There's a scanner in QQmlJS::Lexer::scanDirectives that can parse those, so
let's get rid of extra parser that operates on a string. Instead this way we
can do the scanning all in one shot, avoid detaching a copy of the source code
string and (most importantly) bring the parser closer to the copy in Qt
Creator, which uses the directives approach to extract imports and pragma.
Change-Id: Iff6eb8d91a45d8a70f383f953115692be48259de
Reviewed-by: Fawzi Mohamed <fawzi.mohamed@theqtcompany.com>
This allows QtQuick.Controls 1.x and 2.x imports to co-exist even
if they are two different plugins with the same module directive.
Change-Id: Idee302439e3c2fd6813ba2f41b69144fbae7902c
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Change data type for defsUntyped from QList -> QVector & reserve space when
creating the list. Drops ~250ms off a pretty morbid QML testcase, most, but not
all of which was spent in allocations.
Change-Id: I2ed8c62e7d41ab353a0194da268a2b430f079474
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Profiling the loading of a pretty morbidly large QML file consistently showed
that this was quite slow, around 1300ms in removeSharedExpressions, of which a
good >600-700ms (I didn't count exactly, but it was a very large amount) was
down to allocating and freeing QHash nodes.
As we don't require removals, leaving insertion and lookup as the only two
remaining options, a sorted vector becomes a viable alternative (credit to João
Abecasis for the idea).
An additional benefit of this change is that the two hash lookups are now
compressed into a single 'hash' lookup (via the lower_bound call) instead of
separately using contains() / insert().
Measuring the exact saving is difficult, but it looks like this saves between
700-1000ms off the runtime RemoveSharedExpressions. After this patch, malloc and
free are dominating the optimizer run, instead of any particular method.
Change-Id: I6c0bb8495eac4dd3613ba0274e8802d7bd609460
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
For gadgets/value types we use moc's static_metacall, which doesn't call the
parent class implementation. Therefore before placing a static metacall we
must resolve the indicies and find the right meta-object.
Change-Id: I258e3d9ecfc704498c68772dc42b16134a3bfd83
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
Add a parseVersion() helper function to avoid duplicating
the same version parsing logic three times.
Change-Id: I4e5b6a8c86ee3a26c4eb91c660a81176ac0346cf
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
In the multithreaded case we attempted to create QWindows outside
the gui thread on platforms where QOffscreenSurface is backed by
the hidden QWindow in the absence of a better solution. This has
to be avoided. Therefore, pass a suitable surface from the gui
thread instead.
This will avoid "Attempting to create QWindow-based QOffscreenSurface
outside the gui thread." type of warnings when running on OS X with
--threaded.
Change-Id: Ie3ebeeeaa9e6bdf83e763e40213e2940fbfde667
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
Reviewed-by: Jørgen Lind <jorgen.lind@theqtcompany.com>
Culling and gl_FrontFacing depend on the vertex order matching the
value set through glFrontFace. To match the default counter-clockwise
order we need to preserve the parity of triangles relatively to the
start of the indices argument of glDrawElements, but we break this
parity because of the degenerate triangles between merged strips.
Fix the issue by skipping the first degenerate triangle, the parity
of following strips will be preserved since they are then separated
by two triangles.
This fixes a regression of
tests/manual/scenegraph_lancelot/data/shaders/culling/culling_1.qml
triggered by 38cab579a0.
Change-Id: Iefa7eaab68112d31be9d8646bd288eb000528cd5
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>
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>
Access to QJSEnginePrivate::createCache and the previous
signature of QQmlPropertyCache::invalidate is used.
Change-Id: Ie2578eee33ee4ebb2cdc4574d8d1cb7c8209868e
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Reviewed-by: Thomas Hartmann <Thomas.Hartmann@digia.com>
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>
May help a bit to the reader, even if the line is already there
sometimes it can be confusing if it is a default property like data
Change-Id: I5cb8e8833c78a784ad4f5541d094840477e8f350
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
[ChangeLog][QtQml] Custom C++ value types annotated with Q_GADGET are now fully
accessible in the QML and QJSEngine JavaScript environment.
QJSEngine::toScriptValue can be used for injection and fromScriptValue to
extraction. The QML "built-in" gadget wrappers for QPoint and the gui types are
not exposed this way, toScriptValue(point) will still return an opaque QVariant
wrapper. We could expose the core types right away, but then we would be
lacking an API to enable use of the Gui types that are registered in QtQuick.
It would be better to make the core types in qtbase gadgets and thus enable
them without the need for hooks and init functions to be called by the user.
Task-number: QTBUG-29769
Change-Id: I8179cd599bdc1209ff61cfdbdda419cb400296bb
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The private class is not empty anymore, so calling the wrong
constructor would lead to memory corruption
Change-Id: I6d477fc63bba46be733273f548397c2a50ef2bf4
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Profiling of allocations shows that this is one of the most frequent allocators.
Change-Id: Iae0350c61bfe37381577642b9a93064377f45c1d
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The QQmlEngine has two containers for property caches, one for QML types and
one for all-purpose meta-objects. The latter is rather useful and now being
moved to QJSEngine to be available there.
Change-Id: Ieab65c400b8a2e410e5f9eee6d603162dbb864d9
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The cache is seemingly tied to QQmlEngine, but it isn't. A lot of
times a QQmlEngine parameter is unnecessarily dragged around and the
engine member is option as well as it turns out.
Change-Id: Iffd2a5046e9785249689ebfcbc8a0ad509f76aea
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
- alloca() needs malloc.h
- crtdbg.h couldn't be found on MinGW
Change-Id: Ibbf91a58d39ef1e2572baae3c409393acf7aa5df
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
QQmlValueType doesn't have anything really necessary :). Some code it has, but
that was only called from the wrapper, so it was moved there. This also reduces
one level of memory allocation for the wrapper.
In theory the allocation for the JavaScript wrapper could be changed to be the
size of the wrapper _plus_ the size needed for the gadget. However in
anticipation of wanting to be able to move objects, we can't do that as we
can't memmove() gadgets around - they might contain pointers to memory areas
within.
Change-Id: Icb5a6f5513e3b2c87f6639f6f7b1fb572af08137
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
In order to support calling invokable methods in gadgets from JavaScript, the
wrapper code needs to be able to operate on gadgets and not only QObject
pointers. The minimal abstraction for that - QQmlGadgetOrObject - is passed
through the relevant invocation methods in the wrapper for that.
Change-Id: I94f939c942241f49dce4d32d05bf24822b2d331b
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
These can extract the type information from either a property cache (through
QQmlData from a QObject) or (fallback) from a QMetaObject. The difference
between a QMetaObject and a QQmlPropertyCache is what QQmlMetaObject abstracts
and therefore we can move the functions there. This is in preparation for
gadget support and also avoids the repeated QQmlData::get() dance but allows
the re-use of a QQmlMetaObject instance in the future to do that dance only
once.
Change-Id: Ibff6ce498d09fabc97e0801653edc5b1ff446c6a
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Simon Hausmann
Lars Knoll
J-P Nurmi
Albert Astals Cid
Robin Burchell
Laszlo Agocs
Jocelyn Turcotte
Gunnar Sletta
hjk
Oleg Shparber
Thiago Macieira
Nobuaki Sukegawa
Konstantin Ritt
Tasuku Suzuki