initial WebView geometry incorrect when launched on screen with dynamic grid units

This *might* be qtubuntu's fault. I fear when webbrowser starts while monitor attached, it believes its window is on the phone screen initially, so uses that scale. While setting up its window, then it is informed the window is moved to the external display, and the scale is then corrected.

I need to learn if oxide is able to deal with is window being moved between screens while it is starting up. Its geometry may not be re-calculated...

I'll also try to have Oxide placed on the correct surface at absolute startup. That'll be a qtmir/qtubuntu job, I'm look at it now.

Here is my debugging output (I manually added printfs to qtubuntu wherever scale is requested). I am also not seeing the "screenPropertyChanged" from qtubuntu, probably also a bad thing.

Loading module: 'libubuntu_application_api_touch_mirclient.so.3.0.0'
Added Screen with id 1 and geometry QRect(0,0 768x1280)
Added Screen with id 2 and geometry QRect(768,0 1920x1080)
=======================================
UbuntuScreen(0xb814ec20) - id: 1 geometry: QRect(0,0 768x1280) form factor: phone scale: 2.25
UbuntuScreen(0xb8153380) - id: 2 geometry: QRect(768,0 1920x1080) form factor: monitor scale: 1
=======================================
could not open containers config file "/home/phablet/.local/share/libertine/ContainersConfig.json"
XX: windowProperyuChanged - scale
XX: windowProperty - scale 2.25
Input device added: "keypad_8064" "/dev/input/event1" QFlags(0x1)
Input device added: "pmic8xxx_pwrkey" "/dev/input/event0" QFlags(0x1)
Input device added: "apq8064-tabla-snd-card Button Jack" "/dev/input/event3" QFlags()
Input device added: "apq8064-tabla-snd-card Headset Jack" "/dev/input/event4" QFlags(0x20)
Input device added: "touch_dev" "/dev/input/event2" QFlags(0x8)
Input device added: "hs_detect" "/dev/input/event5" QFlags(0x1|0x20)
Input device added: "unity8-simulated-mouse" "/dev/input/event6" QFlags(0x2)
XX: windowProperyuChanged - scale
XX: windowProperty - scale 1
ubuntumirclient: Got invalid serialized mime data. Ignoring it.
XX: nativeResourceForScreen - scale 2.25
XX: nativeResourceForScreen - scale 2.25
XX: nativeResourceForScreen - scale 2.25
XX: nativeResourceForScreen - scale 2.25
qml: Loaded 8 UA override(s) from file:///usr/lib/arm-linux-gnueabihf/qt5/qml/Ubuntu/Web/ua-overrides-mobile.js
XX: nativeResourceForScreen - scale 1
XX: nativeResourceForScreen - scale 1
XX: nativeResourceForScreen - scale 1
XX: nativeResourceForScreen - scale 1
XX: nativeResourceForScreen - scale 1
XX: nativeResourceForScreen - scale 1

> I am also not seeing the "screenPropertyChanged" from qtubuntu, probably
> also a bad thing.
Nope, this is fine. No Screen properties actually change here

This is the current behaviour of qtubuntu:

1. webbrowser launches & creates a window
2. Qt creates window on the "primary" screen - which is the first screen it ever saw. This is the phone screen.
3. Oxide reads the current screen, and the scale parameter for it: 2.25 here
4. Once the window is shown, Unity8 places the window on the external screen
5. Webbrowser & oxide get notified their window is *moved* to the external screen
6. Oxide reads the current screen, and the scale parameter for it: 1.0 now

My current suspicion is that the Oxide view isn't resizing when its screen is changed. Testing this is tough for me atm, I'm curious if a patch like

+++ qt/quick/oxide_qquick_contents_view.cc
void ContentsView::screenChanged(QScreen* screen) {
  screenChangedHelper(screen);

  if (!proxy()) {
    return;
  }

  proxy()->screenUpdated();
+ proxy()->wasResized();
}

might have an impact. I hope Chris can try this out soon.

A valid question would be: why can't Oxide/Webbrowser know what screen its window will be placed on immediately, or at least sooner? This is something I'm also thinking about

http://bazaar.launchpad.net/~oxide-developers/oxide/oxide.trunk/revision/1458 appears to do the trick

I can confirm this fix works! Thanks for looking into it Chris