Do not downscale the icon used for QSystemTrayIcon even though the msdn
documentation states that SM_CXICON/SM_CYICON should be used as the
size to avoid blurry icons in the notification.
Pick-to: 6.7 6.6
Fixes: QTBUG-103825
Change-Id: I2d466e895254aa3c41682290cc7864632c57d5d7
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Wrap the arguments to \inlineimage in curly braces so that the
describing image text doesn't show on the page.
Fixes: QTBUG-120572
Pick-to: 6.7 6.6 6.5 6.2
Change-Id: I00f4befc64c20fb6de1dd5ddcb6ca2573fa7c355
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Could not connect "org.freedesktop.NetworkManager" to "stateChanged" : Type not registered with QtDBus in parameter list: QNetworkManagerInterface::NMState
Could not connect "org.freedesktop.NetworkManager" to "connectivityChanged" : Type not registered with QtDBus in parameter list: QNetworkManagerInterface::NMConnectivityState
Could not connect "org.freedesktop.NetworkManager" to "deviceTypeChanged" : Type not registered with QtDBus in parameter list: QNetworkManagerInterface::NMDeviceType
Could not connect "org.freedesktop.NetworkManager" to "meteredChanged" : Type not registered with QtDBus in parameter list: QNetworkManagerInterface::NMMetered
These came from the fact that QDBus reimplements connectNotify() to
detect connections to a dbus interface's signals. It was triggered
also when connecting to "normal" signals from the dbus interface
subclass QNetworkManagerInterface, leading to those (new) warnings.
The fix in this commit replaces signals with direct method calls,
given that the two classes are internal and that QNetworkManagerInterface
is only used by QNetworkManagerNetworkInformationBackend (and now
vice-versa too).
Pick-to: 6.7
Change-Id: Ifef3f65ab8e4e21a2f7fdba7fa5f0fc566153649
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(in preparation for the next commit)
Pick-to: 6.7
Change-Id: I1a6771dc953540dfa63cb80306a48122e97eb600
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
Without this change, while using EglFS, showIsFullscreen returns false,
which is not true, because EglFS always runs in fullscreen.
This change requires setting QT_QPA_PLATFORM_PLUGIN_PATH
environmental variable.
Pick-to: 6.7
Task-number: QTBUG-115777
Change-Id: Ie7cf709e002ac70d95fe5e2698d2ee7b52e08094
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
This test is resizing window, which is not supported in EGLFS,
so it needs to be skipped.
Pick-to: 6.7
Task-number: QTBUG-115777
Change-Id: Ie136f6565b4d466bc62c4e2ebe2337488b5003b4
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
e5d591a0d0 focused on document or tree URIs you'd get from the file
dialog, but we can also get ones via Intents from application-specific
content providers (e.g. opening an attachment from an email client).
Those need to be handled like a single document here, parsing them
like a tree URI results in a bunch of Java exceptions down the line.
Change-Id: I19cb1075fd6a3ccad927543f8bc5a63566f4877e
Pick-to: 6.7 6.6 6.5
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
We used deleteLater(), which was triggering ASAN use-after-free error.
Apparently, what could happen is that after the context was destroyed,
we called deleteLater(), but if at this point the previous future got
finished, we still tried to emit watcher->run() to execute the
continuation. And then the watcher got deleted.
This patch replaces deleteLater() with a plain delete call. This looks
safe, because the watcher is only accessed while holding the lock.
Amends 59e21a536f.
Fixes: QTBUG-120302
Pick-to: 6.7 6.6
Change-Id: Ia32f20bfe8daea2e2346f3d446c978ae305d2f68
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
It looks like resource object targets don't need AUTOMOC as well
as other AUTOGEN tools. The only source file skips autogen explicitly,
but AUTOMOC requirement is added to the resource object targets anyway.
This disables all AUTOGEN related properties explicitly for all
resource object targets.
The problem occurs when building top-level Qt with the static library
that uses resources in tree.
Pick-to: 6.2 6.5 6.6 6.7
Change-Id: I5c968a3e34c881e7fbc5589626ab02b3b1e53968
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
getOpenGileContent() takes a parent now, and the compat
API is behind a QT_WIDGETS_REMOVED_SINCE(6, 7) which
the manual tests can't access.
Pick-to: 6.7
Change-Id: I81784baed178bb8efd1a848d32c9ee4510971e26
Reviewed-by: Piotr Wierciński <piotr.wiercinski@qt.io>
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
We were reporting the correct url, but saving to the
root of the file system.
Pick-to: 6.7
Change-Id: I3e1fd63efb77dd75ccf7a8f44a3211965fe11be8
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Verify that the file is saved correctly to the local
file system by loading it and printing the sha1 of
its contents.
Also add indents in order to improve legibility. The
output lines should now be correctly nested.
Pick-to: 6.7
Change-Id: I698c016253656aef29f04129da5f01a9dc35191e
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Update m_active early during DPI change to make scaling
calls made during the DPI change use the new value.
The existing call to updateHighDpiScaling() in QGuiApplicationPrivate::
processScreenLogicalDotsPerInchChange() is supposed to
take care of this, however that call is made after
the geometry change event is crated (with incorrect
geometry)
Pick-to: 6.7
Change-Id: Ia0f6006c90355ec69ef92b88a020089ab21e072d
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
This flag should be set in order to opt out of that Qt Quick Layouts
takes implicit (built-in) size policies into account, a feature that was
introduced in Qt 6.7
Task-number: QTBUG-117597
Pick-to: 6.7
Change-Id: I206cd61534b2103b2c3a9b4abc97fec8c88799d1
Reviewed-by: Santhosh Kumar <santhosh.kumar.selvaraj@qt.io>
The docs claim that QThreadPool always creates at least one thread.
However, the user can (usually by mistake) request zero or a negative
number of threads.
The maxThreadCount() function is simply returning the value, that was
requested by the user.
Since it's a public API, it is used in several places in QtConcurrent,
where it is assumed that the value is always positive. This can lead
to a crash if the user sets zero as a maxThreadCount.
Update all such places with std::max(maxThreadCount(), 1).
Prefer this approach over changing the return value of
maxThreadCount(), because its behavior is documented and tested.
Amends 885eff0537.
Fixes: QTBUG-120335
Pick-to: 6.7 6.6 6.5 6.2
Change-Id: Id3b2087cec7fbc7a2d42febca6586f2dacffe444
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This reverts commit 3bae65e7d6.
Reason for revert: This version requires API >= 34 but we have AGP
7.4.1 which was tested against API 33 max, and updating it to version
8.x is blocked by QTBUG-106907. So, this will be re-stated AGP update
is done.
Task-number: QTBUG-106907
Pick-to: 6.7
Change-Id: I1551ace76392a0532e1ecf431d03afbbd9d4a300
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
The pdf testing relies on the sips tool in macOS to render generated
pdfs into pngs. In recent macOS updates, that tool has changed
behavior, and no longer produces the desired output. Fix this by using
a custom pdf page size, instead of relying on cropping behavior in
sips.
Pick-to: 6.7 6.6 6.5
Change-Id: I9b2e947424774bb4456a6aefe04288b84401a9b7
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
QDoc is now capable of handling const-qualified property types.
Change-Id: Id36b44c7dfc6531d14f7fe31fdd0d69061743bb9
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
The UBSan raises a division by zero in getRadialGradientValues. This can
be avoided by calculating the inverse in
qt_fetch_radial_gradient_template as a checker is done to avoid
division by zero there. (Credit to OSS-Fuzz)
Pick-to: 6.7
Fixes: QTBUG-120332
Change-Id: I798d1efc87ee07df7ca6f401aa476013cdbffe42
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
With ES 3.0 and up the GL_MAX_SAMPLES query is available and is
guaranteed to be at least 4.
With ES 2.0 there is no such query.
While what we report from supportedSampleCounts() has little
significance with OpenGL (for the window's color buffer it is
anyway the QSurfaceFormat that decides, not QRhi), QRhi users
such as Qt Quick may rely on supportedSampleCounts() returning
real values and may, for example, warn if the value attempted
to use is not in the supported list. (even if MSAA is fully
functional)
Avoid getting such warnings on a pure ES 2.0 implementation
with 4x at least.
Pick-to: 6.7 6.6 6.5
Fixes: QTBUG-120474
Change-Id: I8a34a8bc9ce37ce951fdf7794247e8db44435d41
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
One might want to add an accellerator to dockwidget titles to be used in
contexts that isn't directly the title, but still uses the title.
Dockwidget title accelerators doesn't work in dock widgets, so don't
show them.
They are still available in e.g. a right click menu on a dockwidget to
hide/show, and here the accellerator might come handy, and also if a
dockwidget is tabbed.
Pick-to: 6.6 6.7
Change-Id: I196e1aa92a5c53fed735b598653c267509f788b8
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Reviewed-by: Milian Wolff <milian.wolff@kdab.com>
Reviewed-by: Sérgio Martins <sergio.martins@kdab.com>
The feature is needed for screen capturing Qt multimedia in order
to have better performance of getting frames
(avoid recreating fbo). In QtMM, we're going to create FBO
and use it as a hw frame (or just render it to image on the first stage)
Pick-to: 6.7 6.6 6.5
Change-Id: Id08a86b76447faa0f341c6967c2dad8f34c84959
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
ColorScheme is used in Qt as a name for dark/light desktop mode
properties.
The enum value was named incorrectly.
=> Fix it.
Pick-to: 6.7 6.6 6.5
Change-Id: I2aa72f81e36c0997127dd6de0e23d4773509ba40
Reviewed-by: Santhosh Kumar <santhosh.kumar.selvaraj@qt.io>
This reverts commit 4c60a11d8f.
Reason for revert: Breaks QMainWindow::restoreState(). Dock widget is shown, even if restored into hidden state.
Change-Id: I9808de4e0fd48871c3b234ae9bbaf6c64c8e832d
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Take vertical alignment into account in QFusionStyle::subControlRect().
Pick-to: 6.7 6.6
Fixes: QTBUG-67708
Change-Id: I9da23a4689565c8dba9fed9ef75d3c640ce77e3f
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
Remove dead code (since at least Qt5.0) - does not work and is
bitrotting since ages.
Change-Id: Ieb34e42137bc40c4edea47656b153a47f0ae34cd
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
This change updates getOpenFileContent to provide a parent when the
non-WASM fallback to a QFileDialog is created, which avoids issues
where this call is made inside another application. Specifically,
if a QDialog is created lacking a parent to the main window, it will
prevent interaction with that dialog. This patch addresses that problem
with window modality.
This change also updates saveFileContent to provide a parent for
the fallback mechanism, so that both static APIs that interact
with WASM/non-WASM dialogs have comparable behavior.
The new parent argument is updated in the documentation. Documentation
is clarified in terms of usage outside Qt for WebAssembly.
[ChangeLog][QtWidgets][QFileDialog] Adds an overload to the
static methods getOpenFileContent and saveFileContent with a
new parent argument which always no-ops in the WASM environment.
Fixes: QTBUG-118396
Pick-to: 6.7 6.6 6.5
Change-Id: Ic59aee386631172d4a29b42fe11e5af318474a1d
Reviewed-by: Venugopal Shivashankar <Venugopal.Shivashankar@qt.io>
Change the best practices to how-tos, which are categorized
now on a new page.
Fixes: QTBUG-118044
Pick-to: 6.7 6.6 6.5
Change-Id: I78dba2a1fde03b346f110ecd54e11485a0869540
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
This test was disabled during Qt5 porting because it was unstable.
According to the discussion in the bug report, a deprecated function was
used to wait for the window to get exposed. This was later replaced by a
more robust QTest::qWaitForWindowExposed() but the test was not
re-enabled again.
Pick-to: 6.7
Fixes: QTBUG-22544
Change-Id: I3190e38c1f42dc719803a60e66a86c3773cbedcd
Reviewed-by: David Faure <david.faure@kdab.com>
Do not add new paragraphs in the \value description,
which breaks the generated HTML table. Instead, use \br
more often to break overly long lines.
Replace 'option is only valid' or 'supported' by
'effective' and 'used'; if an enum doesn't apply to a
dialog or platform it will just be ignored, but it's
OK to set it.
Fix link to DontUseCustomDirectoryIcons enum.
Pick-to: 6.5 6.6 6.7
Task-number: QTBUG-119551
Change-Id: Iac86a85c7b3b402997fa14f4c4e2fa7111c62e9e
Reviewed-by: Venugopal Shivashankar <Venugopal.Shivashankar@qt.io>
Promises are registered upon entry to the pool and
unregistered upon exit. If all promises are in a 'pending' state,
new promises can't be processed. Upon completion of a registered
promise, it is unregistered, allowing space for a new promise.
The code path responsible for unregistering promises when they
resolve runs each time a promise's callback is called.
Unfortunately, there's no guarantee that the callback will be
invoked upon the promise's resolution. For instance, promises
registered with only a 'catch' callback may never be triggered
when the promise resolves correctly.
This commit ensures that a final callback is always registered,
even if the user did not provide one.
This guarantees that promises are always unregistered upon resolution
Fixes: QTBUG-118161
Pick-to: 6.5 6.6 6.7
Change-Id: Ifea93d692464a6ef40c4bcad60f840ca0cb650c9
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Misc style fixes in preparation for the new overloads taking
QStringView.
Pick-to: 6.7 6.6 6.5
Change-Id: I3b838543aefd08bf115488e571b1bb6eec8d968d
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Make toMimeDataWithFile() handle writing files to the
in-memory file system correctly: this is an async operation
and we need to collect the files as the writes complete
and then invoke the callback once all files have been
saved. There's then no need for a global static QMimeData.
Use toMimeDataWithFile() for both the paste and drop
handling, however QPlatformClipboard::setMimeData() takes
ownership of the passed in QMimeData and the callback API
must be designed accordingly.
An open question is when we should delete the files.
Deleting them right away (after calling the app event
handler) is predictable, however it looks like QPlatformClipboard::setMimeData() retains the current
QMimeData until a new one is set, so maybe we should
follow that.
Pick-to: 6.7
Change-Id: Ia9b825eaef1134ff9a554e51ee7e41d1c2ee779a
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
QScopedScopeLevelCounter was added and it doesn't really need
threads to function, but apps fail to link without it.
Change-Id: I7b355369079bc11b8b8b45b3ee1bf5ba81cd1953
Reviewed-by: Piotr Wierciński <piotr.wiercinski@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Some platforms like VxWorks do not support inotify
or any similar solution for file watching.
Because of that, the implementation falls back to polling.
That means that tests that require a native backend will fail.
Fix the problem by skipping adding a test case for a native backend.
Task-number: QTBUG-115777
Pick-to: 6.7
Change-Id: I0d683e6376a0453bd2fd591302c4f3ef7ceea87c
Reviewed-by: Pasi Petäjäjärvi <pasi.petajajarvi@qt.io>
Also means these mapping can now also handle overflowing perspective
better.
Pick-to: 6.7
Change-Id: Id506634c445521f1b041ec2e9051aa499c24671b
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Use QString as inheader record as the place of definition of both
Qt::Literals and Qt::Literals::StringLiterals. Previously mentioned
headers never exist.
Fixes: QTBUG-120379
Pick-to: 6.5 6.6 6.7
Change-Id: I104f73b338a144ef5f296500a9a4368cd3791750
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The default value in the default manifest has been for few
releases, set as minimal, and since Widgets Android style
is not fully supported anymore, we can set this to default
to minimal under the hood as well to avoid needing to always
explicitly needing to set it to minimal.
Pick-to: 6.7
Change-Id: Id0b2134c572694be1e190347ff75f51ade65f0c4
Reviewed-by: Nicholas Bennett <nicholas.bennett@qt.io>
Reviewed-by: Rami Potinkara <rami.potinkara@qt.io>
Reviewed-by: BogDan Vatra <bogdan@kdab.com>
Q_TESTLIB_PRIVATE_EXPORT is the redefinition of Q_TESTLIB_EXPORT.
Keeping the both makes no sense.
Task-number: QTBUG-117983
Change-Id: Id909530f1453ad092a45f40c1c85fed1f66e38d7
Reviewed-by: Jason McDonald <macadder1@gmail.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>