This saves us some ping-pong between the IDs and the QMetaTypes, and
avoids possible ambiguities if multiple metatypes are registered for the
same C++ type.
Change-Id: I81cec94a9cd05d69927dc884f65574f0ab2ddc22
Reviewed-by: Maximilian Goldstein <max.goldstein@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
It really only works for value types and it's not intended to do
anythign else. The name should reflect this.
Change-Id: Ib73bf7e9655971f7826fe72145e2d2fab363363c
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We implicitly do the same when calling toVariant().
Change-Id: I288326125d88bc658dcaf12d3ee623e0e529bb69
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Due to the quirk of ECMAScript's Date.parse() spec [0] stipulating the
use of UTC for date-only strings, in contrast to most other ways of
creating a Date using local time, reasonable users get surprised by
the behavior of QDate properties initialized from strings. This can't
be avoided without breaking other uses of Date, so document the
work-around needed to cope with it (use UTC-specific methods to access
the Date object).
[0] https://tc39.es/ecma262/#sec-date.parse
Make conversions back to QDate from Date work round the possibility
that the Date, seen as a QDateTime(,, LocalTime), needs to be handled
as UTC when extracting the date, and catch two more places that
conversion from QDate neglected to use UTC's start of day, for
consistency.
Revised tests to call UTC-specific methods instead of the local-time
ones, where appropriate. Drive-by: some tests were (entirely bogusly)
constructing a fresh Date using the UTC-fields of the Date they had,
in order to then test the non-UTC fields of this fresh Date; instead,
simply test that the UTC fields are as expected.
[ChangeLog][QML][Behavior change] Where a QDate is represented in
QML's JavaScript as a Date, it is now more consistently associated
with the start of the UTC day it describes. Previously cases where it
was represented as the start of local time's day could lead to a Date
turning into a QDate for the preceding day. Inconsistencies in the
specified behavior of Date preclude eliminating such problems
entirely, but they should now be limited to cases where (perversely
for a date property or parameter) the date is specified with a local
time late enough to make it coincide with the start of the next UTC
day (in which case that next day's QDate will be its C++
representation).
Fixes: QTBUG-92466
Change-Id: I2306dd9ecef0d5c2d59b562762392e51bb6d66ca
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Use the newer version of QV4::Function::call() that does not require
manual JSCallData setup and is more optimal for AOT function calls
Change-Id: I5a5e2d0477c0603b05b7213f1b2adcc34d156bf5
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Previously passing a QList of a registered enum would result in an array of objects instead of the array of numbers which usually represent enum values in QML.
You now get an array of numbers as you would expect.
[ChangeLog][QtQml][Important Behavior Changes] QJSEngine::toScriptValue() used to return a QVariant containing an enum, now it returns the enum directly. If you still wish to use valueOf() on the resulting value use QJSEngine::toManagedValue() instead.
[ChangeLog][QtQml][Important Behavior Changes] A QList containing enums will now result in an array of numbers instead of an array of objects.
Fixes: QTBUG-85861
Change-Id: I5c28f4489dfd02d8256aa818e27b1dd6b7d3113d
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Some applications that use JavaScript as a scripting language may want
to extend JS through C++ code. The current way to do that is with
global objects.
ES6 provides a better way of encapsulating code: modules.
registerModule() allows an application to provide a QJSValue as a named module.
Developers familiar with Node.js will find this very easy to use.
Example:
```c++
QJSValue num(666);
myEngine.registerModule("themarkofthebeast", num);
```
```js
import badnews from "themarkofthebeast";
```
[ChangeLog][QtQml][QJSEngine] Adds the ability to register QJSValues in
C++ as modules for importing in MJS files.
Change-Id: I0c98dcb746aa2aa15aa2ab3082129d106413a23b
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
URL has become a builtin type. We should support it on the same level as
QString/String and QDateTime/Date.
In order to continue support for comparing URL properties with the
JavaScript equality operators, we still pass URLs as variants when
using them in JavaScript. However, we now create proper URL objects for
QJSValue and QJSManagedValue, and we allow transforming the URL-carrying
variant objects back into QUrls.
Change-Id: I78cb2d7d51ac720877217d2d4b4d0ab17cdd2a4b
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
This way, we can avoid the costly id to metatype lookup in case where we
actually need the full metatype.
Task-number: QTBUG-88766
Change-Id: Ibe29b323007f00d2f8d1807fb9b64f9a8f87e807
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
This is pretty much the same as creating them from QQmlListReference.
Change-Id: I8d873840fc08887655d19a61b028f3eb60eaf938
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
In the good case this is just reading a few members of the relevant
classes. No need to call functions for this.
Change-Id: I9908cd6437cf9a1ca840f9aa0e524d3976272d67
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We should avoid looking up metatypes by ID. That's expensive.
Change-Id: I00ce0a7f95ec82b0db6e7eb976e39e50522a7fe4
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
We need to be able to retrieve QQmlListReference and attached objects
from QJSValues.
Change-Id: I39679317da4066b054e86f767fc5f723ead2b2e7
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Parsing the type from a signature is expensive and we can do better in
most cases.
Change-Id: Iae85f4dec9ad6b8de60efeb3469a253fd0862672
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
When called via the metaobject system, parameters and return values are
passed as void*, with accompanying type information in the form of
QMetaType. The same format is expected when calling an AOT
compiled function.
Previously, we would first convert all the parameters to QV4::Value,
just to convert them back the moment we notice that there is an AOT
compiled function. This is wasteful.
This change provides a second call infrastructure that accepts void* and
QMetaType as parameter and return value format, and passes them as-is
all the way to any AOT compiled functions. If there is no AOT compiled
function, the conversion is done when detecting this, rather than when
initiating the call. This also passes the information "ignore return
value" all the way down to the actual function call. If the caller is
not interested in the return value, we don't have to marshal it back at
all.
For now, we only add the extra "callWithMetaTypes" vtable entry to
ArrowFunction. However, other callables could also receive variants
optimized for calling with void*/int rather than V4 values.
This required changing the way how function arguments are stored in the
property cache. We squeeze the return type into
QQmlPropertyCacheMethodArguments now, and we use QMetaType instead of
integers. In turn, we remove some unused bits.
Change-Id: I946e603e623d9d985c54d3a15f6f4b7c7b7d8c60
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We either have pre-populated arguments and thisObject, then we can just
use them and keep them const. Or, we want to allocate and populate the
arguments and the thisObject. Then, do allocate them in a separate
object, and transform that into JSCallData afterwards if necessary.
Furthermore, avoid alloc(0) as that just returns the current stack top.
Writing to it will clobber other data. Rather, just use nullptr and
crash if it's written to.
Also, remove the useless operator-> from JSCallData. That one just
confuses the reader.
Change-Id: I8310911fcfe005b05a07b78fcb3791d991a0c2ce
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We only need it when generating CallData, or when filling in any
thisObject or arguments that weren't provided. Provide a constructor
that expects thisObject and arguments to be pre-allocated and one that
allocates them in a scope passed as argument.
Change-Id: Iddfba63f4dbc5b09e2b33fb22a94eea88f515902
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
We can cache the QQmlContextWrapper rather than retrieving it twice.
Inline some things, and do not unnecessarily create and destroy ref
pointers.
Change-Id: Ife0980f83b7efe1ea9dc56aacbfbccd029ce77c8
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We almost never need to construct a QVariant to do this. Constructing a
QVariant is excessively expensive if you have something simple like an
integer. This also fixes the unexpected "unwrapping" of variants when we
pass them through QJSValue.
[ChangeLog][QtQml][Important Behavior Changes] If you create a QJSValue
from a nested QVariant (that is, a QVariant containing another
QVariant), then, when retrieving its contents again, the outer variant
is not unwrapped anymore. Rather, you get exactly the value you've
passed in.
Change-Id: I8c16eed4f13e8cfdeced0756eef593b3b8e84dd1
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
We already covered the case where the metatype's flag indicate that it
is a QObject. That is exactly the same check used in
QQmlMetaType::toQObject, so if the test did not succeed before, it won't
succeed now either.
As a drive-by, avoid useless metatype-id to metatype lookup.
Change-Id: Ie36a07587aa2b899d2a932bcb3f4a0b5da8aa282
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Instead of arguments around as a pointer to
[argc, metaTypeId1, metaTypeId12, ...]
pass argc and a pointer to [QMetaType1, QMetaType2] around.
Moreover, make use of the fact that we now carry the metatype instead of
only the id in a few places, to avoid id -> metatype lookups.
Task-number: QTBUG-82931
Change-Id: Ib00e4d793727f85f3358a8162d1aac972daab3d3
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
[ChangeLog][QML][Important Behavior Changes] If a QObject pointer is
passed to the QML engine and subsequently frozen with Object.freeze,
modifying its QObject properties now fails and throws a TypeError. The
TypeError is thrown even in non-strict mode.
[ChangeLog][QML] It is now possible to pass const QObject derived
pointers to QML in properties, and to call Q_INVOKABLE functions which
take such pointers as arguments. If the QML engine receives such a
pointer, it is treated as if the object was frozen.
Fixes: QTBUG-82354
Change-Id: Ib0dbcdfb2370654505936c3cf391d87dd2c9677b
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
It appears that nowadays v4->metaTypeToJS handles QVariant and QObject
derived classes just fine, and in exactly the same way as the custom
code in populateJSCallArguments did.
Task-number: QTBUG-82931
Change-Id: Ic5f97dfc3296a409fdd6a1fcb78d3b9bdba5f3a1
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
You can also cast enums to integers, after all.
Pick-to: 6.1
Change-Id: I283d3dd280eeb44ba22bb45ca9be69e5358d5781
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Add execution function that can evaluate runtime functions available
in the compilation unit. Private API for now as it's unclear what would
be a comprehensive solution to support all existing use cases
Task-number: QTBUG-84368
Task-number: QTBUG-91039
Change-Id: Icf755b53484587d7983eaae4821c1aa0111d5c05
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Move the value type registry into QQmlMetaTypeData. This way we can
conveniently drop the relevant entries when unregistering a type.
Fixes: QTBUG-86946
Change-Id: Id024a34a8b2b622fd9417fc0e52864b43c66cc01
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Previously it would only disable loading of separate cache files.
Change-Id: Iae92fc03d2e5566ef7dc44a6730b788b7512fd3d
Reviewed-by: Maximilian Goldstein <max.goldstein@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
This time, the ValueTypeFactory gets converted. As a consequence, many
callers get touched again.
Task-number: QTBUG-88766
Change-Id: I3a8b7d5cfeb7fac85daf1702febba205971d4256
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Also, allow conversion from UrlObject and String. We allow the string
conversion because we treat string and url as interchangeable in various
places.
Change-Id: Ib229c6d190e1c5d849ea18798925965b8dbeef7e
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
This is a single char16_t, not an array of them.
Pick-to: 5.15
Change-Id: I55d23ebb5f2abebd43cd4160a75d373706392ddf
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
You get to write QML_SEQUENTIAL_CONTAINER(value_type) now, and
qmltyperegistrar will generate a sensible registration call from that.
A registration might look like this:
struct MyStringListForeign
{
Q_GADGET
QML_ANONYMOUS
QML_SEQUENTIAL_CONTAINER(QString)
QML_FOREIGN(MyStringList)
QML_ADDED_IN_VERSION(3, 1)
};
It's unfortunate that we need to use a metaobject to transfer all of
this information, but there is no other sensible way.
Transform the containers defined in qv4sequenceobject.cpp to use the new
style, and move them out of the builtins, into QtQml. Recognize that
only one of them was ever tested, and add tests for the rest.
Task-number: QTBUG-82443
Change-Id: I3a30f9e27266bb575eea26c5daf5dad1ec461cc5
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Quite obviously, the Qt object is a singleton, extended with a
namespace, backed by a member of the JavaScript global object.
Defining all the methods as JavaScript functions is unnecessary and
duplicates the general type transformation code. Also, it makes it
hard to use those same methods from a C++ context as we cannot
properly set up the arguments outside the JS engine.
Rewriting the Qt object reveals some deficiencies in the old
implementation that we need to fix now:
1. The enums of the Qt type were listed as properties of the Qt object,
which means you could iterate them with a for..in loop in in JavaScript.
This is just wrong. Enums are not properties. This functionality
is deleted and the test adapted to check for each enum value separately.
The commit message for the change that introduced the iterability
already mentioned that the author had failed to find any occurrence of
this in the real world.
2. Parsing time objects from strings was done by parsing the string as a
date/time and then picking the time from that. We still support that for
now, but output a (categorized) warning. Parsing the time directly is
preferred where possible.
3. Previously you could create (invalid) dates and times from various
kinds of QML types, like int and color. This does not work anymore as we
now validate the types before calling the functions.
4. Passing more arguments to a function than the function accepted was
unconditionally ignored before. Now, a Q_CLASSINFO on the surrounding
class can specify that the arguments should be checked, in which case a
JavaScript error is thrown if too many arguments are passed. In order
for this to work correctly we also have to ignore JS undefined values as
trailing arguments for overload resolution. This way, if a method
matching the defined arguments exists, it will be preferred over a
method that matches the full argument count, but possibly cannot accept
undefined as parameter.
Consequently a number of error messages change, which is reflected in
the qqmlqt test.
[ChangeLog][QtQMl][Important Behavior Changes] You can not iterate the
enumerations of the Qt object in JavaScript anymore. This does not work
with any other enumeration type either. You can of course still access
them by name, for example as Qt.LeftButton or similar.
[ChangeLog][QtQMl][Important Behavior Changes] The time formatting
functions of the Qt object in QML now allow you to pass an actual time
string, rather than a date/time string as argument. Passing a date/time
string results in a warning now.
[ChangeLog][QtQml][Important Behavior Changes] Functions in the Qt
object for formatting date and time will now throw a JavaScript error
when presented with a value of an incompatible type, such as int or
color.
[ChangeLog][QtQml][Important Behavior Changes] The Qt.resolvedUrl()
function now returns a URL rather than a string. This follows the
documentation.
[ChangeLog][QtQml][Important Behavior Changes] The GlobalColor enum of
the Qt namespace is not exposed to QML anymore. It did not make any
sense before as the enum values could not be used as colors.
Change-Id: I7fc2f24377eb2fde8f63a1ffac5548d652de7b12
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
This avoids the template explosion and makes the mechanism extendable.
You can now register additional anonymous sequential containers.
Fixes: QTBUG-71574
Task-number: QTBUG-82443
Change-Id: I5b9ed9af1533a3b7df8fc5bb37bbb73b8304e592
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
metaSequence() became metaContainer() and we should ask for canConvert()
as there are two ways to convert to a container.
Change-Id: Iba868491ff9d2cc8fc89de1cab29818b834b53f4
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Due to qiterable.h specializing a template declared in qmetatype.h we
temporarily need to include it in a few tests so that the iterables
work.
Change-Id: Ia32392419dead76eaf2b91b2ec4157b726d8de74
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
When converting a QDate to a QDateTime, startOfDay() takes care of
avoiding any gaps in time at the start of the day, where naively
asking for QTime(0, 0, 0) can produce an invalid date-time.
Change-Id: I24f3d230eb1ee7396600b030ad1305e060215cbd
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
They're value types, packaging qint64 and int respectively.
Change-Id: I78a0097f77238751ac3ef9f928537f719a6d05d6
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
The internal QVariant constructor taking a QMetaTypeId has been removed.
Thus, construct QMetaTypes where necessary from the id, or avoid a
QMetaType -> ID -> QMetaType roundtrip where we already have a metatype.
Also fix a few missing includse that were previously transitively
included.
Change-Id: I56ce92281d616108a4ff80fe5052b919d1282357
Reviewed-by: Fawzi Mohamed <fawzi.mohamed@qt.io>
When converting JS arrays to sequence<T> type, check first for the
existence of a QJSValue -> T converter function. This restores the
behavior from Qt <= 5.14. Amends ecdb4ed275
Fixes: QTBUG-84104
Change-Id: I14c86ab37e34a3c8cff072574d4b90fe9e558535
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Now that char16_t is used in Qt (for instance in QChar::unicode()), we
need to support it.
Change-Id: I527a70795524bfd883fc4d729aac714708b51181
Reviewed-by: Liang Qi <liang.qi@qt.io>
Implements URLSearchParams (https://url.spec.whatwg.org/#urlsearchparams),
completing our implementation of the URL object.
Still needs the for..of iterator to get implemented.
Change-Id: Iad33ed2f3fe0b2598ca2b0b21a4743f5f7dc19fd
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Use the unused field in the CachedUnit structure provided by qmlcachegen
to allow for providing function pointers for functions and bindings that
are compiled ahead of time.
Provided is the pointer into an array that is terminated with a {index:
0, functionPtr: nullptr} entry. The array index field in each array
entry allows for gaps.
Change-Id: I7457f5eea5f14e5f94431b9cc6da042cb03517a0
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Pass the address of the entire structure through to the compiler, so
that when adding new members we can easily access them.
Change-Id: I5da75ba4e64d3e0e750a3ff3df4edbb88cdb6937
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Implements the JavaScript URL object (https://url.spec.whatwg.org/#api).
Except that it does not currently implement the searchParams field.
Task-number: QTBUG-54988
Change-Id: I19abc69e075cbf84bd15e6791be195ce16f3fe73
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Remove all code that supported converting between JS RegExp's and
QRegExp, as QRegExp is going away in Qt6.
Change-Id: I4863e68dd87a337d7e836d1b26c28ee3bb914e9f
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Also, fix the check to actually test the correct capabilities by using
the containerCapabilities function; testing _iteratorCapabilities only
worked by chance so far.
Task-number: QTBUG-82743
Change-Id: I64f20c6bf1e47737c7b927f79e1e78c1a1603741
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
This class is not a private detail of QQmlContext. And it is incredibly
hard to see who owns what in there. Let's add some civilization ...
We enforce refcounting for QQmlContextData across the code base, with
two exceptions:
1. QQmlContextPrivate may or may not own its QQmlContextData.
2. We may request a QQmlContextData owned by its parent QQmlContextData.
For these two cases we keep flags in QQmlContextData and when the
respective field (m_parent or m_publicContext) is reset, we release()
once.
Furthermore, QQmlContextData and QQmlGuardedContextData are moved to
their own files, in order to de-spaghettify qqmlcontext_p.h and
qqmlcontext.cpp.
When the QQmlEngine is deleted, any QQmlComponents drop their object
creators now, in order to release any context data held by those.
Before, the context data would be deleted, but the object creators would
retain the dangling pointer.
[ChangeLog][QML][Important Behavior Changes] QQmlContext::baseUrl() does
what the documentation says now: It prefers explicitly set baseUrls over
compilation unit URLs. Only if no baseUrl is set, the CU's URL is
returned. It used to prefer the CU's URL.
Change-Id: Ieeb5dcb07b45d891526191321386d5443b8f5738
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Being careful, we can now save primitive values inline. We use the heap
pointer of QV4::Value as either QString* or QV4::Value* for complex
types. We cannot store persistent managed QV4::Value without the double
indirection as those need to be allocated in a special place.
The generic QVariant case is not supported anymore. The only place where
it was actually needed were the stream operators for QJSValue. Those
were fundamentally broken:
* A managed QJSValue saved and loaded from a stream was converted to a
QVariant-type QJSValue
* QVariant-type QJSValues were not callable, could not be objects or
arrays, or any of the special types.
* Cyclic references were forcibly broken when saving to a data stream.
In general the support for saving and loading of managed types to/from
a data stream was so abysmally bad that we don't lose much by dropping
it.
[ChangeLog][QML][Important Behavior Changes] When saving a QJSValue to a
QDataStream only primitive values or strings will be retained. Support
for objects and arrays was incomplete and unreliable already before. It
cannot work correctly as we don't necessarily have a JavaScript heap
when loading a QJSValue from a stream. Therefore, we don't have a proper
place to keep any managed values. Using QVariant to keep them instead is
a bad idea because QVariant cannot represent everything a QJSValue can
contain.
Fixes: QTBUG-75174
Change-Id: I75697670639bca8d4b1668763d7020c4cf871bda
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
This is needed in a few places outside of declarative, so this change
restores the loc member in DiagnosticMessage and moves
QQmlJS::AST::SourceLocation into common's QQmlJS namespace/directory.
QQmlError is unaffected and retains only line/column.
Amends d4d197d062
Change-Id: Ifb9d344228e3c6e9e26fc4fe112686f9336ea2b2
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Instead of applying a heuristic on when to call drain() in unrelated
code, we check the stack limit on each push(). If the soft limit is
reached we try to drain. As drain() itself can push again, we try to
limit the stack size by allowing at most 65 recursions of drain(). If
none of that helps, we crash with a meaningful error message.
This allows us to remove all the hacky drain() calls in other parts of
the code.
Change-Id: Ib979339470da0e85981de8131e7997755b757c71
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The previous attempt to fix this was lost in a merge resolution.
Change-Id: I0638c434543d231352c44687b06bf429b7be7a04
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
If the provided typeHint is -1, it does not make sense to construct a
QVariant of this type and to check whether it is appendable.
Fixes: QTBUG-81945
Change-Id: I32cbb9e70e210a7eca8d55801c1783338d1173b7
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
We started to convert containers to QJSValues, so that we could use them
as JavaScript arrays. Unfortunately, this would then lead to a type missmatch
when those same values where to be stored in a property of the container
type. This commit fixes this by converting them back to the original
type.
Fixes: QTBUG-80916
Change-Id: I30a3b03e17c34b171d4a6881dfd7801c13e94d80
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
[ChangeLog][QtQml] Added Qt.uiLanguage and QJSEngine::uiLanguage properties
These properties mirror the same value in QML and C++ and can be used
freely. They also provide API symmetry to Qt for MCUs.
QQmlApplicationEngine binds to this property and applies translations
accordingly by constructing a QLocale with the value and using
QTranslator::load(locale).
Change-Id: Id87d6ee64679b07ff3cb47844594e8eeebd8c8b6
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Christian Kamm <mail@ckamm.de>
as type is going to be deprecated.
This change was done automatically with the help of clazy.
In addition, ColumnRoleMetadata was changed to take an int instead
of a QVariant::Type
Change-Id: Ibc02d7b52e7d931a56c19fdebc4788b5e6df2a39
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
MemoryManager::collectFromJSStack did push to the mark stack without
checking if there is actually still space available. To fix this, we now
drain the stack once we hit the limit.
The test case is a slightly modified version compared to the reported
one, removing one loop. This was required as our regular expression does
not throw an exception when there are too many capture groups. However,
to trigger the bug, looping was not actually necessary.
Change-Id: I4d00865f25a989c380f4f5b221f4068c80b71d2b
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
In XMLHttpRequest, we need to get the QNetworkAccessManager from the
engine. However, if the request originates from a WorkerScript, there
exists no qmlEngine. We therefore add a new indirection to access the
QNAM, and set it up accordinly in registerWorkerScript.
Fixes: QTBUG-81055
Change-Id: I8915202b6d6b7139c8386304b3d1d7a22a82045e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
8704c64094 introduced new conversions
via sequentialIterableToJS. Due to that, QVariant properties which
formerly stored e.g. std::vector<QObject*> now would store a QJSValue.
Those would still claim to support a conversion to QVariantList, but
-contrary to what our documentation says-, we were not able to do a
conversion to QSequentialIterable. The default constructed
QSequentialIterable would then crash when calling begin(), as that
function pointer was null.
This patch fixes this by adding the necessary support to convert a
QJSValue containing an array.
Non-array QJSValues will still return an "empty" QSequentialIterable.
Note that this changes what happens when a QJSValue is converted to a
QVariantList, as QVariantValueHelperInterface<QVariantList> will check
first if there is a converter to QSequentialIterableImpl before
attempting to call any directly installed converter to QVariantList. In
order to not change the existing behavior, the QSequentialIterable
returns the QVariant corresponding to the QJSValue at a given array
position, intead of a QVariant containing the QJSValue.
Fixes: QTBUG-80609
Change-Id: I8101229c0d2043b3f2d618ed035b279844802dd8
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
With QV4_CRASH_ON_STACKOVERFLOW set you can use up all the stack
provided by the operating system to parse and execute JavaScript. Once
the stack space is exhausted the program crashes like it would in case
of a C++ stack overflow.
We cannot reliably determine either the maximum stack size or the amount
of stack space currently in use at runtime. Therefore, the guards we
usually put in place are necessarily conservative.
[ChangeLog][QtQml] There is now an option to disable the (necessarily)
conservative stack size checks when parsing and executing JavaScript. If
the environment variable QV4_CRASH_ON_STACKOVERFLOW is set, JavaScript
stack overflows crash the program the same way C++ stack overflows do.
On the flip side, more stack space is made available that way.
Task-number: QTBUG-74087
Change-Id: I5e9d9ec6c0c9c6258c31d9e2d04a5c1819fbf400
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Added the missing lookup for cached .mjs files in
ExecutionEngine::compileModule. This allows using .mjs files in
WorkerScript {} elements in conjunction with the Qt Quick Compiler and
also fixes the use when using QJSEngine::importModule.
[ChangeLog][QtQml] Fix loading of EcmaScript modules when using the Qt
Quick Compiler.
Fixes: QTBUG-77761
Change-Id: I58130b0468f4920b2f6c49b98a2f51d5ae3a0491
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
QMLEngine by default allocates 4 MB for javascript stack and garbage
collection stack takes 2 MB. It is a lot of memory for platforms without
virtual memory.
Change-Id: I1575dd9584898dca33df66704f716c7b5a7c01c1
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
We only need two classes to describe all possible diagnostics:
* A low-level private POD DiagnosticMessage. This is easily copied and
passed around internally. It doesn't need to adhere to a stable API
and it doesn't carry any extra baggage.
* The high-level public QQmlError with its stable interface. This can
internally also use a DiagnosticMessage as storage.
Change-Id: I52be88d9b5d9855a661b8032b01eedb43a0fb0b3
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
I've never seen it used and I've never seen the companion library
required to operate it.
Change-Id: I5a0e6aed9a416f1bd26dea97def9667a11a4d77d
Reviewed-by: Robin Burchell <robin.burchell@crimson.no>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
They all had some interesting bugs and duplicated each other:
a, propertiesFrozen() changed each property individually, creating a lot
of unnecessary intermediate classes. frozen() changed them all at once.
b, If a class happened to contain only properties that matched the
characteristics of being "sealed" or "frozen", sealed(), frozen() and
propertiesFrozen() would set the flags in place and return the same
class. This is bad because it violates the assumption that an
InternalClass is immutable and it breaks the recursive freezing
algorithm we rely on for the global object. It would stop freezing child
objects at any such class, even if the children were not frozen.
c, propertiesFrozen() did not set any of the flags even though it
effectively sealed and froze the class. Therefore, when requesting the
same class as frozen() it would iterate through all the properties
again.
d, frozen() implicitly also sealed the object and made it
non-extensible. sealed() also implicitly made it non-extensible. This is
impractical as we want to allow objects to be extensible even though all
their properties are frozen. Therefore we only set the flag that belongs
to each method now. We do know, however, that a frozen object is
implicitly sealed. Therefore we can short-circuit this transition.
Furthermore, we need to remove the assert in InternalClass::init() as
you can indeed use frozen objects as prototypes for others, but that
needs to be recorded in the original InternalClass via the isUsedAsProto
flag. In order to set this flag, we need to perform a transition and
therefore, derive from the old InternalClass.
The JavaScript isFrozen() method asks for an _implicitly_, "duck typed",
frozen state, which is different from what our "isFrozen" flag denotes.
Therefore we add a separate const method that just checks whether all
properties are frozen.
Task-number: QTBUG-76033
Change-Id: I375fef83fb99035d470490fdf2348766b090831e
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
We need a CompilationUnit that only holds the data needed for
compilation and another one that is executable by the runtime.
Change-Id: I704d859ba028576a18460f5e3a59f210f64535d3
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
This is a better fit for the method. In turn, remove all the
V4_BOOTSTRAP conditions from qv4engine_p.h and make sure we don't
include or compile it in bootstrap mode.
Change-Id: I5933b0724e561313ca20c420b83e4d70e63bddf5
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
For clang we need to use has_feature to detect the presence of an ASAN
build. Amends commit eb363c3a0b
Task-number: QTBUG-75410
Change-Id: I2adb69deb07f8c6b77be8c2f23751fd0a2bbdb95
Reviewed-by: Erik Verbruggen <erik.verbruggen@me.com>
The tracing JIT won't be finished. Therefore, remove the parts that have
already been integrated.
Change-Id: If72036be904bd7fc17ba9bcba0a317f8ed6cb30d
Reviewed-by: Erik Verbruggen <erik.verbruggen@me.com>
ASAN enabled builds require more stack space and therefore our call
depth limits should be lower.
In my measurements with a recursion through arrow functions with the
interpreter, as per the test case in the bug report, different types of
builds require different amounts of stack space. On x86-64 Linux, I
measured, by printing $rsp and subtracting:
Debug: ~6k
Debug with -Og: ~590 bytes
Release with -O2: ~570 bytes
Release (-O2) with ASAN: ~40k
Fixes: QTBUG-75410
Change-Id: I403b261c677b1adb9f349958339b5a1294ae4d5d
Reviewed-by: Erik Verbruggen <erik.verbruggen@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
This way you can enable or disable the JIT when configuring Qt. The
conditions for the availability of the JIT have also been cleaned up.
There is no reason anymore to artificially restrict availability on x86
and x86_64. The reason for the existence of those clauses are old
problems on windows that have been fixed by now. However, on arm and
arm64, we need a specialization of the cacheFlush() function for each OS
to be supported. Therefore, restrict to the systems for which such a
specialization exists. iOS and tvOS are technically supported and you
can enable the JIT via the feature flag now. Due to Apple's policy we
disable it by default, though.
Change-Id: I5fe2a2bf6799b2d11b7ae7c7a85962bcbf44f919
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Technically UINT_MAX is actually a valid array index, although that is
an academic problem right now. However, we do have a method
isArrayIndex() and should just use that to determine if a PropertyKey is
an array index.
Fixes: QTBUG-73893
Change-Id: I302e7894331ed2ab4717f7d8d6cc7d8974dabb4e
Reviewed-by: Erik Verbruggen <erik.verbruggen@qt.io>
Those are "scarce" resources which need to be kept as QVariant.
Fixes: QTBUG-74751
Change-Id: I28381e2a754ed4bbf4e409dc275f6288b64416cc
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
This makes the use of the feature easier and lowers the dependency
on setup of internal structures.
Also, evaluation of expressions is notoriously expensive and unreliable
on the gdb side, so moving the complexity to the compiled side is
an advantage.
Task-number: QTCREATORBUG-22209
Change-Id: Id43d5c2bf4d852d496ceb59189209d167213afcb
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
Ulf Hermann
Edward Welbourne
Andrei Golubev
Maximilian Goldstein
Alex Shaw
Fabian Kosmale
Lars Knoll
Simon Hausmann
Qt Forward Merge Bot
Fawzi Mohamed
Olivier Goffart
Liang Qi
Tor Arne Vestbø
Mikhail Svetkin
Friedemann Kleint
hjk