| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Every batch would have its own buffer with its own memory
and the original idea was that it would stabilize so that
the same batch would be used over and over, so the memory
allocatd for each batch's buffer would be reused. In practice
batches get discarded and recycled in somewhat arbitrary
order so we always end up reallocating.
Instead, use a single buffer for all uploads. This saves us several
reallocations per frame, and also prevents pooling of several
potentially large allocated blocks inside the renderer.
The exception is when using visualization or for drivers
which have broken index buffer support (nouveau only currently),
where we keep the existing behavior as the buffer data is
needed during rendering and visualization.
Change-Id: Id29095e00128ec1ee6d4ac3caa8f2c17cc7bbd27
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
|
|
|
|
|
| |
It is perfectly valid to reparent nodes to a parent
that is part of an existing hierarchy.
Change-Id: Icf2dbe6030d399047e7ebe735e3096d1a8cd8c0e
Task-number: QTBUG-43278
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Since the children are being removed from the parent, using
nextSibling on the child won't work. We need to extract the
first child from the parent until there are no more children
left.
Change-Id: Ifca5f2760e2b1c7ba56c198623c5dc9d82c7560f
Task-number: QTBUG-42530
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
|
|
|
|
|
| |
Change-Id: I0b68c534ea513a7c230b12114f6b42b069f9864b
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
| |
Change-Id: I64ecbf6cea463387a70e909ecc5f9165d22a7b0f
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
| |
Change-Id: I3427129dc7a0e68c8fa0b61f353835e2bee179a0
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
| |
Change-Id: I0f2f77c9cc268a0c5ca3ffe0cd66fc98bb1964b3
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Changed built-in QtQml value types to use gadgets. This is in preparation for
supporting external gadgets.
This replaces the mostly direct inheritance of the concrete value types with
gadgets and "dynamic" inheritance through QQmlValueType being generic. Over
time as some of the value types may become gadgets, we can remove the ones
here. It's important that these "separate" gadgets have the same memory layout
as the actual types (QPointF, etc.).
Also while QQmlValueType remains practically a singleton, it's not required
anymore to be one. Consequently the JS wrappers for value types keep their own
instance of QQmlValueType. This allows eliminating the qobject_cast in various
places that dealt with the singleton nature. This comes at a cost, making the
JS wrappers slightly heavier. However that is meant to be a temporary situation
and finally the value type wrapper should merely store the meta-object in
addition to the data and the type.
Change-Id: I15071ded0a1e54203f29ec1ecf7a9ab646d1168e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
|
|
|
|
|
|
|
| |
This is no longer required, and simply uglifies the code
Change-Id: Iba91a1d7735ebe23a43437f137a488423b6eb743
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It is currently necessary to set supportsAtlasTextures to enable
batching of all ShaderEffect, even if they sample a
ShaderEffectSource or have no sampler uniform at all.
After fccbe57258c10e18ccdbbec405f5522dbd97af66 it is not necessary to
set the RequiresFullMatrix flag based on supportsAtlasTextures since
this now prevents batching altogether. Limiting batching should
however also first verify that there are any sampler of the atlas
used on that ShaderEffect in QQuickShaderEffectMaterial::compare,
since that check wouldn't have any effect in
QQuickShaderEffect::updatePaintNode through RequiresFullMatrix.
Also make sure that using texture()->normalizedTextureSubRect()
as the texture coordinate attribute is synchronized with calls to
removedFromAtlas for all samplers and make sure that this only
applies when the texture count == 1. This fixes a bug where
setting supportsAtlasTextures on a ShaderEffect with more that
one sampler would incorrectly use the texture coordinates of the
first atlas texture in the geometry.
This patch also removes:
- The unneeded check for the RequiresFullMatrix in compare; the same
program can be used in batched geometries and the renderer itself
will prevent merging in that case.
- The possibly heavy string comparison of m_source;
QQuickShaderEffectMaterial::type() and QQuickShaderEffectMaterialKey
are doing that for us more efficiently.
I updated the documentation to clarify the 1-texture rule and
explain the difference with qt_SubRect uniforms and when one or
the other should be used.
Change-Id: I85a86b8e7cb76f1536d5fcd369c7889f22537846
Reviewed-by: Michael Brasser <michael.brasser@live.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
| |
Change-Id: I4f8aa9b209096bbcdabcb689007e4166279f8a4c
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is analogous to GL_TRIANGLES and GL_TRIANGLE_STRIP. There is no
reason why we shouldn't batch it. In the case of GL_LINES this means a
slight behavior change, though: If you provided geometry with an uneven
number of vertices, it would draw lines the way opengl specifies it and
drop the last vertex. Now, with batching, it will draw a line from the
last vertex of an "uneven" node to the first vertex of the next node.
This, however, is exactly like it already works for GL_TRIANGLES if you
specify a number of vertices not divisible by 3. We assume the user
should specify a number of vertices that matches the drawing mode.
Change-Id: I372b9c15a19f065d3aeb3727688cca408ce9608e
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The value type wrapper for QFont tries to keep track whether pixel or point
sizes were set and warns the programmer accordingly if both were set. This is
implemented using separate boolean variables, but this can also be done using
QFont's resolve mask.
This move is motivated by the ability of the value type wrappers to operate
only on the type they wrap in the future, without requiring additional
helper variables.
Change-Id: Ia05093b5dcc950281ef89b2bb6fe96c18329cdfa
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Reorganize the rendercontrol example to demonstrate both the single
and multi threaded approaches.
A small helper function is introduced to the QQuickRenderControl API:
The QSGRenderContext has to live on the render thread. Previously there
was no way for applications to move it to the desired thread. This is
now possible.
Pass --threaded to the rendercontrol example to use a separate render thread.
[ChangeLog][QtQuick] QQuickRenderControl can now be used to render the Qt Quick
scene on a dedicated render thread, similarly to how the built-in threaded render
loop operates.
Task-number: QTBUG-42813
Change-Id: I01c3b2ffca8a174d9d2c267a51f2e484ed7b34b3
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
Reviewed-by: Jørgen Lind <jorgen.lind@theqtcompany.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This appplies only when using opengl32.dll. When using ANGLE or
when using another non-opengl32 DLL, the "windows" single-threaded
loop is chosen, just like before.
One badly written autotest is fixed. We must never assume that
performing some QQuickWindow related work will leave a context
current on the main thread. The assumption worked by accident with
the single-threaded render loop.
[ChangeLog][QtQuick] Changed to use the threaded render loop by
default on Windows when running with desktop OpenGL (opengl32.dll).
Change-Id: I21798622c19ca510a97a96c19c1e70e29f086e3a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Jørgen Lind <jorgen.lind@theqtcompany.com>
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Conflicts:
src/qml/jsruntime/qv4arraydata.cpp
src/qml/jsruntime/qv4context_p.h
src/qml/jsruntime/qv4globalobject.cpp
src/qml/jsruntime/qv4internalclass.cpp
src/quick/items/qquicktext_p.h
src/quick/items/qquicktextedit_p.h
src/quick/items/qquicktextinput_p.h
Change-Id: If07e483e03197cb997ef47a9c647a479cdb09f4c
|
| |\
| | |
| | |
| | | |
refs/staging/5.4
|
| | |\
| | | |
| | | |
| | | | |
Change-Id: I5978bab8a4fde4c2ee33907fd81f49cb69e4fb26
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Change-Id: I4ceb1a969bd4296b82f899088b02b5e8cf100bcd
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Change-Id: I9b6c54b572d4653d609b154000e274b9e64d591f
Reviewed-by: Fawzi Mohamed <fawzi.mohamed@theqtcompany.com>
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
String-based connect is relatively slow, and should be avoided in core
items. This improves performance of the
tst_librarymetrics_performance::instantiation_cached(043) image - empty
test case (approximately halving the time).
Task-number: QTBUG-43096
Change-Id: I02485c515435eceacc95c55f877fc8566e7406d7
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This commit ensures that the value type providers installed by the
QtQuick2 QML module during initialization are uninstalled when the
plugin is unloaded.
It also fixes a bug in the type compiler so that it now works with
types from plugins which get unloaded and then reloaded.
Task-number: QTBUG-43004
Change-Id: I4b3fb75aae65dfbc5de9c88701ed82514087ab7d
Reviewed-by: Matthew Vogt <matthew.vogt@qinetic.com.au>
|
| | |
| | |
| | |
| | |
| | |
| | | |
Task-number: QTBUG-43063
Change-Id: I92a33047ac7fa2afeff4986b6e42c07afbc59918
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Several more methods/slots are inherited from QWindow but are not
documented for the QML Window type. Also some other documentation
improvements.
Task-number: QTBUG-40093
Task-number: QTBUG-42426
Change-Id: Ib753be269cbc41ee540e6556e0ef483758eefe62
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
|
| |\|
| | |
| | |
| | | |
Change-Id: I89bbb2977350a03c156d531f810d08a5560ffbb3
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Fractional scale factors are not as broken as previously
believed, especially for Qt Quick. Keep the door open
for finding a way to support it at some point in the
future.
Change-Id: Ifeadcc53175ac6c25ea0288d5fe1966e3de408f9
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The subpixel shader is probably not a good default here because most
devices are mobile (can change screen orientation) and have high pixel
densities. Furthermore, it breaks text rendering on the Surface RT, where
the graphics hardware is too weak to support the number of uniforms the
subpixel shader uses.
Task-number: QTBUG-41769
Change-Id: I10210af91976ab55e611025c0452a1ee1f3114a9
Reviewed-by: Oliver Wolff <oliver.wolff@theqtcompany.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | | |
Change-Id: Ic8c8e6d7a9d99216292b8b4faa2926d849333a05
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | | |
Change-Id: Ic0e594cb53016e6f68fbfb6e6064707344afefef
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| |\ \
| | | |
| | | |
| | | | |
refs/staging/5.4
|
| | |\|
| | | |
| | | |
| | | | |
Change-Id: I2e7fc085663e00dd0390593a91c30d23d1369c4e
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The font caches can only be used from a single thread at a time.
QFontEngineFT for instance, uses a global static thread storage
which is accessed on releasing and creating engines, and this
causes a crash if the font engine is created on one thread and
released on another.
We use the updatePolish() function to make sure the caches are
empty before entering updatePaintNode(), and then we invalidate
the cache again after updatePaintNode() is done.
[ChangeLog][Text] Fixed uncommon crash in text nodes.
Change-Id: I01dbc2ed58aeebd03d77a157c700330334bdb385
Task-number: QTBUG-38800
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The new documentation style for publishing docs under qt.io
requires dedicated div classes for pages that use a three-
column layout.
The new divs enable the columns to stack and react to changes
in the window width.
Task-number: QTBUG-42086
Change-Id: Id0ccb0ef7e0be237789b8c891db05413efc1f8aa
Reviewed-by: Martin Smith <martin.smith@digia.com>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This reverts commit a9238292145e05d36be7c35bdd50829c6400a3de.
It also reverts follow-up commits:
87755d0437413cfb875c50e3dfa53030601567c9
c2c710e5a8d66fa696276aa69c2e3b00436eefe5
cc8a76ce8ea76afe912902067e95ca2abf9e482f
The new implementation is much too slow because it will cause each
section of the border image to be uploaded to the atlas, and it will
not be shared between different border images nor between different
sizes of the border. The performance regression is so significant
that it's not worth it for the original bug fix.
We leave the tests around, so some of those might regress due to
this.
Change-Id: I3f56f4564255ebb77e6487881a6a11b19ad0234e
Task-number: QTBUG-42288
Task-number: QTBUG-35838
Reviewed-by: Michael Brasser <michael.brasser@live.com>
|
| | | |\
| | | | |
| | | | |
| | | | | |
refs/staging/5.4.0
|
| | | | |\
| | | | | |
| | | | | |
| | | | | | |
Change-Id: I2c69deb3bc69da1a06485aaf72fb298b08756734
|
| | | |/ /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The same old issue surfaces in the windows render loop too.
The basic render loop is already fixed, apply a similar patch
to the windows one too.
Task-number: QTBUG-42213
Change-Id: I07068315f5164014e329b8ce061947c97ae9da61
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
There are new APIs in Qt 5.4, the import version needed is 2.4.
Change-Id: I1c52c6a50362f539670d9ad00e03e57208234a17
Reviewed-by: J-P Nurmi <jpnurmi@theqtcompany.com>
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | | |
Task-number: QTBUG-42777
Change-Id: I8027f100c593ae7c57df7c952e66cc2b3fae1dd9
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Leave it up to the clients of QQuickRenderControl to do this, if they
want it. It is usually not necessary.
In the single-threaded widget world forcing deferred deletes to execute
on every invalidate(), so for example from the hide event handler of
QQuickWidget, is dangerous because widget apps tend to deleteLater()
all sorts of widgets which can then be destroyed at unexpected times.
From windowDestroyed() we continue to send the deferred deletes, just
like all the render loops do.
Task-number: QTBUG-42618
Task-number: QTBUG-40435
Change-Id: I8189124e2e7675361ee97bd8ba3e88b10ef193fa
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
sizeHint is not a property of QQuickView
Task-number: QTBUG-39044
Change-Id: I6e19e975ab71f73ddaf703956afb90f2e607b6f8
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Fix a typo and add a note to the signals.
Change-Id: Ia8810562c5b5f192d7e54bc965807b8e78a26985
Reviewed-by: Karim Pinter <karim.pinter@digia.com>
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
|
| | |/
| |/|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Linear filtering can cause pixels outside the glyph's bounding rect
to be sampled. On drivers where uninitialized texture data is actually
uninitialized, this could cause artifacts in rendering for glyphs at
the right edge of the initiliazed area since they would sometimes sample
random pixels. To avoid this, we add padding between the glyphs in the
cache.
[ChangeLog][Qt Quick] Fixed uncommon artifacts in rendering of distance
field glyphs.
Task-number: QTBUG-42148
Change-Id: I6982b4a150d9459185d50a4362e1ead588d3860f
Reviewed-by: Yoann Lopes <yoann.lopes@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The previous code was wrong, as it cleared focus from the active
focus item directly. By doing so we would only clear focus inside
the focus scope that surrounded the item, but leave the scope itself
with active focus (which would then be the focus object). The intended
result is rather to end up with the root as focus object.
Change-Id: I455a8939f8bc6c48765119b995aa781aee6d1e70
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This makes compilation possible of the saveTexture() function which is
used to store the distance field glyph cache on disk when debugging.
It also changes the warnings output from here to have no prefix, since
this is anyway always available as the context in Qt 5.
Change-Id: If3b64f0921a59c485b5321a5465369708e764afc
Reviewed-by: Yoann Lopes <yoann.lopes@theqtcompany.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Fixes purify warning:
[W] UMR: Uninitialized memory read in QSGAnimationDriver::advance(void) {1 occurrence}
QSGAnimationDriver::advance(void) [qtdeclarative\src\quick\scenegraph\qsgcontext.cpp:221]
Task-number: QTBUG-42213
Change-Id: Ide35830c60c5935644747da5cfbad1cdbd357232
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This is essential to the Qt Quick Controls for implementing floating
text selection handle popups.
Change-Id: Ibae65110a5db6d65dacd197fef3ad567d11342dc
Reviewed-by: Alan Alpert <aalpert@blackberry.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
|
| |/
| |
| |
| |
| |
| |
| |
| |
| | |
Just copy the one from QSGSharedDistanceFieldCache with some
compile fixes to the superclass so we store the default cache
as well.
Change-Id: I1fcc390601eea58f8b7729c9cead418e4c94714c
Reviewed-by: Yoann Lopes <yoann.lopes@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
There are a couple of ways in which this code creates
temporary copies of the window list, m_windows. This is often
benign but there are also places (e.g. startOrStopAnimationTimer)
which get non-const references to items which results
in m_windows being detached from the temporary resulting in a real
copy of the list items. Again the copy is often fairly benign,
however, as the code also relies heavily on pointers to items
in the list, it can also result in crashes.
I think it might be advisable to store a list of pointers to
Window structures rather than store the structure themselves as
it appears really easy to introduce copies of the list accidentally.
The removal of the use of foreach for example is not made here for
aesthetics but because it introduces a hidden temporary copy of
the list.
Change-Id: I504951a897c4fb0cf106f5a4792b5cfcd532ba8f
Reviewed-by: Laszlo Agocs <laszlo.agocs@theqtcompany.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Generate the code from a macro instead of duplicating boiler
plate code. Operate on Heap::Base instead of Managed.
Change-Id: I84c5a705980899be3e5b931a093645e50d3923bf
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
|