When Harfbuzz returns zero glyphs, this will typically be because
the character is ignored. We currently have no way to differentiate
this from actual error cases, so to avoid return the error glyph
for cases which are not really errors, we set the dontPrint flag
for this.
When we add a way to differentiate between error cases and non-error
cases later, we should revert this for the actual errors.
Pick-to: 6.4
Task-number: QTBUG-108799
Change-Id: I2a910d951b71c705fb4dd761fcfe3a942b3afa7c
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
There is no technical reason why qFatal shouldn't support categorized
logging, nor have a streaming version.
There is perhaps an API issue, that is, a streaming syntax may encourage
users to do "too much":
qFatal() << gatherLogs() << saveDatabase() << ...;
and that sounds like a bad idea in case the application is in an
unrecoverable state that requires immedate terminatation (indeed,
through qFatal). I'd err on the side of providing the extra convenience.
This commits adds overloads of QMessageLogger::fatal to support
categorized logging (note that fatal messages cannot be disabled),
the relative qCFatal macro, as well as overloads for streaming.
[ChangeLog][QtCore][QMessageLogger] QMessageLogger::fatal now supports
categorized logging, for instance using the qCFatal(category) macro.
Moreover, qFatal() and qCFatal() now support streaming of values
to be printed in the fatal message.
Change-Id: Ia57f25f5c85fca97e5fcf27eaa760dbde09cba0e
Fixes: QTBUG-8298
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
For both the [4, 7] and [8,15] length cases, we can perform the same
technique: perform two overlapped loads, zero-extend, then perform two
overlapped stores. The 8-character case could be done in a single
load/store pair, but is not worth the extra conditionals. And it should
have the exact same performance numbers whether we use non-overlapping
4-character operations or completely-overlapping 8-character ones (I
*think* the full overlap is actually better).
The 4-character operation is new in this commit. That reduces the
non-vectorized, unrolled to at most 3 characters.
Change-Id: Ib42b3adc93bf4d43bd55fffd16c257ada774236a
Reviewed-by: Lars Knoll <lars@knoll.priv.no>
Add handling for MYSQL_TYPE_JSON by treating it the same as
MYSQL_TYPE_BLOB (which is used by current MariaDB Server for a json
column)
Pick-to: 5.15 6.2 6.4
Fixes: QTBUG-101680
Change-Id: I4d4b0cdad73cd12e0db4df4021fddbd6a649c8ed
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
QWidgetWindow dispatched only tablet presses to the toplevel widget if
no child was found at the position; other events, such as hover events,
were discarded. The tabletTracking test case even documented that
shortcoming in a comment.
Fix that by falling back to the toplevel widget for any event. As
before, only press events initialize the tablet grabbing target widget.
Remove the now unneeded parent widget from the test case, and move the
test class into the only test function that uses it.
Amends ea615b421b and
8fd6cef372.
Pick-to: 6.4
Fixes: QTBUG-108747
Change-Id: I79050f1e063931e439945f64b50712dcc1ecb18c
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
For QSet, the key_type is const, so can't test assiging to it.
Pick-to: 6.4
Change-Id: I9d363ef3fe52646b937d6a422227b19c48fdaf1f
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Because they are too convenient to leave out.
Change-Id: I844cfb794ce0f575c2c65075d9051b0b878a434f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Make sure that touching any file that is part of a qrc resource
actually rebuilds the resource and the target that depends on the
resource.
Pick-to: 6.4
Task-number: QTBUG-107687
Task-number: QTBUG-108113
Change-Id: I1153dc13fee44ffe59d2685a8cb33303538b026c
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Overloading insert is a bit tricky since the size might change after
the conversion so either the tail has to be moved twice or a temporary
buffer is needed. For now, add an ineffective but simple overload as in
the case of the const char *s overload, and do the performance
optimization in a follow-up task (QTBUG-108546).
[ChangeLog][QtCore][QString] Added insert(QUtf8StringView) overload.
Task-number: QTBUG-103302
Change-Id: If01c216ff626da29abb43eb68d4de82824f3bfba
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The += operator is already overloaded to handle QStringView and
QLatin1String - add the missing QUtf8StringView overload.
[ChangeLog][QtCore][QString] Added operator+=(QUtf8StringView)
overload.
Task-number: QTBUG-103302
Change-Id: Iec6940bad7866310c826a130b98accebc3c82aa8
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Add the missing overload, among other things it is needed to
implement QTBUG-103302.
[ChangeLog][QtCore][QString] Added append(QUtf8StringView)
overload.
Task-number: QTBUG-103302
Change-Id: I576f73c1919e3a1f1a315d0f82c708e835686eb1
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
For the case of static builds. Also, remove somewhat funny comment
from the shared tlshelpers.h file.
Fixes: QTBUG-105046
Pick-to: 6.4 6.2
Change-Id: I2fd677a11b64df89c0ad0237e1014f5e80144c24
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The test functions tst_QWidget::optimizedResizeMove and
tst_QWidget::optimizedResize_topLevel use a custom widget to record
paint events in a boolean.
Using QTRY_* or qWait() to wait for the boolean to become true stops
event processing upon consumption of the first paint event.
In case of multiple paint events, the next paint event will be recorded
in the next check and may lead to a wrong result.
This leads to flakiness on platforms generating multiple paint events,
openSuSE Leap 15.4 / XCB being one of them.
This patch replaces the boolean with a counter. It adds a
waitForPaintEvents method to the custom class, which processes events,
until the counter stops increasing. It returns true when paint events
have been recorded and false otherwise. It resets the counter after
each call.
It also removes the QSKIP on wayland, as the failure results from the
same reason.
Fixes: QTBUG-109093
Pick-to: 6.4
Change-Id: I59ee8bb4efeaf5417d5749d21e384bee89301ae0
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Also add optimizations for more string comparisons and add tests and
benchmarks.
[ChangeLog][QtCore][QString] Added utf-8 case-insensitive comparisons
Fixes: QTBUG-100235
Change-Id: I7c0809c6d80c00e9a5d0e8ac3ebb045cf7004a30
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
During the implementation of QString::append(QUtf8StringView) it has
become apparent that the testing is insufficient as it did not warn
about an extra growth. The following tests have been added that append:
- y-umlaut and greek letter small theta (2 UTF-8 code units => 1 UTF-16)
- devanagri letter ssa (3 UTF-8 code units => 1 UTF-16)
- chakma digit zero (4 UTF-8 code units => 2 UTF-16 code units)
- some combinations of the above
Pick-to: 6.4 6.2
Task-number: QTBUG-103302
Change-Id: I981213c296bafc81663b08c0f1f339bbd8a96485
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
During the implementation of QString::append(QUtf8StringView) it has
become apparent that the testing is insufficient as it did not warn
about an extra growth. The following tests have been added that append:
- y-umlaut and greek letter small theta (2 UTF-8 code units => 1 UTF-16)
- devanagri letter ssa (3 UTF-8 code units => 1 UTF-16)
- chakma digit zero (4 UTF-8 code units => 2 UTF-16 code units)
- some combinations of the above
Pick-to: 6.4 6.2
Task-number: QTBUG-103302
Change-Id: I3d81cf10b7eb74433ce5bea9b92ce6bce1230dcd
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
During the implementation of QString::append(QUtf8StringView) it has
become apparent that the testing is insufficient as it did not warn
about an extra growth. The following tests have been added that append:
- y-umlaut and greek letter small theta (2 UTF-8 code units => 1 UTF-16)
- devanagri letter ssa (3 UTF-8 code units => 1 UTF-16)
- chakma digit zero (4 UTF-8 code units => 2 UTF-16 code units)
- some combinations of the above
Note that this also affects operator_pluseq_data, which is basically
a wrapper around append_data.
Pick-to: 6.4 6.2
Task-number: QTBUG-103302
Change-Id: I09ed950e3f0e71ae9ae85a455f42e130887f1109
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Implements an adaptor from the notification signal of a Q_PROPERTY to
QBindable. The Q_PROPERTY does not need to be BINDABLE, but can still
be bound or used in a binding.
[ChangeLog][Core][Q_PROPERTY] Q_PROPERTYs without BINDABLE can be wrapped in QBindable to make them usable in bindings
Change-Id: Id0ca5444b93a371ba8720a38f3607925d393d98a
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
The test function tst_QComboBox::setPalette() tests if the QComboBox
correctly inherits its palette to its QLineEdit child, by directly
comparing both palettes after a palette change on the QComboBox.
If the application palette for both widget types contains brushes which
are either pixmap based or the resolve mask prevents them from being
copied, the direct palette comparison between parent and child fails,
despite of a correct propagation.
This patch sets a simple gray application palette for QComboBox and
QLineEdit at the beginning of the test. The QSKIP for macOS is
removed, because using simple palettes solves both issues.
Change-Id: I61d509745377306d6f8331dfc1ac189fc58cdedb
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
It started being recorded as flakey after making macOS 13 a significant
config in the CI, but it's unclear why the test was not failing during
the nightly testing leading up to that.
http://testresults.qt.io/grafana/goto/-AzSwMKVz
It has also been blacklisted on macOS many times in the past.
Change-Id: I7be5fefb3669e9643a98d2ac82059bf0576c6d5c
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This test causes instability in the CI, but so far nobody could
reproduce it outside of that specific CI environment. Skip it for now,
and investigate the root cause later.
Pick-to: 6.2 6.4
Task-number: QTBUG-106906
Change-Id: Idbbc560236ac475fcb4fdd5980397393f7a95474
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
When a widget's palette has different brushes in Active / Inactive
color groups, or it is pixmap based, multiple paint events are
triggered. This leads to unpredictable, double entries in
QWidgetRepaintManager::dirtyWidgetList().
This patch overrides event() of all test widgets and ignores
activation / deactivation events in order to make the entries of
QWidgetRepaintManager::dirtyWidgetList() predictable.
Pick-to: 6.4
Change-Id: I164d7ab4148551590ac3c50fcc3b9f98c5ac0535
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
QXcbBackingStore::toImage() returns a QImage that differs from what
is displayed on the screen: After multiple moves of children within
their parent widget, the last move is not always properly reflected
on openSuSE 15.4.
This patch blacklists the test functions fastMove, moveAccross and
moveInOutOverlapped on openSuSE Leap.
Task-number: QTBUG-109036
Pick-to: 6.4
Change-Id: Ideb505570df090b25d30762492d9c60beb6641ff
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
They are already slow without ASAN, and risk running into timeouts with
ASAN enabled.
Change-Id: I427b990066d8ffa838ea90a318176fbcba400852
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Both QStringView and QAnyStringView implicitly convert from any
container with a fitting value_type, and working std::data, std::size,
std::begin and std::end.
Add these missing operations (and complementary ones) to XmlStringRef,
so it implicitly converts to QStringView and QAnyStringView, too.
Add a check to that effect and remove the now-superfluous operator
QStringView().
Task-number: QTBUG-103302
Change-Id: I89d586cf64447a82022e06d546d7ee8339fc6dc7
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Avoid using perl in CMake scripts. Remove the syncqt.pl specific
code.
Task-number: QTBUG-87480
Change-Id: I7fcd5cc83d173ec463c275b5b50b84f25044a118
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
I've reserved the IDs for int128, uint128, bfloat16, and float128,
because the mask in qvariant.cpp's qIsNumericType() requires primitives
to be less than 64 to operate properly.
Added a QMetaType/QDataStream test to confirm it is indeed built-in.
Change-Id: I3d74c753055744deb8acfffd17247f7f57bada02
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The C++ equivalent is std::float16_t, defined in P1467[1], and is coming
with GCC 13 both in native mode (for x86, using AVX512FP16) and in
emulated mode. The C and C++ types will be the same type (<stdfloat>
simply typedefs).
qfloat16 will need to remain a wrapper with an integer member to keep
ABI with previous Qt versions. Because it is a trivially-copyable small
type, it gets currently passed in registers; the presence of the integer
member means it gets passed in general-purpose registers, while a single
_Float16 member would be passed in a floating-point register. See:
https://gcc.godbolt.org/z/8fEendjff
[1] https://wg21.link/p1467
Change-Id: I8a5b6425b64a4e319b94fffd161be56397cb48e6
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
... where checks for QByteArray existed before.
The checks we can't add are
- left/right/mid (legacy APIs not implemented in QBAV)
- several relational operators, d/t ambiguities. Created
QTBUG-108805 to track these.
Task-number: QTBUG-108805
Pick-to: 6.4 6.2
Change-Id: I30cc9b29a228d69d32af51234f2c28221478a75c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
It's not intuitive, so check lest people break it.
Pick-to: 6.4 6.2
Change-Id: I2435cd69be7b77a6ae59cdc7b5fb99658cfc42fd
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It was forgotten to assign a simple palette to this object in
1576f82721
This patch adds the missing line.
Pick-to: 6.4
Change-Id: I0acf765d9646ecc8d49791d96b5ea910e11997fe
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
We will need it in QML to determine which signal out of a number of
"overloads" to generate handlers for.
Task-number: QTBUG-108762
Change-Id: Ib73eeb59dbf037bd03faf0d841860d9ebc9e7032
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
The algorithms for calculating the scene's position within the view
did not compensate for scrollbars showing. The scrollbars should be
ignored when positioning hte scene within the view, as alignment
only cares about the positioning of the scene when the view is
larger than the scene anyway.
Add a test case that verifies that items don't jump up or down when
dragging horizontally, and not left or right when dragging
vertically.
Mark variables in the modified function as const where applicable to
make it easier to follow the code.
Done-with: Volker Hilsheimer <volker.hilsheimer@qt.io>
Fixes: QTBUG-46757
Change-Id: If205637dfe124e0034f68201b23f174d6863084d
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
When a widget's palette has different active / inactive colors,
multiple paint events can occur. This makes tst_QWidget functions fail
when they expect a spcific amount of paint events and the platform
theme provides a palette with active / inactive differences.
This patch adds a function to populate test widgets with a simple
palette, to prevent multiple paint events.
Pick-to: 6.4
Change-Id: If9b2faedcc5ca87ba24991cedd5e4ac927b02644
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
After 109e088c7c, cache keys were unique
for palettes with different private or data instances, but the key did
not change when a palette without any shared copies was modified, as
that does not create new private data structures.
To fix this, always increase the counter for the private data structure,
also when not detaching from shared copies.
Augment test case with scenario that broke.
Pick-to: 6.4
Fixes: QTBUG-108709
Change-Id: I606abfb8b1a03e515e46b10dc840a631eb31d496
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: JiDe Zhang <zhangjide@uniontech.com>
Reviewed-by: Alessandro Portale <alessandro.portale@qt.io>
Underflows should be treated the same as overflows.
Fixes: QTBUG-108628
Change-Id: I23aa7bbe1d103778cefca08bd3e584e72f306583
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The helper function TestWidget::waitForPainted returned after the test
widget had consumed the first paint event. In case of multiple paint
events, QRegion paintedRegions did not match the entire region that
was supposed to be painted. The test function children() failed/flaked
due to that.
This patch extends the helper function. After consumption of the first
paint event, it processes events until the painted regions no longer
change.
Pick-to: 6.4
Fixes: QTBUG-108764
Change-Id: I54e14bb07725dd1f602cc93085da13836e3b7494
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Rework QSimpleParsedNumber to store a qsizetype whose sign serves as
ok flag (positive is ok, zero and negative are not) and magnitude is
the number of characters used. This replaces an endptr that was set to
null to indicate !ok, but that deprived us of end-of-parse
information, which is needed for number-parsing. In particular, JS's
parsing of numbers accepts overflow (where qstrntod() flags it as
invalid) as infinity; so qstrntod() does need to say how long the
overflowing (but JS-valid, none the less) number-text was.
Modify all callers of functions using this (recently-introduced) type
and add tests that fail without this fix.
Fixes: QTBUG-108628
Change-Id: I416cd213e1fb8101b1af5a6d43615b970a5db9b4
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
tst_QOpenGLWidget fails on Linux when other windows on the desktop grab
focus. That can happen on CI VMs.
This patch blacklists the test on Linux as a temporary solution
Task-number: COIN-966
Change-Id: Ibdd7e626f92d2da692bd7c44741e0ca4c9dbd1a8
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
tst_gestures flakes in MacOS arm and Linux when delivering a gesture to
a screen. The problem cannot be narrowed down to specific test
functions. When two test functions with gesture delivery to a screen
are called, the second fails.
This patch temporarily blacklists OpenSuSE for tst_gestures.
Task-number: QTBUG-104496
Task-number: QTBUG-106244
Change-Id: I3bc3c9b82567478784a84006d27f5cbed39d9c9a
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This is improved version of previous fix
013c346a8d that was reverted because it
broke some tests for Quick Text. The problem was that it did not work
correctly in the case the text was wrapped to a fixed width.
To deal with this we'll accumulate current line full width (as if it
hadn't been wrapped) in layout data (layoutData->currentMaxWidth).
Then when the next line is explicitly wrapped by line or paragraph
separator, this accumulated width will be used to adjust layout's
maximum width.
Change-Id: Iad7119d9808e1db15fe1fbc5db049c3db928529f
Fixes: QTBUG-89557
Fixes: QTBUG-104986
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
This reverts commit e36abea63b.
Reason for revert: QTBUG-108470
Change-Id: If9aeae8be67594320d00dcdb42c993eda7acaeef
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
- If this string is not shared, modify it directly
- If this string is shared, instead of detaching copy the characters
from this string, except the ones that are going to be removed, to a
new string and swap it. This is more efficient than detaching, which
would copy the whole string including the characters that are going
to be removed.
This affects:
remove(const QString &str, Qt::CaseSensitivity cs)
remove(QLatin1StringView str, Qt::CaseSensitivity cs)
Adjust the unittests to test both code paths.
[ChangeLog][QtCore][QString] Improved the performance of
QString::remove() by avoiding unnecessary data copying. Now, if this
string is (implicitly) shared with another, instead of copying
everything and then removing what we don't want, the characters from
this string are copied to the destination, except the ones that need to
be removed.
Task-number: QTBUG-106181
Change-Id: Id8eba59a44bab641cc8aa662eb45063faf201183
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
In Qt 5, QWin(dows)Mime and QMacMime lived in the respective Extras
modules, which were removed and partially folded into the relevant
modules in Qt. QWindowsMime and QMacMime continued to provide the
abstraction for implementing built-in support for native clipboard
formats and UTIs within Qt, but only as private APIs.
After the recent clean up of those APIs and respective infrastructure,
we can now bring them back as public converter interfaces. Application
developers can subclass those and instantiate an instance of their
implementation to add support for platform or application specific
data formats.
These interfaces are not in the QNativeInterface namespace, as
applications don't call into Windows or macOS using those interfaces.
I.e. there is no class on which an application would call
auto *converter= nativeInterface<QWindowsMimeConverter>();
Also, since applications override those converter types, we do want to
guarantee binary and source compatibility.
[ChangeLog][QtGui][QWindowsMimeConverter] Reintroduced to allow
applications to add support for conversion from and to Windows-native
clipboard formats to MIME-encoded data.
[ChangeLog][QtGui][QUtiMimeConverter] Reintroduced to allow
applications to add support for conversion from and to clipboard data on
macOS and iOS to MIME-encoded data.
Fixes: QTBUG-93632
Change-Id: Iebd909c3970015d203f59d5ab15e306b3d312f6e
Reviewed-by: Yuhang Zhao <2546789017@qq.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Test causes integrations on macOS to fail.
Temporary solution, fixing of auto test required.
Task-number: QTBUG-108402
Change-Id: I9ceef8ec425cdd5131bce0cfffcb4daf739e674d
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Support for 1D textures on Vulkan, OpenGL, Metal, and D3D.
Change-Id: Ie74ec103da9cfcbf83fa78588cf8cfc1bd6e104f
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
The native implementation uses NSAlert, making a best effort to map the
QMessageBox properties to the native dialog, falling back to the cross
platform non-native dialog if the discrepancy is too big.
The initial implementation focuses on the current state of the
native dialog helper "protocol", but there's room for improvement
here, which would allow even more dialog types and properties to
be native.
[ChangeLog][macOS] Message boxes such as QMessageBox now follow
the platform look and feel by using native dialogs if possible.
Change-Id: I4da33f99894194a7b301628cd1fbb44d646ddf18
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Skip on crashing tst_qvulkan test
Blacklisted completer_data on qfiledialog
(8d76c5af51 should have been enough but
it is still failing)
Task-number: QTBUG-108328
Task-number: QTBUG-108329
Change-Id: Iad5573af60cca16d16ba0462293e276186e25653
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
This effectively reverts a4a51f6a64
while solving the problem that change intended to fix by an
alternative approach: Swap the order of checks for rhi-based
flushing. Checking the widgets' wishes first was a mistake. We should
first check what is forced, e.g. via the env.vars. Then only move on
investigating the child widget hierarchy if there was nothing specific
requested.
This way having a QOpenGLWidget in a window and running with
QT_WIDGETS_RHI=1 QT_WIDGETS_RHI_BACKEND=vulkan will prioritize the
forced request (Vulkan) and so the QOpenGLWidget will gracefully not
render anything while printing the expected warning to tell what's
going on.
The expensive recursion plaguing the construction of larger widget
hierarchies is now avoided, that should no longer take seconds due to
walking the entire widget hierarchy of the top-level window every time
a new widget is added to it.
However, this then uncovered a set of problems concerning native child
widgets. The repaint manager seems to have an obvious mistake where
the usage of rhi (and so textures and whatnot) is decided based on
'widget' (which is either a top-level or a native child) instead of
'tlw' (which is the top-level with the backingstore). The usesRhiFlush
flag only really matters for a real top-level, not for native child
widgets. The rhi-based flushing is tied to the backingstore, and the
backingstore comes from the top-level widget.
Finally, make the qopenglwidget autotest to actually exercise
something when it comes to QOpenGLWidgets (or their ancestors) turned
native. The original code from a long time ago does not really test
native child widgets, because it turns the top-level into native which
is quite superfluous since the toplevel is backed by a native window
(and a backingstore) anyway.
Pick-to: 6.4
Task-number: QTBUG-105017
Fixes: QTBUG-108344
Fixes: QTBUG-108277
Change-Id: I1785b0ca627cf151aad06a5489f63874d787f087
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
saveRestoreGeometry() somtimes creates a 29px offset when debugged in
Qt Creator, which makes the test fail.
This patch adds a code comment to make developers aware of this fact.
Pick-to: 6.2 6.4
Change-Id: I920bd02eb7543faf8b25a0a242b888f3a3745e2a
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
multipleToplevelFocusCheck() occasionally failed on XCB because
QApplication::activeWindow() was nullptr immediately after
qWaitForWindowActive() returned true.
This patch replaces QCOMPARE on QGuiApplicaiton::activeWindow() with
QTRY_COMPARE in order to continue spinning the event loop until the
condition has been met. It adds QWidget::activateWindow after show()
to ensure focus is acquired also when another window has received
focus in the meanwhile.
Pick-to: 6.2 6.4
Change-Id: If84eb8b77c5a6b16af271334a1fe5eb92c05644b
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
The test function raise() occasionally failed because of unexpected
paint events being counted.
This is due to a QTRY_VERIFY returning after consumption of the first
paint event. If more than one paint event got posted, it will be
delivered and counted when no more paint events are expected.
This patch ensures that all paint events are consumed before resetting
the count and expecting no more paint events.
Fixes: QTBUG-68175
Pick-to: 6.2 6.4
Change-Id: I3e91a34e851da4bd01c7429e824d2b9101077a06
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
The uses of the click_cocoa_button() function was removed in 2012,
in ba21ca7b5b.
Change-Id: If7abfc56bb307cfbf9f6628cec9c7267a8a1f31f
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
This tests that strings using the first Unicode code-point of such a
multi-character token don't get recognized as "valid" number strings.
This would catch an implementation issue if the parsing code
mistakenly matched against only the first code-point of each "single
character" token.
It also adds tests of integer formatting, with multi-character sign,
and reworks some QStringView().toString()s to use u"..."_s.
Task-number: QTBUG-107801
Change-Id: I7b868ce2955bb322b3ecfc200438a21437090a0c
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Upon programmatic window state changes, windowStateChange was fired
once in QWindow::setWindowStates and once when the visual state had
been changed by the window interface.
This patch adds if guards to ensure that the singal is fired only once.
It adds a corresponding autotest to tst_QWindow.
tst_QWidget::resizePropagation() is adapted to no longer expect double
signal emission.
Fixes: QTBUG-102478
Pick-to: 6.4 6.2
Change-Id: If093c0a883d76d8a676e4fab90db6b0676452267
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When a complex object (i.e. one with children that are themselves not
fully exposed objects) gets focus, then we need to inform the
accessibility system about which child object actually has focus. This
was only done for item views, but not for other complex widgets.
An editable QComboBoxes is the focus proxy for its line edit. The line
edit never gets focus itself (QComboBox forwards relevant events),
and is the accessible child item with index 1. So when an editable
combobox gets focus, it needs to raise the automation event for the
line edit child.
Implement QAccessibleComboBox::focusChild to return the interface to the
lineedit for editable comboboxes so that the UI Automation bridge can
correctly notify about the focus being moved to an editable text input
field.
Fixes: QTBUG-107572
Pick-to: 6.4 6.2
Change-Id: Id60e2791ec859365255baa9bfd01547979cd2b44
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
For a11y purposes, a table needs to be mapped into a logical
accessibility hierarchy. There are several ways of doing this mapping,
and unfortunately macOS expects something different than what
QAccessibleInterface does.
So suppose we have a a 2x2 QTableView with both horizontal and vertical
header like this (the names reflect the QAccessible::Role names):
+-----------+--------------+--------------+
| | ColumnHeader | ColumnHeader |
+-----------+--------------+--------------+
| RowHeader | Cell | Cell |
+-----------+--------------+--------------+
| RowHeader | Cell | Cell |
+-----------+--------------+--------------+
In order to be presented to the screen reader on a platform, it goes
through two rounds of mapping:
QAccessibleInterface will have all headers and cells as direct children of the table:
- Table
+- ColumnHeader
+- ColumnHeader
+- RowHeader
+- Cell
+- Cell
+- RowHeader
+- Cell
+- Cell
macOS expects a deeper hierarchy:
- AXTable [QAccessible::Table]
+- AXRow [Qt:no eqiuivalent]
+- [QAccessible::Cell] (The content of the cell, e.g. AXButton, AXGroup or whatever)
+- [QAccessible::Cell] (The content of the cell, e.g. AXButton, AXGroup or whatever)
+- AXRow
+- [QAccessible::Cell] (The content of the cell, e.g. AXButton, AXGroup or whatever)
+- [QAccessible::Cell] (The content of the cell, e.g. AXButton, AXGroup or whatever)
+- AXColumn (this seems to just store the geometry of the column)
+- AXColumn (this seems to just store the geometry of the column)
+- AXGroup (this represents the column headers)
+- AXSortButton (clicking a header cell will trigger sorting)
+- AXSortButton (clicking a header cell will trigger sorting)
It's unclear to me how RowHeaders are mapped (they are rarer than
ColumnHeaders, I expect to find them in e.g. spreadsheet applications).
I haven't found any native usage of them. So this patch simply ignores
them.
Notice that macOS have a three layer deep hierarchy to represent a table
(Table->Row->Cell), while QAccessibleInterface has a two-layer deep
hierarchy (Table->Row/Cell).
In the macOS bridge we therefore need to "inject" the Row/Column element
to be "between" the table and the cell.
The table will take ownership of all row and column elements that are
children of the table. These elements are not inserted into the cache
(it would be pointless, since the cache is basically just a mapping
between the QAccessibleInterface -> QMacAccessibilityElement, and the
row and column elements does not have a corresponding
QAccessibleInterface to be mapped from).
The rows and columns are therefore also created as soon as the table
element is initialized, and they are stored in two NSMutableArray
members of QMacAccessibilityElement.
A table is constructed like any other accessibility element, with a
valid axid and synthesizedRole set to nil.
Each child row and column element is constructed with the same axid as the
parent table element, and they will have the synthesizedRole set to
either NSAccessibilityRow or NSAccessibilityColumn.
With the synthesizedRole member we can then identify if we are a row,
column or the actual table, and implement their respective behaviors.
Notice that the child row/column is created with the parent's table axid
in order for them to have a way of finding their parent table element.
(there is no 'parent' member variable in QMacAccessibilityElement)
This glorious scheme isn't pretty, but seems to work.
Fixes: QTBUG-37207
Change-Id: I7c2451e629f5331b9a0ed61dc22c6e74a82cc173
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When comparing a glyphIndex with a hard coded number, the number is
cast to an int, whereas the glyphIndex is an unsigned int.
That causes a compiler warning.
This patch forces the numbers to be cast to an unsigned int.
Change-Id: I8a31124c6afacfc4ecfb13caf2cb8133dad44a21
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Debug serials and detatch numbers of deep and shallow detatch in test
function cacheKey, if it fails.
That implicitly removes a compiler warning about these variables being
unused.
Change-Id: I481f4b63e3ed0d50fb442dffc658b97d913059bc
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
The usage of the helpers was removed in 2011,
in bf8dfc394a.
Change-Id: I950812982148fd76bcc65c4781a144c21cb3c901
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Most of them were easy to change. The pair one was a bit of a stretch,
but still worked. I've removed the lines on QPair, since QPair is
std::pair in Qt 6.
Change-Id: I3d74c753055744deb8acfffd17246ec98c583d08
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
qIsNumericType does not return true for enum types, which meant we never
called numericCompare() or numericEquals() when one of the types was an
enum.
Task-number: QTBUG-108188
Change-Id: I3d74c753055744deb8acfffd172449c68af19367
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
The code implementing the C++ rules of type promotion and conversion
was too pedantic. There's no need to follow the letter of the standard,
not when we can now assume that everything is two's complement (this was
true for all architectures we supported when I wrote this code in 2014,
but wasn't required by the standard).
So we can reduce this to fewer comparisons and fewer rules, using the
size of the type, not just the type ID.
Change-Id: I3d74c753055744deb8acfffd172446b02444c0c0
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Generate bc file using the "new way".
Task-number:QTBUG-106331
Pick-to: 6.4
Change-Id: I70d2c0e5026636deee1b8389f50f51e075187b76
Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
qWaitForWindowActive was called in two test functions without
activating the relevant widget.
This patch adds widget activation before calling
qWaitForWindowActive.
Helper function verifyColor:
A loop made six comparison attempts of widget size, pixel color and
image with a 200ms waiting time after each unsuccessful attempt.
The widget size was tested at the beginning of the loop.
The test was failed on the first size mismatch, which occurred when
verifyColor was called before the widget was rendered.
That has lead to flakiness (e.g. on openSuSE).
This patch encapsules each check in a lambda and calls qWaitFor to
ensure event processing until each condition has become true.
Change-Id: Ic98f93c8acf41459bc728f2969fe8b01768048dd
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Follow-up to 8222e06d12, which only reset
the item repaint count.
Flushing the queued paint events will bump numRepaints, and the whole
point of calling reset() is to prepare a consistent state before the
next test, so we need to call it after flushing the events.
Pick-to: 6.2 6.4
Change-Id: Id1fe840c14c0940d7020cf8f8cc6a3aeceaa5fb5
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This patch replaces the first instance of QCOMPARE with QTRY_COMPARE /
QVERIFY with QTRY_VERIFY after a call to QWidget::grabGesture.
It re-groups verifications so that the verification of the highest
event count is on top.
The test function customGesture is skipped if
QGestureManager::deliverEvents cannot establish a target to deliver
the custom event.
Change-Id: I8188559a40ed5be86f3c6e9c82fa54a97ce5d7d6
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The insets used to calculate the correct height were not the best
choice. It used display cutout insets which would work correctly on
most devices, but in the case of an emulator or a device without a
camera, it could fail to calculate correctly.
Task-number: QTBUG-107604
Task-number: QTBUG-107709
Task-number: QTBUG-107523
Pick-to: 6.4 6.4.1 6.2 5.15
Change-Id: I8c4da83ae7359a0c133dbeb02dbd2cd260565f78
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Flushing the queued paint events will bump numRepaints, and the whole
point of calling reset() is to prepare a consistent state before the
next test, so we need to call it after flushing the events.
Pick-to: 6.2 6.4
Change-Id: Iaefc9854caafe82c65c9587e18fd081439e8dda6
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The qfiledialog test was failing on completer_data test. The fix
available for Android 11 also works in Android 7, so removed the if
clause.
Task-number: QTBUG-105377
Change-Id: I76a4c1073a754c9b299cfe731f42b80da1a6f8e2
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
It may take some time before the shown window is available through the
accessible hierarchy, so do an asynchronous test for that to happen.
Pick-to: 6.4
Change-Id: I3f312ae636505b805899973678b1bf10a65f96b3
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
There's no need to split it, we can use the -x compiler flag to build
the .cpp file as Objective-C++, or, as done here, just rename the file.
This allows us to use QVERIFY and friends, giving more precise failure reporting.
Pick-to: 6.4
Change-Id: I6fb1c4454335082c8a39010c5b75c59e6ec6561b
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
When a compound widget is created not directly before its children,
then another widget will be in the focus chain between the compound and
the compound's first child. If one of those children is then made the
focus proxy of the compound, then the widget in between becomes
unreachable by tabbing.
To fix this, detect that we set the focus proxy to be a descendent of
the compound widget, and then move the compound widget directly in front
of its first child in the focus chain. This way we can't have any gaps
in the focus chain.
Augment the test case with a corresponding scenario. As a drive-by, move
the debug helper up in the code so that it can be easier used, and set
object names on relevant widgets.
Pick-to: 6.4 6.2 5.15
Fixes: QTBUG-89156
Change-Id: I17057719a90f59629087afbd1d2ca58c1aa1d8f6
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Basic unitttest and one to verify erase returns iterator, not
const_iterator.
Change-Id: I44c3b82b4686ff3809648063376f5e36fb7e181d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
- If the string isn't shared, don't call detach(), instead remove characters
matching ch, and resize()
- If the string is shared, create a new string, and copy all characters
except the ones that would be removed, see task for details
Update unittets so that calls to this overload of remove() test both code
paths (replace() calls remove(QChar, cs) internally).
Drive-by change: use QCOMPARE() instead of QTEST()
Task-number: QTBUG-106181
Change-Id: I1fa08cf29baac2560fca62861fc4a81967b54e92
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
- If this bytearray isn't shared call d->erase() as needed
- if it's shared, instead of detaching, create a new bytearray, and copy
all characters except for the ones that would be removed
See task for details.
Adjust unittest to test both code paths.
Task-number: QTBUG-106182
Change-Id: I806e4d1707004345a2472e056905fbf675f765ab
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This reverts commit 79a11470f3, which
resulted in QTBUG-105735. The new behavior is worse and affects multiple
screen readers, while the old issue is isolated to Windows Narrator and
could be considered a narrator bug.
Task-number: QTBUG-105735
Pick-to: 6.2 6.4
Change-Id: Ic8be1dbd592a3fdf2c3219ec4c5524bc2c7f0f6a
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
This is a semantic patch using ClangTidyTransformator as in
qtbase/df9d882d41b741fef7c5beeddb0abe9d904443d8, but extended to
handle typedefs and accesses through pointers, too:
const std::string o = "object";
auto hasTypeIgnoringPointer = [](auto type) { return anyOf(hasType(type), hasType(pointsTo(type))); };
auto derivedFromAnyOfClasses = [&](ArrayRef<StringRef> classes) {
auto exprOfDeclaredType = [&](auto decl) {
return expr(hasTypeIgnoringPointer(hasUnqualifiedDesugaredType(recordType(hasDeclaration(decl))))).bind(o);
};
return exprOfDeclaredType(cxxRecordDecl(isSameOrDerivedFrom(hasAnyName(classes))));
};
auto renameMethod = [&] (ArrayRef<StringRef> classes,
StringRef from, StringRef to) {
return makeRule(cxxMemberCallExpr(on(derivedFromAnyOfClasses(classes)),
callee(cxxMethodDecl(hasName(from), parameterCountIs(0)))),
changeTo(cat(access(o, cat(to)), "()")),
cat("use '", to, "' instead of '", from, "'"));
};
renameMethod(<classes>, "count", "size");
renameMethod(<classes>, "length", "size");
except that the on() matcher has been replaced by one that doesn't
ignoreParens().
a.k.a qt-port-to-std-compatible-api V5 with config Scope: 'Container'.
Added two NOLINTNEXTLINEs in tst_qbitarray and tst_qcontiguouscache,
to avoid porting calls that explicitly test count().
Change-Id: Icfb8808c2ff4a30187e9935a51cad26987451c22
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
- If this string isn't shared, don't call detach, instead use ->erase() as
needed
- If this string is shared, create a new string, and copy all elements
except the ones that would be removed, see task for details
Update unittest to test both code paths.
Task-number: QTBUG-106181
Change-Id: I4c73ff17a6fa89ddcf6966f9c5bf789753f6d39e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When 'this' is IPv6 and 'other' is Any then there is no point in testing
'other's IPv6 address.
Added extra tests against QHostAddress::Any*.
Pick-to: 6.4 6.2 5.15
Fixes: QTBUG-108103
Change-Id: I09f32b1b147b1ec8380546c91cd89684a6bebe2e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
With the introduction of QAnyStringView, overloading based on UTF-8
and Latin-1 is becoming more common. Often, the two overloads can
share the processing backend, because we're only interested in the
US-ASCII subset of each.
But if they can't, we need a faster way to convert L1 into UTF-8 than
going via UTF-16. This is where the new private API comes in.
Eventually, we should have the converse operation, too, to complete
the set of direct conversions between the possible three
QAnyStringView encodings L1/U8/U16, but this direction is easier to
code (there are no error cases) and more immediately useful, so
provide L1->U8 alone for now.
Change-Id: I3f7e1a9c89979d0eb604cb9e42dedf3d514fca2c
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Borrowed from tst_qtemporaryfile with some changes.
Change-Id: I596ddd0ac8dbe10edd63e481198064dcec15d3e6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
qhashfunctions.h defines a catch-all 2-arguments qHash(T, seed)
in order to support datatypes that implement a 1-argument overload
of qHash (i.e. qHash(Type)). The catch-all calls the 1-argument
overload and XORs the result with the seed.
The catch-all is constrained on the existence of such a 1-argument
overload. This is done in order to make the catch-all SFINAE-friendly;
otherwise merely instantiating the catch-all would trigger a hard error.
Such an error would make it impossible to build a type trait that
detects if one can call qHash(T, size_t) for a given type T.
The constraint itself is called HasQHashSingleArgOverload and lives in a
private namespace.
It has been observed that HasQHashSingleArgOverload misbehaves for
some datatypes. For instance, HasQHashSingleArgOverload<int> is actually
false, despite qHash(123) being perfectly callable. (The second argument
of qHash(int, size_t) is defaulted, so the call *is* possible.)
--
Why is HasQHashSingleArgOverload<int> false?
This has to do with how HasQHashSingleArgOverload<T> is implemented: as
a detection trait that checks if qHash(declval<T>()) is callable.
The detection itself is not a problem. Consider this code:
template <typename T>
constexpr bool HasQHashSingleArgOverload = /* magic */;
class MyClass {};
size_t qHash(MyClass);
static_assert(HasQHashSingleArgOverload<MyClass>); // OK
Here, the static_assert passes, even if qHash(MyClass) (and MyClass
itself) were not defined at all when HasQHashSingleArgOverload was
defined.
This is nothing but 2-phase lookup at work ([temp.dep.res]): the
detection inside HasQHashSingleArgOverload takes into account the qHash
overloads available when HasQHashSingleArgOverload was declared, as well
as any other overload declared before the "point of instantiation". This
means that qHash(MyClass) will be visible and detected.
Let's try something slightly different:
template <typename T>
constexpr bool HasQHashSingleArgOverload = /* magic */;
size_t qHash(int);
static_assert(HasQHashSingleArgOverload<int>); // ERROR
This one *does not work*. How is it possible? The answer is that 2-phase
name lookup combines the names found at definition time with the names
_found at instantiation time using argument-dependent lookup only_.
`int` is a fundamental type and does not participate in ADL. In the
example, HasQHashSingleArgOverload has actually no qHash overloads to
even consider, and therefore its detection fails.
You can restore detection by moving the declaration of the qHash(int)
overload *before* the definition of HasQHashSingleArgOverload, so it's
captured at definition time:
size_t qHash(int);
template <typename T>
constexpr bool HasQHashSingleArgOverload = /* magic */;
static_assert(HasQHashSingleArgOverload<int>); // OK!
This is why HasQHashSingleArgOverload<int> is currently returning
`false`: because HasQHashSingleArgOverload is defined *before* all the
qHash(fundamental_type) overloads in qhashfunctions.h.
--
Now consider this variation of the above, where we keep the qHash(int)
overload after the detector (so, it's not found), but also prepend an
Evil class implicitly convertible from int:
struct Evil { Evil(int); };
size_t qHash(Evil);
template <typename T> constexpr bool HasQHashSingleArgOverload = /* magic */;
size_t qHash(int);
static_assert(HasQHashSingleArgOverload<int>); // OK
Now the static_assert passes. HasQHashSingleArgOverload is still not
considering qHash(int) (it's declared after), but it's considering
qHash(Evil). Can you call *that* one with an int? Yes, after a
conversion to Evil.
This is extremely fragile and likely an ODR violation (if not ODR, then
likely falls into [temp.dep.candidate/1]).
--
Does this "really matter" for a type like `int`? The answer is no. If
HasQHashSingleArgOverload<int> is true, then a call like
qHash(42, 123uz);
will have two overloads in its overloads set:
1) qHash(int, size_t)
2) qHash(T, size_t), i.e. the catch-all template. To be pedantic,
qHash<int>(const int &, size_t), that is, the instantiation of the
catch-all after template type deduction for T (= int)
([over.match.funcs.general/8]).
Although it may look like this is ambiguous as both calls have perfect
matches for the arguments, 1) is actually a better match than 2) because
it is not a template specialization ([over.match.best/2.4]).
In other words: qHash(int, size_t) is *always* called when the argument
is `int`, no matter the value of HasQHashSingleArgOverload<int>. The
catch-all template may be added or not to the overload set, but it's
a worse match anyways.
--
Now, let's consider this code:
enum MyEnum { E1, E2, E3 };
qHash(E1, 42uz);
This code compiles, although we do not define any qHash overload
specifically for enumeration types (nor one is defined by MyEnum's
author).
Which qHash overload gets called? Again there are two possible
overloads available:
1) qHash(int, size_t). E1 can be converted to `int` ([conv.prom/3]),
and this overload selected.
2) qHash(T, size_t), which after instantiation, is qHash<MyEnum>(const
MyEnum &, size_t).
In this case, 2) is a better match than 1), because it does not require
any conversion for the arguments.
Is 2) a viable overload? Unfortunately the answer here is "it depends",
because it's subject to what we've learned before: since the catch-all
is constrained by the HasQHashSingleArgOverload trait, names introduced
before the trait may exclude or include the overload.
This code:
#include <qhashfunctions.h>
enum MyEnum { E1, E2, E3 };
qHash(E1, 42uz);
static_assert(HasQHashSingleArgOverload<MyEnum>); // ERROR
will fail the static_assert. This means that only qHash(int, size_t) is
in the overload set.
However, this code:
struct Evil { Evil(int); };
size_t qHash(Evil);
#include <qhashfunctions.h>
enum MyEnum { E1, E2, E3 };
qHash(E1, 42uz);
static_assert(HasQHashSingleArgOverload<MyEnum>); // OK
will pass the static_assert. qHash(Evil) can be called with an object of
type MyEnum after an user-defined conversion sequence
([over.best.ics.general], [over.ics.user]: a standard conversion
sequence, made of a lvalue-to-rvalue conversion + a integral promotion,
followed by a conversion by constructor [class.conv.ctor]).
Therefore, HasQHashSingleArgOverload<MyEnum> is true here; the catch-all
template is added to the overload set; and it's a best match for the
qHash(E1, 42uz) call.
--
Is this a problem? **Yes**, and a huge one: the catch-all template does
not yield the same value as the qHash(int, size_t) overload. This means
that calculating hash values (e.g. QHash, QSet) will have different
results depending on include ordering!
A translation unit TU1 may have
#include <QSet>
#include <Evil>
QSet<MyEnum> calculateSet { /* ... */ }
And another translation unit TU2 may have
#include <Evil>
#include <QSet> // different order
void use() {
QSet<MyEnum> set = calculateSet();
}
And now the two TUs cannot exchange QHash/QSet objects as they would
hash the contents differently.
--
`Evil` actually exists in Qt. The bug report specifies QKeySequence,
which has an implicit constructor from int, but one can concoct infinite
other examples.
--
Congratulations if you've read so far.
=========================
=== PROPOSED SOLUTION ===
=========================
1) Move the HasQHashSingleArgOverload detection after declaring the
overloads for all the fundamental types (which we already do anyways).
This means that HasQHashSingleArgOverload<fundamental_type> will now
be true. It also means that the catch-all becomes available for all
fundamental types, but as discussed before, for all of them we have
better matches anyways.
2) For unscoped enumeration types, this means however an ABI break: the
catch-all template becomes always the best match. Code compiled before
this change would call qHash(int, size_t), and code compiled after this
change would call the catch-all qHash<Enum>(Enum, size_t); as discussed
before, the two don't yield the same results, so mixing old code and new
code will break.
In order to restore the old behavior, add a qHash overload for
enumeration types that forwards the implementation to the integer
overloads (using qToUnderlying¹).
(Here I'm considering the "old", correct behavior the one that one gets
by simply including QHash/QSet, declaring an enumeration and calling
qHash on it. In other words, without having Evil around before including
QHash.)
This avoids an ABI break for most enumeration types, for which one
does not explicitly define a qHash overload. It however *introduces*
an ABI break for enumeration types for which there is a single-argument
qHash(E) overload. This is because
- before this change, the catch-all template was called, and that
in turn called qHash(E) and XOR'ed the result with the seed;
- after this change, the newly introduced qHash overload for
enumerations gets called. It's very likely that it would not give
the same result as before.
I don't have a solution for this, so we'll have to accept the ABI
break.
Note that if one defines a two-arguments overload for an enum type,
then nothing changes there (the overload is still the best match).
3) Make plans to kill the catch-all template, for Qt 7.0 at the latest.
We've asked users to provide a two-args qHash overload for a very long
time, it's time to stop working around that.
4) Make plans to switch from overloading qHash to specializing std::hash
(or equivalent). Specializations don't overload, and we'd get rid of
all these troubles with implicit conversions.
--
¹ To nitpick, qToUnderlying may select a *different* overload than
the one selected by an implicit conversion.
That's because an unscoped enumeration without a fixed underlying type
is allowed to have an underlying type U, and implicitly convert to V,
with U and V being two different types (!).
U is "an integral type that can represent all the enumerator values"
([dcl.enum/7]). V is selected in a specific list in a specific order
([conv.prom]/3). This means that in theory a compiler can take enum E {
E1, E2 }, give it `unsigned long long` as underlying type, and still
allow for a conversion to `int`.
As far as I know, no compiler we use does something as crazy as that,
but if it's a concern, it needs to be fixed.
[ChangeLog][Deprecation Notice] Support for overloads of qHash with only
one argument is going to be removed in Qt 7. Users are encouraged to
upgrade to the two-arguments overload. Please refer to the QHash
documentation for more information.
[ChangeLog][Potentially Binary-Incompatible Changes] If an enumeration
type for which a single-argument qHash overload has been declared is
being used as a key type in QHash, QMultiHash or QSet, then objects of
these types are no longer binary compatible with code compiled against
an earlier version of Qt. It is very unlikely that such qHash overloads
exist, because enumeration types work out of the box as keys Qt
unordered associative containers; users do not need to define qHash
overloads for their custom enumerations. Note that there is no binary
incompatibity if a *two* arguments qHash overload has been declared
instead.
Fixes: QTBUG-108032
Fixes: QTBUG-107033
Pick-to: 6.2 6.4
Change-Id: I2ebffb2820c553e5fdc3a341019433793a58e3ab
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If we use winrt's factories we have to make sure to to clear the factory
cache when one of our dlls is unloaded or we will run into dangling
factory entries which might result in crashes. So we have to make sure
that winrt::clear_factory_cache is called on every dll unload.
In order not to increase compile times and dependencies too much
qfactorycacheregistration_p.h needs to be included in Qt code whenever
we use winrt's factory cache. A rule of thumb being: Include
qfactorycacheregistration_p.h whenever including winrt/base.h.
Other Qt modules which use winrt's factories need to be updated too.
Fixes: QTBUG-103611
Pick-to: 6.2 6.4
Change-Id: I7ab24e4b18bffaca653c5b7f56a66ce99212e339
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Some of the entries in QLocale's single_character_data[] table are
not, in fact, single characters; some RTL languages include
bidi-markers in some of the fields, some locales use some denotation
of "times ten to the power" as the exponent separator. There may be
further complications, but let's just get some tests in that verify we
are correctly serializing numbers in these locales. Include some
parsing tests to show that we are indeed failing them.
Task-number: QTBUG-107801
Change-Id: Iab9bfcea5fdcfcb991451920c9531e0e67d02913
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Ievgenii Meshcheriakov <ievgenii.meshcheriakov@qt.io>
%.f should be handled like %.0f. You probably don't want it for strings,
though.
Fixes: QTBUG-107991
Pick-to: 6.2 6.4
Change-Id: I07ec23f3cb174fb197c3fffd1721a941fbcf15e1
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
I'm not entirely sure whether this is a toolchain bug or if this is
intended. This commit ODR-uses all the static inline variables in
QOperatingSystemVersion so they are added to the list of exported
symbols in QtCore.
On Windows:
$ objdump -p bin/Qt6Core.dll | grep Windows11E
[2534] _ZN23QOperatingSystemVersion9Windows11E
On Linux:
$ eu-readelf --dyn-syms lib/libQt6Core.so | grep Windows11E
1985: 0000000000575430 16 OBJECT GNU_UNIQUE PROTECTED 18 _ZN23QOperatingSystemVersion9Windows11E@@Qt_6
Pick-to: 6.4
Change-Id: Ia317fd249bcd80dbd02c198803a3a61178c0c219
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
It seems the value name correction is not needed at all,
and we must not do such correction.
Amends commit 738e05a55a
Task-number: QTBUG-107794
Change-Id: I903a762aafab4b55275beb8438e6769285821567
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
This amends 23780891a5 which moved the.txt
test to tets/auto/corelib/platform/android and kept the old location
mistakenly.
Pick-to: 6.4 6.2
Change-Id: If58422f9a94cfe4d6a941cc5453d8f0506057dcb
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Removes a warning in the build.
Pick-to: 6.4
Change-Id: I07ec23f3cb174fb197c3fffd17215c40b40333cb
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The `rcc_PREFIX` will be set to `/` if it is not passed to the
function and it is not defined in `QT_RESOURCE_PREFIX`.
I also removed an unnecessary check of the `rcc_PREFIX`.
[ChangeLog][QtCore][CMake] The `PREFIX` parameter of the
`qt_add_resources` is now optional. If not passed, and
`QT_RESOURCE_PREFIX` is not defined, `/` will be used as the path
prefix.
Fixes: QTBUG-104938
Change-Id: I6524ab5dc54f035272e4c2e3154eb67591efb650
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
This introduces a new helper function accessibleIdForAccessible
(inspired by Windows' automationIdForAccessible) to synthesize an id out
of the objectNames of the accessible parent chain. The id is then
exposed via the GetAccessibleId D-Bus function for consumption in a11y
tools.
Change-Id: If72b86c5864c43f4ca842aa11423dd8aea0dde4a
Reviewed-by: Aleix Pol Gonzalez <aleixpol@kde.org>
MySQL has a different default setting for case sensitive table names on
linux and windows which makes the test fail on linux but work on
windows when using the database with the default settings. Read out the
respecitive setting so the test will pass every time.
Change-Id: I8651858d47652022ddc4b6386a6153cf70c6fed6
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The QHoverEvent ctor takes two points: pos and globalPos; pos is then
passed as both the scene and global pos to the QSinglePointEvent ctor,
which calls QMutableEventPoint::setScenePosition() on the persistent
QEventPoint instance and then detaches befeore setting ephemeral state.
Therefore, we must construct QHoverEvent with scene position first, not
local position, so that the right value is persisted; it's better to set
local position after the detach(), whereas it's too late to fix the
persistent point then.
Pick-to: 6.4
Fixes: QTBUG-106918
Change-Id: I45726a9ec05bba2fe0cde6f5fb87c269105caca6
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
-1 is the default value for QSurfaceFormat::alphaBufferSize. Changing it
to 0 may result an unexpected pixel format change by ARB OpenGL
extension.
Pick-to: 6.4
Fixes: QTBUG-107629
Change-Id: Ia6a1b90fba6c43b6872b406f4fd946d937135cf8
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
There is no reason to use different style of quotes when printing
messages.
Change-Id: I7d513ec04c803702974054569d28f26947942fbf
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Amends edd983071e.
Remove redundant calls to AAssetDir_getNextFileName() in
AndroidAbstractFileEngine::setFileName(). It's enough to check
if AAssetManager_open() returns null and AAssetManager_openDir() is
valid for the provided asset file name.
As part of this fix, add some unit tests to cover/ensure assets
listing/iterating works as expected.
Fixes: QTBUG-107627
Pick-to: 6.4 6.2 5.15
Change-Id: I37ae9cb64fbbc60699bb748895f77fd6a34fae1f
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
QWindow::requestActivate() is not supported.
This function crashed very often in ci/coin when system is busy.
Task-number: QTBUG-107153
Pick-to: 6.4 6.2
Change-Id: I82523080db40bddce9c9dc000433117d8ef74847
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
QWindow::requestActivate() is not supported.
This function failed in test vm in coin manually very often.
Task-number: QTBUG-107153
Pick-to: 6.4 6.2
Change-Id: I013651b0e5e7618c29742effd85a091ca95a7414
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
To allow the user to customize the C++ code that QDoc sees, so as to be
able to work-around some limitations on QDoc itself, QDoc defines two
symbols: Q_QDOC and Q_CLANG_QDOC, both of which are "true" during an
entire execution of QDoc.
At a certain point in time, QDoc allowed the user the choice between a
custom C++ parser and a Clang based one.
The Q_QDOC symbol would always be defined while the Q_CLANG_QDOC symbol
would be defined only when the Clang based parser was chosen.
In more recent times, QDoc always uses a Clang based parser, such that
both Q_CLANG_QDOC and Q_QDOC are always defined, making them equivalent.
To avoid using different symbols, and the possible confusion and
fragmentation that derives from it, all usages of Q_CLANG_QDOC are now
replaced by the equivalent usages of Q_QDOC.
Change-Id: I5810abb9ad1016a4c5bbea99acd03381b8514b3f
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
I wrongly assumed we can't query a value with an empty name ""
during the previous refactor commit, however, in Windows registry,
an empty name for a value means the default value of a key, we can
read and write it through the "Default" name.
Remove the wrong assert to fix the crash when we are trying to query
a default value of a key.
Add a new test case to test this kind of scenarios.
Amends commit 40523b68c1
Fixes: QTBUG-107794
Change-Id: Idacbcb86df4435a8c1ca1c19121599390ae8f3d3
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The libraryMap only stored the file path, so we couldn't load two
versions of the same library as we'd find the other version already
loaded. Change the map to index by file name and version (using a NUL as
separator, since that can't appear in file names).
[ChangeLog][QtCore][QLibrary] Fixed a bug that caused QLibrary to be
unable to load two different versions of a library of a given name at the
same time. Note that this is often inadviseable and loading the second
library may cause a crash.
Pick-to: 6.4
Change-Id: I12a088d1ae424825abd3fffd171ce3bb0590978d
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
According to the grafana dashboard it has not failed in the
past week.
Reverts ad736e9150
Pick-to: 6.4 6.2
Change-Id: I3eac3c7fd667cfe2cf951b2808dddbfed8eae087
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Regression introduced by commit 8d4eb292b2
in 6.0, when QTaggedPointer was introduced. We set the tag even when the
loading failed and failed to reset it because d = {} retains the tag.
Pick-to: 6.2 6.4
Fixes: QTBUG-103387
Change-Id: Ie4bb662dcb274440ab8bfffd170a07aa9c9ecfca
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
QLibrary intentionally does not unload on destruction, so failing tests
may leave libraries already loaded and cause further tests to fail
because of that. So add a cleanup() method to unload everything we may
have loaded.
Note that QLibrary::unload() sets its state to NotLoaded after one
successful call, so we must recreate the object in case it had been
load()ed multiple times.
Pick-to: 6.2 6.4
Change-Id: I12a088d1ae424825abd3fffd171d133c678f910a
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
The "qhash" test relied on the fact that those four elements would
produce a different order with a zero and a non-zero seed. But since
commit b057e32dc4 removed the setting of a
deterministic non-zero seed, this test had a 1 in 4! chance of failing.
Since 4! = 24, 128 retries should be more than enough to ensure we do
find at least hash seed that provokes a different order.
Fixes: QTBUG-107725
Change-Id: I3c79b7e08fa346988dfefffd171ee61b79ca5489
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
It makes it impossible to rerun it, bad for both CI and local test runs.
As a drive-by name the database file sqlite.db instead of foo.db
Pick-to: 6.2 6.4
Fixes: QTBUG-100245
Change-Id: I2e4ee01189ccbad2a6add5db7771d35fd7248da8
Reviewed-by: Dimitrios Apostolou <jimis@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Various places were knowingly provoking warnings without telling QTest
to check for and suppress those warnings. Some others did check for
this warning, so let's consistently suppress the noise.
Change-Id: I71b9829680c7a513f4d8fbb3c57442875a6c2dc4
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
For some reason the QTest::ignoreMessage() was conditioned on the type
being tested being Array; however, the warning is in fact produced for
all types. So anticipate it for all and make the test log less noisy.
Change-Id: I78681624252ff8a71f080204f8b031609ddac468
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There were two copies of the 0x1D157 row and we can't remember why.
So change one of them to the Chakma digit 3 (a spiral) and annote all
three test-cses with what meaning Unicode assigns to them.
Change-Id: I95837588bd5944f7f2c39c8438d9076e844e4dd0
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Two test-cases had the same name; distinguish them by the part of
their data in which they differ - one closes, the other doesn't.
Change-Id: I37051baf194bf8df742688739ad01e3335e64dc7
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Two rows shared the same name. They claimed the value used was out of
range, but actually that was only true for one of them. The other was
in range, but the test reduced the number of digits allowed after the
decimal point, thereby making it invalid, so rename that one to
reflect this.
Change-Id: I0936ea25ec799c0069cd148b9f9bae5d35906093
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Include the spacings used, to avoid a naming collision.
Change-Id: Iaf78f7142f6780dcf4c7a0b973db9f625af06767
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Avoid duplication by distinguishing similar test-cases.
Change-Id: I1a100d6c9729f0ea356f177535d15c3d36e2da9e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The test used to trigger a lot of QWARN messages; these are clearly
intended, so tell QTest to expect them, so that we get an error here
if those warnings ever don't show up.
Incidentally tidy up a comment and convert a != verify to a
QCOMPARE_NE(), since it's now available to do that job.
Change-Id: I83e225c37abe8446dac06ebe4e75258cb87b71b0
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
In the process, clean up the building of the data tags: use a
range-for loop, albeit we do need an index to show in tags; show it
and the angle in the tags using addRow()'s easier formatting. Change
the low angle tests to show the sign of the angle (which is how they
differ)rather than just labeling them 1 and 2.
Change-Id: Ib5aaa3e22d771c530c9343ba368b0fdfceb264ce
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Jason McDonald <macadder1@gmail.com>
The last three duplicated earlier ones; and their names didn't take
into account the circle that had been added to the path since those
tests. So revise their names to reflect that.
Change-Id: I32d74f21947b4ba0c04eee53daf8efde6b4a6409
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Two rows shared a data tag. Prefixed one of them with the color of the
half-transparent image it involves (the other's is fully transparent).
Change-Id: I1bd174008ed29bcf2f460e683fdf6d1f12ba19d0
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The same data tags were used with two distinct filenames in their
data. Include the basename of each filename in the data tag, to avoid
duplication.
Change-Id: I216fecbd413fab409227ad6f93f8ac3fcc74b059
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
The "a" and "A" rows appeared under Valid, then again under Only Keys.
The two copies were identical, in each case, so drop the latter.
Change-Id: Ib3d84710e772171bb4a5e0aefd20022810fb41cd
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
tst_QDBusType::isValidFixedType_data() called addFixedTypes() and then
addBasicTypes(); but the latter calls addFixedTypes(), too; so those
rows got duplicated. Only add the fixed types once.
Change-Id: If0d6f44ec7defb12117dad251878850ca75beb48
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The row 27 that was positioned before row 01, as if it were meant to
be numbered row 00, was identical to the row 27 that appeared after
row 26. Since row 26 was the other case dealing with the null
QRectF(), I kept the one after it instead of renumbering row 00 and
deleting row 27.
Change-Id: I3585839184233f1f1629280ac9e5b25110c155c0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use key(i) rather than valueToKey(value) as the Sha3_* alias Kekkak_*
or RealSha3_*. This way, we still test all members of the enum,
without duplicating row keys (albeit the aliases duplicate values).
Change-Id: I6acba5ffdf5b68294031d609a76b37ca8fad9d94
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Both countBits() and datastream() had two copies of an all-zeros test
with 35 zeros. Removed the second, in each case.
Change-Id: I5dec4765236ae870c30828dae0f04b8902a100f0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
/Zc:lambda seems buggy. Although in my experiments it works well
for 99% Qt repos, it seems some tests will trigger the bug and it
also blocks some new commits. So disable it for now, it's not stable
enough.
Now that this check is disabled, the workaround for tst_qstringapisymmetry
is also not needed anymore, so remove the workaround as well.
Partially reverts commit 8cb832090a
Change-Id: Icf0ecbbaa6262522470e5f5dea05705985ab18f1
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
setFocus() was called on a double spinbox without calling show()
first. That causes flakiness on XCB when checking focus afterwards.
The test can still fail, when focus is acquired by e.g. a system
popup.
This patch adds a show() call before setFocus() to stabilize normal
behavior. In case the double spin box is shown, but cannot acquire
focus, the test is skipped.
Fixes: QTBUG-70088
Change-Id: If02e88800a31b09a1da63dcc074eb8bb1b0df391
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
The zlib convenience API we've been using so far has two problems:
- On Windows-64, where sizeof(long) == 4, the use of ulong for sizes
meant that we could not compress data compressable on other 64-bit
platforms (Unix). While zstream also uses ulong, being a stream API,
it allows feeding data in chunks. The total_in and total_out members
are only required for gzip compression and are otherwise just
informational. They're unsigned, so their overflow does not cause
UB. In summary, using zstream + deflate() allows us to compress more
than 4GiB of data even on Windows-64.
- On all platforms, we always allocated the output buffer in such a
way as to accommodate the pathological case of random, incompressible
data, so the output buffer was larger than the input. Using zstream
+ deflate(), we can start with a smaller buffer, then let zlib pick
up where it left off when it ran out of output buffer space, saving
memory in the common case that compression meaningfully reduces the
size. To avoid the first few rounds of reallocations, we continue to
use zlib's compressBound() for input less than 256KiB.
This completely fixes the compression side of QTBUG-106542 and
QTBUG-104972.
Pick-to: 6.4 6.3 6.2
Fixes: QTBUG-104972
Fixes: QTBUG-106542
Change-Id: Ia7e6c38403906b35462480fd611b482f05a5c59c
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Add at least a few, so size() isn't completely untested.
Pick-to: 6.4 6.2 5.15
Change-Id: I500d28f7efb30ab578808d8fefb6ea57949edc2e
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
A violation of coding style (requiring braces on multi-line bodies
of conditionals) was accompanied by a mis-indented else block.
Fix a long line while I'm about it.
Change-Id: Ibe9cf15eadbe9ef58138d7876e5e2c5a14a92fd4
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Pull out the arbitrary factor of three as a named constant and
document its arbitrariness once.
Pull out the mask and bit used in each function's loop to the outer
layer of the loop, since they don't depend on the inner loop variable
(or the random value generated in that loop).
Use QTest::addRow() instead of constructing a string to pass to
newRow().
Change-Id: Ifacbcb390e00828fd47f51b0c73d0ad5f6bc8bdb
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The tests for indexOf() and lastIndexOf() had duplicate data row tags,
due to only using the needle and haystack, although some tests
differed only in start position. Include start position where needed.
Change-Id: I197d415265ab1a805f2d36fb88aec92ea8646f7a
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Enclosing one string in each substring of another does not need to
repeat the empty substring of the latter. Extracting the empty
substring from different positions doesn't get different results.
In the process, tidy up the code a bit.
Change-Id: Ic66febbdadeaac0c466f4f1174d831a991d31e20
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
There were two copies of the same line in mid_data(), leading to
duplicated data row tags.
Change-Id: Ia21e855ff781b13fe18c932cff48cb0aabd12750
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The zlib convenience API we've been using so far has two problems:
- On Windows-64, where sizeof(long) == 4, the use of ulong for sizes
meant that we could not uncompress data compressed on other 64-bit
platforms (Unix). While zstream also uses ulong, being a stream API,
it allows feeding data in chunks. The total_in and total_out members
are only required for gzip compression and are otherwise just
informational. They're unsigned, so their overflow does not cause
UB. In summary, using zstream + inflate() allows us to decompress
more than 4GiB of data even on Windows-64.
- On all platforms, if the size hint in the header was too short, we'd
double the output buffer size and try again, from scratch. Using
zstream + inflate(), we still need to reallocate, but we can then
let zlib pick up where it left off when it ran out of output buffer
space. In all but the most pathological cases, copying the
already-decoded data instead of re-decoding it again should be
faster, esp. if QArrayData uses realloc() instead of malloc() +
free() to grow the buffer.
We also now directly allocate at least as much output buffer as we
have input, to cut the first few rounds of reallocations when the
expectedSize was created, as qCompress still does, using modulo
arithmetic mod 4GiB instead of saturation arithmethic.
Factor the growing of the output buffer into a wrapper function,
flate(), which can be reused when porting qCompress().
This completely fixes the uncompression side of QTBUG-106542 and
QTBUG-104972.
Pick-to: 6.4 6.3 6.2
Task-number: QTBUG-104972
Task-number: QTBUG-106542
Change-Id: I97f55ea322c24db1ac48b31c16855bc91708e7e2
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The timestamp will no longer be incremented by 500ms after a mouse
release if the delay has been explicitly specified.
The default delay is 1 ms since f5010c49a3
but the running timestamp was unconditionally post-incremented by 500ms
after every mouse release, to prevent double-clicks, which were always
deemed as unintended (because we have a mouseDClick function for that).
Now, we do that 500ms increment only if the user has not provided a
delay value in the function argument at all. We have often found it
useful in our own tests to generate double-clicks "the hard way", by
sending indivdual events, so as to be able to check state in some target
object at each step, as shown in the new snippet.
[ChangeLog][QtTest] QTest::mouseRelease() and mouseClick() can now be
used to test double-clicks, by specifying a realistic timestamp delay.
Fixes: QTBUG-102441
Change-Id: I8e8d242061f79efb4c6e02638645e03661a9cd92
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Fix two tests which got broken due to the latest changes without
notifying because those tests are not run automatically.
Change-Id: Ibe9d9601f0a2ad4ce8f06ca21e7503e77fa55781
Reviewed-by: Fredrik Ålund <fredrik.alund@mimer.com>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This is a combination of Q_UNREACHABLE() with a return statement.
ATM, the return statement is unconditionally included. If we notice
that some compilers warn about return after __builtin_unreachable(),
then we can map Q_UNREACHABLE_RETURN(...) to Q_UNREACHABLE() without
having to touch all the code that uses explicit Q_UNREACHABLE() +
return.
The fact that Boost has BOOST_UNREACHABLE_RETURN() indicates that
there are compilers that complain about a lack of return after
Q_UNREACHABLE (we know that MSVC, ICC, and GHS are among them), as
well as compilers that complained about a return being present
(Coverity). Take this opportunity to properly adapt to Coverity, by
leaving out the return statement on this compiler.
Apply the macro around the code base, using a clang-tidy transformer
rule:
const std::string unr = "unr", val = "val", ret = "ret";
auto makeUnreachableReturn = cat("Q_UNREACHABLE_RETURN(",
ifBound(val, cat(node(val)), cat("")),
")");
auto ignoringSwitchCases = [](auto stmt) {
return anyOf(stmt, switchCase(subStmt(stmt)));
};
makeRule(
stmt(ignoringSwitchCases(stmt(isExpandedFromMacro("Q_UNREACHABLE")).bind(unr)),
nextStmt(returnStmt(optionally(hasReturnValue(expr().bind(val)))).bind(ret))),
{changeTo(node(unr), cat(makeUnreachableReturn,
";")), // TODO: why is the ; lost w/o this?
changeTo(node(ret), cat(""))},
cat("use ", makeUnreachableReturn))
);
where nextStmt() is copied from some upstream clang-tidy check's
private implementation and subStmt() is a private matcher that gives
access to SwitchCase's SubStmt.
A.k.a. qt-use-unreachable-return.
There were some false positives, suppressed them with NOLINTNEXTLINE.
They're not really false positiives, it's just that Clang sees the
world in one way and if conditonal compilation (#if) differs for other
compilers, Clang doesn't know better. This is an artifact of matching
two consecutive statements.
I haven't figured out how to remove the empty line left by the
deletion of the return statement, if it, indeed, was on a separate
line, so post-processed the patch to remove all the lines matching
^\+ *$ from the diff:
git commit -am meep
git reset --hard HEAD^
git diff HEAD..HEAD@{1} | sed '/^\+ *$/d' | recountdiff - | patch -p1
[ChangeLog][QtCore][QtAssert] Added Q_UNREACHABLE_RETURN() macro.
Change-Id: I9782939f16091c964f25b7826e1c0dbd13a71305
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
For the full list, please refer to [1].
Needed to change the qstringapisymmetry unit test:
In theory we don't need the array to be static and it did compile
without any problems so far, indeed. However, with this patch applied,
MSVC complains that the lambda function below can't access the array.
I don't understand why, because we use [&] in the lambda and it should
capture all the variables in theory, but in reality it failed to
capture this variable in the end. And making the variable static
solves this issue. Maybe it's a MSVC bug.
Already tested locally. Most Qt repos build without any issues,
only very few repos are not tested, as my local environment
can't build them.
[1] https://docs.microsoft.com/en-us/cpp/build/reference/zc-conformance?view=msvc-170
Change-Id: I658427aa171ee1ae26610d0c68640b2f50789f15
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
When deploying into some directory structure where CMAKE_INSTALL_LIBDIR
is different from Qt's lib dir, we need to set the RPATH of installed
plugins such that Qt libraries are found.
We do this using CMake's undocumented file(RPATH_SET) command and pray
that this command is safe to use across current and future CMake
versions. For CMake versions < 3.21, we use patchelf, which must be
installed on the host system.
The adjustment of rpaths can be turned on explicitly by setting
QT_DEPLOY_FORCE_ADJUST_RPATHS to ON.
The usage of patchelf can be forced by setting QT_DEPLOY_USE_PATCHELF to
ON regardless of the CMake version.
Change-Id: I62ced496b4c12bf6d46735d2af7ff35130148acb
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Otherwise we don't properly test whether the deployed executable can run
without adjusting the environment.
We temporarily adjust the test_widgets_app_deployment test and set
CMAKE_INSTALL_LIBDIR to make the test pass. It would now fail on Linux
distros where CMAKE_INSTALL_LIBDIR defaults to "lib64" but Qt is built
with lib dir "lib". The next commit removes this hack.
Change-Id: I63c79ef1ee23ffaeed881337fde6e9d889ecc0fe
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Unix systems have got crash loggers in the past 15-20 years, notably
macOS and Linux (abrtd, systemd-coredumpd, etc.). By setting the core
dump limit to zero, those tools should be mostly inhibited from running
and thus not interfere with the parent process' timeouts. Even for
systems without core dump loggers, disabling the writing of a core dump
to the filesystem should also help.
Pick-to: 6.4
Change-Id: I12a088d1ae424825abd3fffd171d112d0671effe
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
QWindow::requestActivate() is not supported.
We have one tst_selftests binary, and will test it with both xcb and
wayland qpa plugin. A runtime check and skip will have different
restult files, which is not implemented in testlib yet.
Task-number: QTBUG-107578
Pick-to: 6.4 6.2
Change-Id: Idc8cb24c6f42a9f0f4dc9493e3fd1a5803ba7ce0
Reviewed-by: Liang Qi <liang.qi@qt.io>
Allocate the QSplitter on the stack so that it and its child widgets are
cleaned up when the test function finishes.
As a drive-by, replace QString usage with QByteArray to avoid unneeded
conversion from and to latin1, and modernize list construction and for loop.
Pick-to: 6.4 6.2
Change-Id: I2e29961edbab1ec88be356fca6bc100f08894e82
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
In Qt 6, after changes such as 121fddcf5a,
we go through the QPA layer to close widget windows properly. Closing
and hiding of windows is now done in when we receive and handle the
window system's CloseEvent.
Such an event to a modally blocked window should be blocked, so that
users can't close a modally blocked window. However, if the event is the
result of a call to QWindow::close, then it should not be blocked.
Luckily, we know that the event is the result of such a call, so let
such events through. This restores compatibility with Qt 5, where it was
possible to first open a new dialog, and then close the previous dialog.
Add a test case.
Fixes: QTBUG-107188
Pick-to: 6.4 6.2
Change-Id: Id812c1fc36aa0e1a10dfb8d3a16a11d387289b05
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
We've been requiring C++17 since Qt 6.0, and our qAsConst use finally
starts to bother us (QTBUG-99313), so time to port away from it
now.
Since qAsConst has exactly the same semantics as std::as_const (down
to rvalue treatment, constexpr'ness and noexcept'ness), there's really
nothing more to it than a global search-and-replace, with manual
unstaging of the actual definition and documentation in dist/,
src/corelib/doc/ and src/corelib/global/.
Task-number: QTBUG-99313
Change-Id: I4c7114444a325ad4e62d0fcbfd347d2bbfb21541
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
Just in case the same test is being run in parallel. We do that by
creating a listening TCP server in the test process. This test is
supposed to test the address reusability, so a clean close on a server
that never accepted a connection should not cause reusability issues.
Change-Id: I12a088d1ae424825abd3fffd171ccfb9fc5c09ee
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
To try to figure out why QProcess::waitForReadyRead is returning false
so quickly. Though we know it's going to be "Address in use".
FAIL : tst_QTcpServer::addressReusable(WithoutProxy) 'process.waitForReadyRead(5000)' returned FALSE. (Failed to listen: The bound address is already in use
Netid State Recv-Q Send-Q Local Address:Port Peer Address:PortProcess
tcp LISTEN 0 50 127.0.0.1:49199 0.0.0.0:* users:(("crashingServer",pid=40529,fd=4))
)
Pick-to: 6.4 6.2
Change-Id: Ic43a460bfc7c7eb6379405b7a1a064e502b6fef3
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
This corresponds to Unicode version 15.0.0.
Added the following scripts:
* Kawi
* Nag Mundari
Full support of these scripts requires harfbuzz version 5.2.0,
this version adds support for Unicode 15.0:
https://github.com/harfbuzz/harfbuzz/releases/tag/5.2.0
Fixes: QTBUG-106810
Change-Id: Ib06c526e49b0f01ef9f21123bcf875c6b19f2601
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
There were two data8 rows; and no data9, so that was easy to fix.
Change-Id: I8191de142e1a3be57bf1ad97e63d5780f2859fea
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Two test cases were called "base 2, negative"; one of them use -1 as
value, so s/negative/minus 1/ for it.
Change-Id: Ia5da3952d93976262cc8423d4e75ec19dab9a088
Reviewed-by: Mate Barany <mate.barany@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Using simply the pattern didn't work so well when some patters are
used repeatedly, on different haystacks. So include the haystack
in the tag name. Remove one straight up duplicate row.
Change-Id: Ib46364581f23c493e83d75e6d04ab09e4329a3a5
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
One "empty" test was base ten, the other left the code to work out the
base. Change the latter's name to reflect that difference.
Change-Id: I4918eb0d293420df315d86e532787950b8f05be8
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
The addCommonCborData() helper had two identical rows named simple255.
It only needs one.
Change-Id: Ie934c31f373069788c3ef774fde8956b54814e67
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Two of the uuidA test cases had an open-brace for the string and no
close; one of them ended with a space (which, apparently, is valid).
Since the data-tag was constructed by formatting the string in a
fixed-width field, padding with spaces, these two cases coincided.
Fortunately the only uuidB test-case had closing as well as opening
braces, so we can just switch the test for "trailing space is not an
error" to use it, instead.
Change-Id: I7068d40145c6b6b3b72777b029282850b1d1ea81
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Mate Barany <mate.barany@qt.io>
The first "test1 text" test-case related to a file called test1.txt;
but the second related to a file called test2.txt; I suspect a
copy-and-paste with incomplete post-edit. In any case, change the
latter's data tag to reflect the difference.
Change-Id: I8354a3d1bd18715d6717dfd0962aa70faefbee90
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The test was using the same tags twice each, giving no clue to the
difference between the two test-cases for each.
Change-Id: I645b01c0c4008a766e505047cb05cc22640ee129
Reviewed-by: Dimitrios Apostolou <jimis@qt.io>
It's not clear why this test repeats each test-case five times, but
give the duplicates distinct names, at least.
Change-Id: I4a098d90c3fe6f61842745c1d5f62047fe13a9b5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
