This is a regression from Qt 5.1/5.0 introduced with 5.2. Private slots/methods
such as QObject::_q_reRegisterTimers() as well as QObject::deleteLater() are
not suitable for calls from JavaScript. deleteLater() in particular is covered
by the destroy() replacement slot.
Task-number: QTBUG-39744
Change-Id: I9b25f3c0d4095ba8e2e8e5ee47e37903b5def1f9
Reviewed-by: Michael Brasser <michael.brasser@live.com>
This can happen during QObject destruction, when
QObjectPrivate::deleteChildren() itself sets entries
in the children list to zero when deleting. These
zeros cause crash in markChildQObjectsRecursively().
Task-number: QTBUG-38635
Change-Id: I29ad9e793b78ca4e8d73fbb125f46db1b8292f20
Reviewed-by: Alan Alpert <aalpert@blackberry.com>
After commit ac57f185d1 we succeed in selecting
the correct overload based on the supplied arguments. However when calling
slots on objects without a property cache, we end up using the local "dummy"
variable to store the synthetic propert data. We also store the currently best
patch in the "best" variable, which is a _pointer_ to the property data of the
match. Suppose we have 5 overloads to choose from, we find that the 3rd is
the best. Then we try the fourth but find it unsufficient and break out of
the loop. Unfortunately the "dummy" property data at this point contains the
data of the fourth (wrong) overload, and our best match variable points to it.
So then when we finally call the method, we do it based on the wrong property
data.
The easy patch is to simply copy the few bytes of property data, so "best" is
stored by value instead of pointer.
Change-Id: Ie2ebbdb88a117770b6c7b9490e1c634077020e9d
Reviewed-by: Karsten Heimrich <karsten.heimrich@digia.com>
This is an ammendment to commit 60730cbb5e
to further restrict the ability to set dynamic properties on JS wrapped
QObjects only on those that are associated with a qml context. Only one
such association comes with the static property lookup rules of QML and
therefore only those should be prohibited from dynamic properties.
The previous implementation on using the "compiledData" field to detect QML
association or not is not strong and reliable enough.
Change-Id: I10c0e6e58a2727c01a6cb56fdf912bf250333e1f
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
In QtScript you could add properties to a JS object that wraps a QObject.
Depending on the wrap option the property was either stored on the JavaScript
side or as dynamic QObject property.
In QJSEngine/QJSValue, neither was supported - properties could not be added.
For QObjects wrapped in JavaScript that weren't created by QML, we can restore
the behavior of storing dynamically added properties as JavaScript properties.
This makes porting from QtScript to QJS* much easier.
Task-number: QTBUG-37408
Change-Id: I5ef1f379c08c3d84de9bdcac9b6a9397238064de
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Always call ~QObjectWrapper() to the wrapper object in
QObjectWrapper::destroy(), otherwise the m_object member of QObjectWrapper
may still hold a QtSharedPointer::ExternalRefCountData object which will
never been deleted.
I don't know why this will not cause leak in the past, but it seems that
the leak was introduced in 7ae796cb14
Change-Id: I24b49bb11f95b7e3060c7adba1ab80b615da2942
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
The method is only required for the QObjectWrapper, but
there we can instead simply register the deletable in the
destroy method.
Change-Id: I944319d327859cce33ac31a174e8d56fc4babfc1
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
When resolving overloaded slots in a situation without property cache, then we
need to detect that we've reached the end of the overload lists and exit from
RelatedMethod with zero.
Task-number: QTBUG-37157
Change-Id: Ifaab8cc1b377725c99c1bb89bb9bb74d5a0af1b8
Reviewed-by: Michael Brasser <michael.brasser@live.com>
In the test case in the bug, the signal was emitted from the QApplication
destructor (somewhere from the qpa plugin when the platform windows were
destroyed)
Task-number: QTBUG-37351
Change-Id: Ieec59e12be10bab1428743b80eecdf22ef9d8bf6
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
This reduces the size of QQmlBinding by 12 bytes.
Task-number: QTBUG-37134
Change-Id: Id55257edec8cee88d863374e8a96d7eebbeaf523
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
Move all our runtime methods into the QV4::Runtime
struct and give them nicer names without underscores.
Sort them logically and remove a few unused methods.
Change-Id: Ib69b71764ff194d0ba211aac581f9a99734d8180
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
The data is anyway stored in the name property of
the FunctionObject, and is not performance critical.
Change-Id: If1784b0ec6f368bc474c246bb9c2c50d5e56b689
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
If we don't have a property cache, we need to fall back to a slower method of
determining the overload methods. We have the code for that in RelatedMethod,
once we determine that we're calling overloads, but we never hit that code path
because we did not _initially_ determine that the method was an overload.
Task-number: QTBUG-37157
Change-Id: I8ff39156e5668236b3797400b4086ed545624398
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* Don't create prototype Object for bindings and
signal handlers. It is inaccessible and not required.
This saves one Object-sized allocation per binding.
* Shrink the size of QQmlContextWrapper by removing
the v8 member variable.
* Shrink the size of QObjectWrapper by moving the destroy
identifier to the engine.
Change-Id: I76e84e4c0581e97a19d2e959f814ac84d9c431fa
Task-number: QTBUG-37134
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Don't crash when an exception is thrown in a JS slot but we don't have
a Qml engine.
Change-Id: I1530d5c1c8cb9b9b33b9fdd0d45639fd4a0516f7
Reviewed-by: Karsten Heimrich <karsten.heimrich@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Move to a class hierarchy that mirrors the main classes. This will
allow moving functionality over into the Ref classes, as the current
Managed classes become mainly something that holds the data. This
is required to make objects movable by the GC.
Change-Id: I4ca88ab0e5d8c88c8dc56d51937990500a33e0d9
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
First step of removing the templates here and turning this
into a class hierarchy. This is required, so we can move all
member methods into the Ref classes and make objects movable
during GC.
Change-Id: Ie14af07fd3e72a7d84a528d0042189ff12ba21bb
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Commit 04774bb14c long time ago introduced the
concept of fixed order binding initialization with lazy evaluation, where a bit
is reserved for each binding that indicates whether it's been initialized the
first time or not. When reading a property on a QObject, we'd check if the
corresponding binding for the property has been initialized or not and flush
(i.e. execute) the binding if necessary.
As part of the V4/V8 clean-up, commit 1eb4120094
removed the StoreV8Binding instruction, which made the call for setting the
this-binding-is-not-evaluated-yet bit. Nowadays we only use StoreBinding, for
which this optimization was never implemented (and not needed really). Now that
we have a unified JS code path, we need to set the pending binding bit and also
make sure that we call flushPendingBinding for any JS side property access
(accelerated or not).
Also flushPendingBindingImpl had two bugs:
* In an attempt of trying to find the binding to flush, it could happen that
we'd try to flush a previously destroyed binding (m_mePtr is null), so
the b variable would remain the first binding in the object and we'd flush
the wrong one (instead of none). Added a missing check to verify that the
property index matches.
* Also resetting the mePtr must be done through clear(), to ensure that the
pointer in bindValues in the VME is also cleared, to avoid re-enabling the
same binding again in complete();
Task-number: QTBUG-36441
Change-Id: Icdb0c8fb036051fd5d6c4d33b10cd0c0ed9a9d5c
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Remove SafeValue, it was used to port over to an exact GC. Since
we now have that, we can now safely merge it with QV4::Value
again. Also rename SafeString to StringValue for better naming
consistency.
Change-Id: I8553d1bec5134c53996f6b0d758738a0ec8a2e4d
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Pass a pointer to the underlying object instead of the ArrayData
to virtual methods that modify the arrayData. This prepares
for allocating the ArrayData together with the array itself.
Change-Id: I66fe187f8b1e4d382ab243a518dbde5f18a1d16d
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Keep the basic methods in ManagedVTable, but have
the Object related stuff in an ObjectVTable class.
Change-Id: I9b068acf3caef813686227b8d935e7df1a7d1a6e
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
This makes the ArrayData class 'pure virtual'. SimpleArrayData
now contains the implementation of simple arrays. This makes the
separation between simple and sparse arrays a lot cleaner.
It also allows us to move len and offset from the base class into
the SimpleArrayClass. This fixes some bugs where we accessed len
for sparse arrays leading to some buggy behavior.
Added a virtual length() method to ArrayData to query the highes
used index in the Array.
Change-Id: Iab2ba2a48ebe5b7031759eeb4ebe02b4d86233f0
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Split up ArrayData into two classes, one for regular
arrays, one for sparse arrays and cleanly separate
the two cases. Only create array data on demand.
Change-Id: I9ca8d0b53592174f213ba0f20caf93e77dba690a
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
First step of separating the array data from Object.
Change-Id: I5c857397f0ef53cff0807debdb1e405424e1046a
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Encapsulate accesses to the current context, and rework
the way we push and pop this context from the context
stack.
Largely a cleanup, but simplifies the code in the long term
Change-Id: I409e378490d0ab027be6a4c01a4031b2ea35c51d
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
This saves one pointer per object, and willmake other optimizations
easier in the future.
Change-Id: I1324cad31998896b5dc76af3c8a7ee9d86283bfe
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
The changed handlers for aliases are connected lazily in the engine.
QQmlPropertyPrivate::flushSignal is responsible for that and called in other
places, for example when installing a onSomeAliasPropertyChanged handler.
However we were missing a call to flushSignal when doing
onSomeAliasPropertyChanged.connect(...), i.e. using the JavaScript connect API.
Task-number: QTBUG-30493
Change-Id: Ia3f008626fd7af3f2cfbdd30d13fb83158bed4d5
Reviewed-by: Albert Astals Cid <albert.astals@canonical.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
* Resolve lookups in namespaces at compile time and instruct the SSA optimizer
to eliminate reads from the namespace (QQmlTypeWrapper) if possible. For example
access to attached properties of types (i.e. MyNameSpace.ListView.isCurrentItem)
requires neither reading the namespace nor the type.
* Add support for accelerated lookup of attached properties
Change-Id: Ib0b66404ed7e70e1d4a46a1ac8218743a4cc8608
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Otherwise you can end up with some weird situation in which you have
two objects, you print them and then you print the comparison and get
a Foo(0x858480) b Foo(0x858480) a == b false
Task-number: QTBUG-34651
Change-Id: Id2444d919f039be085eb5ed9112345db691540a5
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
At the moment you can pass them as their FinalType* or
as one of their ParentType* but not as QObject* which
does not make much sense to me
Task-number: QTBUG-34617
Task-number: QTBUG-30730
Change-Id: Id5cfb7bbb123456ef43f44f33b450f8966a7641a
Reviewed-by: Alan Alpert <aalpert@blackberry.com>
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
The previous approach of collecting the dependencies through an IR visitor
doesn't work, because it relies on a fixed structure - for example MEMBER(NAME,
prop) - which we can't guarantee (it's usually MEMBER(TEMP, prop)). But it
turns out that we can only pre-calculate dependencies for context, scope or id
properties, so we can do that right away in the QML specific JS codegen, store
that information in the IR function and use it from there in the data structure
generator as well as in the isel as a parameter to getQObjectProperty to tell
the run-time whether capture is required or not.
Change-Id: I33711c3420d6534c653c2a6a4284f0fc12e941cf
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Get rid of the SimpleCallContext, instead simply
use the CallContext data structure, but don't
initialize the unused variables.
Change-Id: I11b311986da180c62c815b516a2c55844156d0ab
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
The engine used to round, but that is inconsistent with ECMAScript's way of
converting doubles to integers by truncation.
With this patch we can also enable the propagation of integer type information
into the IR, but we have to be careful not to utilize it when writing
properties.
Change-Id: I04af4879ba5131349eca2eeff2b27f4598f5267b
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Don't use recursive function calls anymore. Instead, push marked
objects onto the JS stack, and then pop them off when their children
are being marked.
Should reduce stack memory usage, and improves performance by ~5%.
Change-Id: I2d37d97579144fcba87ec8e9fd545dd220c01fbb
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
After the resolution of a property, we can set it by index at run-time instead
of via name resolution.
Change-Id: I479599dabe343cf9e6582dcda12291aebfcce418
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
This avoids having to do a string lookup for ids and in the import cache at
run-time, before we can do a string hash lookup in the property cache. Instead
we resolve final properties in the context and scope object at compile time and
look them up at run-time using their index instead. The dependencies to these
properties are also tracked separately and recorded in the compiled data.
This is merely the initial patch. There's a lot left to do, such as having
specialized getter and setters for specific property types. Setters are missing
altogether right now and will fall back to name lookup.
Change-Id: If3cb4e7c9454ef4850a615f0935b311c9395b165
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Don't write to objects if we have a pending exception to
avoid any side effects.
Change-Id: I9f93a9195a652dbae7033cc6ebb355d5d86e9b5e
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
We don't want to check for exceptions after every single
line on our runtime methods. A better way to handle this
is to add the check in all methods that have direct side
effects (as e.g. writing to a property of the JS stack).
We also need to return whereever we throw an exception.
To simplify the code, ExecutionContext::throwXxx methods now
return a ReturnedValue (always undefined) for convenience.
Change-Id: Ide6c804f819c731a3f14c6c43121d08029c9fb90
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Replace all try/catch statements used when parsing
with checks for engine->hasException.
Change-Id: I4493cb600d5a3eb095c2003bb88bd031403e47c9
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Don't pass a QVariant pointer for the return value when we're
not interested in it and the return type might not even be a
QVariant (that would be only true for QML methods).
Task-number: QTBUG-32801
Change-Id: I8f14e40d8f94caef7e3d086b776735f0484dbf0e
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
QQmlError is public API and shouldn't expose an internal
method.
Change-Id: I7caf06af9340fefec5c96103395fe74acbf19497
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Commit 84627464eb introduced a performance
regression of falling back to reading the QMetaObject, when the lookup for a
property in QML failed, after we've tried in the property cache. This is very
very expensive to do and was only due to QQmlPropertyMap not correctly
invalidating the property cache.
Instead remove the property cache from the property map's QObject and on the
lookup side rely on the property cache being correct in the result (positive or
negative).
Change-Id: I8a013483203f2007d48b71feafa10b3ea02c53fd
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
This makes pretty much all test cases pass with exact
garbage collection.
Change-Id: Ia874e3c17c3984afb7cfe370f9bd3ad8fe46699a
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Frederik Gladhorn
Simon Hausmann
Oleg Shparber
kh1
Liang Jian
Lars Knoll
Michael Brasser
Oswald Buddenhagen
Albert Astals Cid
Erik Verbruggen
J-P Nurmi