Effects are given the interval between two consecutive frames. The main
flaw of this approach is that if the Compositor transitions from the idle
state to "active" state, i.e. when there is something to repaint,
effects may see a very large interval between the last painted frame and
the current. In order to address this issue, the Scene invalidates the
timer that is used to measure time between consecutive frames before the
Compositor is about to become idle.
While this works perfectly fine with Xinerama-style rendering, with per
screen rendering, determining whether the compositor is about to idle is
rather a tedious task mostly because a single output can't be used for
the test.
Furthermore, since the Compositor schedules pointless repaints just to
ensure that it's idle, it might take several attempts to figure out
whether the scene timer must be invalidated if you use (true) per screen
rendering.
Ideally, all effects should use a timeline helper that is aware of the
underlying render loop and its timings. However, this option is off the
table because it will involve a lot of work to implement it.
Alternative and much simpler option is to pass the expected presentation
time to effects rather than time between consecutive frames. This means
that effects are responsible for determining how much animation timelines
have to be advanced. Typically, an effect would have to store the
presentation timestamp provided in either prePaint{Screen,Window} and
use it in the subsequent prePaint{Screen,Window} call to estimate the
amount of time passed between the next and the last frames.
Unfortunately, this is an API incompatible change. However, it shouldn't
take a lot of work to port third-party binary effects, which don't use the
AnimationEffect class, to the new API. On the bright side, we no longer
need to be concerned about the Compositor getting idle.
We do still try to determine whether the Compositor is about to idle,
primarily, because the OpenGL render backend swaps buffers on present,
but that will change with the ongoing compositing timing rework.
QGraphicsRotation and Scale are QObject wrappers. It's not useful in
data structures where we're creating mulitple of these every frame. It's
large enough to appear in hotspot as taking over 1% of a regular frame.
We don't even use the QGraphicsRotation mapping inside scene for a
reason, so it's not giving us much.
It's technically an API break in libkwineffects. Pragamatically no-one
would use these. We also lose QGraphicsScale's origin, but we never
exposed this in PaintData's public header.
On Wayland, a surface must be displayed the same way no matter how the
attached buffer is transformed. In order to guarantee that, we build the
surface-to-buffer matrix, which is used to compute the texture coords.
The surface-to-buffer matrix represents an affine transformation. Thus,
performing linear interpolation between texture coordinates won't end up
in corrupted rendered results. This is the main assumption that we make
during generation of contents window quads. After creating a sub-quad,
the new quad's texture coordinates are computed by interpolating between
the source quad's texture coords.
However, WindowQuad::makeSubQuad() makes a concrete assumption about the
order of texture coords, which might be false if the attached wayland
buffer is rotated 90 or 270 degrees.
This issue went unnoticed after merging the viewporter patches because
the developer who was working on it had been using primarily nested
kwin_wayland for testing purposes. And it appears like kwin schedules
full screen repaints even though it supports buffer age. It still needs
some investigation why that happens.
BUG: 428003
Prefer for() to foreach(), as the latter is deprecated.
Prefer iterating QList using the iteration_expression for() loops,
rather than doing it by index.
The main advantage of SPDX license identifiers over the traditional
license headers is that it's more difficult to overlook inappropriate
licenses for kwin, for example GPL 3. We also don't have to copy a
lot of boilerplate text.
In order to create this change, I ran licensedigger -r -c from the
toplevel source directory.
Summary:
The screenshot made on screens with scale factor were downscaled by their scale factor making them blurry.
It prevents taking screenshots of missing Hidpi related bugs showing the issues under Wayland.
This fix the case of a single screenshot, but not the rest:
Multiscreen screenshot downscales the screen using scale factor.
Spectacle rectangular selection screenshot is broken as soon as some scale factor different than 1 is used on any screen.
Test Plan:
Under Wayland with a scale factor on a screen, take a screenshot using spectacle.
The output image is not downscaled and has the same size as the screen resolution.
No other change to any other screenshot mode, or under X.
Reviewers: davidedmundson, #kwin
Reviewed By: davidedmundson, #kwin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D29010
Summary:
When a plasma session quits, the order of plasmashell and clients quitting is undetermined.
This looks bad as our desktop background disappears quite early. Rather than changing the real order, we can have kwin reference the last pixmap of any desktop and hold on to it.
Lasting till the compositor quits means it stays on the X frontbuffer till X quits and the experience is seamless.
Test Plan: Logged out (with relevant other ksmserver patches)
Reviewers: #kwin, zzag
Reviewed By: #kwin, zzag
Subscribers: zzag, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D24864
Summary:
Won't make things go much faster since everything that was
being passed by value is refcounted but still const & is a bit faster
than refcounting
For shared pointers instead of adding const & we move them into the
destination variable saving some cpu usage but at the same time making
clear the pointer is being stored by not being const &
Reviewers: zzag
Reviewed By: zzag
Subscribers: zzag, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D25022
Summary:
Switch to Q_ASSERT in order to make code a bit more consistent. We have
places where both assert and Q_ASSERT are used next to each other. Also,
distributions like Ubuntu don't strip away assert(), let's hope that
things are a bit different with Q_ASSERT.
Test Plan: Compiles.
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: romangg, davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D23605
Summary:
Since C++11 alignas is a keyword, so we don't need to create our own.
All relevant compilers (gcc, clang, icc) define __SSE2__, so just use that.
There seem to be frequent crashes in the quad version of this code, but
this change should not make any difference from what I can tell.
Reviewers: #kwin, zzag
Reviewed By: #kwin, zzag
Subscribers: zzag, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D23120
Summary:
There is no point in using quint32 and casting back and forth in numerous places.
Fix a bunch of compiler warnings that we implicitly cast between signed and unsigned.
This makes things consistent with what we get from libinput.
Reviewers: #kwin, romangg
Reviewed By: #kwin, romangg
Subscribers: zzag, romangg, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D23086
Summary:
7834bec52a missed to port minimize and
unminimize to dynamic dispatch. In addition to that, we don't benefit
from QMetaObject::invokeMethod so port addRepaint and addLayerRepaint
to dynamic dispatch as well.
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D18049
Summary:
EffectWindow proxies its properties from the client/deleted's
properties.
QObject::property(char*) is a slow string search. It's a loop
of string comparisons not a hash lookup!
QML's use of properties is different, there's a property cache.
It's fetched multiple times for every window in every paint of some
effects (such as blur). Hotspot shows this as a significant amount of
the render pass (X11) with nothing in kwin animating.
This patch replaces the macro that does
parent()->property("propertyName")
with a macro calling the relevant function directly without metaobjects.
This also improves type safety for future changes.
Test Plan:
Existing unit tests
Ran it for a bit
Reviewers: #kwin, graesslin
Subscribers: graesslin, zzag, broulik, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16602
Summary:
The redirect modes control behavior of the timeline when its direction
is changed at the start or target position. For example, consider the
following piece of code:
TimeLine timeLine(1000ms, TimeLine::Forward);
timeLine.setDirection(TimeLine::Backward);
What should happen when the direction of the timeline was changed to go
backward? Should the current value of the timeline go from 1 to 0, or
should the timeline stop its "execution"?
In the relaxed mode, the timeline will go from 1 to 0.
In the strict mode, the timeline will stop its execution.
Different effects may prefer different modes for source and target
positions. For example, most C++ effect would prefer relaxed mode for
source position, and strict mode for target position. On the other side,
scripted effects(AnimationEffect) would prefer strict mode for source
position, and relaxed mode for target position(because of set).
Reviewers: #kwin, graesslin
Reviewed By: #kwin, graesslin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16447
Summary:
As setDesktop was changed to "move" this left unSetDesktop non-symetric.
This replaces it with explicit API to enter/leave.
This also moves new API to the new object based API rather than still
using ints.
Where numbers are used it has been tidied up so that desktop IDs are
uint, which should be used when we have a list of desktops.
int is used only when we have either a desktop ID or NET::OnAllDesktops
(-1)
Effects API cleared up to use this and use a set of x11 IDs, which
avoids any potential complications of handling add and removes any
ambiguity with what happens if you leave all desktops and such.
Test Plan:
testVirtualDesktops passes (with pending kwayland patch)
Moving a window in the desktop grid on X11 behaves
Moving a window in the desktop grid on wayland behaves
Reviewers: #kwin, zzag
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16704
Summary:
With the plasma shell protocol, a ShellClient gets its role/window type
after it's created, so the cached value of the popupWindow property in
EffectWindow can be wrong.
In addition to that, the value of the popupWindow property still can be
wrong because plasma shell surfaces don't have xdg-popup role, so we need
to check NetWM-based window type in ShellClient::isPopupWindow.
Reviewers: #kwin, davidedmundson
Reviewed By: #kwin, davidedmundson
Subscribers: davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16835
Summary:
Type of the x11DesktopIds property was changed from QList<int> to
QVector<int> in KWin core, but not in libkwineffects. That resulted in
having effects like slide not working because Qt can't implicitly
convert QVector<int> to QList<int>.
Test Plan: Slide effect works again.
Reviewers: #kwin, graesslin
Reviewed By: #kwin, graesslin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D16743
Summary:
implement virtual desktop support for Wayland.
use the new virtual desktop protocol from D12820
The VirtualDesktopManager class needed some big change in order
to accomodate it, which is where most changes are.
Other than that, it's mostly connections to wire up
VirtualDesktopsManager and VirtualDesktopsManagement(the wayland protocol impl)
Depends on D12820
Other notable detail, is the client visibility updated to reflect the presence
of the client in the plasmavirtualdesktop.
(and the unSetDesktop concept)
Test Plan: used a bit a plasma session together with D12820, D13748 and D13746
Reviewers: #plasma, #kwin, graesslin, davidedmundson
Reviewed By: #plasma, #kwin, davidedmundson
Subscribers: hein, zzag, davidedmundson, kwin
Tags: #kwin
Maniphest Tasks: T4457
Differential Revision: https://phabricator.kde.org/D13887
Summary:
The Scale effect and the Glide effect have to animate only ordinary
windows(i.e. the ones that are considered to be apps).
On X11, in order to distinguish ordinary windows from combo box popups,
popup menus, and other popups, those effects check whether given window
is managed.
On Wayland, there is no concept of managed/unmanaged windows.
XDG Shell protocol defines 2 surface roles:
* xdg_toplevel;
* and, xdg_popup.
The former can be used to implement typical windows, the ones that can
be minimized, maximized, etc.
The latter can be used to implement tooltips, popup menus, etc. Thus,
that's a good criteria to filter popup windows.
CCBUG: 398100
Reviewers: #kwin, graesslin, davidedmundson
Reviewed By: #kwin, graesslin, davidedmundson
Subscribers: davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D15117
Test Plan:
Won't make a practical difference, anything that used before
would have crashed. Only now it's tidier.
Unit tests still pass
Reviewers: #kwin
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D15171
Summary:
Most effects use QTimeLine in the following manner
```lang=cpp
if (...) {
m_timeline->setCurrentTime(m_timeline->currentTime() + time);
} else {
m_timeline->setCurrentTime(m_timeline->currentTime() - time);
}
```
Because effects do not rely on a timer that QTimeLine has, they can't
toggle direction of the QTimeLine, which makes somewhat harder to write
effects. In some cases that's obvious what condition to use to figure
out whether to add or subtract `time`, but there are cases when it's
not. In addition to that, setCurrentTime allows to have negative
currentTime, which in some cases causes bugs.
And overall, the way effects use QTimeLine is really hack-ish. It makes
more sense just to use an integer accumulator(like the Fall Apart
effect is doing) than to use QTimeLine.
Another problem with QTimeLine is that it's a QObject and some effects
do
```lang=cpp
class WindowInfo
{
public:
~WindowInfo();
QTimeLine *timeLine;
};
WindowInfo::~WindowInfo()
{
delete timeLine;
}
// ...
QHash<EffectWindow*, WindowInfo> m_windows;
```
which is unsafe.
This change adds the TimeLine class. The TimeLine class is a timeline
helper that designed specifically for needs of effects.
Demo
```lang=cpp
TimeLine timeLine(1000, TimeLine::Forward);
timeLine.setEasingCurve(QEasingCurve::Linear);
timeLine.value(); // 0.0
timeLine.running(); // false
timeLine.done(); // false
timeLine.update(420);
timeLine.value(); // 0.42
timeLine.running(); // true
timeLine.done(); // false
timeLine.toggleDirection();
timeLine.value(); // 0.42
timeLine.running(); // true
timeLine.done(); // false
timeLine.update(100);
timeLine.value(); // 0.32
timeLine.running(); // true
timeLine.done(); // false
timeLine.update(1000);
timeLine.value(); // 0.0
timeLine.running(); // false
timeLine.done(); // true
```
Test Plan: Ran tests.
Reviewers: #kwin, davidedmundson, graesslin
Reviewed By: #kwin, davidedmundson, graesslin
Subscribers: romangg, graesslin, anthonyfieroni, davidedmundson, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D13740
Summary:
When windowClosed signal is emitted, effects can't distinguish managed
windows from unmanaged windows(e.g. combo box popups, popup menus, etc).
This leads to dirty hacks like IsXXXWindow. Also, there's a big chance
that such hack can introduce more bugs and overall this makes harder to
write/maintain effects.
This change proposes to save value of managed property during
construction of EffectWindow. So, its value is preserved with Deleted.
Test Plan: Manually.
Reviewers: #kwin, graesslin
Reviewed By: #kwin, graesslin
Subscribers: graesslin, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D13690
Summary:
EffectWindow has keepAbove property, but not keepBelow.
This change adds keepBelow property as a counterpart to keepAbove.
Test Plan: Manually.
Reviewers: #kwin, mart
Reviewed By: #kwin, mart
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D13650
Summary:
When one uses:
* breeze as of 5.12
* wobbly windows
* shaded window
* a distribution building with assert enabled
and starts to move a shaded window, KWin asserts. The root cause for
this is that WindowQuad::makeSubQuad has an assert for y1 being smaller
than y2. With the combination listed above this is not guaranteed. For
the left shadow quad the y1 and y2 are identical and thus trying to
split it, results in the assert condition.
The problem of the shadow quad having an invalid size might be addressed
as well with D10811. Due to that the generation of the quads is not
touched. Instead a sanity check is introduced to not try to split
already invalid sized quads.
BUG: 390953
FIXED-IN: 5.12.3
Test Plan: Added unit test hit the assert, now doesn't hit it any more
Reviewers: #kwin, #plasma
Subscribers: kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D11015
When an application is not responding, its window is desaturated to communicate this.
Also "(Not Responding)" is added to the title bar.
Differential Revision: https://phabricator.kde.org/D5245
Summary:
The effectsConfig is no longer used by any Effect. ScriptedEffects use
KWin's KSharedConfigPtr and the builtin effects all have a kcfg with
kwinrc being hard coded.
The effectsConfig had the big disadvantage that it hardcoded kwinrc
while the rest of kwin uses a KSharedConfigPtr which might not point to
kwinrc.
Reviewers: #kwin, #plasma
Subscribers: plasma-devel, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D3384
Summary:
On Wayland per output rendering is performed and paintScreen is invoked
for every output. Some effects need the information which output is
currently being rendered as otherwise e.g. FBO access could fail.
This change adds the current output geometry to ScreenPaintData. On X11
(all outputs one geometry) this information is not set and a null rect
is returned. That way the effects can also easily check which rendering
mode is used.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D3058
Summary:
So far TabBox used highlight windows by passing window ids around through
an X property. This doesn't work on Wayland where we don't have window
ids for our TabBox and the Wayland windows.
This change introduces a new Effect::Feature for HighlightWindows which
the HighlightWindowsEffect provides. The EffectsHandlerImpl has a new
method to highlightWindows which it delegates to that effect if it is
loaded by invoking a new performFeature method.
The TabBoxHandler now passes the highlighting to the effects system
instead of updating the x11 property. Thus this works on Wayland and
at the same time improves the X11 side by no longer having to go through
the property protocol.
Test Plan: Verified that Alt+Tab highlights the windows on Wayland correctly.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D2630
Summary:
The Effect class is extended by three new virtual methods:
* touchDown
* touchMotion
* touchUp
The methods return a boolean value so that the events can be filtered
out. E.g. an effect which has also a mouse grab installed wants to
filter out all events, other effects don't need the events exclusively.
This is a difference to how e.g. keyboard and pointer events are handled.
But is more close to how KWin's internal input event passing works and
makes it easier to get touch event: one does not explicitly has to grab
the events. It's also closer to Wayland where all input events are
available.
As a first example the Present Windows effect is adjusted and allows to
activate windows through the touch screen. As much code as possible is
shared with pointer input.
Reviewers: #kwin, #plasma_on_wayland
Subscribers: plasma-devel, kwin
Tags: #plasma_on_wayland, #kwin
Differential Revision: https://phabricator.kde.org/D2450
The screen projection matrix is needed if an effect wants to customize
the modelview projection matrix while rendering a window and keeping
the currently applied screen projection and transformation.
With this change a new ctor overload is added to ScreenPaintData
which allows passing a projection matrix through the effects. This
allows effects to put up custom shaders with a shared projection matrix
and without having to calculate it themselves.
The projection matrix is a read-only information for the effects. There
is no way to change or overwrite it.
-use qstringliteral only when necessary (i.e. not in concat or comparison)
-use qbytearray instead of qstring when dealing with latin1 input and output (glplatform)
-use qstringref to extract numbers from strings (glplatform)
-define qt_use_qstringbuilder to optimize all string concatenations
-anidata: use ctor init lists, add windowType member initialization
REVIEW: 125933
A quad might have a null size in case it has a null width or null height.
Not just one dimension.
BUG: 236353
BUG: 210467
FIXED-IN: 5.4.2
REVIEW: 125131
