| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
Task-number: 245650
Reviewed-by: Gunnar
|
|
|
|
|
|
|
|
| |
everything in the source model will be accepted.
Task: 251308
Reviewed-By: Olivier Goffart
|
|
|
|
| |
Reviewed-By: TrustMe
|
|
|
|
|
|
|
|
|
|
|
|
| |
New flag: QGraphicsItem::ItemNegativeZStacksBehindParent, which makes
it easy to toggle stack-behind based on the value of Z alone.
Add interface initializations to QGV classes.
Add a simple internal focus policy to QGraphicsItem to allow derived
items to be focusable without allowing clickfocus.
Reviewed-by: Alexis
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
The signal was being associated with a particular property,
but in many classes, the NOTIFY signal applies to multiple
properties. Added a new function to the PropertyNode class
that adds the signal function without associating it with
any property.
Task-number: 259071
|
|
|
|
|
| |
the test was wrong on the QPrinter pointer
In addition, priv is never null
|
|
|
|
|
|
| |
legacy and completely pointless...
Reviewed-By: Eskil
|
| |
|
|
|
|
| |
Reviewed-by: Eskil
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
setTransformationAnchor(QGraphicsView::AnchorUnderMouse) would not work
properly if viewport margins were set. When centering the view in
QGraphicsViewPrivate::centerView, the viewport margins were not being taken
into account.
Mapping from global cursor coordinates in the viewport instead of the view
fixes the issue.
Task-number: 255529
Reviewed-by: Olivier
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On the Mac, it means "-framework ApplicationServices -framework Carbon
-framework AppKit" are no longer part of the default LIBS in Qt
applications. This required a lot of fixes where we used Mac-specific
code in Qt.
On X11, it was very straightforward, because we apparently use very
little of X11 outside QtGui.
I haven't changed the Windows-specific LIBS paths, because I don't
know how Windows behaves. Windows has DLLs, but it links to static
"import" libraries. So is it static linking or dynamic linking?
Reviewed-By: Marius Storm-Olsen
|
| |
|
|
|
|
|
|
|
|
|
| |
inneficient
As the virtual functions (such as the boundingRect()) are not available
anymore in the destructor, the view has to refresh everything.
Reviewed-by: Alexis
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If you have a floating object which affects the width available to the
text, we need to recalculate the width of the text line. In the code,
the setLineWidth() call to do this would by default have WrapAnywhere
as its wrap mode, even when this was not necessary. The code has now
been moved so that WrapAnywhere is only used if we try to set the
line width to match the available width and detect that the text is
too wide (the natural text width exceeds the available space.)
Task-number: 240325
Reviewed-by: Simon Hausmann
|
|
|
|
| |
we call QVariant::toReal instead of toDouble when needed
|
|
|
|
| |
Reviewed-by: Olivier
|
| |
|
| |
|
|
|
|
|
| |
Some dead code removed
Some member not initialized missing
|
|
|
|
|
|
|
|
| |
Made better use of qreal all over the place. We were previously
using QVariant::toDouble a lot. That is triggering unnecessary
conversions between float and double on embedded.
Reviewed-by: ogoffart
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Autotest: qcoreapplication, qapplication, qtimer
qwidget_window, qwidget
Issue 1: stacking order of modal windows was not
working correctly. With this patch, we remove the
need for rebuilding modal sessions all the time,
and when we do, we rebuild them all in the correct
order.
Issue 2: When running the event processor manually
(that is, just calling processEvents in a loop), we
sometimes spendt 100% cpu if a window was pending to
become modal. The reason was that we need to keep
reposting the QCocoaRequestModal event until we
could block the calling thread (that is, one of
the exec flags was given to processEvents). With this
patch, the need for posting QCocoaRequestModal is
completly removed in favor of an 'interrupt'
approach instead.
Issue 3: If using Qt as a plugin, or just add widget
to a native cocoa application, it would often lead
to closing down the application. The reason is that
the event dispatcher needs to restart [NSApp run]
now and then. But this approach fails if Qt was not
the code that started [NSApp run] in the first place.
This patch removes the need to restart NSApp in this
situation, at the cost of modal windows not beeing
modal if Qt is not spinning the event dispatcher.
Normal QDialog::exec etc will always work.
|
|
|
|
|
| |
it was even a bug when checking against QMetaType::Float because Float
is a user type so it could never be true
|
| |
|
| |
|
|
|
|
|
| |
for unreachable code, break missing and uninitialized members
in constructors
|
|
|
|
|
|
|
| |
We don't get a FontChange event initially, so the control ended up
with the default font instead.
Reviewed-by: Andreas
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change would have been much simpler if either QGraphicsItem
inherited QObject, or if we had some similar QPointer-like class that
supported QGraphicsItem. The issue is this: Each item can delegate
another item to be its focus proxy. That item can be a parent or child,
or something completely unrelated. Either of the two items can be
deleted independently.
The former solution was to store backpointers in a map in the scene.
Problem is, the items may not be in a scene when this happens, they
may be removed from the scene, and the items may be moved between
two scenes.
The bad part about this fix is that it adds another pointer to
QGraphicsItemPrivate.
Reviewed-by: Shane Kearns <shane.kearns@sosco.com>
|
|
|
|
|
| |
Task-number: 258378
Reviewed-by: Maurice
|
|
|
|
|
|
|
| |
Also, try not to regress in functionality (which the last compile fix
did just a little)
Reviewed-by: Thomas Hartmann
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When specifying round corners with QStyleSheetStyle and no border-width
specified, the round corners were not rendered with antialiasing. Furthermore,
if border-width was set to 0, part of the border was rendered in discordance
with CSS3.
The background in now rendered directly instead of drawing a clipped
rectangle. The actual border width is checked before rendering.
A test has been added at tests/auto/uiloader/baseline/css_borderradius_allwidgets.ui
Task-number: 230362
Reviewed-by: olivier
|
|
|
|
| |
Reviewed-by: Joerg
|
|
|
|
|
|
|
|
| |
You can now also set the axis following hte Qt::Axis enum
Note: I'm not 100% sure about the maths in QGraphicsRotation::applyTo
Feel free to fix it.
Reviewed-by: ogoffart
|
|
|
|
|
|
|
|
| |
On windows we were not checking whether the widgets receiving the
mouse events are blocked by another modal widget or not.
Task-number: 255912
Reviewed-by: Thierry Bastian
|
|
|
|
|
|
|
|
| |
Widgets with the WA_DontShowOnScreen attribute must not have a window
decoration.
Autotest: tst_QWidget::initialPosForDontShowOnScreenWidgets
Reviewed-by: thartman
|
|
|
|
|
|
| |
There's no QT_OS_WINCE define.
Reviewed-by: mauricek
|
|
|
|
|
|
| |
Q_WS_SINCE_WM ifdefs were broken
Reviewed-by: Joerg
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
|
|
|
|
|
|
|
| |
Added WA_StyledBackground to QMenuBar when using style sheets. This also implies
that CE_PanelMenuBar (drawing only the menubar border) no longer needs to
be drawn.
Tested in uiloader/baselne/css_borderimage_allwidgets.ui.
Task-number: 230363
Reviewed-by: olivier
|
|
|
|
| |
Reviewed-by: Oswald Buddenhagen
|
|
|
|
|
|
|
| |
string is a multi-length one
Task-number: QT-10
Reviewed-by: Oswald Buddenhagen
|
|
|
|
|
| |
Reviewed-by: Oswald Buddenhagen
Task-number: QT-10
|