| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
| |
If we ever decide we need them later, we can readd. For the moment,
let's keep the QString versions only.
|
| |
|
|
|
| |
It would have been of great help to find this out when I was debugging
the previous commit.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
First of all, make it a lot easier to access individual variables by
having them in an associative container (a QHash). This fixes task
232427, albeit one release later than I had originally planned.
On Windows, the variable names in the environment are
case-insensitive, so a direct QHash isn't a good solution. Implement
code that does the uppercasing on Windows and leaves untransformed on
other platforms.
Since we're doing this anyways, use QByteArray on Unix systems, since,
in theory, the environment could contain any random binary data, which
is not representable in QString.
Task-number: 232427
|
| |
|
|
|
|
|
| |
Changed the constructor of the QGesture to separate the gesture target (the
object/widget that the gesture filters events for), and the parent object.
Reviewed-by: Richard Moe Gustavsen
|
| |
|
|
|
|
|
| |
When Pan is implemented with touch events, make sure it pans only when two
fingers are used.
Reviewed-by: trustme
|
| |
|
|
|
|
| |
Use the proper winid to set gesture configuration to.
Reviewed-by: trustme
|
| |
|
|
|
|
|
|
|
|
|
|
| |
The code used to create native window handle even before checking if the widget
subscribes to gestures, basically breaking alien widgets. Check the position of
the gesture instead when the WM_GESTURE message is received in the GID_BEGIN
state. This will work properly in most cases - but not always as the position
that we get with the WM_GESTURE message is the center point of the multitouch
gesture, which might be outside of the widget depending on positions of
fingers.
Reviewed-by: trustme
|
| |
|
|
|
|
|
|
| |
model indexes
Reviewed-By: Olivier Goffart
Task: 231608
|
| | |
|
| |
|
|
|
|
|
|
|
| |
When the variant is invalid the shared is not destroyed.
We even can avoid the creation of the PrivateShared if we know the
variant is invalid
Reviewed-by: Thierry
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Constructing a variant of a pointer type was previously quite slow:
QVariant had no information it was a pointer. It had to create a
QVariant::PrivateShared, ask the QMetaType for a constructor (slow as it
involve locking mutextes) and allocate a pointer.
By detecting a pointer in qVariantFromValue, we can store the pointer
value dirrectly in the Variant union. We then avoid 2 allocations,
and the expensive locking in QMetaType::construct
Reviewed-by: Thierry
|
| |
|
|
| |
This is meant to speed up loading the landing page. But does it work?
|
| | |
|
| | |
|
| |
|
|
|
|
| |
This is private API
Reviewed-By: Trustme
|
| |
|
|
|
|
| |
This is private API
Reviewed-By: Trustme
|
| |
|
|
|
|
| |
It seems the winId can change during a WindowStateChange event too.
Reviewed-By: Trustme
|
| |
|
|
|
|
|
|
|
| |
The PBO code path causes a crash on certain AMD graphics cards, plus
seems to perform worse than the straight forward glTexImage2D code path
on several configurations.
Task-number: 257353
Reviewed-by: Gunnar Sletta
|
| |
|
|
|
|
| |
eventObject was a horrible name. The documentation already used the
term "event source", so let's call it that.
Agreed with Eskil.
|
| |
|
|
|
|
|
| |
Most Linux distributions already move qdoc3 to that place for exactly
this reason anyway.
Reviewed-by: Marius Storm-Olsen <marius@trolltech.com>
|
| |\ |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
On ELF systems, the static linker isn't allowed to find a linked
library's dependencies using the -L flag. That means if you're linking
lib or app X against a library A, and library A links against library
B, then -L$QTDIR/lib won't apply to B.
Before LIBS_PRIVATE, we had -lB, so it did apply. Now we need to find
another solution.
The solution is one of:
a) install the libraries before linking anything against them
b) set LD_LIBRARY_PATH
c) set -rpath or -rpath-link
Reviewed-by: TrustMe
|
| | | |
|
| | | |
|
| | |
| |
| |
| | |
Reviewed-by: Kent Hansen
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
JSC::Debugger::exceptionThrow event was moved _before_ stacks
unwinding so there is possibility to check stack state before it's
deletion.
Missing functionExit event was added in Interpreter::unwind()
Reviewed-by: Kent Hansen
|
| | |
| |
| |
| | |
Reviewed-by:TrustMe
|
| | |
| |
| |
| | |
Reviewed-by:TrustMe
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
QAbstractScrollArea
We send the Touch* events filtered from the viewport to the scroll
area's default handler so that we can take advantage of the default
handling in QWidget. But we also need to ignore Touch* events in
QAbstractScrollArea::event() so that we don't get the events twice
(once from the filter, and again if propagated).
Reviewed-by: Richard Moe Gustavsen
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The default cursor on QWS is loaded as an 8bit QImage.
Blitting that one is really slow so we better convert
the image into a fast format before using it.
Based on a patch by Lars.
Reviewed-by: Tom
|
| | |
| |
| |
| |
| |
| | |
Previously, length of linuxfb device's path must be 8 and start with "/dev/fb".
Reviewed-by: Paul
|
| | |
| |
| |
| |
| |
| |
| |
| | |
Request focus before requesting a geometry. This makes it much more
likely that the initial paint will be done after the window is active.
Task-number: 255828
Reviewed-by: Tom
|
| | |
| |
| |
| |
| |
| |
| | |
AVR32 doesn't support non-aligned access. (Contributed patch)
Task-number: 255495
Reviewed-by: Paul
|
| | | |
|
| | |\
| | |
| | |
| | |
| | |
| | |
| | | |
git@gitorious.org:+openbossa-developers/qt/openbossa-clone
Conflicts:
src/gui/graphicsview/graphicsview.pri
|
| | | |
| | |
| | |
| | |
| | | |
We decided to remove it because it does not serve its purpose as an
example.(Too much code, does not show how to use the API in a nice way)
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Those are:
* setAnchorSpacing()
* anchorSpacing()
* unsetAnchorSpacing()
Autotests for the two last ones are missing though..
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The changes are:
* Move enums in QGraphicsAnchorLayout::Edge to Qt::AnchorPoint.
Prefix them with Anchor since they are not edges in general.
* Rename anchor() to addAnchor()
* Rename anchorCorner() -> addCornerAnchors()
* Rename anchorWidth() -> addLeftAndRightAnchors()
* Rename anchorHeight() -> addTopAndBottomAnchors()
* Rename anchorGeometry() -> addAllAnchors()
* remove the overloads that take a spacing argument, and add
setAnchorSpacing() to accommodate for that.
* Added anchorSpacing() (implementation missing)
* Added unsetAnchorSpacing(). (implementation missing)
* made sizeHint() protected.
Updated all examples and autotest to reflect this API change.
|
| | | |
| | |
| | |
| | | |
This failed on Alexis' Mac Mini (Leopard)
|
| | | |
| | |
| | |
| | |
| | | |
Also, only include the "name" member to AnchorData if we are compiling
in debug mode.
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | | |
Functionality-wise it should be the same, its just a bit less code.
|
| | | |
| | |
| | |
| | | |
Don't assume that m_item is always a QGraphicsWidget
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This commit implements what's described in the previous commit
QGraphicsAnchorLayout: fix expected values for parallel test
When filling the sizeAt* values (the three points used for interpolation
when setting geometry), now sequential anchors distribute the sizes
to the children in a fair way, i.e. proportionally in relation to the
existing min/pref/max hints.
Each value is defined in relation of the pref (either as a shrinkage in
pref or a grow in pref). In both cases the shrinking/growing factor is
the same for all children.
When we implement support for QSizePolicies, this distribution might be
changed by setting size policies for a certain item or anchor.
This makes two tests work -- so no more expected fail. Also fixed a typo
in one test's expected results.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When creating an anchor we might have to create the edges for an item
if the item is new. In this case both dimensions are affected. The
call is clearly missing since createItemEdges() declared this
dependency via Q_ASSERT.
Patch from Jan-Arve.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org>
|
| | | |
| | |
| | |
| | |
| | |
| | | |
For debugging purposes, if QT_ANCHORLAYOUT_NO_SIMPLIFICATION
environment variable is set, the layout won't use
simplification. Useful for some debuggings.
|
| | | | |
|
