Despite being move-only, std::vector<unique_ptr> advertizes
is_copyable:
https://quuxplusone.github.io/blog/2020/02/05/vector-is-copyable-except-when-its-not/
Our combined reallocation and resizing function, reallocate_impl(),
runs afoul of this when it uses std::is_copyable in a constexpr-if to
implement resize(n, v) without running into problems with move-only
types: the trait is true, but actual instantation runs into a
static_assert in the STL implementation.
To fix, move the problematic resize functionality out of
reallocate_impl() and into the resp. resize_impl overloads. The shrink
functionality remains in reallocate_impl(), because there are many
more users, and it only requires destructible<T>, which isn't
constraining at all.
Amends a00a1d8806.
Fixes: QTBUG-109745
Pick-to: 6.5 6.4
Change-Id: Ibc5b9cf5375108eb3d8f6c8a16d4fd02dadd73b1
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
As discussed in API review, the default-constructed T() returned from
a mismatched data<T>() call is indistinguishable from a real T with
default state.
To make them distinguishable, return optional<T>. Call the new
function value<T>(), mimicking QVariant::value<T>(), and suggested in
API review, because data() is usually used to return raw pointers, not
values.
Remove the qWarning() on requestedType and actualType mismatch, as the
new function can be used in std::get_if/dynamic_cast-like if-then-else
chains, in which failure is part of the normal operation, and a
warning message misplaced:
if (auto loc = perm.value<QLocationPermission>())
~~~ use *loc ~~~
else if (auto con = perm.value<QContactsPermission>())
~~~ use *con ~~~
~~~ etc ~~~
Pick-to: 6.5
Change-Id: I799a58e930307323ebce8f9ac50a42455e9c017f
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Following the QRect, add functions converting the QString to native
emscripten::val and back: fromJsString, toJsString
Change-Id: I2d0625ede3bbf7249e2e91b8de298b5b91df8ba2
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
The new test:
- checks the properties of stateful Q<Typed>Permissions work
- ensures that piping a Q<Typed>Permission through QPermission
maintains state
- also ensures that assignment of a Q<Typed>Permission to a
QPermission works (via QPermission(Type) + move ctor)
Pick-to: 6.5
Change-Id: I340e49b1ecc665702ccab26d9050ca158b0e7885
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The test body has been inside an #if 0 for long, remove the test
altogether. The part that loads libthai does not check anyhing
meaningful.
Task-number: QTBUG-109954
Change-Id: I8ce006a14e4cf926b668e958d4b2343f965a5fb6
Reviewed-by: Ievgenii Meshcheriakov <ievgenii.meshcheriakov@qt.io>
It had various QVERIFY()s of != and >= checks: convert these to
QCOMPARE_NE() and QCOMPARE_GE() checks.
Change-Id: Ida6f7dca726187f7837da0d805549d9c582f946a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
When rows are hidden (implicitly or explicitly), then their layout data
does not get fully updated. If rows get hidden after the layout data has
been calculated once, then we must make sure that their indices are
reset. Otherwise we might access array indices that are out of bounds
when the layout data structure gets resized to fit only visible rows.
For good measure, skip entirely over hidden rows when accessing the
layout data when arranging the widget.
Fixes: QTBUG-109237
Pick-to: 6.5 6.4
Change-Id: I4d6943b6a110edb61f60ce78d31f0fc64b5cc03d
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
QDateTimeEdit has such reset functions for the minimum/maximum
range properties, this makes QCalendarWidget consistent with the
simpler editors.
Fixes: QTBUG-62448
Change-Id: Iabb89f599e4996f00488a885a5b8a009471838b3
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
For most territories, if we specify only that territory, we should get
a locale specific to that territory. There are exceptions for various
reasons, but check that it's true in most cases, at least. In the
process, convert two QVERIFY(... == ...) into QCOMPARE(..., ...)
Task-number: QTBUG-64940
Change-Id: I7590f20f37b0b459aafb3d1d08f6eb77932fa027
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
To be extended as neeeded.
Pick-to: 6.5
Change-Id: I57d2f55f67de073fe3e4916b7ba655342cf661dc
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Just describe the row instead. We'd lose the original input in case of
failure, so I added a class to print that value on destruction. Example:
FAIL! : tst_QAlgorithms::countLeading64(0) Compared values are not the
same
Actual (qCountLeadingZeroBits(value)): 63
Expected (expected) : 64
Loc: [tests/auto/corelib/tools/qalgorithms/tst_qalgorithms.cpp(374)]
QWARN : tst_QAlgorithms::countLeading64(0) Original value was 0x1
Pick-to: 6.4 6.5
Fixes: QTBUG-109958
Change-Id: I69ecc04064514f939896fffd1738b1119cd80cf8
Reviewed-by: Dimitrios Apostolou <jimis@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
tst_qstringapisymmetry needs to use the same collation testability
criteria as tst_qcollator, that is, the locales that may be tested
without ICU and not on Mac nor Windows are only the system locale and
C locale.
Task-number: QTBUG-109954
Change-Id: I69f19ae28b3a16b3827c1eee62ae59fcfdf45209
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This reverts commit 4096667d66.
Reason for revert: <Screen shot workaround is flaky.>
Change-Id: I6e01cc584c094d0d0b8c1544b9daf72d648dd002
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
qcollator_posix.cpp is included if ICU is not used and the OS is not
Win nor macOS. Reflect that fact in tst_qcollator instead of using
alternative means which breaks with new platforms that use the
posix collator.
Task-number: QTBUG-109954
Change-Id: I592500ce9626efbcc9377cecf6641967f978c6da
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There are numerous conflicting symbols in the tst_qstringbuilder\d
variants when batching those together. Remove the linkage from symbols
by putting the common include stringbuilder.cpp in an unnamed
namespace.
Task-number: QTBUG-109954
Change-Id: Ic2a745795b57482c90c9def7667a1145cdb19854
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Due to an XCB library change, QXcbBackingStore::toImage() cannot be
safely assumed to return an image identical to QWidget::grab().
The test functions fastMove(), moveAccross() and moveInOutOverlapped()
relied on QXcbBackingStore::toImage() and failed.
They were backlisted on Linux/XCB.
This patch obtains a screen shot instead of an image from the backing
store on XCB platforms. It processes events until the screen shot
matches QWidget::grab(). It makes the test fail only if the comparison
times out.
The patch also removes the BLACKLIST file, containing only the test
functions mentioned above.
Fixes: QTBUG-109036
Pick-to: 6.5 6.4
Change-Id: I26dd5b89dc62b313db066a285f6ad7d4d92baaf2
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Also replace the C array with a std::array and (some) indexed loops
with ranged-for loops. Most loops need the index in one way of
another, so can't easily be converted to ranged.
Pick-to: 6.5 6.4 6.2
Change-Id: I7fa05f22de9df6c68ec5797c9583476a3881532c
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
(Un)Register the custom recognize in init() and cleanup() instead of
initTestCase() and cleanupTestCase(), so that a new recognizer is used
for each test function.
In the test functions, use a scope guard to unregister the locally
registered recognizers to make sure that in the case of a failing test
and early return, the recognizer is removed.
Pick-to: 6.5 6.4 6.2
Change-Id: I4fe9509f35474514ef55191d799e6707199fe853
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Setting a negative line width does not make much sense, but in
earlier Qt versions, this work the same as if the line width was 0
(just give you the minimal layout given wrapping constraints).
But since 991c056438, we check if
current width > line width at an earlier point, and because 0 > -1,
we would exit immediately before adding any characters to the text
line.
To restore the behavior in earlier versions, we set the minimum
possible line width to 0.
Pick-to: 6.5
Fixes: QTBUG-109474
Change-Id: Iceadd5135681f61b30de8221853834983941c5a4
Reviewed-by: Lars Knoll <lars@knoll.priv.no>
Reviewed-by: hjk <hjk@qt.io>
Clients who called the base-class implementation in QDialog would
as a result start hitting the canBeNativeDialog code path at the
start of QDialog::setVisible(), which would show the native dialog,
but without updating the QWidget visibility state.
To keep things 100% compatible, we shuffle the implementation of
QDialog::setVisible() into QDialogPrivate, which allows us to
override it in QMessageBoxPrivate and QErrorMessagePrivate.
The existing subclasses of QDialog that override setVisible have
been left as is, to not cause any unintended behavior change.
Pick-to: 6.5
Change-Id: Icafe31a7b84a75049365e4e04b80492de08614d2
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
When the test is showing the error message over and over, it's not
waiting for the native dialog to actually become visible, and as
a result, hiding it has no effect and won't result in a call
to processResponse, where we got rid of the native dialog.
To fix this we explicitly release the native dialog when
encountering this corner case.
Add logic to QErrorMessage to test both native and non
native dialogs.
Pick-to: 6.5
Change-Id: I19ac3f463997aed1e66f646fdfcbb4d2459116d1
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
In case of QT_FEATURE_timezone=OFF test should not be built
Change-Id: If667b9edb1d670b9ed8a62f301a7e5e21e7d2b4c
Reviewed-by: Kimmo Ollila <kimmo.ollila@qt.io>
Reviewed-by: Janne Koskinen <janne.p.koskinen@qt.io>
The multi-thread snippet in the documentation, when copied as is,
could actually crash because of the race condition between the main
thread and the thread that generate results for the promise.
This is fixed by explicitly calling QPromise::start().
Actually, the underlying snippet already has this call, it just was
not included in the documentation.
This patch modifies the documentation snippet to include calls to
both QPromise::start() and QPromise::finish().
Fixes: QTBUG-109230
Pick-to: 6.5 6.4 6.2
Change-Id: Ic25f31a6b3b16ba6bc06a0b199289c8c5d50bab6
Reviewed-by: Sona Kurazyan <kurazyan.sona@gmail.com>
Support serializing shaders with specific qsb version. The default
behavior remains the same, using the latest version.
Task-number: QTBUG-101062
Pick-to: 6.5
Change-Id: I090a88c1ccb3be4ac5eee1da4058afaa8bf3111c
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Instead of the more verbose currentDateTime(QTZ::UTC).
Pick-to: 6.5
Change-Id: Ie759f4270b12fca39c458bf85c8296f5342033db
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
To support cancellation of continuations attached via the parent future,
we store a pointer to continuation future's data in parent. This
requires preserving the lifetime of continuation future's data while the
parent is still alive (see 24dedaeaa1).
This is achieved by capturing the promise in the continuation's lambda,
which is only cleaned up after the parent's data is destroyed. This is
already the case for continuations without context, but was overlooked
for continuations with context: they transfer the ownership of the
continuation promise to lambda passed to QMetaObject::invokeMethod(),
which destroys the lambda's context after it's run. As a result, the
continuation's promise (and data, if there are no other copies of it)
is also destroyed, leaving the parent pointing to deleted continuation
data.
To fix this, capture a copy of continuation future's ref-counted data in
the continuation's lambda. This will guarantee that the continuation
data remains alive until the parent is destroyed and the continuation
is cleaned up.
Fixes: QTBUG-108790
Pick-to: 6.5 6.4 6.2
Change-Id: Ief4b37f31e652988d13b03499505ac65c7889226
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Not all locales use ASCII hyphen-minus U+002D as minus sign. On macOS
using the nb_NO locale the U+2212 character is used instead when
displaying negative years. Verify that one of the two characters is
found.
Fixes: QTBUG-109853
Change-Id: I424539cc8d427ac199b4528e44bef98e45312d07
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
As discussed in the latest CMake API Review, we are deprecating the
FILENAME_VARIABLE variable name everywhere, and replacing it with
OUTPUT_SCRIPT.
[ChangeLog][CMake] The FILENAME_VARIABLE option of
qt_generate_deploy_script and qt_generate_deploy_app_script is now
deprecated, use OUTPUT_SCRIPT option instead.
Change-Id: Ic8be33eefbc48540166ea0fcf1d1948b052d4b8a
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
These are time stamps, and they use toMsecsSinceEpoch(), no point
getting the time in Local time zone then converting it.
Change-Id: I2db2db5f9ebc062e65514a592fa7fa00cf1d179d
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This is inherently faster than getting it in UTC from the underlying
native API stat call, then converting it to the Local Time Zone just to
compare them. The same goes for any use-case where you get a QDateTime
then the first thing you do is call t.to{Msec,Secs}SinceEpoch().
Change-Id: Ic13bcfd99b937c9f10f102ea7741832950a553c6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Kai Köhne <kai.koehne@qt.io>
If an item in a list view has a null icon, then the decorationSize gets
calculated as -1, -1. The style would then try to wrap the text to a
lineWidth of -1, ending up in an infinite loop in viewItemTextLayout.
To prevent that, don't set the HasDecoration flag of the style option
when the icon is null, and don't fall back ot the decorationSize unless
the flag is set.
Add a test for this particular item configuration. This also fixes the
widget baseline test with styles that don't provide all standard icons.
Pick-to: 6.5
Change-Id: I691db6abede9a9b2ad300f3ee7fbfdae5fb6097f
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Reviewed-by: Santhosh Kumar <santhosh.kumar.selvaraj@qt.io>
Wayland omits optimizations tested in tst_QWidget::optimizedResizeMove()
and optimizedResize_topLevel() under certain circumstances, e.g. on
Ubuntu 22.04 / Gnome. This makes the test functions fail.
This patch skips the test functions on wayland platforms, if an
omission is detected.
This amends 2ec7a6322f.
Fixes: QTBUG-109746
Pick-to: 6.5 6.4
Change-Id: If0df6b1cf451a7f0dccfc1bb7411e895d7ae29a3
Reviewed-by: Liang Qi <liang.qi@qt.io>
QWindow::requestActivate() is not supported.
This amends c95de359b4.
Task-number: QTBUG-107153
Pick-to: 6.5 6.4
Change-Id: I45f53b5e9de85049ca41cc139a78a82450f53bed
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The test function used to flake on Linux occasionally.
8e6ede7cd1 provided a fix. While it
seemed to work, further analysis has shown that the root cause is
event sequence in case of multiple paint events.
This patch re-engineers the test function:
1. Allocate test widget on the stack instead of the heap.
2. Send layout requests posted by QHBoxLayout constructor and
QLayout::addWidget() before showing the widget.
3. Remove calls to QCoreApplication::processEvents().
They are unnessecary, because
- the size hint request counter is supposed to increase (by any number)
- QTRY_VERIFY processes events anyway until the counter increases or it
times out.
Pick-to: 6.5 6.4 6.2
Change-Id: I54998483725cbdd4899ba6f5469d7dae0980ab1d
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The test shows an application modal QMessageBox, but assumes that doing
so will be non-blocking, which for macOS is not the case (yet). Instead
of making the dialog window-modal, which would potentially affect the
logic of the test, we disable native dialogs. This should be fine, as
the purpose of the test is to test the is_closing logic of
closeAllWindows, which lives on a layer above the native dialogs.
Pick-to: 6.5
Change-Id: I6d627984a6ca452b876f34404b669fce41a00851
Reviewed-by: Doris Verria <doris.verria@qt.io>
With the methods that use helpers from qstring.cpp defined in the
latter.
Change-Id: I11d6b0bfb95efe34e56d33d2ecbfe8f4423a9e6c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Makes the pre-existing QFutureInterface functionality available via
the public QPromise API.
[ChangeLog][QtCore][QPromise] Added addResults() to report multiple
results at once.
Change-Id: I18e6ef2781df422020b9022d78d6c45107b01668
Reviewed-by: Sona Kurazyan <kurazyan.sona@gmail.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
In preparation to their deprecation / removal.
Change-Id: Ia073a9f7caabbc06063a1e416b23cdb12788b283
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If a type is trivially default constructible, QMetaType (and QVariant)
think that it can be built and value-initialized by zero-filling a
region of storage and then "blessing" that storage as an actual instance
of the type to build. This is done as an optimization.
This doesn't work for all trivially constructible types. For instance,
on the Itanium C++ ABI, pointers to data members are actually
value-initialized (= zero-initialized, = initialized to null) with the
value -1:
https://itanium-cxx-abi.github.io/cxx-abi/abi.html#data-member-pointers
This means that a type like
struct A { int A::*ptr; };
is trivially constructible, but its value initialization is not
equivalent to zero-filling its storage.
Since C++ does not offer a type trait we can use for the detection that
we want to do here, and since we have also decided that Q_PRIMITIVE_TYPE
isn't that trait (it just means trivially copyable / destructible), I'm
rolling out a custom type trait for the purpose.
This type trait is private for the moment being (there's no
Q_DECLARE_TYPEINFO for it), and limited to the subset of scalar types
that we know can be value-initialized by memset(0) into their storage
(basically, all of them, except for pointers to data members).
The fix tries to keep the pre-existing semantics of
`QMetaType::NeedsConstruction`. Before, the flag was set for types which
were not trivially default constructible. That included types that
aren't default constructible, or types that cannot do so trivially.
I've left that meaning unchanged, and simply amended the "trivial" part
with the custom trait. A fix there (to clarify the semantics) can be
done as a separate change.
Change-Id: Id8da6acb913df83fc87e5d37e2349a4628e72e91
Pick-to: 6.5
Fixes: QTBUG-109594
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
tst_QDockWidget::createTestWidgets did not set a minimum size for
test widgets. Upon unplugging, that can lead to a size correction in
connection with 8687c3f938cac56c4a2518576a5c9fac150d8de1, if the widget
gets too narrow to correctly display window handles. Test functions
checking the size after unplugging can fail in that case.
This patch sets a minimum size to each test widget created. The minimum
size corresponds to QStyle::PM_TitleBarHeight.
Task-number: QTBUG-106531
Pick-to: 6.5 6.4 6.2
Change-Id: I3e552b37416d57d8ed30e0b9de4eec55f07ea158
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Turns out we don't support QStringView/QUtf8StringView comparison, and
the only reason the corresponding test succeeded was because it
contained a typo (QStringView instead of QUtf8StringView).
Fix the typo and disable the now-failing test.
Pick-to: 6.5 6.4 6.2
Change-Id: I2210a247aac66743851e53578172a563ee1e96f7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The test function places a QSpinBox in a QHBoxLayout, which has a
QWidget parent. The spin box is expected to be shown with the widget.
While the widget is the layout's parent, the layout is not explicitly
set. That makes the test function flaky in some cases.
This patch adds QWidget::setLayout() to explicitly set the layout on
the widget.
Pick-to: 6.5 6.4 6.2
Change-Id: I3a1cc77c302c5ba96d3628d035139f9718dda9e5
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
The existing symLinkTarget() always resolves the symlink target to an
absolute path; readSymLink() provides access to the relative path when
that is how the symlink references its target.
[ChangeLog][QtCore][QFileInfo] Added readSymLink() to read the symlink's
raw target, without resolving to an absolute path.
Fixes: QTBUG-96761
Change-Id: I360e55f1a3bdb00e2966229ea8de78cf29a29417
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
testWhiteSpaceNames() returns true for all current database drivers so
it's useless and can be removed. safeDropView() and getPSQLVersion()
is not called anywhere, getMySqlVersion() is only used for a check for a
MySql version we no longer support.
Change-Id: I8d02f17f475821e81d309ee96897e772cdfb895d
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Marc Mutz
Mikolaj Boc
Edward Welbourne
Volker Hilsheimer
Thiago Macieira
Axel Spoerl
Eskil Abrahamsen Blomfeldt
Tor Arne Vestbø
Tatiana Borisova
Ivan Solovev
Kaj Grönholm
Ahmad Samir
Sona Kurazyan
Øystein Heskestad
Christian Ehrlicher
Amir Masoud Abdol
Liang Qi
Alexey Edelev
Giuseppe D'Angelo
Wang Fei