[03:57] <duflu> robert_ancell: Hey just FYI, that scrolling speed problem with Xmir, while it has a fix may not get Fix Released as quickly as Xmir itself. We might end up with more people seeing the problem for a while till the next Unity8 update
[03:58] <robert_ancell> duflu, is that going to affect the SRU to Xenial?
[03:59] <duflu> robert_ancell: Yes, probably. But only for users of X apps in Unity8 on xenial. I'm not sure that's a major audience
[03:59] <robert_ancell> duflu, probably not
[04:00] <duflu> I'm just warning everyone so they expect the duplicate bug reports when they happen
[04:00] <duflu> It is however a bug that already has a fix
[10:54] <alan_g> greyback: do the #pragmas have a purpose? g++-6 rightly rejects them, and I don't find a need anywhere. https://code.launchpad.net/~alan-griffiths/miral/compatibility-with-mir-trunk/+merge/302137
[10:55] <greyback> alan_g: it used to silence compiler warnings due to the UAL header files
[10:56] <greyback> which made compilation with -Werror fail
[10:56] <greyback> that situation may have been fixed now
[10:56] <greyback> though probably not for the qt private header
[10:57] <alan_g> Where should I compile to check? V+O and yakkety desktop build clean
[10:58] <greyback> alan_g: yep, v+o most likely culprit
[11:00] <alan_g> So they're just obsolete.
[11:02] <greyback> quite probably
[14:17] <aquarius_> Phone question: how can I launch an application on the phone if Unity 8 isn't running?
[14:17] <aquarius_> Unity 8 does some jiggery-pokery with setting up its own mir sockets and so on, as far as I can tell...
[14:20] <alan_g> aquarius_: what are you using as a shell if you're not using U8?
[14:21] <aquarius_> alan_g: at the moment, assume nothing and that I'm launching apps from the command line. (I'm exploring the idea of a different shell, but the first part is to have some way I can launch apps at all without unity around :))
[14:22] <alan_g> well, if you mean phone apps (not command line programs) they need a shell/mir server to talk to
[14:23] <alan_g> It would be like launching an X app without an X server
[14:23] <alan_g> Doomed
[14:23] <aquarius_> ya, graphical apps -- qml, whathaveyou. So, question: an app can't launch and talk to mir itself? That's not how Mir works, I take it; something needs to be the "display server", and that's Unity 8?
[14:23] <anpok> you can launch unity-system-compositor
[14:23] <aquarius_> so if I don't have unity running, I need to write a separate "dummy window manager" to act as a shell?
[14:23] <anpok> or you might have it still running..
[14:24] <alan_g> U8 is the Mir server
[14:24] <aquarius_> ah! unity-system-compositor is an alternative mir server?
[14:24] <anpok> u-s-c owns the screen and unity8 connects as a fullscreen client to it
[14:25] <alan_g> aquarius_: you might find lp:miral a good starting point for an alternative server
[14:25] <anpok> but in general it seems that you should write an alternative server
[14:25] <anpok> u-s-c could serve as an intermediate replacement but it only accepts full screen clients
[14:26] <alan_g> But IIUC you can find what you need in the "running the software" bit of this: http://www.octopull.co.uk/sw-dev/WorkingOnAPhone.html
[14:26] <aquarius_> full screen clients are fine for my use case (which is very simple at the moment! It might get more complex later, but i'm just exploring)
[14:26] <anpok> so you are better of launching miral-shell from lp:miral as a client to u-s-c or as a u-s-c replacement.
[14:27] <aquarius_> so, if I have u-s-c running, I should be able to launch (for example) qmlscene and it should work?
[14:27] <alan_g> No
[14:28] <aquarius_> ah, so I still need miral-shell or similar running as well?
[14:28] <alan_g> USC is very fussy about what connects to it.
[14:29] <alan_g> aquarius_: have a skim through the above article - you don't need to build things, but it also covers getting control of the screen
[14:29] <aquarius_> am indeed reading it as we speak :)
[14:29] <ogra_> wasnt there a tiling WM demo in the mir code somewhere ?
[14:29] <ogra_> you might be able to spawn that and run your app in it
[14:31] <alan_g> ogra_: there is, there's also a non tiling one. (But the one in MirAL is a better starting point)
[14:37] <aquarius_> alan_g: that demo isn't on the phone already, I imagine?
[14:37] <aquarius_> Certainly I'll take a look at miral, which seems like the right thing to do
[14:37] <aquarius_> I've just gotta set up a build environment and whatnot to play around with that :)
[14:38] <alan_g> aquarius_: is isn't installed by default. But if you've e.g. a chroot set up you can install it
[14:38] <aquarius_> kk
[14:38] <aquarius_> that makes sense
[14:38] <aquarius_> I was just sneaily hoping to avoid work ;)
[14:38] <alan_g> Or if you've made thing RW you can just use apt
[14:39]  * alan_g doesn't recommend doing that though
[14:39] <aquarius_> yeah. As it happens I have made this RW so I can experiment with things, but I don't really want to install loads of mad stuff and break the system images just yet
[14:41] <alan_g> aquarius_: in that case installing mir-demos shouldn't do much harm. And gives you example servers and apps to play with
[14:44] <aquarius_> oh really? that sounds useful
[14:44] <alan_g> $ mir_demo_server --test-client mir_demo_client_egltriangle
[14:46] <alan_g> But to get serious set up a chroot
[14:46] <aquarius_> certainly I will do when I get serious :)
[14:46] <ogra_> when will *that* ever happen ?
[14:47] <aquarius_> ha! shush, ogra ;)
[14:47] <ogra_> :)
[14:47] <alan_g> when the first phone bricks
[14:47] <aquarius_> hm, mir-demos isn't available in the repos. do I need to have a different archive enabled?
[14:47] <aquarius_> the phone points at vivid stuff
[14:48] <alan_g> oh, I guess that's in universe
[14:48] <aquarius_> it is. bah :)
[14:49] <aquarius_> and vivid stuff isn't available in the archives any more. Can I use the wily version, or does it change in lockstep with newer mir?
[14:50] <aquarius_> huh. libmirclient9 on the phone is the version from yakkety. So mir stuff is more up to date than it pointing at "vivid" archives would imply :)
[14:51] <anpok> vivid stuff is in the stable-phone-overlay ppa
[14:51] <anpok> and xenial stuff too..
[14:53] <aquarius_> huh, darn. mir-demos 0.21 and 0.23 will both not install for dependency reasons
[14:58] <aquarius_> I bet I have to be running rc-proposed on the phone to match the version of mir-demos in yakkety.
[14:59] <alan_g> Don't mix with yakkety - the gcc isn't compatible
[14:59] <alan_g> Add "universe" in /etc/apt/sources.list, apt update && apt-install mir-demos
[15:00] <alan_g> Or is that too serious?
[15:00] <aquarius_> I already have that, though
[15:00] <aquarius_> deb http://ports.ubuntu.com/ubuntu-ports/ vivid universe
[15:00] <alan_g> Shouldn't that be deb http://ports.ubuntu.com/ubuntu-ports vivid main universe
[15:01] <aquarius_> deb http://ports.ubuntu.com/ubuntu-ports/ vivid-updates main restricted is in there too
[15:02]  * alan_g is out of suggestions
[15:03] <aquarius_> I suspect this is because mir-demos *was* in vivid universe, but vivid is out of support now :)
[15:04] <aquarius_> (checks packages.gz file)
[15:04] <aquarius_> hey! it's there.
[15:04] <aquarius_> mir-demos, 0.12.1+15.04.20150324-0ubuntu1
[15:05] <aquarius_> but apt-cache search doesn't find it. That's weird.
[15:05] <alan_g> It ought to be in the stable-phone-overlay ppa
[15:05] <alan_g> Probably 0.23
[15:06] <aquarius_> hm, have I ever actually run apt update on this device?
[15:07] <alan_g> that would help
[15:07] <aquarius_> it would rather, wouldn't it :)
[15:08] <aquarius_> and there it is. Good one, Langridge, you moron
[15:08] <aquarius_> installed :)
[15:09] <aquarius_> ERROR: Dynamic exception type: St12system_error
[15:09] <aquarius_> std::exception::what: Enable multithreading to use std::thread: Operation not permitted
[15:12] <aquarius_> https://bugs.launchpad.net/mir/+bug/1460868 seems relevant here
[15:13] <alan_g> aquarius_: that's familiar (a bug - IIRC there's an LD_PRELOAD workaround)
[15:13] <alan_g> not that bug
[15:13] <aquarius_> the bug sorta suggests that i need a mir-graphics-drivers package installed, but I have mir-graphics-drivers-android
[15:14] <alan_g> https://bugs.launchpad.net/mir/+bug/1577357/comments/5
[15:14] <aquarius_> aha!
[15:14] <aquarius_> LD_PRELOAD=libpthread.so.0
[15:14] <aquarius_> woo, spinning triangle :)
[15:15] <aquarius_> very very flickery spinning triangle.
[15:15] <alan_g> You "forgot" $ sudo stop lightdm
[15:15] <aquarius_> ah, because unity-system-compositor is still running. I don't want that, right?
[15:16] <alan_g> they fight
[15:16] <aquarius_> if I stop lightdm entirely, then the screen goes black
[15:16] <aquarius_> and running egltriangle doesn't show anything on screen
[15:17] <alan_g> mirbacklight
[15:17] <aquarius_> ah, as per the doc
[15:17] <aquarius_> I did forget that this time :)
[15:17] <aquarius_> winner!
[15:17] <aquarius_> that works.
[15:18] <alan_g> :)
[15:18] <aquarius_> brill. Now to convince qmlscene to start up :)
[15:19] <aquarius_> (thank you, alan_g, this is really helpful, btw!)
[15:19] <alan_g> yw
[15:21] <aquarius_> so the mir_demo_server is enough to run mir-understanding clients on top, excellent; is there a layer wich goes in between a "normal" app (say, qmlscene, or a compiled Qt app) and mir-demo-server, or does Qt know how to talk to Mir itself?
[15:21] <anpok> aquarius_: QT_QPA_PLATFORM=ubuntumirclient needs to be set
[15:22] <anpok> and the user needs to have read write access rights on the MIR_SOCKET opened by the server..
[15:22] <alan_g> Or more generally: https://bazaar.launchpad.net/~mir-team/miral/trunk/view/head:/miral-shell/miral-run.sh
[15:22] <anpok> or use the --launch-client <clientbinary> functionality of the various servers.. to avoid having a socket file
[15:23] <aquarius_> I'm using --test-client
[15:24] <alan_g> True. --launch-client also deals with the environment variables
[15:24] <anpok> then qmlscene still needs QT_QPA_PLATFORM
[15:24] <anpok> oh does it now?
[15:24] <anpok> ok
[15:25] <alan_g> That got added in 0.22
[15:26] <aquarius_> progress!
[15:27] <aquarius_> LD_PRELOAD=libpthread.so.0 mir_demo_server --launch-client simp.sh, where simp.sh is QT_QPA_PLATFORM=ubuntumirclient qmlscene /opt/click.ubuntu.com/org.kryogenix.fontbrowser/current/main.qml , gives me a tiny mouse pointer in the top left of the screen :)
[15:28] <aquarius_> ah. org.freedesktop.DBus.Error.NoServer: Failed to connect to socket /tmp/dbus-DTpyAMgnHQ: Connection refused
[15:29] <aquarius_> and indeed there is no dbus socket there
[15:29] <aquarius_> although $DBUS_SESSION_BUS_ADDRESS is set. Interesting.
[15:30] <alan_g> Prefix with dbus-run-session?
[15:31] <aquarius_> ha haaaaaa!
[15:31] <alan_g> Or, maybe, go for an app that doesn't use dbus?
[15:31] <aquarius_> and I have an app running! with no unity 8 at all! score. :)
[15:31] <aquarius_> brilliant.
[15:32] <aquarius_> thank you very much alan_g and anpok
[15:33] <aquarius_> it does still have a tiny mouse pointer :)
[15:33] <anpok> and you dont want that?
[15:33] <aquarius_> nope. this is on the phone
[15:35] <alan_g> aquarius_: tough. This is an *example* of how to use the mir library, not a fully function mir server.
[15:35]  * aquarius_ laughs
[15:35] <aquarius_> it wasn't a complaint
[15:35] <aquarius_> I have gone from zero to demonstrable in about fifteen minutes with your help, which was exactly what I wanted. Doing this properly needs more than just wiring together existing demos
[15:36] <aquarius_> hence I shall look into miral and so on!
[15:37] <alan_g> Good luck
[15:37] <aquarius_> really helpful. Thank you, people.
[15:41] <alan_g> anpok: that make me think MirAL ought to have a way to detect pointing devices and a way to disable the cursor.
[15:41] <anpok> yes
[16:10] <alan_g> anpok: https://code.launchpad.net/~alan-griffiths/miral/more-todo/+merge/302301
[17:24] <anpok> alan_g: :)..