/srv/irclogs.ubuntu.com/2014/09/01/#ubuntu-unity.txt

=== alecu__ is now known as alecu_
=== robru_ is now known as robru
=== me4oslav_ is now known as me4oslav
memekahi, can i run unity on top of qtwayland?06:38
Saviqmemeka, neither unity7 or unity8 support qtwayland right now, no06:51
Saviqunity7 probably never will and we're investing heavily into Mir, so it's unlikely we will spend time to port it to qtwayland06:52
memekaSaviq: how about the qt apps made for ubuntu touch?06:53
memekathey are all qt and qml .... they should start, right?06:53
memeka*qt506:53
Saviqmemeka, well, not under *qtwayland* directly, but under a shell that implements qtwayland, sure, they work just fine under X11 too06:54
memekaSaviq: qml-compositor comes with qtwayland ... and the only other I know of is GreenIsland/Hawaii ...06:55
Saviqmemeka, yes, sure, I see no reason why the apps wouldn't start there06:55
memekaSaviq: does Mir works now with hwcomposer via libhybris ?06:55
memekathat's the only reason I am using qtwayland now, since I am under the impression that Mir works only with fbdev06:56
Saviqmemeka, didn't know qtwayland supports hwc? when did that happen?06:56
Saviqmemeka, in any case, the way that unity8 uses Mir right now, not yet, I believe if you use Mir as compositor, yes06:57
memekaSaviq: not officially... but there is this: https://github.com/mbrasser-jolla/qt5-qpa-hwcomposer-plugin06:57
Saviqmemeka, I don't know all the details, you might wanna check #ubuntu-mir06:58
memekaSaviq: hence my question if Mir supports hwcomposer...06:59
Saviqmemeka, AFAIK, if you use Mir's composition, then yes06:59
memekaso then I can install Mir - Unity8 - and all the rest07:00
Saviqbasically if Mir draws more than one surface (which is not currently the case with Unity8), then it can do hwc07:00
memekaSaviq: the problem is that not even qtwayland worked for me with the libhybris version that is currently used by Ubuntu - which is from 2013 -- it could work only with more recent versions07:02
memekaSaviq: however, given the newer libhybris installed, and since i am using 14.10 ... is it as simple as apt-get install mir and apt-get install unity ?07:03
Saviqmemeka, what do you want to achieve though?07:03
Saviqthere is not a package called "mir" btw, and unity8, not unity (that one is the current desktop version)07:04
memekaSaviq: I have an ARM board with PowerVR gpu, with android drivers only - and I wanna get unity to work on it07:04
memeka(yeah, I can find the right package names :P)07:04
Saviqmemeka, so yeah, apt-get install unity8 in general, you'll need to explicitly install qtdeclarative5-qtmir-plugin, too07:05
Saviqah no, that will get pulled in automagically07:05
memekaSaviq: so I have the android hwcomposer running in my linux install, and libhybris installed .... I can run test_hwcomposer and it works07:05
memekabut X won't run at all for example, since the display is taken by the android hwcomposer07:06
Saviqmemeka, you mean surfaceflinger?07:06
Saviqmemeka, you can't have surfaceflinger running if you want to run unity8, we need direct access to the GPU07:07
Saviqmemeka, you might wanna check out https://wiki.ubuntu.com/Touch/Porting07:07
memekaits not that07:08
memekaSaviq: I am running android-media-server and android-service-manager from android07:08
Saviqmemeka, sure, we do the same I believe07:08
memekaSaviq: and also I am running pvrsctl, which is the PowerVR hwcomposer07:09
memekabut yeah, I guess that's not direct access to the GPU ... it's only through libhybris07:09
Saviqmemeka, ok, I'm out of my depth here, you'll have to wait for an answer in #ubuntu-mir07:09
memekaI will ask there too...07:09
Saviqmemeka, we do definitely talk android hwc07:10
Saviqbut they'll know more details07:10
memekaright, so then it should work ....07:10
memekaSaviq: last time when I checked the porting guide, it was building stuff from CM base07:11
memekaI have already all the android stuff installed and working07:11
tsdgeosis jenkins qa down?07:19
tsdgeosi can't reach http://jenkins.qa.ubuntu.com/job/generic-click-autopilot-runner-mako/444 for example07:19
tsdgeospaulliu: any reason not to top approve https://code.launchpad.net/~aacid/unity8/removeUnusedHasAttributes ?07:23
Saviqtsdgeos, Paul is on debconf, might be touch to reach07:34
tsdgeosah ok07:35
tsdgeosSaviq: so then maybe you can top approve given his comment?07:35
SaviqACKed07:35
tsdgeosSaviq: can you access the jenkins?07:36
Saviqtsdgeos, s-jenkins?07:36
tsdgeosit's kind of bad if i can't check if the unstables in there are because of the popover thing of something else07:36
tsdgeosyeah07:36
Saviqtsdgeos, you on VPN? works fine?07:36
tsdgeoshttp://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-utopic/117107:37
tsdgeosno, vpn not on07:37
tsdgeosjust stays there forever07:37
Saviqah that one07:37
Saviqtsdgeos, yeah, that's dead07:37
tsdgeosok, let's go to the vpn one07:37
Saviqtsdgeos, yeah, http://s-jenkins.ubuntu-ci:8080/job/unity-phablet-qmluitests-utopic/?07:38
tsdgeosCimi: top approve https://code.launchpad.net/~aacid/unity8/clickBlackStuffOverview/+merge/232431 ?07:42
=== _morphis is now known as morphis
tsdgeosSaviq: ping08:10
Saviqtsdgeos, here08:10
tsdgeosSaviq: all these tests http://s-jenkins.ubuntu-ci:8080/job/unity-phablet-qmluitests-utopic/1181/#showFailuresLink are failing because of the network module thing we made a runtime dep not being there08:11
Saviqtsdgeos, when did I say we should make it not being there?08:11
Saviqtsdgeos, I said ideally it would be mocked out08:11
Saviqtsdgeos, but obviously until it is, we need it in both build and run08:12
tsdgeosok08:12
tsdgeosi misunderstood you then08:12
Saviqtsdgeos, but TBH if we don't mock it out, can we depend on the tests?08:12
tsdgeosnot sure i understand what you mean08:13
Saviqtsdgeos, like I'm worried it could affect our results08:13
tsdgeosyou mean tests are wrong because that is not mocked?08:13
Saviqtsdgeos, because we don't know what network status will be08:13
Saviqwe can't rely on it08:13
Saviqtsdgeos, if it works, fine, but I just mean we need to expect this to break if we don't mock it out, is all08:14
tsdgeoshmmm08:16
tsdgeosyeah i guess08:16
tsdgeoslet's see if i can mock it out then08:18
tsdgeosWellark: simple fix https://code.launchpad.net/~aacid/connectivity-api/unused_qthread_include/+merge/23285908:35
=== sonne_ is now known as sonne
tsdgeosWellark: also shouldn't you for limitedBandwith a similar connect in the qml side like you do for statusChanged?08:42
=== Guest35996 is now known as karni
=== Estilanda_ is now known as Estilanda
mzanettihmpf... why wouldn't run.sh work now...10:32
anpokah so qtmir reuses the context created by mir and just switches to Opengl ..10:42
anpokthat explains everything10:42
facundobatistaHolas10:49
=== MacSlow is now known as MacSlow|lunch
tsdgeosuh11:20
tsdgeosgreat11:20
tsdgeosnow all the builds are failing because https://jenkins.qa.ubuntu.com is down11:21
tsdgeosSaviq: can't we really find anyone that cares?11:22
Saviqtsdgeos, there's the ci vanguard in #ubuntu-ci-eng11:23
Saviqtsdgeos, they are the ones to ping11:23
tsdgeosSaviq: https://bugs.launchpad.net/unity8/+bug/134120511:40
ubot5Ubuntu bug 1341205 in Unity 8 " Top left icon of my apps can be placed in the middle of the apps" [Medium,Triaged]11:40
Saviqtsdgeos, there isn't a see all any more, let me try and repro here and update the description11:42
Saviqtsdgeos, it was enough for me to just go to the bottom of apps and go to store and back11:43
Saviqtsdgeos, one thing I noticed there is why would the apps page move at all?11:43
Saviqtsdgeos, because it got scrolled up by itself as I was going into/back from the store11:44
tsdgeosSaviq: hidden height stuff probably11:45
tsdgeoshttps://code.launchpad.net/~aacid/unity8/category_view_invisible_in_preview_mode changes may make that not happen11:45
Saviqtsdgeos, yeah, I expected that11:46
Saviqtsdgeos, but yeah, I can reproduce reliably, just go into store, back out from store, broken11:48
tsdgeosSaviq: not here at all11:48
tsdgeosSaviq: krilin?11:48
Saviqtsdgeos, mako11:48
tsdgeoswhich image?11:49
Saviqtsdgeos, devel-proposed11:49
Saviqtsdgeos, I think it might be related to the number of app11:49
Saviqs11:49
tsdgeosmost probably11:49
tsdgeosthat's why asked for the number of apps too ;)11:49
Saviqtsdgeos, I've 8*3 + 1 == 25 apps11:49
tsdgeoscan you comment on the bug?11:49
tsdgeosi've more11:50
tsdgeosbut i haven't installed anything in this new image11:50
tsdgeosor do click apps are installed somewhere else so they persist between reflashings?11:50
Saviqtsdgeos, they do11:54
Saviq(persist)11:54
Saviqtsdgeos, I got the header in the middle now11:55
Saviqtsdgeos, https://docs.google.com/a/canonical.com/file/d/0B32jwBcbaPloUEdpRFRKVlNycnM/edit11:55
Saviqtsdgeos, same steps, krillin, 23 apps11:56
* Saviq wonders if 2048 causes this :D11:56
Saviqoh and now I got it to get stuck...11:57
anpokwas there a recent change that could explain why usc does not show unity8.. but unity8 receives user input..12:19
anpok(on desktop)12:19
anpokor unity8 renders just black12:19
Saviqah nasty scopes again12:20
Saviqanpok, I just tried and IIUC QT_QPA_PLATFORM wasn't exported properly12:21
Saviqanpok, can you check /proc/`pidof unity8`/environ12:21
Saviqanpok, for QT_QPA_PLATFORM=ubuntumirclient12:21
Saviqanpok, to confirm you can always look in ~/.cache/upstart/unity8.log and see if Xcb complains about display not being set12:21
greybackunity8 needs QT_QPA_PLATFORM=mirserver12:23
Saviqgreyback, ah then that's what got broken12:23
anpokah yes here it has ubuntumirclient too12:28
anpokthat is supposed to be done in the upstart session confs?12:28
Saviqanpok, /usr/bin/lightdm-unity8-session12:29
Saviqgreyback, not mirclient? doesn't unity8 deal with the s/client/server/ itself?12:29
anpokoh thats wrong12:30
anpokit really says ubuntumirclient there..12:30
Saviqanpok, yeah, should say "mirclient"12:30
greybackSaviq: the name of the QPA plugin for unity8 is mirserver. For applications, it is ubuntumirclient12:30
Saviqgreyback, ah then ubuntumirclient is correct12:31
Saviq    if (qgetenv("QT_QPA_PLATFORM") == "ubuntumirclient") {12:31
Saviq        setenv("QT_QPA_PLATFORM", "mirserver", 1 /* overwrite */);12:31
Saviqanpok, ↑ we're doing that in our main12:31
Saviqso ubuntumirclient is good, but I didn't have anything in QT_QPA_PLATFORM in the unity8 environ12:31
greybackkinda for legacy, and kinda as clients use platform-api (and thus ubuntu-specific functionality), not mirclient directly12:31
Saviqgreyback, sure, but that should still work12:32
anpokok even when putting mirserver it still is all black12:32
Saviqanpok, anything in ~/.cache/upstart/unity8.log ?12:33
Saviqtsdgeos, ok, ruled out 2048 as being the reason, but it seems to require a particular number of rows in the dash (which supports the height suspicicon), on mako it has to be 9 rows, on krillin it was 8 here12:34
Saviqtsdgeos, more or less than that probably changes the calcs enough to work around the issue12:34
=== MacSlow|lunch is now known as MacSlow
anpokSaviq: unity8 log http://paste.ubuntu.com/8205954/12:46
anpokbut not dash log .. instead a crash file12:46
Saviqanpok, ah, so it started with mir for you at all12:46
Saviqanpok, I think it might be the GstFoo thing... I can't launch camera-app under X11 here either12:47
Saviqwith the exact same messages12:47
anpokbut the unit8-dash failure is in ubuntumirclient qpa12:48
Saviqanpok, yeah, but probably just because it can't connect to mir12:49
anpokbut then unity8 would still display the indicators and the greeter part?12:49
Saviqanpok, but it doesn't start12:51
Saviqanpok, it hangs at the GstFoo messages12:52
Saviqanpok, I get the same for camera-app: http://pastebin.ubuntu.com/8205997/12:52
=== _salem is now known as salem_
Saviqand no camera app12:52
anpokso log messages like 'unhandled motion event action 9' and similar fit in the u8 does not start picture?12:55
Saviqanpok, that I don't know where it comes from12:56
dandradermzanetti, greyback, shouldn't we have a cap at the number of applications showing up in the spread? Even if they get killed due to OOM, their snapshot will still hang over there, taking memory. I was thinking like removing the oldest ones past that number cap. Also, if the number of apps in the spread is too big, it might just get unwieldy to the user to have to many cards over there12:56
Saviqdandrader, as for memory considerations, sure, we should unload the screenshots if we can12:57
Saviqdandrader, but design-wise we can't drop any apps of12:58
Saviqf12:58
Saviqif it's unwieldy, the user should just drop some of them, we shouldn't12:58
mzanettidandrader: in any case, such things need to be discussed with design. you can suggest those things either directly to Vesa or create a bug for ubuntu-ux12:59
dandraderSaviq, greyback, mzanetti  btw, I think we don't have a bug on that "unload screenshots to disk" do we?13:00
dandraderwe should have one or we might just forget about it13:00
Saviqdandrader, we might not indeed13:00
* dandrader volunteers13:00
dandradermzanetti, right13:01
anpokhm ok that blocks main13:01
* greyback concurs13:01
mzanettigreyback: to what?13:01
mzanetti:)13:01
greybackmzanetti: the replies to dandrader's query13:02
tsdgeosSaviq: can reproduce with 913:02
Saviqtsdgeos, good, finally some info, this was rather crazy ;)13:04
mzanettigreyback: :) wasn't sure if the query or the replies13:04
tsdgeosSaviq: i'm fearing it may be a gridview bug itself, since we do no positioning of that lost item13:10
tsdgeosbut i'm going to try if lp:~aacid/unity8/category_view_invisible_in_preview_mode that has a few fixes for sizing doesn't make the jump happen and then at least we don't see it13:11
Saviqtsdgeos, yeah I'd imagine that's the case13:12
tsdgeosSaviq: yes it doesn't fail anymore, we're happy enough with the "workaround", right?13:22
tsdgeosSaviq: is there any reason https://bugs.launchpad.net/unity8/+bug/1356410 is not marked as release other than "we forgot"?13:36
ubot5Ubuntu bug 1356410 in Unity 8 "Category header links don't work in overview" [Critical,In progress]13:36
Saviqtsdgeos, yeah, good with workaround, and yeah we forgot13:38
Saviqtsdgeos, one of the reasons why I'm leaning towards dropping the non-ubuntu unity8 bugs altogether (as in only allowing unity8 (ubuntu) bugs)13:38
tsdgeosom26er: Saviq: https://bugs.launchpad.net/unity8/+bug/1195707 is long ago fixed, no?13:39
ubot5Ubuntu bug 1195707 in Unity 8 "[Dash] Installed scopes icons are shown in the dash" [High,New]13:39
Saviqtsdgeos, done13:39
om26ertsdgeos, yes13:40
tsdgeosthis is hight? https://bugs.launchpad.net/unity8/+bug/1199423 really? :D13:40
ubot5Ubuntu bug 1199423 in Unity 8 "Log output is excessive" [High,In progress]13:40
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
Saviqdandrader, dednick, how close are we to landing silo 2?14:22
dandraderSaviq, today if all goes well I think. Worst case tomorrow lat in the day I guess14:24
dandraders/lat/late14:24
Saviqdandrader, k thanks, I'll start a separate silo then14:24
MacSlowSaviq, is silo 2 the potential place for the visual-tweaks for notifications too?14:30
SaviqMacSlow, as I won't touch it under dandrader's feet, no, I'll do a separate one14:30
MacSlowor just the prompt-in-prompt14:30
Cimidednick, can we filter only the list of wifi from unity menu model without using the sortfilterproxymodel?14:34
dednickCimi: not that i know of14:36
Cimidednick, I think the sortfilterproxymodel might confuse index and stuff14:37
dednickCimi: you need to translate the index14:37
Cimidednick, yes, but we have issues of tapping sometimes opens the wrong access point14:37
Cimidednick, might be bindings or how the filter works (not perfectly dynamic)14:38
dednickCimi: logging out which one gets activated might help14:38
dednickCimi: pehaps there's a double tap happening or something and it's re-sorting in-between14:39
CimiI cannot have listview with different height for delegates, right?14:40
SaviqCimi, sure you can14:40
dednickCimi: i put a bug in for indicator-network about not changing the "active connection" until it's actually connected14:40
CimiSaviq, I am wondering if I can not use sortfilterproxymodel and hide delegates of network properties I don;t want14:41
dednickCimi: i think at the moment it just changes immediately when you tap the network14:41
SaviqCimi, you could, but it's obviously better if they're not even there14:41
CimiSaviq, we have issues when tapping networks and connecting to wrong APs or such, I believe could be due to this filtering we do14:41
SaviqCimi, belief has nothing to do with it ;)14:41
CimiSaviq, I am trying to test indeed14:42
Cimidednick, what is unitymenumodelstack?14:43
dednickCimi: the unitymenumodel is multi-level (model in model). the stack keeps track of which level you're looking at14:48
Saviqtsdgeos, ah, wanted to say14:48
tsdgeosSaviq: yes?14:48
Saviqtsdgeos, while doing the thing for overview on left edge /  BFB...14:48
Cimidednick, they added it in the settings, I am wondering if I need it here14:48
Saviqtsdgeos, you could get rid of the DashCommunicator thing14:48
Cimiin the wizard14:48
Saviqtsdgeos, as we now support scope:// urls14:48
dandraderdednick, prompts are supposed to be killed/closed when the app (root session) is suspended, right?14:49
tsdgeosSaviq: hmmm, but we'll need it again if we need cancellable stuff..14:49
dednickCimi: it depends on how your menu is contructed. if it's flat, or a tree.14:49
CimiI see14:49
Saviqtsdgeos, TBH with the cancellable we'll need to be quite tricky there14:50
Saviqtsdgeos, I'm thinking ShaderEffect tricky14:50
dednickdandrader: depends on the app implementation. in qtmir they are.14:50
Saviqtsdgeos, because we won't be able to recover the whole thing14:50
Saviqtsdgeos, like search history etc., I think we'll just need to have a snapshot of the apps scope and display it, and hide on dash focus14:51
Saviqtsdgeos, but yeah, ok, leave it be for now14:51
dandraderdednick, I'm following those steps: https://code.launchpad.net/~nick-dedekind/qtmir/prompt-in-prompt/+merge/232417/comments/56691214:51
dednickdandrader: ya? are they not being closed?14:51
dandraderdednick, they are14:52
dednickright. that's good then :)14:52
dandraderdednick, But https://code.launchpad.net/~nick-dedekind/qtmir/prompt-in-prompt/+merge/232417/comments/56786514:52
dednickdandrader: hm. does it fade?14:53
dandraderdednick, no14:53
dandraderdednick, it really disappears instantly14:53
dednickdandrader: huh. give me a mintue, i'll give it another try14:54
tsdgeosSaviq: this is sad https://code.launchpad.net/~aacid/unity8/gettext_textdomain/+merge/23289914:54
tsdgeosworse function name ever14:54
dednickdandrader: you using local build or silo?14:55
dandraderdednick, local build of unity8 and qtmir14:55
* dandrader is old school14:55
dednickdandrader: :) me too.14:55
Saviqtsdgeos, OH14:55
dednickdandrader: wasn't sure if silo had been updated since session.live14:56
Saviqtsdgeos, do we need the Binding to i18n in Shell.qml then?14:58
tsdgeosSaviq: hmmm, not sure, let me see14:58
dednickdandrader: hm. it's working for me. the page is pretty hard to tell if it's fading quickly (a lot of white). can you increase the zombie animation duration in SurfaceContainer.qml to 5 seconds and try again?14:59
dednickdandrader: oo. actually now that i increased it its not sliding down! haha15:02
dednickhm15:02
dednickit was just fading out...15:02
dednickdoh15:02
tsdgeosSaviq: yeah seem unneedeed now15:05
dednickdandrader: ah. it's the lack of session.parent since it's just a qobject15:08
dandraderdednick, and more bad news :-(. check latest qmir comments15:09
dandraderqtmir15:09
anpokhm who is the one to blame if the qt5 plugins libgstcamerabin and libgstmediaplayer link against gst0.10 while UbuntuComponents use gstreamer 1.015:10
dednickdandrader: damn. kind of need to move the app screenshot onto the session...15:14
dednickand fade to that on zombie15:14
dednickor something...15:14
dednickno, nevermind15:15
dandraderdednick, must also figure out why it doesn't happen in the qmltest15:15
dandraderdednick, would be good to get it covered there somehow15:16
dednickdandrader: i think it's the just the lag of mir.15:16
dednicki'll work it out15:17
dednickdandrader: will it already have a screenshot when the app is stopped?15:19
dandraderdednick, no15:21
dandraderdednick, it takes the screenshot *once* the suspended app gets stopped15:21
dandraderdednick, and once the screenshot is ready. it crossfades to it. once that animation is done it finally releases the zombie surface15:22
Saviqtsdgeos, so, what does that i18n change give us then? setting the domain early?15:22
tsdgeosSaviq: and makes it work from c++15:22
Saviqtsdgeos, right15:22
tsdgeosthat i have no idea why it didn't work before15:22
tsdgeosnot sure what the qml stuff was doing15:22
tsdgeosmaybe was using dgettext all the time15:22
tsdgeosSaviq: don't approve yet15:22
tsdgeosthere's another patch comming up on the same branch15:23
Saviqk15:23
dednickdandrader: http://paste.ubuntu.com/8207107/15:23
dednickthe session is being release.15:23
dednickdandrader: i think it's the same problem as the prompt disappearing15:24
tsdgeosSaviq: ok, pushed the other change15:25
tsdgeosmzanetti: you may want to have a look, to it too sicne i think your rework branch changes this15:26
tsdgeosmzanetti: https://code.launchpad.net/~aacid/unity8/gettext_textdomain/+merge/23289915:26
Saviqtsdgeos, right, we should read from Name[$locale] first, clicks won't have the gettext domain set15:27
mzanettitsdgeos: interesting... yep, will check it out15:27
Saviqtsdgeos, wonder if QSettings even supports that15:28
Saviqmzanetti, yeah, I'll leave that branch to you15:28
mzanettiack15:28
mzanettiwell... I'm probably gonna drop this one but redo on top of the rework branch15:28
mzanettias launcherbackend.cpp is no more15:28
mzanettitsdgeos: ok? ^15:28
dednickgreyback: you said you don't want to expose Application to qml right?15:29
greybackdednick: not via the model15:29
dednickgreyback: ok, so by Session is ok?15:29
dednicklike it's "attached to an applicaiton"15:30
greybackdednick: since you rely on session for your implementation, I guess you need to15:30
tsdgeosmzanetti: well i don't see why we should not merge this *now* and yours later15:31
tsdgeosso we can have stuff working sooner than later15:31
tsdgeosi guess it's easier to get a 5 lines fix than a 2000 lines refactor15:31
tsdgeosmore over when the bug for this has been one month ignored15:32
=== Trevinho_ is now known as Trevinho
=== dandrader is now known as dandrader|lunch
tsdgeosmzanetti: ↑↑↑15:45
=== Wellark is now known as Wellark|holiday
=== Wellark|holiday is now known as Wellark
=== boiko_ is now known as boiko
dednickdandrader|lunch: that problem should be fixed now16:10
dednicks/that/those (same issue)16:14
mzanettidandrader|lunch: hey, are you working on the bug with the lost input (touch vs mouse etch)?16:15
=== dandrader|lunch is now known as dandrader
dandradermzanetti, not anymore. proposed a qtmir fix for it last week. why?16:32
mzanettidandrader: found a way to reproduce it16:33
mzanettifor the dash at least...16:33
dandradermzanetti, might be a separate bug, but still interesting16:34
dandradermzanetti, what is it?16:34
mzanettidandrader: tap and hold the dash, not releasing, then launch another app from the launcher with the other hand16:35
mzanettidandrader: seems the touchEnd event gets lost and doesn't recover16:35
dandradermzanetti, if you press power twice to bring in the greeter, can you still swipe it away?16:37
mzanettidandrader: yeah. input only gets lost in the app16:38
mzanettiin this case16:38
mzanettiI know there's also issues with input getting lost in unity16:38
dandraderunity = unity8 or unity = unity8-dash?16:38
dandradermzanetti,16:38
mzanettiunity == unity816:39
mzanettidash == unity8-dash16:39
mzanettidandrader: interestingly: now that mouse input is stuck in the dash, I can still use the bottom edge in the dash (touch input).16:39
mzanettiso it seems to be very similar to the issue in unity16:40
dandradermzanetti, yes, it triggers the same issue in Qt16:40
mzanettidandrader: can't we just send a cancel input event on focus switches?16:40
dandradermzanetti, that would be a work around at best16:41
mzanettihm... not sure...16:41
mzanettidandrader: what would be the proper solution then?16:42
dandradermzanetti, ensure that MirSurfaceItem is forwarding a sane stream of touch events to the mir client16:42
mzanettidefine "sane stream"16:43
dandradermzanetti, maybe the existing qtmir fix for that bug also solves that. but maybe not. it has to be checked16:43
dandradermzanetti, every touch has a begin and an end16:43
mzanettiright... that's what I said, no? :D16:43
dandradermzanetti, ending a touch is different from canceling it16:44
mzanettidandrader: wait... if the user doesn't release on the dash I don't think there should be a mouse End event16:44
mzanettiin fact a cancel event is more correct in that case16:44
dandradermzanetti, let's not mix touch with mouse16:45
dandradermzanetti, once a window gets ownership of a touch it will get all events from that touch until it ends16:45
mzanettibut this specific issue I see in the dash is a mouse input issue16:45
dandradermzanetti, because qt internally generates mouse events out of the touch events it receives16:46
dandradermzanetti, check that qt review link I have on that bug16:46
dandradermzanetti, its bug 129562316:46
ubot5bug 1295623 in Unity 8 "Sometimes input breaks and only edges are responsive" [Critical,In progress] https://launchpad.net/bugs/129562316:46
dandradergreyback, Saviq, btw, when are we releasing that?  https://code.launchpad.net/~dandrader/qtmir/missingTouchEnd-lp1295623/+merge/23241016:47
greybackdandrader: I'll request a silo for it16:48
dandradergreyback, thanks!16:50
greybackdandrader: I'm gonna hang on for mir0.7 to land though16:50
greybackbut will after16:50
dandraderSaviq, dednick, prompts-in-prompts is good to go17:17
dednickdandrader: woop17:19
=== salem_ is now known as _salem
=== Pici` is now known as Pici

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