This way, we can correctly write to multiple levels of QDebug with QML context
information. A followup change will port all existing callers, and
subsequently change qmlInfo's message level to QtInfoMsg.
[ChangeLog][QtQml] Introduced qmlDebug & qmlWarning functions to qqmlinfo.h, in
addition to the pre-existing qmlInfo function. As a side effect, QQmlError has
also gained messageType() and setMessageType().
Change-Id: I04ced5952c5c3c58293a89a6767c7b545c03cc0a
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
In preparation for the binop generator to work with different
assemblers, we can bring some of the externally prefixed Assembler::
types into the Binop class namespace. That makes the code easier to read
(less visual noise) and centralizes the dependency to the external type
to one place (the using statements).
Change-Id: Ib7700689196f8e54bba1f3ab6de3bf85ed93ff87
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
The QV4::Assembler does not need a reference to the ISel that uses the
assembler, a reference to the JS unit generator for string registration
is sufficient.
Change-Id: I9e929e4d0aecbf0144296f85cb8d208296f28003
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Move code from the Assembler helper-class into the .cpp file that has
the rest of the Assembler class implementation.
Change-Id: I4e9fb264216a078a3ee11119d7610fd108ddf8c7
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reuse the same const QString in both places, in each of the functions
that repeated this pattern.
Change-Id: Ic20281692d84b3ad5257af9837957e0893452b1a
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Nine tests all used setDate and setTime on a default QDateTime, where
just initializing it right, in one line, is simple, terse and easy to
understand. This also allows it to be const.
Change-Id: I77fd8b36b705cafc49f9020ae9280dfdcfece0d8
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
While it departs from common date numbering, as used by QDateTime, the
0=Jan numbering used by JS's Date was copied from Java's Date, which
(probably) copied it from ANSI C's struct tm; and C likes to count
from zero. So don't call it weird; and let's comment on it where it's
actually relevant (in JS code with the offset, rather than against
uses of QDate, which uses 1=Jan numbering) and only once.
Change-Id: I926138e88583339bec8641e3b28a642794dc217d
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
tst_qqmllocale::timeZoneUpdated() was recording the prior time zone,
frobbing it for the duration of its test, then restoring it. However,
if TZ was previously unset, it got set to empty instead of being
unset. This did not quite have the same effect; UTC ends up being
used instead of the system default, with potential for conflict
between system functions (that have attended to tzset) and third-party
libraries (e.g. ICU) with internal state saved from before the change.
Change-Id: I9f2147cd8858316e9ba5fd84d95dfc2c4538b0ab
Reviewed-by: Matthew Vogt <matthew.vogt@qinetic.com.au>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Use booleans as booleans, it's what they're for.
Change-Id: I68c0ba5efcfd3556eeef043b179769e91c77cf03
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Instead of relying on the somewhat "cryptic" ILP32 macro on x86/x86-64,
we can also use the macro that the JIT generated code _really_ relies
on, which is the size of pointers. For x32 those will be 4 despite
Q_PROCESSOR_X86_64, and the JIT will end up disabled. For regular x86
and x86-64 it should get enabled.
The same logic is applied to ARM32/ARM64 to disable the JIT once
machines start showing up with ILP32 data model on ARM64.
Change-Id: Id8dfb203c9b75e8ed5a61dcae28482f8f99be75f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Dmitry Shachnev <mitya57@gmail.com>
So far uniqueId and rotation are mainly applicable to TUIO.
Deprecate the area property in favor of ellipseDiameters.
Also improve the mpta-crosshairs manual test to show this information.
Change-Id: I16ea6618ae21ce66dac45638d6e2bb3c0a3b1818
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
acos cannot return any possible angle, but atan2 can.
Change-Id: Ic277bc1d3616900775b6076dce6b05bdf3c35da3
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
After change dc6b73390b262b9554599cbf40539763b1280261 in qtbase,
the qml runtime was failing at an assertion in
QQmlTypeLoader::getType() which tries to avoid relative URLs
and relative directory paths. It's important to convert
relative paths to fully-qualified paths when converting to
an URL, because QQmlTypeLoader uses it as a cache key, and
we want to avoid ambiguity about which file is being cached.
Task-number: QTBUG-57870
Change-Id: Ib984cf722009f5f04cb67fffbc52d12bcc98df89
Reviewed-by: David Faure <david.faure@kdab.com>
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Window.contentItem was not associated with any QQmlContext and
QQmlEngine. A valid pointer to QQmlEngine is needed for callback
function execution.
Task-number: QTBUG-57175
Change-Id: Iab7730bfc8860521ff2e8c1631a11d0e1fe0cf94
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
The initial goal is to verify the behavior of text, preeditText
and displayText properties and their associated signals
on various platforms.
Change-Id: Ie50c78a83930e05158a197f38087db9ae3cba534
Reviewed-by: J-P Nurmi <jpnurmi@qt.io>
[ChangeLog][QtQuick][TextInput] Added textEdited() signal to
distinguish user edits from programmatical text changes.
Change-Id: I1d78499e3e11f9f1cab80ce3b0a6d9f2713219f4
Task-number: QTBUG-57203
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
[ChangeLog][QtQuick][Flickable] Added horizontalOvershoot and
verticalOvershoot properties that can be used for implementing
boundary actions and effects.
Task-number: QTBUG-38515
Change-Id: I06379348a67d03507b56788d6fc7020bbb2d375f
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
It could be argued that all callers should be fixed, but this warning has
existed since 2012 and even our own tests do not all set sizes, so I think that
battle is a lost fight.
This helps make real problems with the tests more visible.
Change-Id: Ie29521695ef11943154dae40726d194dc72316c2
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
We can maintain the spoken about compatibility without the performance penalty
by inlining what qguivariant.cpp does.
Change-Id: I95a8b8d660fb9104ae2d30d0ece943bccf169679
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
The isOpenGLES() condition is plain wrong: the code always uses ushort as
the index type.
Change-Id: Ibc5989227f61201913ff124485df4890b3490db5
Task-number: QTBUG-57515
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
This matches the behavior seen from Component's createObject()
function.
Change-Id: I83fe73a588d04c5efd30c49059bb19e7584bef48
Reviewed-by: J-P Nurmi <jpnurmi@qt.io>
This seems to date back to much earlier times, and does not seem to be of much
use anymore.
Change-Id: Ia0047c280102da9e85de0d10133df3135433dc45
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Objects created with createTemporaryObject() and
createTemporaryQmlObject() can be destroyed manually before the test
ends. Before this patch, there would be a type error when trying to
check if the object still exists in qtest_destroyTemporaryObjects().
The error seems to stem from the fact that objects that are stored in
an array and later destroy()'d are not nullified, but rather left in
a corrupt state. See the linked bug for more information about this.
We work around this issue by checking for the existence of the
destroy() function on the object before attempting to call it.
Task-number: QTBUG-57749
Change-Id: I0f6ddd47d86af6fb87392c2992f9f6143af6aab8
Reviewed-by: J-P Nurmi <jpnurmi@qt.io>
Was still displaying a screenshot of the previous example
\image --> \borderedimage
Change-Id: Ifb23d0ffe1216c7d79170b6bbbd7da1511afe79c
Reviewed-by: Topi Reiniö <topi.reinio@qt.io>
Change 3c5e438890 enabling
using the target path version of a parent module made it harder
to determine the deployment target path from the module name.
Output the relative path in addition to remove the need to
do that in deployment tools.
Task-number: QTBUG-57494
Task-number: QTBUG-52556
Change-Id: Ib76f40f19b530f9351205df9109f2fef6191979d
Reviewed-by: J-P Nurmi <jpnurmi@qt.io>
QQuickWidget::grabFrameBuffer() was not polishing its items nor
syncing the scene graph compared to standard QQuickWindow::grabWindow().
This lead to QQuickWidget grabbed content to be outdated (render
the previous frame as a new frame).
Task-number: QTBUG-57596
Change-Id: I94f5e0aa5b096fb9e21259267c0e50473e8ed5bd
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Before this patch, this code:
import QtQuick 2.7
ListView {
keyNavigationEnabled: true
}
Would cause this error:
"ListView.keyNavigationEnabled" is not available in QtQuick 2.7.
This is because ItemView was missing a revision:
qmlRegisterUncreatableType<QQuickItemView, 7>(
uri, 2, 7, "ItemView", QQuickItemView::tr("ItemView is an abstract base class"));
Task-number: QTBUG-57621
Change-Id: Ia00cb7446a0d83278760f3aa361db5a864661bc6
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
We want the IDs to reflect the location and the type of event that is
happening. Thus all instances of a particular component should share
the same IDs for their bindings. Otherwise the aggregated views of the
QML profiler list the same locations again and again for each object.
Change-Id: I3f8a7531589b42f785e40c7c5b04b229f639d696
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
If the trace data is the only thing holding a reference to a particular
compilation unit and we are copying the last FunctionCall from that
compilationUnit onto the second last, we have to first addref() and
then release() as otherwise it might disappear in between.
Also we don't need to check m_function for null, as the ctor guarantees
it is never null.
Change-Id: I6eaf64abe71f9785d4e76e18dabd9a0676564da1
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
We want to cleanly crash if they fail, rather than return a null
pointer from vertexData().
Change-Id: Ie88adaa9f0977c70ef4a47a59463509dcd15bdcf
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
This allows an item to accept the shortcutOverride event, preventing
e.g. Shortcut from stealing key events.
The original use case that prompted the creation of this patch was
using a Popup from Controls 2 to create a keyboard shortcut editor.
When the user wanted to cancel the shortcut that they were editing,
they could press escape, but Popup would grab the shortcut and close
itself. As the test case demonstrates, the same problem occurs with
the Shortcut type in Qt Quick.
[ChangeLog][QtQuick][Keys] Added shortcutOverride signal to Keys
attached object to allow prevention of e.g. Shortcut from stealing
key events.
Task-number: QTBUG-57098
Change-Id: I594e4ea17ec417d8c7d93c6cf347c1a1a2e62b93
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Robin Burchell <robin.burchell@viroteck.net>
Introduce pressAndHoldInterval to allow setting the pressAndHold delay
per-MouseArea.
[ChangeLog][QtQml][MouseArea] Introduce pressAndHoldInterval property,
which controls the elapsed time before pressAndHold is emitted.
Task-Id: QTBUG-47662
Change-Id: Ic2173335033a6ed0d4b652333020f030de63a8e7
Reviewed-by: Michael Brasser <michael.brasser@live.com>
A typical pattern with QML tests is to dynamically create
an item and then destroy it at the end of the test function:
TestCase {
id: testCase
name: "MyTest"
when: windowShown
function test_click() {
var item = Qt.createQmlObject("import QtQuick 2.0; Item {}", testCase);
verify(item);
// Test item...
item.destroy();
}
}
The problem with this pattern is that any failures in the test function
will cause the call to item.destroy() to be skipped, leaving the item
hanging around in the scene until the test case has finished. This can
result in interference with future tests; for example, by blocking input
events or producing unrelated debug output that makes it difficult to
follow the code's execution.
By calling e.g. createTemporaryQmlObject() instead, the object is
guaranteed to be destroyed at the end of the test function:
TestCase {
id: testCase
name: "MyTest"
when: windowShown
function test_click() {
var item = createTemporaryObject("import QtQuick 2.0; Item {}", testCase);
verify(item);
// Test item...
// Don't need to worry about destroying "item" here.
}
}
[ChangeLog][TestCase] Added createTemporaryObject() and
createTemporaryQmlObject() functions that can be used to ensure that
dynamically created objects are destroyed at the end of each test
function.
Change-Id: I483342052462aa86464c86bf2082892712dceb58
Task-number: QTBUG-56361
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Michael Brasser <michael.brasser@live.com>
OpenVG also needed to ping-pong between two offscreen surfaces like the
OpenGL renderer.
Change-Id: I41f671d98d361eb114be5c00ebfe62710a971aaa
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
A qHash function was added for QRawFont, so use that instead of our own.
Change-Id: Iad8deb8a0e2a727039ec752e38e42cef2127735c
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
The current approach to rendering paths (used by Rectangles and Glyph
nodes) does not support rendering with non-affine transformations. That
is because OpenVG does not support passing a non-affine transformation
matrix when rendering paths. So instead when using a non-affine
transform we will fallback to rendering to an offscreen VGImage, then
rendering that as an image which can have a perspective transform.
Change-Id: I01508bcb67b339323cb6400c7ff6d885b62c5e02
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
This is an OpenVG backend for the Qt Quick 2 scenegraph.
Should be feature complete now, but there are still some issues that
could be improved in future commits:
If Rectangle nodes are rendered with a non-affine transform, they will
be rendered incorrectly. This is because paths expect affine
transformations.
The Glyph cache is a bit cheeky in that it's caching paths, but doing
so per font size. It shoudln't need to, but right now I've not though
up a good way of getting the transform/scale needed when rendering yet.
Change-Id: Ie3c4f2df35d14279b0f9f55e0e10a873328c025b
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Error msg: Can't assign to existing role 'distance' of different
type [String -> Number]
Change-Id: I28842ed25a56077e52496a229294ef992406ca1d
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Private Use Area characters are quite valid input characters when used
in combination with a custom font. Joiners also serve an important language
purpose in semitic writing systems.
[ChangeLog][QtWidgets][Input] Support characters in Private Use Area, as well as
zero-width joiners and zero-width non-joiners in input in TextInput and TextEdit.
Task-number: QTBUG-42074
Task-number: QTBUG-57003
Change-Id: I4a9d6b9ea172d29375c20abfb7aebb6748c5d300
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Robin Burchell
Simon Hausmann
Albert Astals Cid
Edward Welbourne
Liang Qi
Shawn Rutledge
Oleg Yadrov
J-P Nurmi
Andy Nichols
Robin Burchell
Laszlo Agocs
Mitch Curtis
Nico Vertriest
Friedemann Kleint
Filipe Azevedo
Ulf Hermann
Olivier JG
Eskil Abrahamsen Blomfeldt