According to [X.org
docs](http://www.x.org/releases/X11R7.7/doc/compositeproto/compositeproto.txt)
regarding NameWindowPixmap:
"This pixmap will remain allocated until freed, even if 'window' is
unmapped, reconfigured or destroyed. However, 'window' will get a new
pixmap allocated each time it is mapped or resized, so this request will
need to be reinvoked"
So, pixmap needs to be discarded not only on XCB_CONFIGURE_NOTIFY event
but on XCB_MAP_NOTIFY event also.
The svg elements cache isn't really a "configuration" and it's not
something that should be overridden by global config settings.
Loading it as a simple config skips the cascading and saves reparsing
kdeglobals.
REVIEW: 128966
when the view is in SizeRootObjectToView mode, the root
object is resized in the event handler, that is too late at startup.
resize the root object right after having announced
the new containment, so the view subclass can have
the view resized to the proper size beforehand,
removing an useless containment graphicsobject resize.
REVIEW:128915
running tasks need to have some visual indicator,
otherwise is impossible to distinguish between
tasks and launchers in icontasks
Change-Id: Ib1c319c675855a605d8a268bad2bb43ad19a6e4d
WindowThumbnail did some open GL operations, discarding old textures, in
the GUI thread. Whislt it's not going to cause a threading issue (as
updatePaintNode always ran when the main thread was blocked) we're not
meant to mix threads with openGL contexts.
It also seems to have a GL leak on nvidia, which was previously masked
by the double delete fixed in
https://git.reviewboard.kde.org/r/126131/diff/2/
It seems only one worked, and in the applied version we went with the
wrong one.
This patch makes use of QQuickItem::releaseResources to delete the GL
textures on window change and destructor; it's then removed from
stopRedirecting so that start/stop redirecting handles xcb on the GUI
thread and updatePaintNode/discardPixmap is the GL stuff on the render
thread.
See http://doc.qt.io/qt-5/qquickitem.html#graphics-resource-handling
REVIEW: 128763
BUG: 368066
The code already had logic in there but since UI ready is usually emitted later
the client (in this case plasmashell) had to manually show the panels and would
usually show them when they're not ready yet leading to visual glitches on startup.
REVIEW: 128708
setting the domain by just using kdeclarative would lead to
setting the domain on the wrong QQmlContext,
sometimes overwriting it on the global context, causing
unexpected crashes deep in qml
BUG:361513
reviewed-by: Ivan Čukić <ivan.cukic@kde.org>
Similar to KPasswordWidget in kwidgetaddons
Also fixes:
- the clear button still being clickable when not enabled
- RTL icons
- Updating the right margin correctly
REVIEW: 128660
This flips the tooltip when using right-to-left languages
CHANGELOG: The standard tooltips are now mirrored when in a right-to-left language
REVIEW: 128662
LayoutChanged causes QML to destroy all delegates and re-create them.
Since the number of days doesn't change but only the numbers, we can just call dataChanged on all roles.
Furthermore, when agenda-related data changes, only the containsEventItems role
might have changed, so only emit a change for it.
CHANGELOG: Performance when changing months in the calendar has been greatly improved
REVIEW: 128648
Commit bb70febdbe changed the slot for
adding createApplet, turning the final argument from QPoint to QRect.
If the rectangle size is nothing, it acts like the original code.
Despite this being private API (ish) there is a hacks in plasma-
workspace that call methods on the view via QMetaObject invoke. This
obviously fails. As we need compatibility for Plasma/5.7 and earlier a
compatibility slot needs to stay.
REVIEW: 128640
The QSGTextures are created with
window()->createTextureFromId(m_texture, QSize(w,h),
QuickWindow::TextureOwnsGLTexture));
this means we don't want to be deleting textures ourselves too, it will
be deleted when we delete the QSGTexture, which is a scoped pointer
inside our QSGNode.
BUG: 355644
BUG: 365946
REVIEW: 126131
applets will be deleted by the parent, as they are
acontainment's children. by not deleting them by hand
also every signal will be disconnected, not causing appletRemoved
to be emitted, so not causing removal of config
pieces at shutdown, this removes the need of the
explicit disconnect in aboutToQuit from contaiment interface
Change-Id: Iddf6ded60ae849ca03c363be6e736da496310b82
reviewed-by: David Edmundson
Currently the code gets the icon name from the QIcon and tries to do
some Plasma theming with it.
However if that fails it then loads the QIcon::fromTheme again.
This is pointless in most cases and will break any icons that have a
custom loader (all SNIs)
REVIEW: 128580
Panel popup dialogs along with other applications of Plasma::Dialog
(e.g. Kicker's submenus) currently don't correctly set WM_TRANSIENT_FOR
to the id of their parent window on X11. This causes them to interact
badly with auto-hide panels which do not set struts, e.g. Kicker's
submenus open behind the panel.
Internally, Dialog makes calls to QWindow::setFlags when its window
type is "Normal" (the default) and otherwise uses KWindowSystem's
setType. (Kicker's subclass, SubMenu, does an additional
KWindowSystem::setType call for NET::Menu). Neither CompactApplet
nor Kicker change their Dialog's 'type' property to anything else,
so their dialogs are "Normal", thus going the setFlags route with
no setType calls.
Dialog also sets its transient parent to the window the visualParent
item is inside of.
Now here is where things break down: QXcbWindow will update
WM_TRANSIENT_PARENT for in show(), but only when an inline function
called isTransient returns true. isTransient decides this based on
window type, which is determined from the flags set with
QWindow::setFlags. Calls to KWindowSystem::setType would have no
bearing on this; there seems to be no mapping back from external
state. This is why setting Dialog.type to e.g. "DialogWindow" does
nothing.
This patch takes the route of adding Qt::Dialog to the starting
flags - after all Dialog is a dialog. That means isTransient()
will consider the window to be a transient and those
setTransientParent calls Dialog does will not be ignored. In the
case of CompactApplet, no further calls to KWS::setType are done;
Kicker continues to call it with NET::Menu to get desired window
manager behavior.
In light testing everything still seems too work, with the added
benefit of fixing:
BUG:366278
That said, the weird mess of setFlags and setType and state in the
windowing system vs. Qt is horrible and sad.
REVIEW:128571
Since nobody bothered to fulfill my request for this icon (cf. Review 128332)
I just did it myself, with no warranty.
Plasma IconItem finds icons from Plasma theme only in the SVG file of the first
section of the name (ie. all icons in network must start with network- prefix
which the flightmode icons did not).
The old icon is kept as is for compatibility.
This is done by ContainmentInterface::mousePressEvent, but not in this
code path, which is hit for applet panel popups.
Folder View relies on this to update actions ahead of the menu being
opened.
BUG:366294
QtQuick.Layouts by default left aligns content and because the Layout is
anchors.fill'd the icon would show off-center in a button when no text has been set.
This fixes this.
REVIEW: 128493
BUG: 365947
The containment's status is the max of all applets.
Unfortunately, HiddenStatus, while semantically lower, has the highest value,
so as soon as there's a hidden applet (eg. pager with just one virtual desktop),
the panel will always auto-hide even when popups are open as its status will be HiddenStatus.
REVIEW: 128301
screen ids are going to not be continuous anymore
as screen id is going to correspond 1:1 to connectors,
it will be possible to have "holes", therefore valid ids
that are bigger than screenNum()
REVIEWED-BY:sebas@kde.org
Change-Id: I1c0b1fb827dba4d95f228d32209403150c089c77
- Initially set state (and type, and flags) in response to PlatformSurfaceCreated.
We know reliably this will run before the window is mapped.
- Drop the comment about removing setState() form showEvent handler, as
we need it to avoid state loss in this scenario:
<mgraesslin> the window gets mapped first time: everything is fine
<mgraesslin> window gets unmapped: kwin removes the state as per spec
<mgraesslin> qt gets change event and removes the states it doesn't care about
<mgraesslin> qt maps window again and sets states
<mgraesslin> we lost the state
<mgraesslin> which means we need to set the state again from our side before(!) Qt sets it
<mgraesslin> and before Qt maps the window
In order to have support for tooltip window type in Wayland, the
window type NET::Tooltip is also set through KWindowSystem API once
the platform surface is created.
REVIEW: 128300
Even though QQuickItem only has a width and height accessor there is a
usable public setSize method.
This gets rid of a lot of potential re-evaluation as internal geometry
is updated before widthChanged is emitted.
REVIEW: 128316
Do to so means the System Tray applet ends up being placed inside the
system tray container, which ends up in a parent loop, causing a stack
overflow every time parent() is called recursively
This happens whenever one drags a sys tray on the desktop or simply adds
a new system tray whilst running.
BUG: 361777
We also need to do a hack for Plasma/5.7 in plasma-desktop somehow in
all cases where this is used.
the llvmpipe driver doesn't have a good quality enough for the
taskbar thumbnails components, it will have many crashes in the driver.
disabale thumbnails with that driver, as kwin disables many effects as well.
BUG:363371
ensure the scriptengine gets created also for c++ applets
that have a valid scriptengine, move the script init
out of applet::init that can be reimplemented.
remove the second appletscript constructor that is
not necessary anymore
REVIEWED-BY: Kai Uwe Broulik
Summary:
The notifications applet installs the .so into the toplevel plugins dir,
and my earlier code removal broke the loading of that applet.
Reviewers: lbeltrame, mart
Reviewed By: mart
Differential Revision: https://phabricator.kde.org/D1851
Summary:
The name of the actual plugin was missing, making the warning rather useless.
Since this is only used within PluginLoader, I added a private function
isPluginVersionCompatible(KPluginLoader), taking code from
Plasma::isPluginVersionCompatible(uint). That one could probably be deprecated,
although no public replacement would be available, just "use PluginLoader".
Test Plan: plasmoidviewer -a org.kde.plasma.digitalclock
Reviewers: mart
Reviewed By: mart
Tags: #frameworks
Differential Revision: https://phabricator.kde.org/D1825
