| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a mouse move event is received, a lot of time is spent looking
for items with cursors, recursively.
With this patch, it will only recur into item hierarchies that
contain cursors.
Not having cursors is much more common than having them.
Benchmark included:
Before: 15 msecs per iteration (total: 62, iterations: 4)
After: 0.000064 msecs per iteration (total: 68, iterations: 1048576)
Task-number: QTBUG-27054
Change-Id: I3a5441652ca1c0b8d2cbc5683013562174f5af4a
Reviewed-by: Alan Alpert <416365416c@gmail.com>
|
|
|
|
|
|
|
| |
This function has been moved into QtV8 module.
Change-Id: Ibc5b0a52bd158e8880a10300eb6b89b7f43db0f6
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
| |
The receiver count needs to be correct in connectNotify() to be
compatible with ordinary connections.
Fix this and add test.
Change-Id: Icff0486dac49876b8c5d8836a85cff9284368a52
Reviewed-by: Alan Alpert <416365416c@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Unless result is an insert iterator, qCopy()
assumes that all iterators in the range
[result + (finish - start)] are dereferenceable.
This is not the case for an empty list.
Using foreach instead.
Change-Id: I2212a88e9b462ae2220ba5aeb43c662592cec57f
Reviewed-by: Marco Bubke <marco.bubke@digia.com>
|
|
|
|
|
|
|
|
| |
QmlJS::Visitor was missing the visit for UiParameterList.
Change-Id: I3c2cb76b934059ebcef53c1a0faad3f239b33359
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Christian Kamm <kamm@incasoftware.de>
|
|
|
|
|
|
|
| |
-'qmlmodules' group contains the QML APIs in Qt
Change-Id: I797e6ee0c9e643a92b9e16d476c93b34e79d67d8
Reviewed-by: Geir Vattekar <geir.vattekar@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Same for originXChanged.
Conflicts:
src/quick/items/qquickflickable.cpp
src/quick/items/qquickflickable_p_p.h
Change-Id: I1a316f326b924da4722b692cf27db48312fc9236
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The propertyTypeToken was not initilized in this case.
Not having the source location of this token leads to
bugs in the tooling. e. g.: QTCREATORBUG-7931
I also added the initializer for program keeping the parser
and the grammar in sync.
Change-Id: Id7bfb4b6dd1b9689e701a858b6c769d3addab5dc
Reviewed-by: Christian Kamm <kamm@incasoftware.de>
|
|
|
|
|
|
|
|
|
|
|
| |
To make QQuickWindowPrivate::setRenderWithoutShowing more useful
for autotesting we can allow grabbing such windows. This is safe
if the platform window have been created and the window has a valid
size.
Change-Id: I8b9a2aaeb93935f662b75ef29651730b890441d5
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
We need to use a resource guard for the FBO in case there is no
current context when the glyph cache is deleted.
This reverts commit b3264e2cb6a8fe87754aa1335ab9f8d5e3910c14 which
was implemented as a band-aid for this crash.
Change-Id: I5b3a09a3998da38836ea851cd0978d3ddadcd2cc
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the top type of a QML component is a C++ type registered with
qmlRegisterExtendedType, its QObjectPrivate::metaObject is already a
QQmlProxyMetaObject that gets chained as the parent of the
QQmlVMEMetaObject of the component.
When QQmlPropertyCache::findProperty iterates over the parents chain
of a child item, our QQmlProxyMetaObject eventually gets static_casted
to QQmlVMEMetaObject and causes a crash.
This patch implements a poor man's dynamic_cast in
QQmlVMEMetaObject::parentVMEMetaObject to fix the crash. Other casts
of parent.asT1() are changed to use parentVMEMetaObject as well even
though in those cases the static_cast is guaranteed by the context.
Task-number: QTBUG-27334
Change-Id: I5982fc273ccf466960ce54974cff5662e6ab605a
Reviewed-by: Matthew Vogt <mattvogt2@gmail.com>
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
|
|
|
| |
QtDeclarative is using private meta type api to register custom types,
but the helper functions were updated to a new signature.
Change-Id: I467b283b5b1ccdad8ebd1a2bb651e8ad59b9c55d
Reviewed-by: J-P Nurmi <jpnurmi@digia.com>
|
|
|
|
|
|
| |
Change-Id: I9f8b98e40340a5f871f7a66a3d7ea63675e21d34
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The new AA algorithm which is implemented in the vertex shader of
SmoothColorMaterialShader and SmoothTextureMaterialShader exposes
a driver bug in the glsl compiler/optimiser for the GPU used in
the BlackBerry PlayBook.
The bug results in the if (scale < 0.0) scale = 1.0 code always
being executed even when the condition is false. This leads to
massive corrupion of Image, Rectangle, and BorderImage elements
when the antialiasing property is enabled.
This commit works around the compiler bug by refactoring the
vertex shader to explicitly include an else clause. The check for
negative values is now performed on the numerator only as the
denominator is dot(dir, dir) which is always positive (square of
magnitude of dir vector).
This gives the correct behaviour on all platforms (that I have
access to).
Change-Id: I236542c9b59ff2915e95cbd9300b442be316bba9
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
|
|
|
|
|
| |
Change-Id: I883109cf2528a5e10487b38f453c2eed96cc3204
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
|
|
|
|
|
| |
Change-Id: Ic1299cd8949c37c6909b8d2ab36e930097335a4d
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
| |
Really it should just be "to".
Change-Id: I3263c8d6ce2f0ad33ab12d7ddb9202beac578f91
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
|
|
|
|
|
|
| |
Change-Id: I0b0e63b572617a56874e27361479de2690644aa3
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
| |
Windows CE has a define for CONST,
so undef it.
Change-Id: I13d363657e91956c611dee4cf90cd17f6a0fa234
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
| |
Change-Id: I5f61a443b49d40c66669fa0c679df7eaab761e73
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
| |
Change copyrights and license headers from Nokia to Digia
Change-Id: Ie7f5d49ed8235d7a7845ab68f99ad1c220e64d5c
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
|
| |
Unit test is testing for it
Change-Id: Ie37a0f7103dbe8c19a4cfbca2d4fe026c805ba0b
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>
Reviewed-by: hjk <qthjk@ovi.com>
|
|
|
|
|
|
|
| |
modified path to doc/global
Change-Id: I646b69c38381556f786e0b4346118d035cbd3648
Reviewed-by: Martin Smith <martin.smith@digia.com>
|
|
|
|
|
|
|
|
|
| |
If 'count' is not initialized, it could be anything and if the
call to glGetIntegerv() fails and leaves this value unchanged
then the loop could go through any number of iterations.
Change-Id: Ibdcfd018b70e265ef6aeab87a5df8c0530c653a7
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
|
|
|
|
|
|
|
| |
-added plugins and removed examples directory
Change-Id: I66a5037334ec7249aa44e8bfa65fe5905b296f9d
Reviewed-by: Martin Smith <martin.smith@digia.com>
|
|
|
|
|
|
|
|
| |
Fixed paths to sources
Added doc/global
Change-Id: I377b31eba3b790ff3e6a69b2d1fbffce0229aac6
Reviewed-by: Martin Smith <martin.smith@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In case the content was flicked horizontally so that the vertical
position did not change handleMouseReleaseEvent performed a fixup
for the vertical position to be an integer _before_ starting the
horizontal flick. For some large content the rounding fixup had
finished before the flicking animation started which resulted
in a premature emission of the movementEnded() signal.
This race condition did not affect vertical flicking because the
vertical animation had always been started before the horizontal
fixup was scheduled.
The issue was revealed by QQuickWebView which above a certain
velocity threshold locks the movement in one direction, thus
produces the exact same vertical position throughout a horizontal
pan gesture.
This patch removes the race condition by starting the flick
animation in either direction first then performing the
position fixup if needed.
There does not seem to be a reliable way to auto-test this.
Change-Id: Idc8fb59b0a6ead71c6cfc7083a386d51048ec50c
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
|
|
|
|
|
|
|
|
| |
The code can't reach there anyway, so let's help the compiler out a
little.
Change-Id: Icee6ebe426c1116a7baaf2c5766c6d8f8276f3ac
Reviewed-by: Roberto Raggi <roberto.raggi@nokia.com>
|
|
|
|
|
|
|
| |
It's never going to be modified, so it should be const.
Change-Id: Ibc3bac0583f72ccbf7caa69aeab029bae03d3cdd
Reviewed-by: Roberto Raggi <roberto.raggi@nokia.com>
|
|
|
|
|
| |
Change-Id: Ia5d8967e241e0697f2903f26899d878210dd5570
Reviewed-by: Geir Vattekar <geir.vattekar@nokia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
.qmltypes files are used by e.g. Qt Creator to provide code completion etc
for QML types defined in plugins. The added files where generated with
qmlplugindump, e.g.
qmlplugindump QtQuick.Particles 2.0 > plugins.qmltypes
(ideally make qmltypes would do the same, but it's not working as expected
right now).
The exception is the QtQuick plugin: Here the file is generated with
qmlplugindump -builtins > plugins.qmltypes
and contains also the types registered in the qml library itself.
Change-Id: I1a6b6641cb5ec3ecffe08e8926d8c1bc082ae6de
Reviewed-by: Christian Kamm <christian.d.kamm@nokia.com>
|
|
|
|
|
| |
Change-Id: Ide2e054e9b5b58976410d16ca4f76fd55e61676d
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
|
|
|
|
|
|
|
|
| |
The sync time was not taken into account
Change-Id: I3d8adb637572c72438e20729a80513850a80d17e
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Alan Alpert <416365416c@gmail.com>
|
|
|
|
|
|
|
|
|
| |
This is quite useful to see how much time the rest of the app takes.
The same information has been available in QtQuick1.
Change-Id: I31ae91bfa764a4d8013af80de9459514ed72314f
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
|
|
|
|
|
|
| |
Change-Id: Ia3958c7c26bf2fd9dd72a07fc6f5ff3a28b0a349
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
|
|
|
|
|
|
|
|
| |
This makes it easier to grep for.
Change-Id: I7630e05c4afddb4ab3ce1a7691466d7307325988
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
|
|
|
|
|
| |
Change-Id: I7fa055049b9e5900d597754c6004febb153de12b
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This is needed following change
https://codereview.qt-project.org/#change,33148
Change-Id: I836d4037306d1efbff48668ff8dd2b7b5ce97516
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@nokia.com>
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
|
|
|
|
|
| |
Change-Id: Ica7685aefde84ec80d8af7a67541af454de4adce
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>
|
|
|
|
|
| |
Change-Id: I03c4faa1aee60cd1ad706cb6c1b983f03311d251
Reviewed-by: aavit <qt_aavit@ovi.com>
|
|
|
|
|
| |
Change-Id: Ice5d60b06ec9ab81fbd98fd1679c8834f3018938
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
|
|
|
|
|
| |
Change-Id: I4d8d12cfce58e49789146639cc6734b6cf1643fa
Reviewed-by: David Faure <faure@kde.org>
|
|
|
|
|
|
|
| |
Depends on the integration of #34077
Change-Id: I188d7617c4805d00226d0fb8d5768a02d24c6933
Reviewed-by: Geir Vattekar <geir.vattekar@nokia.com>
|
|
|
|
|
| |
Change-Id: I0f7e60fd6060381bd8790dfaacc3a39a890a0fe7
Reviewed-by: aavit <qt_aavit@ovi.com>
|
|
|
|
|
|
|
|
|
|
| |
Tie incubation directly to wether animations are running
or not. When animations are running, we incubate for
a third of a frame only. When animations are not running
we incubate whatever is left as the scene is anyway idle.
Change-Id: I88c9248e9d7b3b35c99fca52aeb9c681d1b4a306
Reviewed-by: Martin Jones <martin.jones@nokia.com>
|
|
|
|
|
|
|
|
|
| |
This reduces memory leaks on exit when the engine is destroyed
after exec() has finished.
Change-Id: I917d103966d55b4dd3ba4e986ff902e29d8fb0ac
Reviewed-by: Chris Adams <christopher.adams@nokia.com>
Reviewed-by: Kent Hansen <kent.hansen@nokia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of just evaluating bindings in a fixed order, and possibly
having to evaluate a single binding multiple times, prior to reading
a property, we check if there are any bindings "pending" on it and
evaluate them then. A pending binding is one that has been assigned
to the property, but not yet evaluated.
To minimize side effects we only do this for "safe" bindings. A
safe binding is one that has no side effects, which we currently
define as not calling functions or otherwise assigning values during
its evaluation. This isn't an entirely foolproof way to ensure that
the evaluation has no side effects, but it should be good enough.
Change-Id: I98aa76a95719e5d182e8941738d64f8d409f404a
Reviewed-by: Michael Brasser <michael.brasser@nokia.com>
|
|
|
|
|
| |
Change-Id: I44eb0113374c221a166609c4fd29b5dad05ecc94
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
|
|
|
|
|
|
|
|
|
| |
Calling console.log(), console.debug(), print() etc would only
print out "Object" if a JS Object was passed as an argument.
This patch calls the toString() on the object.
Change-Id: Iadf8b4d1fe81c3e2c7bd65e3c153a930fd994bef
Reviewed-by: Kai Koehne <kai.koehne@nokia.com>
|
|
|
|
|
|
|
|
| |
Function overriding is required by QmlTest, and is likely to be used by
other existing QML users.
Change-Id: I04086a933456145bda5cede74aba753799feb555
Reviewed-by: Michael Brasser <michael.brasser@nokia.com>
|