[03:58] <RAOF> Fun fact! mg::DisplayConfigurationOutput will be full of unititialised values if connected=false!
[04:20] <bschaefer> RAOF, yup ran into that before
[04:20]  * bschaefer just assumed it was undefined behavior 
=== balloons is now known as Guest87130
=== chihchun is now known as chihchun_afk
[11:33] <anpok> alan_g: the needs fixing in https://code.launchpad.net/~andreas-pokorny/mir/probe-libinput-platform/+merge/273046 was for the update of the symbols.map, the not yet existing am-i-root check or the mircommon pormotion discussion?
[11:35] <alan_g> anpok: the "Needs Fixing" one was the the symbols.map comment. The others are just comments.
=== alan_g is now known as alan_g|lunch
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
[12:58] <dandrader> alan_g|lunch, "[...] probably better to call override_the_cursor() [...]" you mean overriding mir::Server::the_cursor()?
=== alan_g|lunch is now known as alan_g
[13:12] <alan_g> dandrader: I mean calling mir::Server::override_the_cursor()
[13:12] <dandrader> alan_g, mir::Server doesn't have a override_the_cursor() method (at least not in version 0.16).
[13:15] <alan_g> dandrader: oh. (It takes so long for a release to hit archive I sometimes lose track)
[13:19] <alan_g> dandrader: BTW the only reason hide() is working for you at all is that you're bypassing Mir input and as a result that doesn't e.g. update the cursor position.
[13:20] <dandrader> alan_g, well, I see the mir cursor moving around
[13:21] <dandrader> alan_g, any way I can override the cursor in mir 0.16?
[13:21] <alan_g> not with the published APIs. You need -c 2953
[13:24] <alan_g> dandrader: I'll have a think. But we may want to roll a 16.2 with -c 2953 (and some other stuff)
[13:24] <dandrader> alan_g, ok, thanks
=== davmor2_ is now known as davmor2
[14:32] <alan_g> dandrader: I was wrong. The reason it seems to work is that the cursor never changes (e.g. by mouse over a surface that has set the cursor) - there's an optimization that avoids the show() I was thinking of. Not quite sure why your hot-plugging scenario leads to an update. Calling hide() again might be a workaround.
[14:40] <dandrader> alan_g, pushed a workaround for it some minutes ago. I call hide() on compositor::start()
[14:40] <dandrader> alan_g, which happens everytime you connect or disconnect a monitor
[14:40] <dandrader> alan_g, or so it seems
[14:43] <alan_g> dandrader: yes. we (currently) restart compositing on display config changes.
=== dandrader is now known as dandrader|afk
[15:03] <alan_g> dandrader|afk: on a related point, why can't you use the Mir cursor? Is there functionality missing?
[15:06] <alan_g> greyback: do you know? ^
[15:17] <greyback> alan_g: missing functionality. Not major things, but a bunch of minor things we'd like, e.g. pointer barrier, better rotation support, pointer trapped on a single screen in multimonitor case...
[15:18] <greyback> we had thought to implement pointer acceleration ourselves too, but libinput will fix that
=== dandrader|afk is now known as dandrader
[15:36] <anpok> greyback: will you really need to imeplement pointer barrier .. or added friction to screen boundaries?
[15:37] <greyback> anpok: not in the recent future
[15:44] <alan_g> greyback: so that leaves "better rotation support" and "pointer trapped on a single screen" as the key issues currently?
[15:45] <greyback> alan_g: yes, those are only things coming to mind short-term
[15:46] <greyback> dandrader: did I forget anything? ^^
[15:48] <dandrader> sound ok
[15:48] <dandrader> sounds
[15:48] <alan_g> What does "better" mean for rotation support?
[15:49] <alan_g> 90 degree rotations to match orientation?
[15:55] <dandrader> alan_g, when shell ui rotates, mouse pointer should rotate accordingly (rotate image and movement axes)
[15:56] <dandrader> alan_g, we get it for free and pretty neatly when the mouse pointer is just another qml item in shells scene
[15:56] <dandrader> alan_g, but with he mouse pointer being an extraneous element, it's a bit of a pain
[15:57] <greyback> rotating mouse in mir right now requires display configuration update, which is hard for unity8 to deal with at runtime
[15:58] <greyback> as it resizes the scene
[15:58] <alan_g> OK, I'm not yet familiar with all the issues, but have something to go on.
[17:10] <anpok> dandrader, alan_g: afaik rotation does not work with drm/mesa... while with android the cursor is done as a scene element so there rotation should work..
[17:10] <anpok> (rotation of the cursor ..)
=== alan_g is now known as alan_g|EOD
[17:11] <dandrader> anpok, so on android mir does not provide a hardware cursor?
[17:11] <dandrader> anpok, (or android doesn't have hardware cursors)
[17:17] <anpok> dandrader: it is using overlay hardware for the cursor when possible
[17:18] <anpok> it is just already present as scene content.. hence the scene rotation will take care of the detail.. (either by rotating when falling back to gl, or by applying the attributes to the overlays)
[17:18] <anpok> for mesa drm it is just something separate.. but yes some day we will switch to drm planes and the same will be true there
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
[19:58] <robert_ancell> Saviq, what video card did you have for bug 1501941?
[19:58] <ubot5`> bug 1501941 in xserver-xorg-video-intel (Ubuntu) "Screen turned off after X server exits" [High,Triaged] https://launchpad.net/bugs/1501941
[19:58] <robert_ancell> olli, also for you?
[20:03] <olli> robert_ancell, we have the same XPS12
[20:05] <robert_ancell> olli, can you paste the result of "grep Intel /var/log/Xorg.0.log"
[20:07] <robert_ancell> olli, thanks. You have a 4400 like me.
=== ljp is now known as lpotter