Window management features were written with synchronous geometry
updates in mind. Currently, this poses a big problem on Wayland because
geometry updates are done in asynchronous fashion there.
At the moment, geometry is updated in a so called pseudo-asynchronous
fashion, meaning that the frame geometry will be reset to the old value
once geometry updates are unblocked. The main drawback of this approach
is that it is too error prone, the data flow is hard to comprehend, etc.
It is worth noting that there is already a machinery to perform async
geometry which is used during interactive move/resize operations.
This change extends the move/resize geometry usage beyond interactive
move/resize to make asynchronous geometry updates less error prone and
easier to comprehend.
With the proposed solution, all geometry updates must be done on the
move/resize geometry first. After that, the new geometry is passed on to
the Client-specific implementation of moveResizeInternal().
To be more specific, the frameGeometry() returns the current frame
geometry, it is primarily useful only to the scene. If you want to move
or resize a window, you need to use moveResizeGeometry() because it
corresponds to the last requested frame geometry.
It is worth noting that the moveResizeGeometry() returns the desired
bounding geometry. The client may commit the xdg_toplevel surface with a
slightly smaller window geometry, for example to enforce a specific
aspect ratio. The client is not allowed to resize beyond the size as
indicated in moveResizeGeometry().
The data flow is very simple: moveResize() updates the move/resize
geometry and calls the client-specific implementation of the
moveResizeInternal() method. Based on whether a configure event is
needed, moveResizeInternal() will update the frameGeometry() either
immediately or after the client commits a new buffer.
Unfortunately, both the compositor and xdg-shell clients try to update
the window geometry. It means that it's possible to have conflicts
between the two. With this change, the compositor's move resize geometry
will be synced only if there are no pending configure events, meaning
that the user doesn't try to resize the window.
This is to improve code readability and make it easier to differentiate
between methods that are used during interactive move-resize and normal
move-resize methods in the future.
We need to emit the clientFinishUserMovedResized signal to notify
effects such as translucency that the interactive move-resize is
finished. Otherwise, the set() animation won't be cancelled and the
window will get stuck frozen.
BUG: 409376
The order in which Xwayland surfaces are associated with X11 windows is
undefined, meaning that we cannot assume that a newly created X11 window
won't have a surface associated with it already.
On X11, the lockscreen greeter is an override-redirect window so the
scale and the glide effect ignore it.
On Wayland, the lockscreen greeter is a regular window so both effects
try to animate it upon the screen being unlocked, which looks bad.
This reduces the number of usages of xStackingOrder(), which simplifies
the reasoning about when it can be marked as dirty.
Since internal windows are now in the regular stack, InternalWindowTest
can use stackingOrder().
As for X11ClientTest, there's no specific reason why it uses the x stack
instead of the regular one.
We need to make sure that the information from
toplevelConfigureRequestedSpy is in place to be used, otherwise we get
an empty size and it doesn't work.
We were expecting a tooltip to be closed when clicking its
transientParent, but it's explicitly not something we are after. We
close popups when we click either other clients or the actual client on
the decoration.
This change makes it so we end up clicking another window instead of the
parent one that is unrelated.
When debugging modifier_only_shortcut_test in _waylandonly mode I saw
that it was failing, among other things, because some aspects were not
initialised.
This changes every test we have to run the new
Test::initWaylandWorkspace() that calls waylandServer()->initWorkspace()
but also makes sure that WaylandServer::initialized is emitted before we
proceed.
Starting with 48c3376927e5e9c13377bf3cfc8b0c411783e7f3 in kglobalaccel,
KGlobalAccel won't work in desktop environments other than Plasma.
We need to set XDG_CURRENT_DESKTOP=KDE to ensure that global shortcuts
still work.
Currently, the fullscreen state is update synchronously, but it needs to
be done in asynchronous fashion.
This change removes some tests as they don't add any value, testFullscreen()
covers them all.
Xcursors don't support hidpi so if a hidpi cursor is needed, kwin will
scale the desired size by the scale factor and ask Xcursor helpers to
load a theme with the given name and the size.
However, the theme loading code doesn't take into account that Xcursor
theme loading helpers may not return cursor sprites of size size * scale
if the theme has no such a size.
For example, if the cursor theme only provides 24, 36, and 48 sizes and
kwin attempts to load cursors of size 48 with a scale factor of 2, we
will get cursors of size 48 instead of 96. Unfortunately, this will
result in the issue where the cursor shrinks when hovering decorations
because kwin doesn't know that the effective scale factor (1) is
different from the requested scale factor (2).
In order to fix loading of HiDPI cursors, we need to approximate the
effective scale factor of every cursor sprite as we load it.
If a decoration is created for an already mapped maximized window, check
the workspace position to ensure that the window still fits the maximize
area.
BUG: 432326
Re-use Qt's implementation of handling non-Latin layouts here
For full ASCII range support (Alt+`, etc.) Qt needs to be patched still,
see QTBUG-90611
BUG: 375518
At the moment, the session code is far from being extensible. If we
decide to add support for libseatd, it will be a challenging task with
the current design of session management code. The goal of this
refactoring is to fix that.
Another motivation behind this change is to prepare session related code
for upstreaming to kwayland-server where it belongs.
This provides the compositor a way to indicate what output is being
rendered. The effects such as the screenshot can check the provided
screen object in order to function as expected.
Summary:
QScriptEngine is deprecated for years and suffers bitrot.
Plasma hit one super major bug with it in 5.11.0 and has now ported
away.
Main porting notes:
- creating low level functions no longer exists
The old global functions are exposed on the ScriptedEffect instance
and then the QJSValue wrappers of the globalObject are modified to
trampoline the methods at a wrapper level.
- We can then use QJSEngine to automatically do argument error checking
rather than unmarshalling a QJSValue manually which significantly
reduces a lot of code.
- We can't make FPX2 a native type, so these are QJSValue args and
unboxed there.
Long term I want overloads for animate that take int/QSize/QPoint which
are native JS types, but that might be an API break.
Test Plan:
Hopefully comprehensive unit test which passes
Tested fade/fadeDesktop manually.
It's a very invasive change, so I expect some things will be broke
please help test any JS effects.
Reviewers: #kwin, mart, fvogt
Subscribers: fvogt, zzag, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D14536
Otherwise the input method test seems to fail with the following error
"The name org.kde.kwin.testvirtualkeyboard was not provided by any
.service files"
According to the spec, when the pointer enters a surface, the contents
of the cursor becomes undefined. The client should call set_cursor() to
make sure that the cursor image is correct.
In case the compositor wants to cancel a touch sequence, we need to
ignore subsequent touch motion and touch up events until a new sequence
is initiated by the user.
Previously, it was implicitly handled by clearing the mapping table
between the touch slots and touch ids generated by kwayland-server.
Once in a while, we receive complaints from other fellow KDE developers
about the file organization of kwin. This change addresses some of those
complaints by moving all of source code in a separate directory, src/,
thus making the project structure more traditional. Things such as tests
are kept in their own toplevel directories.
This change may wreak havoc on merge requests that add new files to kwin,
but if a patch modifies an already existing file, git should be smart
enough to figure out that the file has been relocated.
We may potentially split the src/ directory further to make navigating
the source code easier, but hopefully this is good enough already.
Occasionally, I see complaints about the file organization of kwin,
which is fair enough.
This change makes the source code more relocatable by removing relative
paths from includes.
CMAKE_CURRENT_SOURCE_DIR was added to the interface include directories
of kwin library. This means that as long as you link against kwin target,
the real location of the source code of the library doesn't matter.
With autotests, things are not as convenient as with kwin target. Some
tests use cpp files from kwin core. If we move all source code in a src/
directory, they will need to be adjusted, but mostly only in build
scripts.
This allows running Xwayland apps as root. Xwayland started with an
empty Xauthority file. After kwin has received the display number, the
file is updated with an actual authority entry.
BUG: 432625
As it was pointed out in 6acf35e4cc, it is
better to return raw pointers than qpointers because returning a qpointer
is equivalent to constructing a new one.
Warning messages are not the kind of messages that should be ignored,
they indicate that something is off or wrong.
Also, this makes triaging bugs easier as we no longer have to ask people
to run kwin with the QT_LOGGING_RULES environment variable set.
This logs to a tracefs filesystem which can be viewed in tools such as
gpuvis to see precise timings of activities in relation to other trace
markers in X or graphic drivers.
This patch is loosely based on D23114. Though modified with thread
safety, support for string building, and a RAII pattern for durations.
Ultimately that expanded it somewhat.
At the moment, our frame scheduling infrastructure is still heavily
based on Xinerama-style rendering. Specifically, we assume that painting
is driven by a single timer, etc.
This change introduces a new type - RenderLoop. Its main purpose is to
drive compositing on a specific output, or in case of X11, on the
overlay window.
With RenderLoop, compositing is synchronized to vblank events. It
exposes the last and the next estimated presentation timestamp. The
expected presentation timestamp can be used by effects to ensure that
animations are synchronized with the upcoming vblank event.
On Wayland, every outputs has its own render loop. On X11, per screen
rendering is not possible, therefore the platform exposes the render
loop for the overlay window. Ideally, the Scene has to expose the
RenderLoop, but as the first step towards better compositing scheduling
it's good as is for the time being.
The RenderLoop tries to minimize the latency by delaying compositing as
close as possible to the next vblank event. One tricky thing about it is
that if compositing is too close to the next vblank event, animations
may become a little bit choppy. However, increasing the latency reduces
the choppiness.
Given that, there is no any "silver bullet" solution for the choppiness
issue, a new option has been added in the Compositing KCM to specify the
amount of latency. By default, it's "Medium," but if a user is not
satisfied with the upstream default, they can tweak it.
At the moment, the Screens class is convoluted with ifdefs because of
MockScreens.
The goal of this change is to reduce the number of usages of the
MockScreens class so it is possible to get rid of the ifdefs.
Since the Screens class is a convenience wrapper around AbstractOutput
objects that come from the Platform, it should not be platform-specific.
By dropping createScreens(), output-related code becomes simpler.
This change introduces a new component - ColorManager that is
responsible for color management stuff.
At the moment, it's very naive. It is useful only for updating gamma
ramps. But in the future, it will be extended with more CMS-related
features.
The ColorManager depends on lcms2 library. This is an optional
dependency. If lcms2 is not installed, the color manager won't be built.
This also fixes the issue where colord and nightcolor overwrite each
other's gamma ramps. With this change, the ColorManager will resolve the
conflict between two.
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.
Vlad Zahorodnii