| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When creating a single dialog in the main function, and tell it
to exec, we run a modal dialog. But there is really no other
window on screen to be modal for. So in that case, since this is
a rather common pattern for Qt applications, we allow users to
quit the application from the dock. But this action is sendt as
an apple event. And and that point in time, cocoa has the the
apple event handler, and refuses to close down the application
because it detects a modal window.
Our solution is to install/overwrite the apple event handler for
kAEQuit _after_ cocoa has finished its own installation. But in
order to do this, we need to wait until [NSApplication run] has
started, otherwise it will not take effect. And that is what this
patch essentially does.
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
QFileSystemModel provides the same features as QDirModel but in a
efficient way so no need to keep QDirModel around.
This commit mark as obsolete QDirModel. Examples has been ported to
use QFileSystemModel instead. QCompleter also gain the support of
QFileSystemModel to complete nicely a line edit.
This commit also add a new signal in QFileSystemModel when the directory
is loaded, i.e. the thread has finished to read the current dir.
Task-number:QTBUG-3884
Reviewed-by:ogoffart
Reviewed-by:gabi
Reviewed-by:mbm
|
| |\
| | |
| | |
| | |
| | |
| | |
| | | |
Conflicts:
src/corelib/io/qfsfileengine.cpp
src/opengl/gl2paintengineex/qpaintengineex_opengl2.cpp
src/opengl/opengl.pro
|
| | |\
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-multimedia-team:
QAudioInput loses data on Maemo5/other linux platforms.
Fix warning in qfilesystemmodel_p.h
Frequency to SampleRate and channels to channelCount.
Added setChannelCount() to QAudioFormat.
|
| | | |\
| | | | |
| | | | |
| | | | |
| | | | | |
Conflicts:
examples/multimedia/audiodevices/audiodevices.cpp
|
| | | | |
| | | | |
| | | | |
| | | | | |
Reviewed-by: Rhys Weatherley
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
QSet is a hash internally, using Iterator::begin while erasing elements
inside the set might create holes and then the complexity increase.
We now use the return value of erase (the next element) so the
complexity is linear.
For those who create/delete item in the polish event (BAD), _q_polishItem
might be slower than the normal call.
Task-number:QTBUG-6958
Reviewed-by:olivier
|
| | | |/
| | |/|
| | | |
| | | |
| | | | |
Task-number: QTBUG-7161
Reviewed-by: Daniel Pope
|
| | |\ \
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/oslo-staging-2: (42 commits)
QBoxLayout::setGeometry would not respect the widget min/max width
Revert "QStyleSheetStyle: Fixed some text croped when having padding with native border."
Use QFile:rename when moving items in QFileystemModel.
Revert "Add GLfloat[2][2] & GLfloat[3][3] uniform setters to QGLShaderProgram"
Fix default filter selection when using HideNameFilterDetails option.
Don't write out fo:word-spacing if its the default value.
Improved initial startup time for a QGLWidget ontop of EGL/X11.
Document the QGraphicsView::IndirectPainting flag
Display broken symlinks in the filesystem model.
Fix typo in autotest testcase name.
Fixed a bug with distribution of spans.
Make unit test more robust
Compile with QT_NO_DOCKWIDGET
Removed temporary QGLWidget created during QGLWidget/X11 initialization.
Fix test: The bug is now fixed
Fix auto-test failure on Windows
QScript: Lookup the native setter from the prototype
Implement QScript::QObjectDelegate::getOwnPropertyDescriptor
fix compilation in GL2 paint engine for Windows
Move QGLTextureGlyphCache into it's own file
...
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
When calling heightforWidth after a geometry change, the width actually used
could be outside the widget's width bounds. The height could then be smaller
than needed to fit the widget's contents resulting in a clipped widget being
drawn. Auto-test included.
Reviewed-by: Olivier
Task-number: QTBUG-7103
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
native border."
This reverts commit 6e90192b599cee9b903177a0978198326f667613.
his change the behaviour a little bit. It is better to wait for Qt 4.7
Reviewed-by: jbache
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
We got for free the recursive move if the directory that we move has
subdirectories. Unfortunately copying a directory to an another is
not easy in Qt so the copy still doesn't work if the directory contains
directories.
Task-number:QTBUG-5855
Reviewed-by:ogoffart
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
When you have HideNameFilterDetails on, comparing the default filter
given in parameter for selectNameFilter has to be done with the striped
version of the filter, i.e. without the details.
Task-number:QTBUG-4842
Reviewed-by:gabriel
Reviewed-by:olivier
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
the property fo:word-spacing is still not in an official ODF spec so
a strict schema check will fail in any doc that uses it. Lets make sure
we limit the writing of that property to only those usecases where the
property was really set by the user.
Task-number: QTBUG-5725
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
And that the QGraphics{View,Scene}::drawItems function are now obsolete.
Reviewed-by: Alexis
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
A broken symlink has a -1 size so we need to special case that.
Task-number:QTBUG-7119
Reviewed-by:olivier
|
| | | |\ \ |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
If a span required more size than the rows/columns it spanned, the
size of the span was not distributed to the rows it spanned. The
result was that the size hints of the layout was not correct,
causing the layout to be potentially smaller than the spanning item.
Task: QT-2261
Reviewed-by: Alexis
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-7133
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This option was simply not implemented at all so it didn't work.
Task-number:QTBUG-6558
Reviewed-by:ogoffart
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
We should check for the fully opaque and fully transparent special
cases, like we do in the dedicated image blend functions.
Reveiewed-by: Gunnar Sletta
|
| | | | | | |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Certain boards are not support StretchBlit very well. This patch enables
them to define QT_NO_DIRECTFB_STRETCHBLIT to fall back to the raster
engine for stretchblits.
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
QX11PixmapData::transformed initializes a new QX11PixmapData object but
doesn't set its serial number.
Reviewed-by: Donald Carr <donald.carr@nokia.com>
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-7047
Reviewed-by: Benjamin Poulain
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The problem was that when the background cache was invalidated, it was
entirely recreated but only the exposed area of the view was repainted
in it, causing the cache to be partly empty in some cases. Now the
background cache is always fully repainted when it is invalidated.
Task-number: QTBUG-6935
Reviewed-by: ogoffart
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-6474
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-6637
|
| | |\ \ \ \ |
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
When accessing the global input context from the QWidget destructor access it
directly instead of calling a helper function. Don't even bother to create an
input context if QApplication is being destroyed (just in case if the user is
calling the QApplication::inputContext manually from the destructor).
Task-number: QTBUG-7105
Reviewed-by: Simon Hausmann
|
| | |/ / / /
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-6763
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-6810
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-7118
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-6992
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Fix warning given by cocoa
(cherry picked from commit 875afab977005b03d307040fb3be15c7524a37ff)
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
I removed a line as an optimization a while ago. Luckily, the
autotest cought a failure doing so. So we put the line back in.
Reviewed-by: Prasanth
|
| | |\ \ \ \
| | | |/ / / |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Task-number: QTBUG-5046
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This is because the light() color of the palette is used and it was
not modified.
Task-number: QTBUG-7216
Reviewed-by: jbache
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This patch draws the focus frame for widgets other than QLineEdit (e.g.
for line edits in Webkit). The frame is drawn outside the rectangle
passed in the option-rect.
Task-number: QTBUG-5530
Reviewed-by: MortenS
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The signal was being emitted both from the carbon event handler and the
platform independent code of QMenu. The one in the event handler is now
removed.
Task-number: 6741
Reviewed-by: Richard Moe Gustavsen
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This problem has been around before. At that time it had to do
with what part of the application received the kEAQuit apple event.
In Qt-4.7 we ensure that NSApplication gets run at least once
before we do any other event processing in Qt. That way the menu bar
gets fully initialized. The backside of the coin is that NSApplication
will not let the quit menu send a terminate action to the menu nibs file
owner (qmenuloader_mac.mm) when a modal window is showing on screen.
This is normally expected behaviour, but in Qt we allow the user to quit
when just a single window/dialog is showing on screen.
This fix remaps the quit items target from files owner to first responder.
That way we can catch the terminate at the window/responder beeing that,
and terminate as needed.
Task-number: QTBUG-7197
Reviewed-by: Prasanth
|
|/ / / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This will catch releases done between creating the
QApplication object and calling QApplication::exec.
(The Cocoa event loop creates release pools for us
when it's running.)
The memory used by the objects stored in this pool
won't be released until the QApplication instance
is deleted at application exit. Because of this it
still makes sense to still add local release pools
when doing potentially large allocations.
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
QImage::bits() and QImage::scanLine() can create deep copies
if the QImage is not const. The constBits() and constScanLine()
functions make things more explicit and less error-prone.
Reviewed-by: Daniel Pope
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The invalidate connection is made before connecting the source model,
which means it gets called first.
If a second proxy model is used with it, the sourceLayoutAboutToBeChanged
of the second proxy is therefore called before its sourceReset slot, but
after the persistent indexes of the first model have been invalidated.
The invalidate call is not needed because clear_mappings is called
when the source model is reset anyway.
Merge-request: 416
Reviewed-by: Olivier Goffart <ogoffart@trolltech.com>
|
|\ \ \ \ \
| | |/ / /
| |/| | |
| | | | |
| | | | | |
Conflicts:
src/opengl/gl2paintengineex/qpaintengineex_opengl2.cpp
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Happens for example if a DockWidget is undocked and has a child
whith the WA_StaticContents attribute.
The parent does not change (so newParent is false) but still, the
top level widget change. So staticWidget need to be moved to the
new backingstore.
Reviewed-by: Benjamin Poulain
Task-number: QTBUG-6883
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Calling metric() to get the device height in mm of a pixmap when using
the raster engine would actually return the width..
Task-number: QTBUG-6985
Reviewed-by: Carlos Duclos
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The native border width need to be taken into account in QRenderRule::boxRect()
Task-number: QTBUG-6855
Reviewed-by: Benjamin Poulain
|