacSlow
larsuWellark: uris. For themed icons, image://theme/<icon-name> (if someone approved my merge, that is)12:31
larsuWellark: what do you need gicon for?12:31
Wellarklarsu: well, I've seen it being used to get themed icons, but what I need it for is loading icons for launcher items based on the .desktop file Icon key12:32
Wellarkor, more precisely giving mzanetti a icon URL he can load with qml12:33
Wellarkwhat ever providers we have12:33
larsuWellark: ya, use image://theme for that. And bother someone to approve https://code.launchpad.net/~larsu/ubuntu-ui-toolkit/add-unity-theme-icon-provider/+merge/17901112:33
Wellarkimage://gicon/<icon name> seems to do the trick just fine12:33
Wellarkmzanetti: ^12:34
larsuWellark: yes it does, but don't use it!12:34
larsuWellark: consider it deprecated.12:34
Wellarklarsu: that loads also from /usr/share/pixmaps ?12:34
Wellarkfor some reason gicon loader doesn't do that12:34
larsuWellark: why should it?12:34
Wellarkwell, I was under the impression that pixmaps belongs to the theme spec and gicon would support it :)12:35
Wellarklarsu: but as long as your loader supports it I don't care12:35
larsuit doesn't12:35
larsuit supports themes as per the themeing spec12:35
larsudesktop entries may be themed icons or absolute paths12:36
larsuthat means for you: find out if it's an absolute path. If it is, generate a file:/ uri, else use image://theme/12:36
larsuor are you using GDesktopAppInfo?12:36
mzanettisounds sane to me ^^12:36
Wellarklarsu: http://standards.freedesktop.org/icon-theme-spec/icon-theme-spec-latest.html#directory_layout12:36
Wellark"and in /usr/share/pixmaps (in that order)"12:37
Wellarkfirefox and thunderbird install their icons under pixmaps12:37
larsuWellark: if it's in the spec, then my loader supports it (unless qt has a bug)12:37
larsuWellark: interesting, I thought everybody just used hicolor12:38
larsuoh well, TIL.12:38
Wellarklarsu: what about click? I just make sure click icons have absolute paths?12:38
larsuthanks for pointing me to that, Wellark12:38
mzanettiWellark: yeah, absolute for click sounds fine12:39
larsuWellark: either that, or install the icon in a directory that the theming spec mentions12:39
mzanettiWellark: probably we don't even need the clickimageprovider12:39
larsumzanetti: ya, that would be wrong.12:39
greyback_Wellark: did you get time to finish the .desktop file parser?12:39
* larsu acts as if he didn't just read what greyback_ wrote and goes back to hacking12:40
greyback_Wellark: or what would you recommend as best tool to use to read .desktop files then?12:40
mzanettigreyback_: https://code.launchpad.net/~kaijanmaki/unity8/launcher-backend/+merge/17966312:41
mzanettigreyback_: we probably need to move it to a library12:41
greyback_mzanetti: +112:41
mzanettigreyback_: feel free to review the parser part and add that ass a comment12:42
greyback_mzanetti: sure12:42
Wellarklarsu: I'm generating some pressure on #sdk to get your MR approved.12:43
larsuWellark: thanks!12:43
CimiSaviq, I find it hard not having a loop here: if the status is Error I have to change the source, but this could create a binding loop12:46
Wellarklarsu: please, join #sdk12:46
SaviqCimi, don't use a binding, then, use signal handlers12:46
CimiSaviq, I am :-\12:46
CimiSaviq, still complains12:46
SaviqCimi, complains how? if there's no binding, how can it complain that there's a binding loop?12:47
CimiSaviq, onStatusChanged: if status == error -> source = defaultbg12:47
CimiSaviq, if defaulting is error, it loops maybe12:48
SaviqCimi, well, you do protect against that in the if()12:48
SaviqCimi, what does it complain about?12:49
CimiSaviq, indeed, but still complains12:49
CimiSaviq, QWARN  : qmltestrunner::Shell::test_wallpaper_wrong() file:///home/cimi/Development/unity/unity8.fix-wallpaper/Shell.qml:150:9: QML CrossFadeImage: Binding loop detected for property "status"12:49
CimiSaviq, unless I have a different handler in crossfadeimage12:50
Cimiand they collide12:50
SaviqCimi, it's difficult to see, push to the branch, please12:50
CimiSaviq, pushed in background ugly12:51
SaviqCimi, it doesn't build/run12:55
CimiSaviq, error? build here12:55
SaviqCimi, tests/mocks/GSettings isn't there12:59
SaviqCimi, clean build won't work12:59
CimiSaviq, ouch13:04
CimiSaviq, pushed13:05
SaviqCimi, file:///home/michal/dev/canonical/unity8/repo/Shell.qml:123:9: Cannot assign to non-existent property "onPictureUriChanged"13:08
Saviq             onPictureUriChanged: backgroundImage.source = pictureUri13:08
SaviqCimi, there is no "onFooBarChanged" on GSettings13:08
Saviqlarsu, right ↑? could there be?13:08
larsuSaviq: no there isn't, because these properties are loaded at runtime13:09
larsuSaviq: there's a onChanged(keyname) on a GSettings object13:09
Cimiso I need an extra property for that13:09
SaviqCimi, ↑13:09
tsdgeosSaviq: there?13:13
Saviqtsdgeos, yup13:13
tsdgeosSaviq: can you add to https://code.launchpad.net/~unity-team/unity8/app-preview-data/+merge/179348 the steps to get the click scope?13:13
Saviqtsdgeos, yeah, doing13:13
tsdgeosi can't find them (you told them on friday) and i guess they'll be useful to others if they want to give it a look13:13
dednicklarsu: ping13:18
fgintherdidrocks, hello! Is there a way to tell what bzr revision was built by a daily release build?13:19
dednicklarsu: should remove theme icon provider from qmenumodel?13:19
dednicklarsu: since you've proposed it to ubuntu-ui-toolkit13:20
=== alan_g|lunch is now known as alan_g
didrocksfginther: it's in debian/changelog13:20
didrocksyou will see it13:20
didrocksas well as when it's merging back13:20
Saviqtsdgeos, added to description13:20
didrocksit's part of the commit message13:20
fgintherdidrocks, thx13:20
CimiSaviq, I skip the hangout and keep focusing here...13:29
CimiSaviq, I don't want to lose the concentraiton13:29
SaviqCimi, k13:29
Saviqnic-doffay, standup13:31
Saviqmterry, can you hear us?13:33
Cimiis the background property needed or not?13:34
mterrySaviq, no, hmm13:34
mterrySaviq, can now13:35
dednicklarsu: also, do we need to bump qmenumodel version? or is that done automatically?13:39
larsudednick: I wanted to remove it once it lands in ui-toolkit, but it looks like that won't happen soon, because it misses testing13:43
larsudednick: maybe Wellark will write them...13:43
tsdgeosSaviq: on the app ppreview thing13:44
tsdgeosit's sloowwwwwwwwwwwwwwww13:44
Saviqtsdgeos, meaning?13:44
tsdgeosi.e. i click on the "Fake Calculator", wait around 2 seconds, and then the preview opens13:44
Saviqtsdgeos, yeah, we need a spinner13:44
larsudednick: good point about the version. I'll bump it.13:44
tsdgeosis it waiting to fetch all the data for the interwebs?13:44
Saviqtsdgeos, and fix the scope, too13:44
tsdgeosyeah a spinner would help13:44
Saviqtsdgeos, yeah13:44
larsudednick: ugh, how do I even do that? Can't find the proper line in CMakeLists.txt13:45
dednicklarsu: no idea.13:49
=== dednick is now known as dednick|lunch
seb128larsu, what version?13:54
larsuseb128: qmenumodel13:55
larsuseb128: we want to add unitymenumodel, and unity8 should depend on the version that has it13:55
seb128larsu, do you want to change the abi version then?13:55
larsuseb128: no,  I don't think that's necessary13:57
seb128larsu, ok, so just edit unity8's control to version the depends?13:57
larsuseb128: right, but I can't figure out where cmake sets the dep in qmenumodel13:58
larsuseb128: s/dep/version13:58
seb128larsu, just version in the packaging?13:58
seb128larsu, I'm not sure qml has proper versionning/check of versions13:59
larsuseb128: hm. Where in the packaging is the version set?13:59
seb128larsu, out of bumping the abi/import number13:59
seb128$ grep qmenumodel unity8/debian/control14:00
seb128Depends: qmenumodel-qml,14:00
seb128 14:00
seb128when unity8 starts usuing unitymenumodel change that line in14:00
seb128Depends: qmenumodel-qml (>> 20130812)14:01
seb128or whatever is the version that got the feature landed14:01
larsuseb128: ah! okay. So no changes in qmenumodel needed14:01
larsudednick|lunch: ^^14:01
larsuseb128: thank you!14:01
=== salem_ is now known as _salem
=== _salem is now known as salem_
CimiSaviq, I'm doomed, still the binding loop even with the if14:19
tsdgeos#1  0x49c06090 in OPENSSL_cpuid_setup () from /lib/arm-linux-gnueabihf/libcrypto.so.1.0.014:19
Saviqtsdgeos, yikes14:19
Saviqmzanetti, can you please help Cimi?14:20
mzanettiCimi: what up?14:20
Cimimzanetti, still the problem with the background fallback14:20
tsdgeoswhat's wrong with my libcrypto? :(14:20
Cimimzanetti, had multiple issues...14:20
mzanettiCimi: ah ok14:20
mzanettiCimi: what's the branch again?14:22
greyback_tsdgeos: the NSA are after you, run quick!14:22
* tsdgeos throws the phone out of the window14:22
Cimimzanetti, I'm going to have lunch, s14:22
Cimimzanetti, yes, now...14:22
=== alan_g is now known as alan_g|tea
tsdgeosSaviq: mzanetti: greyback_: can you guys do ./run_on_device -g on a Nexus 4 (if you have one)?14:28
tsdgeosi'm getting that ssl crash14:28
tsdgeosworks without the .g14:28
* mzanetti tries14:28
mzanettiCimi: when you're back, let me know what the issues are...14:31
=== alan_g|tea is now known as alan_g
Saviqtsdgeos, greyback_ yeah we have an image14:42
greyback_Saviq: now to see if it works..14:42
* tsdgeos tries14:43
=== daek_ is now known as daekdroom
dednick|lunchlarsu: approved.14:55
=== dednick|lunch is now known as dednick
larsudednick: thanks!14:56
sil2100Trevinho, andyrock: ping15:06
Cimimzanetti, back15:09
Cimimzanetti, binding loop first of all15:11
andyrocksil2100, pong15:12
andyrocksil2100, welcome back! :D15:12
sil2100andyrock: thanks! ;) This week let's take care of the new compiz for saucy - but besides that, do you think you could take a look at an unity test failing?15:12
sil2100AP one15:12
dednickfginther: ping15:14
andyrocksil2100, sure15:14
andyrocksil2100, have you got a link?15:15
andyrocksil2100, with all the failures?15:15
sil2100andyrock: it's just one failure, let me fetch it15:15
* andyrock missed the "an"15:15
sil2100andyrock: for instance here
fgintherdednick, otp, I'll be available in a few minutes15:17
andyrocksil2100, fails here too15:22
sil2100andyrock: can you fix it? ;) It's blocking platform even!15:24
andyrocksil2100, i can try... maybe something changed in bamf and wait_utils_application_is_runnig does not work15:25
sil2100Could be!15:25
larsudidrocks: any idea what this error is? https://jenkins.qa.ubuntu.com/job/qmenumodel-saucy-armhf-autolanding/2/console15:25
* larsu read daily build and thought of didrocks :)15:25
didrocksas everyone :p15:26
didrocksbut I guess this just updated at the wrong time15:26
didrocksbetween an index refresh15:26
didrocksso no need to bother the QA team, just retry and it should be fine :)15:26
sil2100andyrock: thanks!15:26
dednicklarsu, didrocks: i re-approved already15:26
larsudidrocks: thanks!15:27
didrockslarsu: yw ;)15:27
Cimimzanetti, ?15:35
mzanettiCimi: in a hangout atm15:37
mzanettiCimi: but I pushed one fix already15:37
* Cimi looks15:37
tsdgeosSaviq: greyback_: bad news, still black screen :'(15:39
greyback_tsdgeos: damn. You tested Mir alone, and same problem?15:39
tsdgeosgreyback_: yeah, mir_demo_server + mir_client_accelerated15:40
greyback_tsdgeos: something more fundamental broken so. No messages in stdout/stderr either?15:40
tsdgeosnothing :-/15:40
Cimimzanetti, we still have the loop though :|15:51
mzanettiCimi: I'll check it out after the hangout...15:52
dednickSaviq: ping15:56
* dednick has a present for you15:56
=== andy__ is now known as andyrock
sil2100jamesh: if anything, the new lucene++ is in saucy now16:06
sil2100jamesh: so we can proceed with merging in all the bits and pieces for media-scanner16:06
tedglarsu, Woot!  UMM!  :-)16:08
sil2100jamesh: I'm re-running CI for your merge16:08
tedgdednick, You're now on my harass list :-)16:08
larsutedg: ;)16:08
andyrocksil2100, fixed16:11
andyrocksil2100, is there already a bug on lp?16:11
dednicktedg: lucky me.16:11
sil2100andyrock: let me check if Mirv filled one in16:11
dednicktedg: is there anything in particular you want to harass me about? :)16:11
tedgdednick, Just the UMM branch into unity8 trunk.16:11
sil2100andyrock: yes, there seems one!16:12
sil2100andyrock: https://bugs.launchpad.net/unity/+bug/121117416:12
ubot5Launchpad bug 1211174 in Unity "test_home_lens.HomeScopeSearchTests.test_quick_run_app failing" [Critical,New]16:12
sil2100andyrock: thanks a lot for fixing! Any branch I could take a look? Where was the problem?16:12
didrockssil2100: btw, the number of AP tests failing on unity is quite high, but let's see once you have those in16:12
dednicktedg: i c. waiting on a review. :) need to harass some on from unity8 ui team.16:12
didrockssil2100: so, I hope you can release this unity easily ;)16:12
andyrocksil2100, Trevinho changed the dbus path16:13
tedgdednick, Then we need to work on moving the indicators over as we solve integration issues.16:13
didrocksandyrock: can you check if nothing else is impacted?16:13
tedgdednick, I'd like to figure those out and get those bugs fixed ASAP.  But it's going to be harder for us until that lands.16:13
andyrockit was something like that before /Application34123416:13
andyrocknow it's /Application/2342416:13
andyrockdidrocks, sure16:13
didrocksthanks! :)16:13
tedgdednick, Ah, so I just need to harass someone to review it.... Saviq, do you need me to call anyone to cancel your plans for the evening?   ;-)16:13
didrockssil2100: given that information, I think you can force the publication of platform16:14
didrocks(and everything stuck on it then ;))16:14
fgintherdednick, pong16:16
sil2100didrocks: hm, can I do it the old way? i.e. run cu2d-platform-head-3.0publish manually first? Since forcing publication would also force it if there were any packaging changes, right?16:16
didrockssil2100: you can run with auto_publication16:16
didrocksor just the publish step16:16
didrocksright :)16:16
dednickfginther: sorry, ment to unping.16:16
fgintherdednick, no worries16:17
sil2100didrocks: if, by any chance, the publish job goes green, will it trigger all the rdeps when I do it by just running the publish job?16:17
sil2100Ok, today is not a good day for me and jenkins16:20
sil2100It's already the second time I double clicked something ;/16:20
andyrocksil2100, https://code.launchpad.net/~andyrock/autopilot/fix-1211174/+merge/17974216:21
didrockssil2100: all the rdepends will run if they were in manual publishing mode16:21
sil2100didrocks: I re-ran the publish job (twice, the second time I hope I aborted in time) ;/ Sorry about that16:22
didrockssil2100: no worry16:23
didrockssil2100: seems ati worked that time16:23
didrocksah, it's still in the apps16:23
didrocksnot mirslaves yet16:23
sil2100andyrock: what makes me happy is that it's in autopilot \o/16:28
sil2100andyrock: so we can get it released much faster!16:28
andyrocksil2100, :D16:28
andyrockdidrocks, btw I don't see any other broken things in ap because of that bamf change16:29
didrocksandyrock: ok, so the broken tests are because of something else?16:29
andyrockdidrocks, what tests?16:29
didrockssil2100: ah, it's in AP? nice :)16:29
sil2100didrocks: yes ;) Trevi changed a path in bamf, and AP wasn't changed to fit that16:29
didrocksgreat ;)16:30
mzanettiCimi: re16:34
Cimimzanetti, here I am16:34
mzanettiCimi: that was a meeting...16:34
mzanettiCimi: so... binding loop you said16:34
mzanettilemme check16:34
Cimimzanetti, if defaultBackground is Image.Error basically...16:34
Cimibut actually not16:35
sil2100andyrock: top approved!16:35
andyrocksil2100, sweet thanks"16:35
mzanettiCimi: right... I see where it happens16:37
mzanettiCimi: actually I think its a false positive from the QML engine16:37
CimiSaviq, I can kill myself16:37
CimiSaviq, two qml bugs in one day16:37
mzanettibecause the state changes to Error, then we set another image which triggers the state change again16:38
mzanettiif we would continue to set invalid images it'd be a valid loop16:38
Cimimzanetti, no16:38
Cimimzanetti, because if it's error the loop stops16:38
Cimimzanetti, since you set to defaultBg16:38
Cimimzanetti, at that point it's either image.ready16:38
Cimimzanetti, or image.error but doesn't go in the loop because there's source != defaultBg16:39
mzanettiexactly... so we're not really having a loop because we know we're setting a good one the second time16:39
mzanettiCimi: so I'd say we can either ignore this warning or set the corrected image delayed to get rid of it16:39
Cimimzanetti, delayed? how?16:40
Cimimzanetti, with extra prop?16:40
mzanettiCimi: as setting something in a delayed fashion is quite ugly in QML (we'd need a Timer etc) I'd vote for ignoring this warning in this case16:40
mzanettiunless Saviq objects ^^16:40
Cimimzanetti, Saviq it actually happens only if you set an invalid image16:41
Cimiso it's not like spamming the debug every time...16:41
mzanettihmm... well, let me try something16:43
mzanettiCimi: ok... I just tested what happens in case the defaultBackground would be invalid (e.g. file deleted from disk)16:44
mzanettiCimi: in that case the binding loop detection seems to do the right thing and stops calling everything16:44
Cimimzanetti, what? :)16:45
Cimimzanetti, if is invalid, it loops?16:45
mzanettiCimi: well, in case we get Image.Error we set the defaultBackground16:46
Cimibut after that16:46
Cimiit doesn't proceed16:46
Cimimzanetti, source will be == defaultBg16:46
andyrocksil2100, didrocks  this one should be fixed too now16:46
mzanettiCimi: assuming that would be invalid too (because the user deleted the file) we also get this warning, but in that case its valid and the image stays in Image.Error16:46
mzanettiCimi: so that's good too16:46
mzanettiCimi: and once we are in this state we perfectly recover because when the setting changes, we set source through an assignment (not a binding) and we enable all the stuff again16:47
mzanettiCimi: so I can't see an error right now.16:47
sil2100andyrock: the same issue?16:48
mzanettiCimi: just that warning from QML which is understood and doesn't hurt us. Now its just waiting for Saviq to tell us if he's fine with it or we need to get rid of the warning by using some more stuff (like a Timer to set it delayed or add back the other "useless" image just to check if it can be loaded)16:49
andyrocksil2100, yeah that test is using the same function16:50
mzanettiCimi: were there any other issues? I fixed the one with the empty setting in the beginning16:50
sil2100andyrock: two birds with one stone ;)16:52
Cimimzanetti, nope, now I need to see if autopilot crashes16:54
mzanettiCimi: huh? why would that happen?16:54
Cimiwe can check that early tomorrow when saviq comes back16:54
Cimiwe were having autopilot issues with the greeter16:54
CimiI don't know16:54
Cimimichal knows16:54
didrocksandyrock: ah great! :)17:02
=== alan_g is now known as alan_g|EOD
SaviqCimi, not crashes, fails17:58
Saviqdednick, a gift? gimme gimme! :D18:00
WellarkSaviq: is X-Ubuntu-StageHint a list?18:41
Wellarktedg: ^18:41
tedgWellark, Not sure, sorry.  Perhaps greyback_ would know.18:41
Wellarkgreyback_: ^18:42
Wellarksomebody who knows please comment here: https://code.launchpad.net/~kaijanmaki/unity8/launcher-backend/+merge/179663/comments/40605218:42
Wellarkand also provide a list of expected values if possible18:42
greyback_Wellark: I'm not 100%, but I think no. It is a string.18:42
greyback_but best get someone who is definite to say18:43
Wellarktedg: are there any other custom keys for the click .desktop files?18:43
Wellarkgreyback_, Saviq: or for the unity8 in general?18:44
tedggreyback_, Is Unity going to register for the name "com.canonical.Unity.WindowStack" ?18:44
tedgWellark, Yeah, mostly housekeeping though: http://bazaar.launchpad.net/~indicator-applet-developers/upstart-app-launch/trunk.13.10/view/head:/desktop-hook.c#L19818:45
Wellarktedg, greyback_, Saviq: all the keys I know about that are currently used in existing .desktop files are here: http://bazaar.launchpad.net/~kaijanmaki/unity8/launcher-backend/view/head:/plugins/Unity/Launcher/backend/desktop-entry.y#L10218:46
Wellarkplease, inform me if I'm missing any18:46
tedgWellark, Why do we need all of them?18:47
tedgWellark, Is there not an keyfile parser in Qt?18:47
greyback_tedg: it can, just need to change the code.18:48
Wellarktedg: because all the keys that we know about are "free"18:48
tedggreyback_, Thoughts?  I was thinking perhaps that way we could move it around if need be.18:48
Wellarktedg: X-Ubuntu-Application-ID was a new one18:48
Wellarktedg: that's the $APP_ID18:48
tedgWellark, Yeah18:49
Wellarktedg: and yes, there is a "ini" file reader in Qt18:49
Wellarkbut it can't handle the existing .desktop files18:49
Wellarknot all of them18:49
greyback_tedg: sorry, I lost you. Move what around?18:49
Wellarksome translations get it confused18:49
tedggreyback_, The window list dbus infomation.  We could move it to a different process if it had its own name.18:49
Wellarktedg: what is X-Ubuntu-Old-Path ?18:50
tedgWellark, Hmm, okay.  I would have suggested using the GLib one then :-)18:50
tedgWellark, We override the path variable, so it just puts the value in incase that's useful.18:50
tedgWellark, Don't think anyone would use it.18:50
greyback_tedg: it needs to be part of unity, as that's the only process which can know the app & window list.18:51
tedggreyback_, Well, it needs to be part of the piece of Unity that connects to Mir, but what we call "Unity" may shift over time.18:52
tedgIt may become less... unified...18:52
Wellarktedg: gdesktopappinfo does not support custom groups18:54
Wellarknor Action definitions18:54
tedgWellark, Sure, but GKeyFile can do the parsing.  And the latest version I think does Actions.18:55
tedgIt was on the TODO list, not sure where it is.18:55
greyback_tedg: if that happens, then we'll deal with it. But right now this is the easiest implementation, and it is done. Bring it up on the ML if you want to propose it further, get other people's input18:55
Wellarktedg: and GKeyFile does not do translations18:55
tedgWellark, https://developer.gnome.org/glib/2.30/glib-Key-value-file-parser.html#g-key-file-get-locale-string18:56
Wellarktedg: if you know the locale18:57
tedgWellark, " If locale is NULL then the current locale is assumed. "18:57
Wellarktedg: that does not handle the in-file translations18:58
Wellarkwhich suck18:58
tedggreyback_, To be clear, I'm not suggesting changing it today.  Just making that possible in future.18:58
Wellarkbut we have to support them18:58
tedgWellark, Hmm, I'm pretty sure it does.  We've distro patched it quite a bit there.18:58
tedgWellark, I believe it does both language packs and in file.18:59
Wellarkand gkeyfile does not know anything about X-[Ubuntu|GNOME]-GettextDomain18:59
greyback_tedg: sure, it's only code, we can always change it :)18:59
tedgWellark, Not upstream, we distro patch.18:59
greyback_tedg: can I ask you one thing about https://code.launchpad.net/~ted/upstart-app-launch/libupstart-flesh/+merge/178296/comments/40614218:59
Wellarkand the last thing: gkeyfile deals with string keys in it's API and thus relies on extensive string comparison19:00
tedggreyback_, What's up?19:00
Wellarkmy parser uses bison/flex with zero string comparison for the preknown keys19:00
greyback_tedg: app can launch another PID. That other PID can create create a new Mir Surface. To authorize it, shell needs to know that new process' PID19:00
Wellarktedg: do you need more reasons? :)19:01
greyback_tedg: am I right, or am I missing something?19:01
tedgWellark, Eh, I'd take a well worn parser over a faster one :-)19:01
tedgWellark, It doesn't matter really.19:01
tedggreyback_, Sure, I think it probably does need to know that PID, but it will be the same app_id.19:01
tedggreyback_, We don't have the cgroup stuff setup today, so I don't know exactly how that'll work though.19:02
tedggreyback_, I think that the cgroup ID will be the PID of the main process.19:02
greyback_tedg: ok, so 1 app_id can have multiple PIDs, and those PIDs can all have individual surfaces if they wanted.19:02
tedggreyback_, So you'll be able to check the PIDs cgroup.19:02
tedggreyback_, Yes19:03
greyback_tedg:  If one of those PIDs creates a surface, I need to check that the PID belongs to that app_id's cgroup19:05
SaviqWellark, re: StageHint, I'd say a list, yes19:05
Saviqgreyback_, after all apps need to be able to support both stages ↑19:05
greyback_upstart has all those PIDs internally, no?. Would it be possible to have upstart-app-lib make that easy for me? :)19:05
greyback_Saviq: I thought hint was what was the preferred stage19:06
greyback_ok, I misunderstood19:06
Saviqgreyback_, the first one can be the preferred19:06
greyback_Saviq: ok19:06
Saviqgreyback_, but if app supports both, we need to know19:06
Saviqas we can only assume that "no hint" == "only main"19:06
tedggreyback_, In theory, I'm just not sure.  We get a set of PIDs but I'm not sure if Upstart will have all or a subset.19:07
tedggreyback_, But we could put a helper function in the library.  And then fix it there.19:07
tedggreyback_, Something like "pid_in_appid" and we can look.19:07
tedggreyback_, Then eventually we'll get cgroup support, and it should be transparent for you.19:08
greyback_tedg: yep, that would be perfect19:08
tedggreyback_, Does the final comment here work?  http://paste.ubuntu.com/5978379/19:09
WellarkSaviq: can you reply on the MR and provide at least some values we currently know about?19:11
SaviqWellark, MainStage, SideStage - that's it, for the foreseeable future19:11
SaviqWellark, with the first one being preferred, if both are given19:11
dednickSaviq: ping19:31
tedggreyback_, Hmm, so I was thinking differently.19:31
tedggreyback_, I did it as you provide the app_id and the PID19:31
Saviqdednick, pong19:31
tedggreyback_, Would that work?  In theory if they're requesting a surface, they give you the app_id, no?19:31
dednickSaviq: did you have a link to a wiki regarding the lifetime of qml objects?19:32
tedggreyback_, http://bazaar.launchpad.net/~ted/upstart-app-launch/libupstart-flesh/revision/7819:32
Saviqdednick, http://qt-project.org/wiki/SharedPointersAndQmlOwnership19:33
dednickSaviq: thanks19:33
greyback_tedg: all I get from Mir is "hey, I've been asked to create a surface by a process with pid x"19:36
tedggreyback_, Huh, interesting.19:41
greyback_tedg: yep, it's pretty limiting. But if you don't think my request is possible, I'll see if I can lean on the Mir guys for more info.19:42
tedggreyback_, It's not as easy... let me investigate a bit.19:43
greyback_tedg: ok19:43
greyback_tedg: I'm going offline now, but catch you tomorrow19:43
Saviqdednick, >>= means something else than >=20:11
dednickSaviq: looked at the debian control spec. didnt seem to be documented.20:12
dednick>> is scrictly greater20:12
Saviqdednick, hmm indeed20:12
Saviqdednick, you're probably right20:12
dednickSaviq: whoop. do you think you can take a look at that MP tomorrow? thomas is pretty eager to get it in.20:13
Saviqdednick, about the unitymenumodel one?20:13
Saviqdednick, I think so, yeah20:14
dednickSaviq: yeah20:14
dednickSaviq: cool. thanks. it might have to change a bit first though. I htink it's leaking like a sieve.20:14
dednickunitymenumodel is anyway20:14
dednickthanks to my additions :(20:15
dednickreturning new from model::data = bad.20:15
Saviqdednick, yeah, indeed ;)20:26
Saviqdednick, unless there's no parent, in which case the deleted delegate would clean them up (assuming they wont dataChanged() first, which might lead to leaks anyway)20:27
dednickSaviq: yeah.. i dont think the parent is being set on the objects that i return at all. they're being used like properties rather than objects.20:28
dednickSaviq: i'm just getting rid of the fancy stuff in unitymenumodel. not really needed anymore anyway20:29
Saviqdednick, k20:29
seb128dednick, Saviq: sorry I typoed that one, it's either >= or >> ;-)20:51
=== jhodapp is now known as jhodapp|afk
=== tvoss is now known as tvoss|eod
Sharp1Hi, I'm using Ubuntu 13.04 and if I set up a hotcorner to switch workspaces in Unity Tweak Tool, it doesn't work after I log out and log in again. The settings stays the same in Unity Tweak Tool, but the corner does nothing. Does anyone have a solution?22:39
=== salem_ is now known as _salem

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!