kwin/screenedge.cpp

1299 lines
40 KiB
C++
Raw Normal View History

/********************************************************************
KWin - the KDE window manager
This file is part of the KDE project.
Copyright (C) 2011 Arthur Arlt <a.arlt@stud.uni-heidelberg.de>
Copyright (C) 2013 Martin Gräßlin <mgraesslin@kde.org>
Since the functionality provided in this class has been moved from
class Workspace, it is not clear who exactly has written the code.
The list below contains the copyright holders of the class Workspace.
Copyright (C) 1999, 2000 Matthias Ettrich <ettrich@kde.org>
Copyright (C) 2003 Lubos Lunak <l.lunak@kde.org>
Copyright (C) 2009 Lucas Murray <lmurray@undefinedfire.com>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*********************************************************************/
#include "screenedge.h"
// KWin
#include "atoms.h"
#include "client.h"
#include "cursor.h"
#include "input.h"
#include "main.h"
#include "screens.h"
#include "utils.h"
#include "workspace.h"
#include "virtualdesktops.h"
// DBus generated
#include "screenlocker_interface.h"
// Qt
#include <QSharedPointer>
#include <QTimer>
#include <QVector>
#include <QTextStream>
#include <QtDBus/QDBusInterface>
#include <QtDBus/QDBusPendingCall>
namespace KWin {
// Mouse should not move more than this many pixels
static const int DISTANCE_RESET = 30;
Edge::Edge(ScreenEdges *parent)
: QObject(parent)
, m_edges(parent)
, m_border(ElectricNone)
, m_action(ElectricActionNone)
, m_reserved(0)
, m_approaching(false)
, m_lastApproachingFactor(0)
, m_blocked(false)
, m_pushBackBlocked(false)
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
, m_client(nullptr)
{
}
Edge::~Edge()
{
}
void Edge::reserve()
{
m_reserved++;
if (m_reserved == 1) {
// got activated
activate();
}
}
void Edge::reserve(QObject *object, const char *slot)
{
connect(object, SIGNAL(destroyed(QObject*)), SLOT(unreserve(QObject*)));
m_callBacks.insert(object, QByteArray(slot));
reserve();
}
void Edge::unreserve()
{
m_reserved--;
if (m_reserved == 0) {
// got deactivated
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
stopApproaching();
deactivate();
}
}
void Edge::unreserve(QObject *object)
{
if (m_callBacks.contains(object)) {
m_callBacks.remove(object);
disconnect(object, SIGNAL(destroyed(QObject*)), this, SLOT(unreserve(QObject*)));
unreserve();
}
}
bool Edge::triggersFor(const QPoint &cursorPos) const
{
if (isBlocked()) {
return false;
}
if (!m_geometry.contains(cursorPos)) {
return false;
}
if (isLeft() && cursorPos.x() != m_geometry.x()) {
return false;
}
if (isRight() && cursorPos.x() != (m_geometry.x() + m_geometry.width() -1)) {
return false;
}
if (isTop() && cursorPos.y() != m_geometry.y()) {
return false;
}
if (isBottom() && cursorPos.y() != (m_geometry.y() + m_geometry.height() -1)) {
return false;
}
return true;
}
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
bool Edge::check(const QPoint &cursorPos, const QDateTime &triggerTime, bool forceNoPushBack)
{
if (!triggersFor(cursorPos)) {
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
return false;
}
// no pushback so we have to activate at once
bool directActivate = forceNoPushBack || edges()->cursorPushBackDistance().isNull();
if (directActivate || canActivate(cursorPos, triggerTime)) {
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
markAsTriggered(cursorPos, triggerTime);
handle(cursorPos);
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
return true;
} else {
pushCursorBack(cursorPos);
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
m_triggeredPoint = cursorPos;
}
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
return false;
}
void Edge::markAsTriggered(const QPoint &cursorPos, const QDateTime &triggerTime)
{
m_lastTrigger = triggerTime;
m_lastReset = QDateTime(); // invalidate
m_triggeredPoint = cursorPos;
}
bool Edge::canActivate(const QPoint &cursorPos, const QDateTime &triggerTime)
{
// we check whether either the timer has explicitly been invalidated (successfull trigger) or is
// bigger than the reactivation threshold (activation "aborted", usually due to moving away the cursor
// from the corner after successfull activation)
// either condition means that "this is the first event in a new attempt"
if (!m_lastReset.isValid() || m_lastReset.msecsTo(triggerTime) > edges()->reActivationThreshold()) {
m_lastReset = triggerTime;
return false;
}
if (m_lastTrigger.isValid() && m_lastTrigger.msecsTo(triggerTime) < edges()->reActivationThreshold()) {
return false;
}
if (m_lastReset.msecsTo(triggerTime) < edges()->timeThreshold()) {
return false;
}
// does the check on position make any sense at all?
if ((cursorPos - m_triggeredPoint).manhattanLength() > DISTANCE_RESET) {
return false;
}
return true;
}
void Edge::handle(const QPoint &cursorPos)
{
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if (m_client) {
pushCursorBack(cursorPos);
m_client->showOnScreenEdge();
unreserve();
return;
}
if ((edges()->isDesktopSwitchingMovingClients() && Workspace::self()->getMovingClient()) ||
(edges()->isDesktopSwitching() && isScreenEdge())) {
// always switch desktops in case:
// moving a Client and option for switch on client move is enabled
// or switch on screen edge is enabled
switchDesktop(cursorPos);
return;
}
if (Workspace::self()->getMovingClient()) {
// if we are moving a window we don't want to trigger the actions. This just results in
// problems, e.g. Desktop Grid activated or screen locker activated which just cannot
// work as we hold a grab.
return;
}
if (handleAction() || handleByCallback()) {
pushCursorBack(cursorPos);
return;
}
if (edges()->isDesktopSwitching() && isCorner()) {
// try again desktop switching for the corner
switchDesktop(cursorPos);
}
}
bool Edge::handleAction()
{
switch (m_action) {
case ElectricActionDashboard: { // Display Plasma dashboard
QDBusInterface plasmaApp(QStringLiteral("org.kde.plasmashell"), QStringLiteral("/PlasmaShell"));
plasmaApp.asyncCall(QStringLiteral("toggleDashboard"));
return true;
}
case ElectricActionShowDesktop: {
Workspace::self()->setShowingDesktop(!Workspace::self()->showingDesktop());
return true;
}
case ElectricActionLockScreen: { // Lock the screen
OrgFreedesktopScreenSaverInterface interface(QStringLiteral("org.freedesktop.ScreenSaver"),
QStringLiteral("/ScreenSaver"),
QDBusConnection::sessionBus());
if (interface.isValid()) {
interface.Lock();
}
return true;
}
default:
return false;
}
}
bool Edge::handleByCallback()
{
if (m_callBacks.isEmpty()) {
return false;
}
for (QHash<QObject *, QByteArray>::iterator it = m_callBacks.begin();
it != m_callBacks.end();
++it) {
bool retVal = false;
QMetaObject::invokeMethod(it.key(), it.value().constData(), Q_RETURN_ARG(bool, retVal), Q_ARG(ElectricBorder, m_border));
if (retVal) {
return true;
}
}
return false;
}
void Edge::switchDesktop(const QPoint &cursorPos)
{
QPoint pos(cursorPos);
VirtualDesktopManager *vds = VirtualDesktopManager::self();
const uint oldDesktop = vds->current();
uint desktop = oldDesktop;
const int OFFSET = 2;
if (isLeft()) {
const uint interimDesktop = desktop;
desktop = vds->toLeft(desktop, vds->isNavigationWrappingAround());
if (desktop != interimDesktop)
pos.setX(displayWidth() - 1 - OFFSET);
} else if (isRight()) {
const uint interimDesktop = desktop;
desktop = vds->toRight(desktop, vds->isNavigationWrappingAround());
if (desktop != interimDesktop)
pos.setX(OFFSET);
}
if (isTop()) {
const uint interimDesktop = desktop;
desktop = vds->above(desktop, vds->isNavigationWrappingAround());
if (desktop != interimDesktop)
pos.setY(displayHeight() - 1 - OFFSET);
} else if (isBottom()) {
const uint interimDesktop = desktop;
desktop = vds->below(desktop, vds->isNavigationWrappingAround());
if (desktop != interimDesktop)
pos.setY(OFFSET);
}
if (Client *c = Workspace::self()->getMovingClient()) {
if (c->rules()->checkDesktop(desktop) != int(desktop)) {
// user attempts to move a client to another desktop where it is ruleforced to not be
return;
}
}
vds->setCurrent(desktop);
if (vds->current() != oldDesktop) {
m_pushBackBlocked = true;
Cursor::setPos(pos);
QSharedPointer<QMetaObject::Connection> me(new QMetaObject::Connection);
*me = QObject::connect(QCoreApplication::eventDispatcher(),
&QAbstractEventDispatcher::aboutToBlock, this,
[this, me](){
QObject::disconnect(*me);
const_cast<QSharedPointer<QMetaObject::Connection>*>(&me)->reset(nullptr);
m_pushBackBlocked = false;
}
);
}
}
void Edge::pushCursorBack(const QPoint &cursorPos)
{
if (m_pushBackBlocked)
return;
int x = cursorPos.x();
int y = cursorPos.y();
const QSize &distance = edges()->cursorPushBackDistance();
if (isLeft()) {
x += distance.width();
}
if (isRight()) {
x -= distance.width();
}
if (isTop()) {
y += distance.height();
}
if (isBottom()) {
y -= distance.height();
}
Cursor::setPos(x, y);
}
void Edge::setGeometry(const QRect &geometry)
{
if (m_geometry == geometry) {
return;
}
m_geometry = geometry;
int x = m_geometry.x();
int y = m_geometry.y();
int width = m_geometry.width();
int height = m_geometry.height();
const int size = m_edges->cornerOffset();
if (isCorner()) {
if (isRight()) {
x = x - size +1;
}
if (isBottom()) {
y = y - size +1;
}
width = size;
height = size;
} else {
if (isLeft()) {
y += size + 1;
width = size;
height = height - size * 2;
} else if (isRight()) {
x = x - size + 1;
y += size;
width = size;
height = height - size * 2;
} else if (isTop()) {
x += size;
width = width - size * 2;
height = size;
} else if (isBottom()) {
x += size;
y = y - size +1;
width = width - size * 2;
height = size;
}
}
m_approachGeometry = QRect(x, y, width, height);
doGeometryUpdate();
}
void Edge::checkBlocking()
{
if (isCorner()) {
return;
}
bool newValue = false;
if (Client *client = Workspace::self()->activeClient()) {
newValue = client->isFullScreen() && client->geometry().contains(m_geometry.center());
}
if (newValue == m_blocked) {
return;
}
m_blocked = newValue;
doUpdateBlocking();
}
void Edge::doUpdateBlocking()
{
}
void Edge::doGeometryUpdate()
{
}
void Edge::activate()
{
}
void Edge::deactivate()
{
}
void Edge::startApproaching()
{
if (m_approaching) {
return;
}
m_approaching = true;
doStartApproaching();
m_lastApproachingFactor = 0;
emit approaching(border(), 0.0, m_approachGeometry);
}
void Edge::doStartApproaching()
{
}
void Edge::stopApproaching()
{
if (!m_approaching) {
return;
}
m_approaching = false;
doStopApproaching();
m_lastApproachingFactor = 0;
emit approaching(border(), 0.0, m_approachGeometry);
}
void Edge::doStopApproaching()
{
}
void Edge::updateApproaching(const QPoint &point)
{
if (approachGeometry().contains(point)) {
int factor = 0;
const int edgeDistance = m_edges->cornerOffset();
// manhattan length for our edge
const int cornerDistance = 2*edgeDistance;
switch (border()) {
case ElectricTopLeft:
factor = (point.manhattanLength()<<8) / cornerDistance;
break;
case ElectricTopRight:
factor = ((point - approachGeometry().topRight()).manhattanLength()<<8) / cornerDistance;
break;
case ElectricBottomRight:
factor = ((point - approachGeometry().bottomRight()).manhattanLength()<<8) / cornerDistance;
break;
case ElectricBottomLeft:
factor = ((point - approachGeometry().bottomLeft()).manhattanLength()<<8) / cornerDistance;
break;
case ElectricTop:
factor = (qAbs(point.y() - approachGeometry().y())<<8) / edgeDistance;
break;
case ElectricRight:
factor = (qAbs(point.x() - approachGeometry().right())<<8) / edgeDistance;
break;
case ElectricBottom:
factor = (qAbs(point.y() - approachGeometry().bottom())<<8) / edgeDistance;
break;
case ElectricLeft:
factor = (qAbs(point.x() - approachGeometry().x())<<8) / edgeDistance;
break;
default:
break;
}
factor = 256 - factor;
if (m_lastApproachingFactor != factor) {
m_lastApproachingFactor = factor;
emit approaching(border(), m_lastApproachingFactor/256.0f, m_approachGeometry);
}
} else {
stopApproaching();
}
}
/**********************************************************
* ScreenEdges
*********************************************************/
WindowBasedEdge::WindowBasedEdge(ScreenEdges *parent)
: Edge(parent)
, m_window(XCB_WINDOW_NONE)
, m_approachWindow(XCB_WINDOW_NONE)
{
}
WindowBasedEdge::~WindowBasedEdge()
{
}
void WindowBasedEdge::activate()
{
createWindow();
createApproachWindow();
doUpdateBlocking();
}
void WindowBasedEdge::deactivate()
{
m_window.reset();
m_approachWindow.reset();
}
void WindowBasedEdge::createWindow()
{
if (m_window.isValid()) {
return;
}
const uint32_t mask = XCB_CW_OVERRIDE_REDIRECT | XCB_CW_EVENT_MASK;
const uint32_t values[] = {
true,
XCB_EVENT_MASK_ENTER_WINDOW | XCB_EVENT_MASK_LEAVE_WINDOW | XCB_EVENT_MASK_POINTER_MOTION
};
m_window.create(geometry(), XCB_WINDOW_CLASS_INPUT_ONLY, mask, values);
m_window.map();
// Set XdndAware on the windows, so that DND enter events are received (#86998)
xcb_atom_t version = 4; // XDND version
xcb_change_property(connection(), XCB_PROP_MODE_REPLACE, m_window,
atoms->xdnd_aware, XCB_ATOM_ATOM, 32, 1, (unsigned char*)(&version));
}
void WindowBasedEdge::createApproachWindow()
{
if (m_approachWindow.isValid()) {
return;
}
if (!approachGeometry().isValid()) {
return;
}
const uint32_t mask = XCB_CW_OVERRIDE_REDIRECT | XCB_CW_EVENT_MASK;
const uint32_t values[] = {
true,
XCB_EVENT_MASK_ENTER_WINDOW | XCB_EVENT_MASK_LEAVE_WINDOW | XCB_EVENT_MASK_POINTER_MOTION
};
m_approachWindow.create(approachGeometry(), XCB_WINDOW_CLASS_INPUT_ONLY, mask, values);
m_approachWindow.map();
}
void WindowBasedEdge::doGeometryUpdate()
{
m_window.setGeometry(geometry());
m_approachWindow.setGeometry(approachGeometry());
}
void WindowBasedEdge::doStartApproaching()
{
m_approachWindow.unmap();
Cursor *cursor = Cursor::self();
connect(cursor, SIGNAL(posChanged(QPoint)), SLOT(updateApproaching(QPoint)));
cursor->startMousePolling();
}
void WindowBasedEdge::doStopApproaching()
{
Cursor *cursor = Cursor::self();
disconnect(cursor, SIGNAL(posChanged(QPoint)), this, SLOT(updateApproaching(QPoint)));
cursor->stopMousePolling();
m_approachWindow.map();
}
void WindowBasedEdge::doUpdateBlocking()
{
if (!isReserved()) {
return;
}
if (isBlocked()) {
m_window.unmap();
m_approachWindow.unmap();
} else {
m_window.map();
m_approachWindow.map();
}
}
/**********************************************************
* AreaBasedEdge
*********************************************************/
AreaBasedEdge::AreaBasedEdge(ScreenEdges* parent)
: Edge(parent)
{
connect(input(), SIGNAL(globalPointerChanged(QPointF)), SLOT(pointerPosChanged(QPointF)));
}
AreaBasedEdge::~AreaBasedEdge()
{
}
void AreaBasedEdge::pointerPosChanged(const QPointF &pos)
{
if (!isReserved()) {
return;
}
const QPoint p = pos.toPoint();
if (approachGeometry().contains(p)) {
if (!isApproaching()) {
startApproaching();
} else {
updateApproaching(p);
}
} else {
if (isApproaching()) {
stopApproaching();
}
}
if (geometry().contains(p)) {
// we don't push the cursor back as pointer warping is not supported on Wayland
// TODO: this clearly needs improving
check(p, QDateTime(), true);
}
}
/**********************************************************
* ScreenEdges
*********************************************************/
KWIN_SINGLETON_FACTORY(ScreenEdges)
ScreenEdges::ScreenEdges(QObject *parent)
: QObject(parent)
, m_desktopSwitching(false)
, m_desktopSwitchingMovingClients(false)
, m_timeThreshold(0)
, m_reactivateThreshold(0)
, m_virtualDesktopLayout(0)
, m_actionTopLeft(ElectricActionNone)
, m_actionTop(ElectricActionNone)
, m_actionTopRight(ElectricActionNone)
, m_actionRight(ElectricActionNone)
, m_actionBottomRight(ElectricActionNone)
, m_actionBottom(ElectricActionNone)
, m_actionBottomLeft(ElectricActionNone)
, m_actionLeft(ElectricActionNone)
{
QWidget w;
m_cornerOffset = (w.physicalDpiX() + w.physicalDpiY() + 5) / 6;
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
connect(workspace(), &Workspace::clientRemoved, [this](KWin::Client *client) {
deleteEdgeForClient(client);
QObject::disconnect(client, &Client::geometryChanged,
ScreenEdges::self(), &ScreenEdges::handleClientGeometryChanged);
});
}
ScreenEdges::~ScreenEdges()
{
s_self = NULL;
}
void ScreenEdges::init()
{
reconfigure();
updateLayout();
recreateEdges();
}
static ElectricBorderAction electricBorderAction(const QString& name)
{
QString lowerName = name.toLower();
if (lowerName == QStringLiteral("dashboard")) {
return ElectricActionDashboard;
} else if (lowerName == QStringLiteral("showdesktop")) {
return ElectricActionShowDesktop;
} else if (lowerName == QStringLiteral("lockscreen")) {
return ElectricActionLockScreen;
} else if (lowerName == QStringLiteral("preventscreenlocking")) {
return ElectricActionPreventScreenLocking;
}
return ElectricActionNone;
}
void ScreenEdges::reconfigure()
{
if (!m_config) {
return;
}
// TODO: migrate settings to a group ScreenEdges
KConfigGroup windowsConfig = m_config->group("Windows");
setTimeThreshold(windowsConfig.readEntry("ElectricBorderDelay", 150));
setReActivationThreshold(qMax(timeThreshold() + 50, windowsConfig.readEntry("ElectricBorderCooldown", 350)));
int desktopSwitching = windowsConfig.readEntry("ElectricBorders", static_cast<int>(ElectricDisabled));
if (desktopSwitching == ElectricDisabled) {
setDesktopSwitching(false);
setDesktopSwitchingMovingClients(false);
} else if (desktopSwitching == ElectricMoveOnly) {
setDesktopSwitching(false);
setDesktopSwitchingMovingClients(true);
} else if (desktopSwitching == ElectricAlways) {
setDesktopSwitching(true);
setDesktopSwitchingMovingClients(true);
}
const int pushBack = windowsConfig.readEntry("ElectricBorderPushbackPixels", 1);
m_cursorPushBackDistance = QSize(pushBack, pushBack);
KConfigGroup borderConfig = m_config->group("ElectricBorders");
setActionForBorder(ElectricTopLeft, &m_actionTopLeft,
electricBorderAction(borderConfig.readEntry("TopLeft", "None")));
setActionForBorder(ElectricTop, &m_actionTop,
electricBorderAction(borderConfig.readEntry("Top", "None")));
setActionForBorder(ElectricTopRight, &m_actionTopRight,
electricBorderAction(borderConfig.readEntry("TopRight", "None")));
setActionForBorder(ElectricRight, &m_actionRight,
electricBorderAction(borderConfig.readEntry("Right", "None")));
setActionForBorder(ElectricBottomRight, &m_actionBottomRight,
electricBorderAction(borderConfig.readEntry("BottomRight", "None")));
setActionForBorder(ElectricBottom, &m_actionBottom,
electricBorderAction(borderConfig.readEntry("Bottom", "None")));
setActionForBorder(ElectricBottomLeft, &m_actionBottomLeft,
electricBorderAction(borderConfig.readEntry("BottomLeft", "None")));
setActionForBorder(ElectricLeft, &m_actionLeft,
electricBorderAction(borderConfig.readEntry("Left", "None")));
}
void ScreenEdges::setActionForBorder(ElectricBorder border, ElectricBorderAction *oldValue, ElectricBorderAction newValue)
{
if (*oldValue == newValue) {
return;
}
if (*oldValue == ElectricActionNone) {
// have to reserve
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if ((*it)->border() == border) {
(*it)->reserve();
}
}
}
if (newValue == ElectricActionNone) {
// have to unreserve
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if ((*it)->border() == border) {
(*it)->unreserve();
}
}
}
*oldValue = newValue;
// update action on all Edges for given border
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if ((*it)->border() == border) {
(*it)->setAction(newValue);
}
}
}
void ScreenEdges::updateLayout()
{
const QSize desktopMatrix = VirtualDesktopManager::self()->grid().size();
Qt::Orientations newLayout = 0;
if (desktopMatrix.width() > 1) {
newLayout |= Qt::Horizontal;
}
if (desktopMatrix.height() > 1) {
newLayout |= Qt::Vertical;
}
if (newLayout == m_virtualDesktopLayout) {
return;
}
if (isDesktopSwitching()) {
reserveDesktopSwitching(false, m_virtualDesktopLayout);
}
m_virtualDesktopLayout = newLayout;
if (isDesktopSwitching()) {
reserveDesktopSwitching(true, m_virtualDesktopLayout);
}
}
static bool isLeftScreen(const QRect &screen, const QRect &fullArea)
{
if (screens()->count() == 1) {
return true;
}
if (screen.x() == fullArea.x()) {
return true;
}
// the screen is also on the left in case of a vertical layout with a second screen
// more to the left. In that case no screen ends left of screen's x coord
for (int i=0; i<screens()->count(); ++i) {
const QRect otherGeo = screens()->geometry(i);
if (otherGeo == screen) {
// that's our screen to test
continue;
}
if (otherGeo.x() + otherGeo.width() <= screen.x()) {
// other screen is completely in the left
return false;
}
}
// did not find a screen left of our current screen, so it is the left most
return true;
}
static bool isRightScreen(const QRect &screen, const QRect &fullArea)
{
if (screens()->count() == 1) {
return true;
}
if (screen.x() + screen.width() == fullArea.x() + fullArea.width()) {
return true;
}
// the screen is also on the right in case of a vertical layout with a second screen
// more to the right. In that case no screen starts right of this screen
for (int i=0; i<screens()->count(); ++i) {
const QRect otherGeo = screens()->geometry(i);
if (otherGeo == screen) {
// that's our screen to test
continue;
}
if (otherGeo.x() >= screen.x() + screen.width()) {
// other screen is completely in the right
return false;
}
}
// did not find a screen right of our current screen, so it is the right most
return true;
}
static bool isTopScreen(const QRect &screen, const QRect &fullArea)
{
if (screens()->count() == 1) {
return true;
}
if (screen.y() == fullArea.y()) {
return true;
}
// the screen is also top most in case of a horizontal layout with a second screen
// more to the top. In that case no screen ends above screen's y coord
for (int i=0; i<screens()->count(); ++i) {
const QRect otherGeo = screens()->geometry(i);
if (otherGeo == screen) {
// that's our screen to test
continue;
}
if (otherGeo.y() + otherGeo.height() <= screen.y()) {
// other screen is completely above
return false;
}
}
// did not find a screen above our current screen, so it is the top most
return true;
}
static bool isBottomScreen(const QRect &screen, const QRect &fullArea)
{
if (screens()->count() == 1) {
return true;
}
if (screen.y() + screen.height() == fullArea.y() + fullArea.height()) {
return true;
}
// the screen is also bottom most in case of a horizontal layout with a second screen
// more below. In that case no screen starts below screen's y coord + height
for (int i=0; i<screens()->count(); ++i) {
const QRect otherGeo = screens()->geometry(i);
if (otherGeo == screen) {
// that's our screen to test
continue;
}
if (otherGeo.y() >= screen.y() + screen.height()) {
// other screen is completely below
return false;
}
}
// did not find a screen below our current screen, so it is the bottom most
return true;
}
void ScreenEdges::recreateEdges()
{
QList<Edge*> oldEdges(m_edges);
m_edges.clear();
const QRect fullArea(0, 0, displayWidth(), displayHeight());
for (int i=0; i<screens()->count(); ++i) {
const QRect screen = screens()->geometry(i);
if (isLeftScreen(screen, fullArea)) {
// left most screen
createVerticalEdge(ElectricLeft, screen, fullArea);
}
if (isRightScreen(screen, fullArea)) {
// right most screen
createVerticalEdge(ElectricRight, screen, fullArea);
}
if (isTopScreen(screen, fullArea)) {
// top most screen
createHorizontalEdge(ElectricTop, screen, fullArea);
}
if (isBottomScreen(screen, fullArea)) {
// bottom most screen
createHorizontalEdge(ElectricBottom, screen, fullArea);
}
}
// copy over the effect/script reservations from the old edges
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
Edge *edge = *it;
for (auto oldIt = oldEdges.constBegin();
oldIt != oldEdges.constEnd();
++oldIt) {
Edge *oldEdge = *oldIt;
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if (oldEdge->client()) {
// show the client again and don't recreate the edge
oldEdge->client()->showOnScreenEdge();
continue;
}
if (oldEdge->border() != edge->border()) {
continue;
}
const QHash<QObject *, QByteArray> &callbacks = oldEdge->callBacks();
for (QHash<QObject *, QByteArray>::const_iterator callback = callbacks.begin();
callback != callbacks.end();
++callback) {
edge->reserve(callback.key(), callback.value().constData());
}
}
}
qDeleteAll(oldEdges);
}
void ScreenEdges::createVerticalEdge(ElectricBorder border, const QRect &screen, const QRect &fullArea)
{
if (border != ElectricRight && border != KWin::ElectricLeft) {
return;
}
int y = screen.y();
int height = screen.height();
const int x = (border == ElectricLeft) ? screen.x() : screen.x() + screen.width() -1;
if (isTopScreen(screen, fullArea)) {
// also top most screen
height -= m_cornerOffset;
y += m_cornerOffset;
// create top left/right edge
const ElectricBorder edge = (border == ElectricLeft) ? ElectricTopLeft : ElectricTopRight;
m_edges << createEdge(edge, x, screen.y(), 1, 1);
}
if (isBottomScreen(screen, fullArea)) {
// also bottom most screen
height -= m_cornerOffset;
// create bottom left/right edge
const ElectricBorder edge = (border == ElectricLeft) ? ElectricBottomLeft : ElectricBottomRight;
m_edges << createEdge(edge, x, screen.y() + screen.height() -1, 1, 1);
}
// create border
m_edges << createEdge(border, x, y, 1, height);
}
void ScreenEdges::createHorizontalEdge(ElectricBorder border, const QRect &screen, const QRect &fullArea)
{
if (border != ElectricTop && border != ElectricBottom) {
return;
}
int x = screen.x();
int width = screen.width();
if (isLeftScreen(screen, fullArea)) {
// also left most - adjust only x and width
x += m_cornerOffset;
width -= m_cornerOffset;
}
if (isRightScreen(screen, fullArea)) {
// also right most edge
width -= m_cornerOffset;
}
const int y = (border == ElectricTop) ? screen.y() : screen.y() + screen.height() - 1;
m_edges << createEdge(border, x, y, width, 1);
}
Edge *ScreenEdges::createEdge(ElectricBorder border, int x, int y, int width, int height, bool createAction)
{
Edge *edge;
if (kwinApp()->operationMode() == Application::OperationModeX11) {
edge = new WindowBasedEdge(this);
} else {
edge = new AreaBasedEdge(this);
}
edge->setBorder(border);
edge->setGeometry(QRect(x, y, width, height));
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if (createAction) {
const ElectricBorderAction action = actionForEdge(edge);
if (action != KWin::ElectricActionNone) {
edge->reserve();
edge->setAction(action);
}
}
if (isDesktopSwitching()) {
if (edge->isCorner()) {
edge->reserve();
} else {
if ((m_virtualDesktopLayout & Qt::Horizontal) && (edge->isLeft() || edge->isRight())) {
edge->reserve();
}
if ((m_virtualDesktopLayout & Qt::Vertical) && (edge->isTop() || edge->isBottom())) {
edge->reserve();
}
}
}
connect(edge, SIGNAL(approaching(ElectricBorder,qreal,QRect)), SIGNAL(approaching(ElectricBorder,qreal,QRect)));
if (edge->isScreenEdge()) {
connect(this, SIGNAL(checkBlocking()), edge, SLOT(checkBlocking()));
}
return edge;
}
ElectricBorderAction ScreenEdges::actionForEdge(Edge *edge) const
{
switch (edge->border()) {
case ElectricTopLeft:
return m_actionTopLeft;
case ElectricTop:
return m_actionTop;
case ElectricTopRight:
return m_actionTopRight;
case ElectricRight:
return m_actionRight;
case ElectricBottomRight:
return m_actionBottomRight;
case ElectricBottom:
return m_actionBottom;
case ElectricBottomLeft:
return m_actionBottomLeft;
case ElectricLeft:
return m_actionLeft;
default:
// fall through
break;
}
return ElectricActionNone;
}
void ScreenEdges::reserveDesktopSwitching(bool isToReserve, Qt::Orientations o)
{
if (!o)
return;
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
Edge *edge = *it;
if (edge->isCorner()) {
isToReserve ? edge->reserve() : edge->unreserve();
} else {
if ((m_virtualDesktopLayout & Qt::Horizontal) && (edge->isLeft() || edge->isRight())) {
isToReserve ? edge->reserve() : edge->unreserve();
}
if ((m_virtualDesktopLayout & Qt::Vertical) && (edge->isTop() || edge->isBottom())) {
isToReserve ? edge->reserve() : edge->unreserve();
}
}
}
}
void ScreenEdges::reserve(ElectricBorder border, QObject *object, const char *slot)
{
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if ((*it)->border() == border) {
(*it)->reserve(object, slot);
}
}
}
void ScreenEdges::unreserve(ElectricBorder border, QObject *object)
{
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if ((*it)->border() == border) {
(*it)->unreserve(object);
}
}
}
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
void ScreenEdges::reserve(Client *client, ElectricBorder border)
{
auto it = m_edges.begin();
while (it != m_edges.end()) {
if ((*it)->client() == client) {
if ((*it)->border() == border) {
if (client->isHiddenInternal() && !(*it)->isReserved()) {
(*it)->reserve();
}
return;
} else {
delete *it;
it = m_edges.erase(it);
}
} else {
it++;
}
}
createEdgeForClient(client, border);
connect(client, &Client::geometryChanged, this, &ScreenEdges::handleClientGeometryChanged);
}
void ScreenEdges::createEdgeForClient(Client *client, ElectricBorder border)
{
int y = 0;
int x = 0;
int width = 0;
int height = 0;
const QRect geo = client->geometry();
const QRect fullArea = workspace()->clientArea(FullArea, 0, 1);
for (int i = 0; i < screens()->count(); ++i) {
const QRect screen = screens()->geometry(i);
if (!screen.contains(geo)) {
// ignoring Clients having a geometry overlapping with multiple screens
// this would make the code more complex. If it's needed in future it can be added
continue;
}
const bool bordersTop = (screen.y() == geo.y());
const bool bordersLeft = (screen.x() == geo.x());
const bool bordersBottom = (screen.y() + screen.height() == geo.y() + geo.height());
const bool bordersRight = (screen.x() + screen.width() == geo.x() + geo.width());
if (bordersTop && border == ElectricTop) {
if (!isTopScreen(screen, fullArea)) {
continue;
}
y = geo.y();
x = geo.x();
height = 1;
width = geo.width();
break;
}
if (bordersBottom && border == ElectricBottom) {
if (!isBottomScreen(screen, fullArea)) {
continue;
}
y = geo.y() + geo.height() - 1;
x = geo.x();
height = 1;
width = geo.width();
break;
}
if (bordersLeft && border == ElectricLeft) {
if (!isLeftScreen(screen, fullArea)) {
continue;
}
x = geo.x();
y = geo.y();
width = 1;
height = geo.height();
break;
}
if (bordersRight && border == ElectricRight) {
if (!isRightScreen(screen, fullArea)) {
continue;
}
x = geo.x() + geo.width() - 1;
y = geo.y();
width = 1;
height = geo.height();
break;
}
}
if (width > 0 && height > 0) {
Edge *edge = createEdge(border, x, y, width, height, false);
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
edge->setClient(client);
m_edges.append(edge);
if (client->isHiddenInternal()) {
edge->reserve();
}
} else {
// we could not create an edge window, so don't allow the window to hide
client->showOnScreenEdge();
}
}
void ScreenEdges::handleClientGeometryChanged()
{
Client *c = static_cast<Client*>(sender());
deleteEdgeForClient(c);
c->showOnScreenEdge();
}
void ScreenEdges::deleteEdgeForClient(Client* c)
{
auto it = m_edges.begin();
while (it != m_edges.end()) {
if ((*it)->client() == c) {
delete *it;
it = m_edges.erase(it);
} else {
it++;
}
}
}
void ScreenEdges::check(const QPoint &pos, const QDateTime &now, bool forceNoPushBack)
{
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
bool activatedForClient = false;
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
if (!(*it)->isReserved()) {
continue;
}
if ((*it)->approachGeometry().contains(pos)) {
(*it)->startApproaching();
}
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if ((*it)->client() != nullptr && activatedForClient) {
(*it)->markAsTriggered(pos, now);
continue;
}
if ((*it)->check(pos, now, forceNoPushBack)) {
if ((*it)->client()) {
activatedForClient = true;
}
}
}
}
bool ScreenEdges::isEntered(xcb_enter_notify_event_t *event)
{
return handleEnterNotifiy(event->event,
QPoint(event->root_x, event->root_y),
QDateTime::fromMSecsSinceEpoch(event->time));
}
bool ScreenEdges::isEntered(xcb_client_message_event_t *event)
{
if (event->type != atoms->xdnd_position) {
return false;
}
return handleDndNotify(event->window,
QPoint(event->data.data32[2] >> 16, event->data.data32[2] & 0xffff));
}
bool ScreenEdges::handleEnterNotifiy(xcb_window_t window, const QPoint &point, const QDateTime &timestamp)
{
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
bool activated = false;
bool activatedForClient = false;
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
WindowBasedEdge *edge = dynamic_cast<WindowBasedEdge*>(*it);
if (!edge) {
continue;
}
if (!edge->isReserved()) {
continue;
}
if (edge->window() == window) {
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if (edge->check(point, timestamp)) {
if ((*it)->client()) {
activatedForClient = true;
}
}
activated = true;
break;
}
if (edge->approachWindow() == window) {
edge->startApproaching();
// TODO: if it's a corner, it should also trigger for other windows
return true;
}
}
Screenedge show support for Clients This provides a new protocol intended to be used by auto-hiding panels to make use of the centralized screen edges. To use it a Client can set an X11 property of type _KDE_NET_WM_SCREEN_EDGE_SHOW to KWin. As value it takes: * 0: top edge * 1: right edge * 2: bottom edge * 3: left edge KWin will hide the Client (hide because unmap or minimize would break it) and create an Edge. If that Edge gets triggered the Client is shown again and the property gets deleted. If the Client doesn't border the specified screen edge the Client gets shown immediately so that we never end in a situation that we cannot unhide the auto-hidden panel again. The exact process is described in the documentation of ScreenEdges. The Client can request to be shown again by deleting the property. If KWin gets restarted the state is read from the property and it is tried to create the edge as described. As this is a KWin specific extension we need to discuss what it means for Clients using this feature with other WMs: it does nothing. As the Client gets hidden by KWin and not by the Client, it just doesn't get hidden if the WM doesn't provide the feature. In case of an auto-hiding panel this seems like a good solution given that we don't want to hide it if we cannot unhide it. Of course there's the option for the Client to provide that feature itself and if that's wanted we would need to announce the feature in the _NET_SUPPORTED atom. At the moment that doesn't sound like being needed as Plasma doesn't want to provide an own implementation. The implementation comes with a small test application showing how the feature is intended to be used. REVIEW: 115910
2014-02-20 15:39:23 +04:00
if (activatedForClient) {
for (auto it = m_edges.constBegin(); it != m_edges.constEnd(); ++it) {
if ((*it)->client()) {
(*it)->markAsTriggered(point, timestamp);
}
}
}
return activated;
}
bool ScreenEdges::handleDndNotify(xcb_window_t window, const QPoint &point)
{
for (auto it = m_edges.begin(); it != m_edges.end(); ++it) {
WindowBasedEdge *edge = dynamic_cast<WindowBasedEdge*>(*it);
if (!edge) {
continue;
}
if (edge->isReserved() && edge->window() == window) {
updateXTime();
edge->check(point, QDateTime::fromMSecsSinceEpoch(xTime()), true);
return true;
}
}
return false;
}
void ScreenEdges::ensureOnTop()
{
Xcb::restackWindowsWithRaise(windows());
}
QVector< xcb_window_t > ScreenEdges::windows() const
{
QVector<xcb_window_t> wins;
for (auto it = m_edges.constBegin();
it != m_edges.constEnd();
++it) {
WindowBasedEdge *edge = dynamic_cast<WindowBasedEdge*>(*it);
if (!edge) {
continue;
}
xcb_window_t w = edge->window();
if (w != XCB_WINDOW_NONE) {
wins.append(w);
}
// TODO: lambda
w = edge->approachWindow();
if (w != XCB_WINDOW_NONE) {
wins.append(w);
}
}
return wins;
}
} //namespace