When the user selects all of the types or none of them, the "types"
property must be set to a special value (`NET::AllTypesMask = -1`),
different than the sum of all the flags together.
This re-implements this behaviour as the old KCM, fixing some heuristics
that prevented finding the rule corresponding to the current window.
The enum name that handles this property has been changed to `NetTypes`
to make it more explicit.
BUG: 423214
FIXED-IN: 5.19.3
TEST PLAN:
1. Open the `Application specific settings` on a window via menu
2. Select every "Window Type" (or none of them)
3. Give the rule a different name than the default and save
4. Open it again and check that the same rule is found
Depending on the current focus stealing prevention level, it's possible
for kwin to call XSetInputFocus() on a window that already has the input
focus. In which case, we won't receive the corresponding FocusIn event
and the client will remain inactive from kwin's perspective even though
it isn't.
In order to work around this issue, we can move the input focus to the
null window. By doing so, it's guaranteed that we're going to receive
the matching FocusIn event for the client.
This commit indirectly fixes a bug where fullscreen games are displayed
below panels.
Property `wmclass` of a window rule works in a special way, as it can have
two meanings depending on `wmclasscomplete` flag:
- false: only matches `resourceClass`
- true: matches `resourceName` and `resourceClass`
This MR fixes two subtle bugs when detecting the properties of a window:
In the first case, `resourceName` was being wrongly set instead, which
prevented window matching for windows where those two values differ.
Also, the `wmclass` field was always set to `resourceName` independently
of `wmclasscomplete` property.
BUG: 423138
FIXED-IN: 5.19.3
Due to a bug in the XI2 protocol, clients have to reset scroll valuators
on XI_Enter because the scroll valuators might have changed while the
pointer was elsewhere. The XI_Enter event is usually sent when an input
device enters the window, but it can also be generated by a passive grab.
If an XI_Enter event has been generated by a passive grab, the client
should not reset scroll valuators. Unfortunately, there is no any
reliable way for the client to determine if an XI_Enter event has been
sent in response to a deactivated passive grab. A correct fix for the
scroll issues in GTK apps would involve changes in the XI2 protocol.
As a work around, we can hold a passive grab only if the current mouse
wheel action is either "Activate and scroll" or "Activate, raise, and
scroll."
BUG: 394772
FIXED-IN: 5.19.3
WaylandCursorImage::loadThemeCursor(const T &shape) uses templates.
These templates are used by input.cpp but the deifnition is in
pointer_input.cpp
On some setups this creates a compilation problem.
This patch introduces an explicit non-templated declaration with the
defintion explicitly complied into the same class.
In master this has been refactored away anyway, so this is a minimal
patch to make things working in the least invasive way.
BUG: 423052
Since ScreenEdge no longer uses physicalDpiX() and physicalDpiY() from
the QPaintDevice class in favor of our own helpers, we need to ensure
that the X11 platform provides valid output physical dimensions.
BUG: 422816
FIXED-IN: 5.19.2
Every `AbstractItemDelegate` was instantiating a `Kirigami.ActionToolBar`,
which is a very expensive component, rising CPU usage specially on long lists.
Use `Kirigami.SwipeListItem` instead, as is seems the proper component.
BUG: 421537
The first time the list in `Add properties` is shown, the position of the first
delegate is miscalculated as (0, 0) and gets hidden behind the section header.
It only repositions when the delegate is destroyed and re-created, for example, when
setting and deleting a filter text. (https://bugs.kde.org/show_bug.cgi?id=422289)
This works-around it by delaying the `overlayModel` filtering until the list is first
shown, which makes the delegate position correctly.
BUG: 421583
FIXED-IN: 5.19
The recommended way to get all EGL extension defines is to include
EGL/eglext.h. EGL/eglmesaext.h is a private header that compositors
should not use.
BUG: 422131
Since the port to KConfigXT, the placement setting in `kwinrulesrc` now stores the enum value instead of a string equivalent.
The strings and enum values are taken from `placement.h/.cpp`
This was suggested by @zzag in https://phabricator.kde.org/D29790
We don't need to bind disabled outputs to surfaces that overlaps them.
This prevents error down the line and warnings about ignoring surface.enter events with Qt.
BUG: 419749
Summary:
Present windows works as follows:
- It moves all windows about until nothing is overlapping with any
other window.
- This doesn't resize anything so ultimately we end up with a new
co-ordinate space that's bigger than the screen depending on the amount
of overlap.
- We then render this whole view transformed to the screen
The rectangle "bounds" is in overviewpixels, with "scale" being the
ratio to convert to screen pixels.
When adjusting the new bounds there's an attempt to centre align things.
As bounds is in "overviewpixels" we multiply references to the previous
bounds by scale, and divide everything through at the end. bounds.x/y
were missed.
This is mostly unoticable except on massive super-ultra-wide monitors
which will otherwise have a tendency to shift to the left.
Test Plan:
Kai created a whole new test framework for this code that copy pasted
this algorithm then showed mock windows as rectangles
Reviewers: #kwin, apol, broulik, zzag
Reviewed By: #kwin, apol, broulik, zzag
Subscribers: zzag, apol, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D29705
Currently, the spinners used to edit position and size rules have an upper
limit of 4098, which may be not enough for wide/hdpi/mutiple screens.
The property itself has no such limit in the `.kcfg` schema
This is just an issue with the editor.
BUG: 421892
FIXED-IN: 5.19
Even though the frame geometry and the client geometry have a connection
between themselves, it doesn't mean that the frame geometry will change
if the client geometry has changed and vice versa. For example, it can be
the case when the border size changes. The frame geometry won't change,
but the client geometry will.
In general, similar to the frameGeometryChanged signal, we need another
signal that is emitted when the client geometry has been changed that
can be used then in DecoratedClientImpl.
Unfortunately, 5.19 release is around the corner and I would prefer not
to do any geometry-related changes in order to avoid introducing new
regressions.
In order to fix the propagation of the client size to decorations, this
change ports DecoratedClientImpl from frameGeometryChanged to a signal
that is emitted whenever geometry of any kind has been changed.
BUG: 419080
FIXED-IN: 5.19.0
In previous code we used to update/show cursor even when the pointer
device was present, e.g mobile device. Guard it behind the cursor
enabled check.
BUG: 418977
FIXED-IN: 5.19.0
In order to generate window quads for a window, the scene needs a valid
WindowPixmap tree. If it's time to render the window, the scene will
build window quads for the contents, the server-side window decoration,
the drop-shadow and cache the resulting window quads. With this way of
generating window quads, we need the window pixmap tree to be valid,
or else no contents window quads will be generated.
While the window pixmap tree is guaranteed to be valid at the time of
generation of window quads for Wayland and X11 clients, this is not the
case for Xwayland clients.
When an Xwayland client is created, some time may pass between the
moment it's been created and the moment when a regular wayland surface
has been associated with the xwayland window. If the compositor decides
to render the Xwayland client in that short period of time, the window
quads cache won't have window quads cache.
In order to work around the weird asynchronous behavior of Xwayland
clients, this change makes the scene discard the window quad cache when
a new window pixmap has been created. This will ensure that the current
window quads are always in sync with the current window pixmap tree.
The list of virtual desktops is retrieved using a DBus method, instead of
`KWindowSystem`, to make it work both in Wayland (as well as X11).
FIXED-IN: 5.20
BUG: 416165
Summary:
Since the port to KConfigXT, the setting "placement" is stored using the enum value, instead of a string equivalent.
This updates the kcm model to be able to communicate this property with the config schema.
BUG: 421586
FIXED-IN: 5.19.0
Test Plan:
- In a rule set the "Initial Placement" property
- The value is stored in `~/.config/kwinrulesrc`
- Close/reopen the kcm. The placement value is correctly loaded.
Reviewers: #kwin, #plasma, zzag
Reviewed By: #kwin, #plasma, zzag
Subscribers: zzag, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D29790
Summary:
Use `QSize`/`QPoint` to handle and store coordinate values (size and position)
Previously, the rules model stored the "coordinate" type properties as a
`QString` with format `x, y`.
This fails when setting the properties to the config schema, as it requires
a proper `QPoint` or `QSize` value, specially the latter which can't be
convert from such a string.
BUG: 421055
FIXED-IN: 5.19.0
Test Plan:
- Add a new rule and set its position and size properties
- Hitting apply stores the right values in `~\.config\kwinrulesrc`
- Close the kcm and reopen, the values are loaded
- Property detection still works for size and position
Please note that there is a pre-existing bug of some position/sizes not being
applied to the windows in some cases, when using `Apply Initially`.
Better try using the `Force` policy.
Reviewers: ngraham, #kwin, #plasma, zzag
Reviewed By: #kwin, #plasma, zzag
Subscribers: zzag, ltoscano, yurchor, kwin
Tags: #kwin
Differential Revision: https://phabricator.kde.org/D29764
