| === chihchun_afk is now known as chihchun | ||
| duflu | greyback: I'm auditing devices to see if we're ready for turning off input resampling. It's looking OK so far (mostly as the touchscreens are emitting events at the same rate as they redraw now). Do you have any knowledge of relevant software changes? | 09:30 |
|---|---|---|
| greyback | duflu: there were no software changes on my end | 09:30 |
| duflu | Oh, except we need to go slower than the refresh rate to overcome the nesting penalty :( | 09:30 |
| alan_g | greyback: is there any way I can help with the Qt integration? | 09:42 |
| greyback | alan_g: until I've the foundations set up, not really. I will propose a series of MPs today | 09:54 |
| duflu | Hmm, my touch screen test was flawed. I forgot Mir's Android platform still has https://bugs.launchpad.net/mir/+bug/1369763 | 10:06 |
| ubot5 | Launchpad bug 1369763 in Mir "[performance] Android overlays don't honour eglSwapInterval 0" [Medium,Triaged] | 10:06 |
| === hikiko is now known as hikiko|ln | ||
| greyback | alan_g: when you return, I need a hand with one thing | 12:43 |
| greyback | https://code.launchpad.net/~gerboland/miral/qt-add-window-model-qt-side/+merge/301261 | 12:43 |
| greyback | please grab that and test the "qml-demo-shell" with a client | 12:43 |
| greyback | problem is that as I'm overriding the MultiThreadedCompositor implementation, there is nothing who calls WindowManager::add_display | 12:44 |
| greyback | do you know any handy way I can get access to the WindowManger object? | 12:45 |
| === chihchun is now known as chihchun_afk | ||
| alan_g | greyback: A hack (that we would need to address later) would be to use a customized version of SetWindowManagmentPolicy that keeps a weak_ptr<> to the object it builds that you can grab. | 13:00 |
| greyback | alan_g: ack. Was what I was thinking, but wanted to run it by you | 13:01 |
| alan_g | I can thing of a couple of uglier hacks. But really its the Mir wiring that is wrong - if you can override MTC you ought to be able to do the stuff the default does. | 13:03 |
| alan_g | *it's | 13:03 |
| alan_g | greyback: a (slightly) better hack: std::dynamic_pointer_cast<AbstractShell>(server.the_shell()) | 13:27 |
| anpok | you need to replace the shell? | 13:28 |
| greyback | only just :) | 13:28 |
| anpok | the AbstractShell.. | 13:28 |
| greyback | I'm gonna rewrite it in Qt | 13:29 |
| greyback | because, you know | 13:29 |
| greyback | :D | 13:29 |
| greyback | (kidding) | 13:29 |
| anpok | Thats not funny. | 13:33 |
| alan_g | anpok: he just needs to call a method on AbstractShell and Mir doesn't make it easy | 13:36 |
| alan_g | /o\ The downcast isn't needed | 13:39 |
| alan_g | greyback: you can just grab server.the_shell() and call add_display() Shell IsA DisplayListener | 13:40 |
| greyback | alan_g: ah, not so bad | 13:40 |
| * alan_g must do penance for thinking Mir got it wrong | 13:41 | |
| anpok | hehe | 13:46 |
| anpok | inconceiveable thought | 13:46 |
| alan_g | You keep using that word. I do not think it means what you think it means. | 13:47 |
| anpok | kdub: is there an easy way to figure out if a BufferStreamId is represented by a buffer stream and not a buffer chain | 13:59 |
| anpok | kdub: I dont want the client to regularly throw and catch on resize events | 14:00 |
| greyback | alan_g: ok, I've MPs up for your reviewing pleasure. | 14:00 |
| greyback | it's all a bit rough and WiP, but the main idea is there | 14:00 |
| greyback | and gedit is usable | 14:01 |
| alan_g | greyback: thanks | 14:02 |
| kdub | anpok, no by design, because the server has no distinction between the two | 14:07 |
| kdub | its really just for the benefit of the client that there's a distiction | 14:08 |
| anpok | so they also share the same id space | 14:08 |
| kdub | right, server-side they are both mc::Stream | 14:08 |
| alan_g | greyback: nothing really bad. Looks like progress. | 14:31 |
| alan_g | But I did crash it rather easily: By connecting gnome-terminal and then starting mir_demo_client_egltriangle from that. | 14:32 |
| greyback | alan_g: ack. I'm not claiming it is very stable | 14:33 |
| greyback | the model syncing stuff isn't great | 14:33 |
| alan_g | That's something I can help with once we have a skeleton POC | 14:47 |
| === dandrader is now known as dandrader|afk | ||
| === dandrader|afk is now known as dandrader | ||
| alan_g | greyback__: not pushed qt-add-window-model-qt-side? | 16:52 |
| greyback__ | alan_g: pushed now | 16:52 |
| alan_g | ta | 16:52 |
| greyback__ | sorry | 16:52 |
| greyback__ | alan_g: just one nitpick with https://code.launchpad.net/~alan-griffiths/miral/add-indirection-to-stabilize-WindowManagerTools-ABI/+merge/301299 | 17:01 |
| alan_g | greyback__: thanks - I'll get to it in the morning. (And yes, renames work pretty well in CLion.) | 17:02 |
| greyback__ | alan_g: even -> to member access? | 17:03 |
| greyback__ | anyway, have a nice evening | 17:03 |
| alan_g | that was just a global text substitution. | 17:07 |
| === dandrader is now known as dandrader|afk | ||
| === dandrader|afk is now known as dandrader | ||
| === dandrader is now known as dandrader|afk | ||
| === dandrader|afk is now known as dandrader | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!