| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Check for the QSGTextureProvider not being 0 before asking for its texture,
we are doing the same in the loop of QQuickShaderEffectMaterial::compare so
it was forgotten to do it here.
Valgrind trace of the crash i get without it
==26317== Invalid read of size 8
==26317== at 0x652B73A: hasAtlasTexture(QVector<QSGTextureProvider*> const&) (qquickshadereffectnode.cpp:49)
==26317== by 0x652BABC: QQuickShaderEffectMaterial::compare(QSGMaterial const*) const (qquickshadereffectnode.cpp:396)
==26317== by 0x63D1BF6: QSGBatchRenderer::Renderer::prepareOpaqueBatches() (qsgbatchrenderer.cpp:1525)
==26317== by 0x63DE7A7: QSGBatchRenderer::Renderer::render() (qsgbatchrenderer.cpp:2611)
==26317== by 0x63E9D3E: QSGRenderer::renderScene(QSGBindable const&) (qsgrenderer.cpp:208)
==26317== by 0x63EA58A: QSGRenderer::renderScene(unsigned int) (qsgrenderer.cpp:168)
==26317== by 0x63FAA7D: QSGRenderContext::renderNextFrame(QSGRenderer*, unsigned int) (qsgcontext.cpp:558)
==26317== by 0x644540A: QQuickWindowPrivate::renderSceneGraph(QSize const&) (qquickwindow.cpp:383)
==26317== by 0x641541A: QSGGuiThreadRenderLoop::renderWindow(QQuickWindow*) (qsgrenderloop.cpp:378)
==26317== by 0x6416520: QSGGuiThreadRenderLoop::event(QEvent*) (qsgrenderloop.cpp:474)
==26317== by 0x605F488: QCoreApplication::notify(QObject*, QEvent*) (in /usr/lib/x86_64-linux-gnu/libQt5Core.so.5.5.1)
==26317== by 0x605F5BA: QCoreApplication::notifyInternal(QObject*, QEvent*) (in /usr/lib/x86_64-linux-gnu/libQt5Core.so.5.5.1)
==26317== Address 0x0 is not stack'd, malloc'd or (recently) free'd
Change-Id: I3b07450438d98910fbbff9f8b7a3d9d851ed4e5d
Reviewed-by: Michael Brasser <michael.brasser@live.com>
|
|
|
|
|
|
|
|
|
|
| |
Take device pixel ratio into account (pixWidth and pixHeight are scaled)
while calculating source and target rects for QQuickImage::Pad.
Change-Id: I4a8b4fc305d0af93921bca900c683927d6bfd0e3
Task-number: QTBUG-52043
Reviewed-by: Mitch Curtis <mitch.curtis@theqtcompany.com>
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Subjects each *.png file that matched grep -law "sRGB" to:
pngcrush -ow -brute -rem allb -reduce -force
Various tools grumble about sRGB tables in PNG images; and our
handling of them doesn't pay attention to these, so purging them
makes the images smaller with no loss to the images.
Change-Id: I23d7a43ba6c6ce6cafa11c1950a118a73f4d07cd
Reviewed-by: Topi Reiniö <topi.reinio@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Subjects src/particles/particleresources/noise.png to:
pngcrush -ow -brute -rem allb -reduce
Various tools grumble about sRGB tables in PNG images; and our
handling of them doesn't pay attention to these, so purging them
makes the images smaller with no loss to the images.
Change-Id: I534f7c16830912fe0ee17159e50094e490b62b70
Reviewed-by: Topi Reiniö <topi.reinio@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I36d22343976a2164c396bb17c13526b63ab7579c
Reviewed-by: Martin Smith <martin.smith@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: I677ab9fce527b5220317d4d30f48d55b7f081e6a
Reviewed-by: Ulf Hermann <ulf.hermann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
For some image formats it is not possible to rely on scanLine() so it is
best to use pixel() for this check as it will take into account the format
of the image correctly.
Change-Id: Iee753b79291f210d243a5aed9d61359c67a78931
Reviewed-by: aavit <eirik.aavitsland@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
Handle a duration of zero in animators using the same approach as
animations.
Task-number: QTBUG-39766
Change-Id: I1628292ad259ec5ec5d3bb0770440efa5fa6dad0
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
|
|
|
|
| |
To do this easily while retaining the meaning of the -parent flag, we add a
private export to QQuickView so that we can set the root object.
Change-Id: Iabb2b998816a6fdfcc8417f679c96f04910b8202
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
|
|
|
|
|
|
|
|
|
|
| |
Worker thead may have started iterating through filesystem before
component was completely initialized, resulting in unnecessary and
unexpected results.
Task-number: QTBUG-49898
Change-Id: Ibaa2572139b0aed2fdc9bd3bbcdffa58791e38d7
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
|
|
| |
Found by ICC.
Change-Id: I0c94a5c2846b48c8aea7ffff14352fd8573cbf66
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
| |
Change-Id: Ib9bdd42ca47853e343c26bc9d1b9c563b75af4f7
Task-number: QTBUG-43810
Reviewed-by: Topi Reiniö <topi.reinio@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: Ic1bf883003f6e20efa3fdba9ad29a8c8276b043e
Task-number: QTBUG-51713
Reviewed-by: Topi Reiniö <topi.reinio@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
| |
Cutting the text in Password mode led to the text being deleted.
Cut and copy operations should not be allowed when echoMode is
set to a mode other than Normal.
Task-number: QTBUG-51231
Change-Id: If75cdaedac7478ecc1a5126ad4a5ecbf32acd1ab
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
|
|
|
|
|
|
| |
QQuickSprite is used in QtQuickParticles. Export unconditionally.
QQuickScaleGrid is used in auto-tests. AUTOTEST_EXPORT it.
Unbreaks UBSan build, which requires access to the classes'
type_info.
Change-Id: I65b538fe0a4255d1b5c4e2e1212001d356382cb4
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Says GCC 6:
qtdeclarative/src/qml/types/qqmllistmodelworkeragent_p.h: In member function ‘QQmlListModelWorkerAgent::VariantRef& QQmlListModelWorkerAgent::VariantRef::operator=(const QQmlListModelWorkerAgent::VariantRef&)’:
qtdeclarative/src/qml/types/qqmllistmodelworkeragent_p.h:94:34: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
if (a) a->release(); a = o.a;
^
qtdeclarative/src/qml/types/qqmllistmodelworkeragent_p.h:94:13: note: ...this ‘if’ clause, but it is not
if (a) a->release(); a = o.a;
^~
This one is correct.
Change-Id: I6315e4afa6a0cf8bb4dd3a599bd7ffe7b6a5c1e4
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
This in theory happen for attached property types.
Fixes Coverity CID 154272.
Change-Id: I113797dea8949877cbeac82bae57655170878d4c
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
Added explicit where it was missing. This is not a source-
incompatible change, because code that breaks by this is
a bug. Let's not have this sitting around in an LTS.
Change-Id: Ic198750717799126b4e37817845a8ee4684d631f
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It’s a bad idea to store a scene graph paint node as a member variable.
First of all, it should not access the node outside updatePaintNode(),
that is, outside the render thread. Secondly, the node is owned by the
scene graph and may be nuked whenever the scene graph feels so. Some
creative re-parenting easily triggers a case where AnimatedSprite ends
up accessing a node that was already deleted by the scene graph.
Change-Id: I89205ac36333a2fcb094121afa61b6409fda5883
Task-number: QTBUG-51162
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
|
|
| |
... or equivalent.
QtBase 5.6 headers already compile that way, so let the other
modules follow suit.
Task-number: QTBUG-45291
Change-Id: Ied526fb2d7adc3a68946d6843a6bd6f475ebf864
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: Iea662214988c35473a8526c546f01ac5629308dc
Reviewed-by: Venugopal Shivashankar <venugopal.shivashankar@digia.com>
|
|
|
|
|
|
|
|
| |
We use 64bit value encoding there, so gcc complained that memcpy would
write 8 bytes to a 4 byte pointer value.
Change-Id: Ib44372fb0e4bbacf0279930f7f5fcf4bbb5d48bf
Reviewed-by: Nikolai Kosjar <nikolai.kosjar@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Given a simple ShaderEffect with supportsAtlasTexture:
ShaderEffect {
property variant source: myImage
supportsAtlasTextures: true
}
ensure the atlas subrect used by the ShaderEffect is updated if
myImage.source changes.
Change-Id: I4560b80ba38c3487b16f53411451fb9891920738
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
| |
invalidate() releases resources, but doesn't clean up the materialCache,
resulting in a memory leak
Change-Id: I3cf7b7a3b977bb9aa295ef1824c384bcdf9494f2
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before change f253f4c in qtbase, when an ordinary mouse wheel is
rotated the phase would be ScrollUpdate continuously, without going
through the other phases; whereas on a gesture-interpreting trackpad
there would be a ScrollBegin first. Now, an ordinary wheel mouse
uses NoScrollPhase continuously, which makes it easier to tell the
two cases apart. But we're required to handle the new phase
explicitly. To avoid change in existing behavior, we handle it
the same as ScrollUpdate phase was handled before.
Change-Id: I2d51a89a965704cf843303d523463805b4fe6ff8
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
|
|
|
|
|
|
| |
Change-Id: I3eb57baf1812f831335429cc7d2b4424f3cfa785
Task-number: QTBUG-50929
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: Ie9541cb83397fe2eccd69cd09cf8d9217e6905e2
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
this fixes static builds by ensuring that all dependencies are exported.
Task-number: QTBUG-51071
Change-Id: I5928f853a1d13b6a73533f9c6f6eae9da0e52a26
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
| |
the QSGGeometryNode owns the geometry, so we need to set the flag to avoid
a memory leak
Change-Id: Ifaf1e88c437b765a76d7ab4028c63d66413440bb
Reviewed-by: Jan Arve Sæther <jan-arve.saether@theqtcompany.com>
|
|
|
|
|
|
| |
Task-number: QTBUG-46758
Change-Id: I14e394021c231bda5552c8d1c98f20c903a62f12
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Not validate binding if it is disabled and validate when it became
enabled. This helps to get read of warnings about not existing property
that started to appear since 61ce37de40711ef2d4a6b4989d8183e1711fc47d.
Task-number: QTBUG-48623
Change-Id: I4aa92896c04b146a79e606842c1d33661b3f5a70
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Tobias Koenig <tobias.koenig@kdab.com>
Reviewed-by: Mitch Curtis <mitch.curtis@theqtcompany.com>
|
|
|
|
|
| |
Change-Id: I0c94a5c2846b48c8aea7ffff14352f8a247c8cec
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
| |
Linking to "QAbstractItemModel" will take the user to the class
documentation, when it was intended to take them to the
"Using C++ Models with Qt Quick Views" page.
Change-Id: Id705353bf729fe9990ca6f77e3d4e430cff33918
Reviewed-by: Venugopal Shivashankar <venugopal.shivashankar@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was found while researching QTBUG-49596. The selection
ranges in the QGlyphRun are inclusive, so the length needs
to be end - start + 1. For a node of 1 glyph with either
preceding or succeeding node, we would detect an overlap
always, since the initial rangeLength would be 0. This
is reproduced by the textinput_selected_fallback_font.qml
test, but the bug was hidden by a different bug in
QTextLayout.
Change-Id: I65d70b1223eebeb5cfbb277fade7f4753465364f
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 2e75be5f64fb21cbbdff3353dbd507c2ca26946a.
This patch was originally written by Andrew den Exter in
222e06bf4ed509e72c1533cbe1d4859ca96933f3, externally from the main Qt tree. I
upstreamed this as part of our porting efforts.
Some time later, this was accidentally reverted in
2e75be5f64fb21cbbdff3353dbd507c2ca26946a: we can't go back in time to examine
exactly what happened, but presumably Andrew didn't notice that I had upstreamed
this, attempted to apply the patch and ignored the "already applied" warning by
accident - and continued on dutifully with the patch accidentally reverted.
This change is correct, though, and is thus reinstated.
Change-Id: Idfe6ab39ad011f0401de25fe056aa3eb3fb8b424
Reviewed-by: Andrew den Exter <andrew.den.exter@qinetic.com.au>
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
|
|\
| |
| |
| | |
Change-Id: Ib4eb5c3572cb5ec11eb744572ec796dc7e70456d
|
| |
| |
| |
| |
| | |
Change-Id: I3c122c0fe56b12a8e91c68ceff357600adcfc2bf
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The whole purpose of native rendering is to optimize precisely for a
given pixel grid.
DPI scaling can change when either settings of the current screen are
changed, or when the window is moved to a different screen.
Details:
- add an ItemDevicePixelRatioHasChanged item change event
- detect DPI scaling changes by watching screen (identity) and screen
config changes
- when DPI scaling changes, recursively send an
ItemDevicePixelRatioHasChanged signal to all items with content
- when a natively renderet TextItem catches such an event, call
updateLayout() which automatically picks up the new logical DPI
Task-number: QTBUG-49019
Change-Id: I9f4f8d1a7f2c172ed26c276294ab143161c4a48b
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
IFW uses QQmlV4Function in a QJSEngine environment, so there is no QML context.
Up to commit cc98678f404cd49750076795f39b31bfa36c80c3 that would work. Let's
fall back to the root context as scope if we don't have a qml context.
Change-Id: Ib240746a95a0d16d61620f97abc31e4c74f36723
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| |\
| | |
| | |
| | | |
Change-Id: I971229ff050c98258729ad70d9e06b89e5bbf617
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Task-number: QTBUG-50516
Change-Id: I6a1513b22401b0fe45da758a239ad82038b83264
Reviewed-by: Zsombor Egri <zsombor.egri@canonical.com>
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This is basically the same patch as 35d8d060b8621cfd, but this
time for GridView instead of ListView.
The major difference is that GridView seems to do the right thing
when the model insertions happen after the visible index. That
explains the missing part of the patch in applyInsertionChange()
as compared to the same function in QQuickListView.
Also, QQuickGridView auto-tests are a bit more robust than their
ListView counterpart. So no changes were necessary to existing
test cases.
Task-number: QTBUG-48870
Change-Id: I19b24c4d84a1a4cef4fdb3ddd3381d0c6b93a76a
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Same as 51443673dce131f2.
Change-Id: I785f777e761aba369014d1e4ad8020066f35fbe0
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
readImage already assigns to errorString if something goes wrong, so there's no
need to overwrite it with a less useful error message (which doesn't tell us
what actually went wrong).
Before: Invalid image data: test.qml
After: Error decoding: test.qml: Unsupported image format
Change-Id: I7df3d24497d001e18453034c6f9f63bb51bd0ecc
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
In itemview/positioner transitions, Animator's target-property is never
explicitly set. QQuickItemViewTransition passes the animation target as
a "default target" instead. This change adds missing handling for the
default target.
QQuickItemViewTransition sets up default "x" and "y" state actions.
If Animator drives opacity, scale, or basically anything else than x/y
animations, it failed to extract the from/to values from the state
action list.
This change fixes the issue that if the default state actions do not
match the animator property (x/y vs. scale/opacity), it uses from/to
values specified on the animator itself. Before, it did that only if
the default state action list was empty. This is not the case with
itemview/positioner transitions.
Change-Id: I0f15e20bc860ddec23e59efebbc9cd346317f4de
Task-number: QTBUG-50908
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
| | |
| | |
| | |
| | |
| | | |
Change-Id: Ifa5c046d389f8f4b538b5b7727b99a0dd26a6121
Reviewed-by: Topi Reiniö <topi.reinio@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | | |
Change-Id: Idb5dc9e68b92cb6b20b4e8df44ed6ac55aaa9568
Task-number: QTBUG-45745
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
... in string comparisons. It's more efficient.
Change-Id: Ia157618cf870870f24c4a2962a385962da819275
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
m_fbo is null when using a threaded render loop.
Change-Id: I297ba651f9605f1718dbe9d09bd30e9682fb8401
Task-number: QTBUG-50085
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | | |
Change-Id: Ieb74ab1a1369ca8c70243002e6b6deec86c03dd6
Task-number: QTBUG-50250
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
|