| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
|
| |
This widget is treated as a window by Qt, but with the embedded flag ON.
setGeometry should not clip this window, it will be cliped automatically
by the parent view of QMacNativeWidget.
Task-number: QTBUG-9199
Reviewed-by: Denis
|
|
|
|
|
| |
Instead of updating the bounding box, we iterate over the rectangles of
the region. We also keep early alien support.
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
4.7-integration
* '4.7' of scm.dev.nokia.troll.no:qt/oslo-staging-1: (72 commits)
different approach to fixing "the other" aliasing issue
fix aliasing issue in node_construct()
detach in fewer cases, remove redundant calculation
SSL: Fix memleak related to local certificate
Improve keyboard layout detection on X11
Compile on ARM with -Werror -Wold-style-cast
Use the vista-style native dialog for QFileDialog::getExistingDirectory
Apply the stdset attribute for resource properties
doc: Completed sentence about HideNameFilterDetails
Doc fix in QLocale
Doc for for QGestureRecognizer::create.
Fix broken setCollectionFile (creating multiple models instead reusing)
Enable the fast paths when converting to Rgb565
qdoc: ifdefed out all the debug code.
Add missing QtScript API shims
Further fix license whitespace.
Remove NetworkManager test-bed application.
Remove configure test for NetworkManager.
Fix QDir::entryList regression
Add DEFINES to mingw32/windres.exe command line.
...
|
| | |
|
| | |
|
|\ \
| |/
|/|
| |
| | |
Conflicts:
src/gui/styles/qs60style_s60.cpp
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The problem is not related to the Splitter handle but related to an
optimization in the paintengine.
We were comparing if the size of the widget had changed before setting
the new mask, which might not be correct. We could have changed the mask
without changing the widget size.
Task-number: QTCREATORBUG-753
Reviewed-by: Morten
Reviewed-by: Samuel
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
4.7-integration
* '4.7' of scm.dev.nokia.troll.no:qt/oslo-staging-1: (57 commits)
fix corewlan detection error when building for 10.5 when 10.6 is also
fixed treatment of zlib on Mac when crossbuilding
Partial overloading support for qdbus cli tool.
Allow empty authority in QUrl::setAuthority as per docs.
Added test for QTBUG-6962: Empty authority ignored by QUrl::setAuthority.
fixed case of GL include directory
check in MAC_APPLICATION_MENU translations
Re-generate the Unicode tables after updates to the program that generates them
Fix the code after merge: DerivedNormalizationProps has two or more columns
add some usefull definitions to qunicodetables_p.h
qchar.cpp: fix identation
finish last commit
prefer DerivedNormalizationProps.txt over CompositionExclusions.txt
improve error reporting a bit more
improve error reporting
fix incorect condition
check if string to int conversions were done w/o errors
improve error reporting for unassigned grapheme/word/sentence break classes
avoid using of qunicodetables_p.h in generator
use QHash for line break map
...
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This was never implemented, and as it stands, just creating a
QFontDialog and calling show on it will not show anything. This
patch makes sure that show works, and that one can have more than
one font dialog showing at the same time (but only the first one will
be native)
Reviewed-by: cduclos
|
|\ \ \
| |/ /
|/| /
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
configure.exe
examples/multimedia/audioinput/audioinput.cpp
src/corelib/io/qfsfileengine.cpp
src/gui/egl/qegl_wince.cpp
src/gui/egl/qeglproperties.cpp
src/gui/egl/qeglproperties_p.h
src/gui/embedded/directfb.pri
src/gui/kernel/qapplication_win.cpp
src/gui/painting/qdrawutil.cpp
src/opengl/qgl_p.h
src/sql/drivers/odbc/qsql_odbc.cpp
src/sql/drivers/odbc/qsql_odbc.h
tests/auto/auto.pro
tests/auto/qgl/tst_qgl.cpp
translations/assistant_adp_ru.ts
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Before doing anything we need to make sure that we don't leave anything
in a non-consistent state. When hiding a widget we need to make sure that
no mouse_down events are active, because the mouse_up event will never be
received by a hidden widget or one of its descendants. The solution is
simple, before going through with this we check if there are any
mouse_down events in progress, if so we check if it is related to this
widget or not. If so, we just reset the mouse_down and then we continue.
In X11 and Windows we send a mouse_release event, however we don't do that
here because we were already ignoring that from before. I.e. Carbon did
not send the mouse release event, so we will not send the mouse release
event. There are two ways to interpret this:
1. If we don't send the mouse release event, the widget might get into an
inconsistent state, i.e. it might be waiting for a release event that
will never arrive.
2. If we send the mouse release event, then the widget might decide to
trigger an action that is not supposed to trigger because it is not visible.
Task-number: QTBUG-8604
Reviewed-by: denis
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The problem occurs because we were drawing the pixmap and then clearing
the buffer. This patch basically moves the clearing of the buffer to
the start of the process instead of having it in the middle.
Task-number: QTBUG-7800
Reviewed-by: Morten Sorvig
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This commit makes alien widgets opt in on a
per-widget basis on Mac, set the Qt::WA_NativeWindow
flag when creating the widget to enable. Setting
this flag on widgets that have native child or sibling
NSViews is not supported.
The main use case for alien widgets on Mac is to
improve performance for applications that have complex
user interfaces. Qt can handle thousands of widgets
per window, while Cocoa is designed to use a smaller
number of NSViews in combination with NSCells and
custom control implementations. This commit moves
us in the direction of having a few main NSViews with
"leaf" qwidgets implemented as a custom control.
|
|\ \
| |/
| |
| |
| |
| | |
Conflicts:
qmake/generators/symbian/initprojectdeploy_symbian.cpp
qmake/generators/symbian/symmake_abld.h
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The optimization created some glitches. It now
works properly but it could be improved
performance-wise.
Task-number: QTBUG-5926
Task-number: QTBUG-5994
Reviewed-by: Morten Sorvig
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It turns out that setParent_sys is wiping out opacity for the
window regardless of what the value is from before. This patch
does the correct thing, namely look at the WA_WState_WindowOpacitySet
flag.
Task-number: QTBUG-5100
Reviewed-by: cduclos
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Window modal dialogs are shown as sheets on Mac, reusing them is showing
some painting artificats. So make sure we create a new window everytime
a sheet is shown. This only applies to Carbon, the Cocoa versions dont
have this problem.
Task-number: QTBUG-8198
Reviewed-by: MortenS
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If one opens a menu with a submenu, the submenu will be shown at a
wrong position on screen. The reason is that upon show, the top menu
will be centered on screen. But the child menu will be shown first, and
tries to center of the parent that has not yet been shown. So moving the
top menu upon show will fool the subwindow.
This patch will move the center code into the setGeometry_sys function
so that any resize on the window before it has been shown will re-center
the window correctly.
Task-number: QTBUG-2196
Reviewed-by: prasanth
|
|\ \
| |/
| |
| |
| |
| | |
Conflicts:
src/gui/kernel/qcocoapanel_mac.mm
src/gui/kernel/qcocoasharedwindowmethods_mac_p.h
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This fixes the qmlviewer "sluggish animations and lost
mouse events" issue by making sure we don't block and
wait for for the screen refresh when flushing the
backing store to the screen.
NB: This commit fixes build issues found in f5f62c0bed.
Review: msorvig
Details:
- Don't force repaints, flush the backingstore in response
to a Cocoa paint/display events only.
- Flush once per window.
- Get the CGContext from the window (don't create a new one)
- Don't call CGContextiFlush on the context.
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
See change: f842ec12706. Needed some ekstra checks for
the cocoa port as well.
Task-number: QTBUG-4227
Reviewed-by: Prasanth
(cherry picked from commit ebf9c943b789bb4ce1e1222ed17cc37bd0b1f1fe)
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The crash occurs because we at one point try to assign a widget that
is marked as 'about to be deleted' to a QPointer, after clearguards
has been called. The correct fix is to avoid such an assignment in
the first place.
Task-number: QTBUG-4227
Reviewed-by: Prasanth
(cherry picked from commit f842ec12706b70f94ab5f634dc2aa025ba2cf3f2)
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The example that shows this off is tools/inputpanel. This example is
strangely written, and calls activateWindow on the main window in an
activate callback to another window. So a bit recursion will happend,
and this caused troubble for the cocoa port. In addition, in Qt several
windows can be active at the same time, expecially tool windows. This
means that we need to be careful when returning early from the
becomeKeyWindow methods for NSWindow. This patch fixes up on this stuff.
Task-number: QTBUG-6002
Reviewed-by: msorvig
|
|\ \
| |/
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/gui/kernel/qeventdispatcher_mac.mm
src/gui/kernel/qt_cocoa_helpers_mac.mm
src/gui/widgets/qmenu_mac.mm
tests/auto/qgraphicswidget/tst_qgraphicswidget.cpp
tools/assistant/tools/assistant/centralwidget.cpp
tools/linguist/lupdate/main.cpp
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
menu hiding.
The problem here was the way we entered Full Screen Mode. We were using
"kUIModeAllSuppressed" which does the following according to the manual:
kUIModeAllSuppressed
All system UI elements (including the menu bar) are hidden. However,
these elements may automatically show themselves in response to mouse
movements or other user activity.
I changed it to "kUIModeAllHidden", which does the following:
All system UI elements (including the menu bar) are hidden.
To prevent a change of behavior I added the following option to the
SetSystemUIMode:
kUIOptionAutoShowMenuBar
This flag specifies that the menu bar automatically shows itself when
the user moves the mouse into the screen area that would ordinarily be
occupied by the menu bar.
Only valid for the presentation mode kUIModeAllHidden.
Task-number: QTBUG-7625
Reviewed-by: Richard Moe Gustavsen
|
| |
| |
| |
| |
| |
| |
| |
| | |
See change: f842ec12706. Needed some ekstra checks for
the cocoa port as well.
Task-number: QTBUG-4227
Reviewed-by: Prasanth
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The crash occurs because we at one point try to assign a widget that
is marked as 'about to be deleted' to a QPointer, after clearguards
has been called. The correct fix is to avoid such an assignment in
the first place.
Task-number: QTBUG-4227
Reviewed-by: Prasanth
|
| |
| |
| |
| |
| |
| |
| | |
This have never worked in the Cocoa port, it seems. This patch gives
it a try.
Reviewed-by: msorvig
|
| |
| |
| |
| | |
Just forgot an ifdef...
|
|\ \
| |/
| |
| |
| | |
Conflicts:
tools/assistant/lib/qhelpsearchquerywidget.cpp
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The problem here was exiting the full screen mode. In Cocoa we don't
activate windows as we do in Carbon, therefore we were not exiting from
the full screen mode. This patch adds a check when hiding a window, if
the window is in full screen mode then we go through the list of top
level windows checking if there are any other visible and not-minimized
windows that are also in full screen mode. If none if found, the we exit
the full screen mode.
Task-number: QTBUG-7312
Reviewed-by: Prasanth
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The crash happens when a QMainWindow with a native toolbar is reparented.
While re-parenting, the QToolBar triggers a new setParent() with the
MacWindowToolBarButtonHint. This messes up the internal state, hence the
crash. To avoid re-entering setParent() for the QMainWindow, this hint
is set depending on the unifiedTitleAndToolBarOnMac property.
This patch also fixes a related issue in Cocoa, we need to remove the
reused NSToolbar from the old window while recreating the QMainWindow.
Task-number: QTBUG-7162
Reviewed-by: Richard Moe Gustavsen
|
| |
| |
| |
| |
| |
| | |
Seems like this were never implemented. This patch does.
Task-number: QTBUG-7375
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
At one point in history, we decided that QDialogs should stack just
above normal windows. This patch reverts this behaviour, as cocoa
gets confused if we stack a dialog at this level without telling it
to be modal. So instead, we let cocoa set the modal window level when
we create a modal session, and also let cocoa restore the level to
what it were when the sessions ends. That way, we avoid doing the work
that cocoa is allready designed to do.
Task-number: QTBUG-5057
Reviewed-by: msorvig
|
|\ \
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
bin/syncqt
doc/src/deployment/deployment.qdoc
src/corelib/io/qfsfileengine_win.cpp
src/corelib/xml/qxmlstream.cpp
src/opengl/gl2paintengineex/qpaintengineex_opengl2_p.h
tools/assistant/tools/assistant/centralwidget.cpp
tools/linguist/lupdate/main.cpp
|
| |
| |
| |
| | |
Reviewed-by: Trust Me
|
| |
| |
| |
| |
| | |
Fix warning given by cocoa
(cherry picked from commit 875afab977005b03d307040fb3be15c7524a37ff)
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This is because a modal flag on the window wins over the type when
determining window class. This fix will respect the type first, before
looking at the window flag.
Task-number: QTBUG-1455
Reviewed-by: Prasanth
|
|\ \
| |/
| |
| |
| |
| |
| | |
Conflicts:
examples/webkit/fancybrowser/main.cpp
src/opengl/gl2paintengineex/qpaintengineex_opengl2.cpp
tools/assistant/tools/assistant/bookmarkmanager.cpp
|
| |
| |
| |
| |
| |
| |
| | |
The icon was not released when destroying the window.
Task-number: QTBUG-6973
Reviewed-by: Richard Moe Gustavsen
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Under some circumstances, the event dispatcher needs to stop
NSApplication, let the control return back to the QEventLoop that
called processEvents, which very often will just call processEvents
again, which in turn will restart NSApplication.
After stopping NSApplication, Cocoa seems to throw away at least some
of the posted cocoa events that are still pending. This will offcourse
have inpact on things like window updates etc. that never happends.
This patch will be more careful about when to stop NSApplication, and
delay this until the run loop has finished one more spin after interrupt
has been called. The same goes for modal sessions. Introducing this delay
also made it neccessary to sometimes block cocoa from updating the
stacking order of the windows.
Autotest: qcoreapplication, qapplication, qtimer
|
| |
| |
| |
| | |
Fix warning given by cocoa
|
|\ \
| |/
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/io/qfsfileengine.cpp
src/network/access/qnetworkrequest.cpp
tests/auto/qgraphicswidget/tst_qgraphicswidget.cpp
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The QPlainTextEdit can change the scroll ranges while dragging the
scrollbar. This will eventualy call QWidget::raise(), on Cocoa it was
done by removing the NSView and adding it back. This causes problems
like resetting internal state while a mouseDragged was active on the
view. The fix we will now sort the views based on their Qt-z-order.
lower() & stackUnder() also fixed like this.
Reviewed-by: Denis
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
| |
It turns out that registering drag types for each NSView that can
receive drop events is _really_ slow. And many widget in Qt
subscribe for DnD (QTextEdit, QScrollArea, etc), so the result is
an application that will spend startup time preparing for DnD. For
some edge cases, we're talking several seconds!
This patch removes this overhead by moving drag type registering
out of NSView, and into NSWindow (that is, QCocoaWindow and
QCocoaPanel).
Task-number: QT-1586
Reviewed-by: Prasanth
|
|
|
|
|
|
|
|
|
|
|
|
| |
On Mac QWidget::destroy() sends an AcceptDropsChange event after
clearing the guards for QPointer. This was used to store a QPointer to
the widget being deleted & that will never be cleared.
The fix removed the setAcceptDrops() from destroy. And as an extra
protection make sure designer will not treat that event as interesting.
Task-number: QTCREATORBUG-307
Reviewed-by: Denis Dzyubenko
Reviewed-by: Friedemann Kleint
|
|
|
|
|
|
|
|
|
|
|
| |
We build the release packages for Carbon against the 10.4 SDK.
This means that the constants used for dealing with gestures are
missing, and currently the implementation will be ifdeffed out.
This patch defines these constants when building with an
SDK < 10.6. Gestures will therefore work when running applications
on 10.6. The same fix is also implemented for cocoa.
Rev-By: MortenS
|
|
|
|
|
|
|
|
|
| |
When we reparent a widget, it has to be removed from the old view before
it is added to the new one. Otherwise this can cause the old view to
have an invalid first responder, causing crash.
Task-number: QTBUG-5327
Reviewed-by: Denis
|