QTRY_COMPARE doesn't work well with how we do our wayland event
dispatching.
We know the client hasn't processed any events yet, so we can safely do
a normal wait.
On X Kwin's lifespan used to be managed by XSMP. This accidentally broke
in 5.20 as we made kwin start before ksmserver for speed purposes.
This leaves kwin to be killed by the display manager as the X connection
closes, but this can lead to deadlocks.
BUG:428817
On wayland kwin's lifespan is mapped to the lifespan of the ksmserver
binary. This is problematic as it makes our entire xwayland robustness
redundant if we ultimately rely on an application that /has/ to run in
xwayland.
BUG: 427688
On wayland, we know we're always going to load our internal QPA. Instead
of shipping a plugin and loading it dynamically we can use Qt static
plugins.
This should result in slightly faster load times, but also reduce the
number of moving pieces for kwin.
This also prevents anyone outside kwin loading our QPA which wouldn't
have made any sense and just crashed.
On X11, it creates a dead pixel on the left edge which interferes with
left panels, the left-nost widget on bottom and top panels, and scroll
events on the left-most pixel in various apps.
It works properly on Wayland, but unfortunately turning it off by
default only on X11 but not Wayland is not easily doable. So let's turn
it off by default everywhere.
BUG: 387775
FIXED-IN: 5.21
Various fixes:
- Remove unneeded PlasmaCore import
- Re-arrange imports to be alphabetical and grouped consistently
- Use standard import names
- Use Kirigami units explicitly
- Remove unused ids
- Remove unneeded cases of setting aproperty to its default value
- Use Layouts where appropriate instead of lots of anchors
- Reduce number of unnecessary items
- Fix inappropriate mutation of `width` and `height` inside layouts
Visual changes are extremely minor and of a bugfixy nature.
BUG: 428883
FIXED-IN: 5.21
We should use the geometry restore to check the full screen workspace
to make sure the window is on the right screen.
Step:
1.plug extend display, change to Extend mode
2.start a window, move to extend display, change to fullscreen
3.unplug the extend display
4.re-plug the extend display
Now, the full screen window do not restore to the extend display
if the dock is on the top,and the dock is not close together with screen edge.
for example,the dock is 10 pixels taller then sceen edge,the "position" will be "Bottom".
Autually,the dock is on the top of screen.
Correctly update the thumbnails' border sizes when the theme
changes and `Theme's Default` is selected,
as different themes may have different border defaults.
Before 9cf52340e, this was implicitly done on QML side.
Do it explicitly now on C++ side.
Use a single combobox instead of checkbox + combobox.
Adds a new property that acts as a proxy for the configuration value,
while internal configuration properties are kept unchanged.
BUG: 426157
BUG: 417430
FIXED-IN: 5.21
Slight UX and code improvents:
- Disable the detection button when property sheet is open.
- Add a minimum limit (smallDuration) to the delay before detecting
a window. Otherwise we freeze the window before even showing the
button release feedback, which can feel unexpected
- Prefer Q_INVOKABLE over public slot to expose method to QML
Currently, the OpenGLBackend and the QPainterBackend have hooks to
indicate the start and the end of compositing cycle, but in both cases,
the hooks have different names. This change fixes that inconsistency.
The purpose of Compositor::aboutToSwapBuffers() is notify the compositor
about a pending buffer. It's totally safe to call it multiple times if you
have multiple outputs. Ideally, it should also take an AbstractOutput or
screen id, but with the introduction of render loops per each output, we
won't need Compositor::aboutToSwapBuffers(). So, for the time being, this
change drops the annoying assert to make necessary per screen rendering
related refactorings easier to perform.
In order to allow per screen rendering, we need the Compositor to be
able to drive rendering on each screen. Currently, it's not possible
because Scene::paint() paints all screen.
With this change, the Compositor will be able to ask the Scene to paint
only a screen with the specific id.
Qt 5.15 introduced new syntax for defining Connections. Fix warnings like this one:
QML Connections: Implicitly defined onFoo properties in Connections are deprecated. Use this syntax instead: function onFoo(<arguments>) { ... }
All platforms that provide support for the QPainter render backend use
per screen rendering. Since there is no any way to test Xinerama-style
rendering, it's better to drop the dead code.
A window id generated by WaylandServer may reference an X11 window
with the same id, which can result in undefined behavior.
The main reason why we needed windowId() was because of the task
switcher. However, since tabbox uses internal ids now, the window id
property can be dropped.
For builtin effects the information comes from the EffectData struct.
For plugin effects the X-KDE-ConfigModule key is read from the plugin metadata.
For plugins that do not yet make use of it we fall back to the old way.
For scripted effects this is empty since the config is loaded in a different way.
Add a new field to EffectData that stores an effect's config module.
We currently determine an effect's configModule using KPluginTrader and the X-KDE-ParentComponents metadata.
IMO it's much more straight forward to let the effect specify its config module directly instead.
