Try to free our large items for every 8 megabytes of additional large item
memory we allocate. This helps in particular on 32-bit builds and was noticable
in tst_qqmlecmascript::push_and_shift().

Change-Id: I4cc0b188e58ccaf32026e38c7aaf1cfadc83148b
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

This actually violates the C++ standard that defines that
you aren't allowed to call member functions on an invalid
object.

Instead insert the 0 pointer checks on the caller side where
required.

Change-Id: I8be3c3831594bb6482e9ef6de6e590ec437ac0f8
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

The pointer to the real data is not required anymore, as
it always follows the class itself. Like this we save one
pointer of overhead, and one indirection when doing reads
and writes of array data.

Change-Id: If6afdac8e97b57420b50e7b7eb8979f77e8dbbcf
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Implement the simple array as a circular buffer instead
of an array with head room. This fixes a couple of severe
issues with performance and memory management if the array
is being used as a queue.

Task-number: QTBUG-41421
Change-Id: I146ad8a874407c108aa8fe1eae68e9957e154847
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Drop the refcount added in QQmlTypeLoader::getType() to prevent object
leakage in resolveQmlType.

Change-Id: I8bd9c486294912cc00ce5feb350c3ff79c6aac09
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

With this change, instead of writing:

    function accessiblePressAction() { submit() }

You should write:

    Accessible.onPressAction: { submit() }

For the moment, only 4 actions are added:
press, toggle, increase and decrease.

The old style action handlers are deprecated, and removed from the
documentation. New style action handlers will be preferred in case an
item declares both styles.

Change-Id: I11919e631d8476d55540f94252757b911c44ade4
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@theqtcompany.com>

Make sure that ImageData instances, m_shadowData as well as the
QSGGeometry of particle nodes are destroyed together with their
QQuickImageParticle.

Also implement the assignment operator for QQuickParticleData to
avoid its v8Datum pointer to be copied over to the shadow datum
in getShadowDatum. This would cause a double delete of the
QQuickV4ParticleData when trying to call clearShadows() in the
destructor.

Task-number: QTBUG-36782
Change-Id: Ie03f2be0415daeb7f4f6e5f92295a3ab26a62155
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>

org.qt-project.org is redundant

Change-Id: Id6a6794e1fc1e4403b3bfcdebbe06a487f904041
Reviewed-by: Sebastian Sauer <sebastian.sauer@kdab.com>
Reviewed-by: Brett Stottlemyer <bstottle@ford.com>
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>

There are many places where this doesn't make sense (a button in a
search field, a line in a list represented as button with a child button
to remove the list item...).

Change-Id: Iab1d411adda696a01b1dcfdebb0a250066d6a7c2
Reviewed-by: Jan Arve Sæther <jan-arve.saether@theqtcompany.com>

Change-Id: I9230deaa75d1e9a9614f364d4adf92349c8a1901
Reviewed-by: Jan Arve Sæther <jan-arve.saether@theqtcompany.com>

JavaScript knows a year 0. That is correctly translated into QDateTime
terms when creating a Date object, but it's not correctly translated
back when converting the JavaScript date to a string.

Task-number: QTBUG-29491
Change-Id: I46b200a144434187656d08e87f422f97523acd0e
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

The grammar did not allow for the declaration of

    readonly property QtObject foo: QtObject { ... }

and it required a workaround through an alias:

    readonly property alias foo: _foo
    property QtObject _foo: QtObject { ... }

This was merely a glitch in the grammar, I see no reason not to support this.
The semantics are like a const pointer in C++, the property itself is read-only
but the object pointed to has per-property defined read/write semantics.

Task-number: QTBUG-41971
Change-Id: I99e2e7ed58731e387a38e46ec39922d280a21ceb
Reviewed-by: Michael Brasser <michael.brasser@live.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

The grammar file itself wasn't updated accordignly and the license headers that
are embedded in the grammar that will be copied into the generated files were
also oudated. Re-generating the parser would produce files with the wrong
license headers.

Change-Id: I1db83df8a9c4bddfb58901f41a4d40f6b1396507
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

When QML declares sprites and goalSprite in wrong order the goalsprite
attempts to set it on null spriteEngine. This patch ensures that the
order doesn't matter anymore.

Task-number: QTBUG-40595
Change-Id: I57f1c8754b2e2af91e0c7f72d1d67fec3ad4ede5
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Using Date() as a regular function without the new operator
returns a string and not a date object.

Change-Id: I083bb62c0cb3041a053d43e3085c3d0d10423db6
Task-number: QTBUG-41712
Reviewed-by: Mitch Curtis <mitch.curtis@digia.com>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@digia.com>

QLineEdit closes input method on enter key press, TextInput
should also do the same. Commit and hide input method on enter
key. When Qt::ImhMultiLine input method hint set only commit.

Task-number: QTBUG-37850
Change-Id: I5a06e3eb777d7f794dd1493b307f0b05a2caf281
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>

There is no point in waiting any further for the remaining bytes to be
written as it will never succeed. We might get luckier by creating a
local event loop and repeatedly calling processEvents(), but as that is
considerably worse style and because you shouldn't rely on the
connection to send on exit anyway we don't do that.

Task-number: QTBUG-42158, see also QTBUG-24451
Change-Id: I79ffd5f5a4a7c41ff8dc0c4f6f1ca7e091844c9d
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

The use of the qDebug() macro to construct a QDebug object fails when
QT_NO_DEBUG_OUTPUT is defined when building Qt.
This patch aims to fix this.

[ChangeLog][General][Build] Can now build with QT_NO_DEBUG_OUTPUT
defined

Change-Id: If807ee3439db2a98b4d146f75860a98f40c247ec
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

Creating the network access manager can involve blocking calls to DBus,
which apparently can take quite long in pathological cases. Using the
same mutex in this case like for instance registering objects can
therefore result in a freezing UI.

Mitigate this by introducing a separate mutex.

Task-number: QTBUG-41183
Change-Id: I55bd8453d8e4bbc0bda1591eb3fd48b50ca921c1
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Change 63e6c9ad

 introduced snapping to
the pixel grid in the vertex shader for native text, but this code was
broken on retina displays because it assumed integer only positions.
Fix it by including the retina scale factor into the rounding.

Task-number: QTBUG-38702
Change-Id: I84492b02d64f263c9fe030790e04cf79b0dc4e2f
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>

In updateSize() we were trying to force relayouts for RichText
when the width changes by always setting widthExceeded to true.
But further down in the same function, we overwrote this
with textWidth() < idealWidth(), which doesn't work, because
both properties are the wrapped width of the document and
should only differ if the text cannot be wrapped properly.

The result was that when increasing the width of a Text element,
we would hit the optimization and skip the relayout.

[ChangeLog][Text] Fixed Text with the RichText format to
correctly update wrapping when the width of the element grows.

Change-Id: I5fd87052a5ba7e8ee2549be0cfac4adc8ddf8290
Task-number: QTBUG-33020
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Excessive makeCurrent - doneCurrent pairs should be avoided. We already
do this in QOpenGLWidget and the QPlatformBackingStore composition code.
Remove the doneCurrent from QQuickWidget too.

Change-Id: I6f998d381c33880c470f34d7c8462b6ed8cd8ae9
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

Also updated the C++ snippet to look like a
gui app that shows the QML example, instead of
a console app that just provides the context data
to the QML example.

Change-Id: I17aca9f03521cfcadba1a965a4924e87cbf14c62
Task-number: QTBUG-41699
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>

This commit ammends 59ed8c35

 to always query
the Qt meta-type registry to retrieve the QMetaObject for a QObject pointer
type.

Change-Id: I70d876a5acfa23967fd1a57c96fcd5ac853eaf49
Task-number: QTBUG-39614
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Ulf Hermann <ulf.hermann@digia.com>

Change-Id: I08b22766b3e389b7d27ca4c56729f550b0647a08
Reviewed-by: Jens Bache-Wiig <jensbw@gmail.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>

Task-number: QTBUG-42006
Change-Id: I3c59b0ba27518750cba667f414f58edaa136d250
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>

Change-Id: I58aa163bd17fae7190161641d94f51887d8f88a6
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

As the example in QTBUG-42051 demonstrates, QML may produce a lot of binding
objects and each of them produce a QV4::Persistent. During the mark phase we
may run out of JS stack space. One fix (for the future) is to reduce the number
of QV4::Persistent objects, but in the meantime we can also reduce the pressure
on the stack by draining it earlier.

Task-number: QTBUG-42051
Change-Id: Iea73f8a869048ea0bf3f4a64dbd24b6fb8c68f6b
Reviewed-by: Ulf Hermann <ulf.hermann@digia.com>

Task-number: QTBUG-40574
Change-Id: If1fcc4cb00c8e477b3afff92eb4a3099424fcfa6
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

It's not obvious that calling setPosition() directly, for example,
won't trigger animations defined in behaviors that react to that
item's x and y properties changing, because a quick glance at the
code shows that they have almost identical code paths and both
emit geometryChanged().

Change-Id: Ic7f7019b1bbba24732569c0de777ee02d79c1d3b
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

The documentation of SHParseDisplayName doesn't mention that you need
to free the ppidl parameter, but the documentation of the ITEMIDLIST
does mention that you need to ILFree any ITEMIDLIST passed as
PIDLIST_ABSOLUTE.

Reproduced the leak by repeatedly showing and closing the window on
the task's example.

Task-number: QTBUG-41588
Change-Id: I91d08728fc907c59e56ae344a2d12f0865031120
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Having a retina and non-retina screen connected resulted in getting
no output from QQuickRenderControl and QQuickWidget on the
non-retina screen.

This is caused by the fact that Quick is blindly calling
QWindow::devicePixelRatio(). This approach is wrong when using
QQuickRenderControl since the QQuickWindow does not have an actual
native window and so devicePixelRatio() merely returns some default value
which will definitely be wrong for one of the screens.

The patch fixes the problem by introducing
QQuickWindow::effectiveDevicePixelRatio(), which, via
QQuickRenderControl::renderWindowFor, supports the redirected case too.

Task-number: QTBUG-42114
Change-Id: I057e01f0a00758dde438fc9b10af3a435b06bb0b
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>

Change-Id: I5fa93d559cdfde5e6b84ab0f3f35fd600e6bec4f
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>

Ignoring items with children will make the children appear as children of
the parent of the ignored item.
Since setAccessibleFlagAndListener now only sets the flag we also rename
the function to just setAccessible

Change-Id: I79fc311509a3e454b4698274c63ad0e879fb93e3
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@theqtcompany.com>

There are lists that won't accept a component as element. For example
QQuickItem's children will only accept QQuickItems.

Task-number: QTBUG-41848
Change-Id: I0fc7b0d1a4770d596caf681be92dff216f48d32b
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

m_ignored is not really used.

(cherry-picked from dev branch: 2c42d6af

)
Change-Id: Ic2926fbdf22a6a5bc82fe9f4475b7b8058eacae5
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@theqtcompany.com>

QQuickCustomParticle::buildCustomNodes() assumes there is a valid OpenGL
context, but when there is not it will simply crash.  Instead we check
for a valid current OpenGL context first, and return 0 if it is not
availalbe.  This needed for the Qt Quick 2d Renderer.

Change-Id: I6bfcfc8fa9581bfd27015f719fc527c36492eade
Reviewed-by: aavit <eirik.aavitsland@digia.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

In the case that QQuickSpriteEngine::assembledImage() is called when
there is no current OpenGL context, return a null QImage instead of
crashing.  This is needed for the Qt Quick 2d Renderer case.

Change-Id: I75b9b1f31f05cc0800293435d660e498fecc4d20
Reviewed-by: aavit <eirik.aavitsland@digia.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

With the upcoming introduction of unignoredChildren, there can
exist many top level items (if the root item is ignored).

Change-Id: If7aaea08fdd4d1f5a0a5109e1239c53e0af9b61e
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@theqtcompany.com>

The hit testing won't work very well with the upcoming patch that
changes which items that can be ignored.
(basically it doesn't consider the isAccessible flag,
so childAt_helper might return a node that was supposed to be ignored)

Earlier this was a sensible optimization in order to avoid too many heap
allocations and deallocations of interfaces, but these are cheap now, so
we can do it the do it the 'proper way' (i.e. before this patch we
didn't respect the a11y hierarchy as given by QAccessibleInterface
child() and parent(). This also uses the QAccessibleInterface::rect()
directly now instead of using the itemScreenRect() function, which
was shared between QAccessibleQuickWindow and QAccessibleQuickItem.

Since this changes the order of child interfaces to paint order (i.e.
second child interface is on top of first child interface), we need to
ensure that we hit test child interfaces in the correct order.
They should now always be processed with the last interface first, and
then return as soon as something is hit.

Change-Id: Ie951fc3b48b7affd9f7e98687a1cbbe008857d2a
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@theqtcompany.com>