Use nullptr instead of 0 for null pointers.
Compare OCIHandleAlloc() return values to OCI_SUCCESS rather than 0.
Initialize instance variables in the class where possible, rather than
in the constructor. Conform to coding style o long lines and braces.
Assert pointers are null before allocating and overwriting them, so we
might have a chance of catching leaks.
Change-Id: Ia885ec18f46de5219a51fb6f9f23f474b3046585
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
QOCIResult::~QOCIResult() was doing part of the tidy-up for
QOCIResultPrivate, whose own destructor took care of the rest.
So move that part to the private, where it makes more sense.
Also correct an error message in part of the private's existing
tidy-up and eliminate a needless local variable.
Change-Id: I09a51c72afd7a30bcee7f6127c59d703650f1c41
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
If d->sql is non-null, it gets freed; later in the same function, it
gets reallocated, unless the query is empty, in which case the
now-freed value was still recorded, so that later clean-up might find
it and mistakenly think it needs to be freed again. Clear when freeing.
Pick-to: 6.2 6.1
Change-Id: I8d37d2ba1fcaa320916eaf30dceaa720bbf62f38
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Transaction handles were allocated but nowhere freed. Thanks to Stefan
Latsch for pointing this out and suggesting the fix. Make the handle
yet another member of QOCIDriverPrivate so that close() can free
it. At the same time, also free the service context handle d->svc when
failing do open(); and shuffle the order of frees to be the reverse of
the order of allocations.
Fixes: QTBUG-94246
Pick-to: 6.2 6.1 5.15
Change-Id: I45818ada6d884b06028056d28635390a300e2def
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Make Emscripten generate a global constructor function
("createQtAppInstance()") instead of a global javascript
module object.
This enables more fine-grained control over module
instantiation; previously the module object would be
created when the runtime javascript was evaluated, and
the number of emscripten module/instances was limited
to one per page.
Set EXPORT_NAME to “createQtAppInstance” which avoids
collisions with other non-Qt Emscripten modules on
the same page. A further improvement would be to include
the app name in EXPORT_NAME, but this is not done at
this time.
Update the code in qtloader.js to call the constructor
function instead of working on a global module object.
The qtloader.js API is functional before the wasm and
Emscripten modules have been instantiated; store properties
and forward to the Emscripten module when it's created.
Change-Id: I12c49a5b9a4a932bbc46fcc5e5ecc453fd0fe7f0
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Depending on build options, the module can have a different
name, or be a non-global object. We were already using
module_property in many places, but some were missing.
In the case of the clipboard code, there is actually
no need to export all of the C++ functions to JavaScript
and then resolve them from the module; instead call them
directly.
Change-Id: I83aa3ad01ad961d48e21f0994e0c205d833cbe8a
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
Remove event listeners on shutdown. For emscripten
event callbacks this can be done by passing nullptr
as the callback, for DOM event listeners use the
“removeEventListener” API.
Change-Id: I5a32a6a8feb906082db5ef3d0a0cd7ecb3b54292
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
m_compositor and m_eventTranslator are owned by the
screen object and can be held in a unique_ptr.
Pick-to: 6.2
Change-Id: I7d5e19e7ac6f762f360d6af0fc9500e6964f737e
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
The setlocale call will only give useful results if the program
had previously set the current locale using setlocale...
See also "Compose Locale" section in xkbcommon doc:
https://xkbcommon.org/doc/current/group__compose.html#compose-locale
Fixes: QTBUG-85529
Pick-to: 6.2 6.1 6.0 5.15
Change-Id: I65b1ac86ea54445bc3a2e1707df79bd9f732ab46
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Remove the configure.json and configure.pri files that
were used for the qmake-based configure.
Remove the .prev_*.cmake files that were a by-product of
configurejson2cmake.py.
Pick-to: 6.2
Task-number: QTBUG-89536
Change-Id: Ie827562f7fd2513d59f69234d77b8b93124ea78e
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
QAbstractItemView opens the editor and changes focus in response to the
first QEvent::InputMethod event. Changing focus however cancels the
current composition, so the key press is lost and needs to be repeated by
the user. What QAbstractItemView does is probably not that unique, so
fixing this there - if possible at all - is not a general solution.
In the platform specific input method handling code, we can generally
discover that the focus object changed in response to the InputMethod
event that gets sent to the focus object. On macOS, we can then pass the
key event to the input method once more, which works fine.
On Windows and X11, the situation is more complicated. That the focus
changed is easily discovered, but re-sending the key press in such a way
that the input method responds doesn't seem possible. On Windows, using
SendInput simply delivers the exact key to the focus object, without
starting the expected composition.
On X11 with IBus, it should be possible to call ProcessKeyEvent again on
the IBus input context when we discover in filterEventFinished that the
focus object changed while delivering the event. However, any attempt to
do so has no effect.
So, for now only fixing this for macOS.
Task-number: QTBUG-54848
Change-Id: I0f7b786cfa57430e6a1e7d605415265cf53eb675
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Abort the system move/resise at XCB_INPUT_TOUCH_END.
Limit the behavior only on supported platforms, such as KDE and
OpenBox.
Change-Id: I53c86979ca56f4de8c5cf2807f781abdad6987b2
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
We can't get mouse release event from master pointers after
QXcbWindow::doStartSystemMoveResize() which calls xcb_ungrab_pointer(),
it looks like most X11 WMs work as that.
So we try to get mouse release event from slave pointers.
Based on https://specifications.freedesktop.org/wm-spec/1.4/ar01s04.html
, we need to send _NET_WM_MOVERESIZE_CANCEL when we get mouse release
event.
Task-number: QTBUG-91077
Change-Id: I01e74a01c87b381ee7cd6f20d51a1fa61c0e98fc
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The iOS port creates one QIOSViewController per connected
screen. And each view controller listens for changes to
the application state. The problem is that we never
disconnect this connection again. So if a screen is removed, and
the corresponing view controller is deallocated, the
connection is still kept alive. This will cause crashes to
occur when the signal emits, since the slot will then be accessing
deleted memory.
Fixes: QTBUG-76948
Pick-to: 6.2 6.1 5.15
Change-Id: I758e51af9297cd62de193aae825f4475a2c7c3e5
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Previously they all returned the runtime versions and one string
function did not include the backend's name.
The NTDDI_VERSION macro is what we use to base certain
feature-availability on during compilation so it makes the most sense to
use for the build-string.
Pick-to: 6.2
Change-Id: I96b025a5a05c0bbb4db3d5ee68656e0df5f4eb07
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
To better match other plugins in Qt
Pick-to: 6.2
Change-Id: I6ab19603f13c80a8afe4fe69f64669559a0eea15
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Yuhang Zhao <2546789017@qq.com>
Dock widgets enable the QTabBar::drawBase property, but the clip region
left a gap of 3 pixels on each side between the base line and the tabs.
Correct the size of the hole cut into the clip region accordingly.
Visually tested on macOS 10.15.7 and macOS 11.2.1 using the dockwidgets
and tabdialog example.
As a drive-by, fix a spelling mistake in related documentation.
Fixes: QTBUG-86362
Pick-to: 6.2 6.1
Change-Id: Iada027999467741d0590576ee4508e6f41fb056d
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
NSWindowStyleMaskMiniaturizable also controls programatic
minimize, in addition to the window decoration button.
Fixes: QTBUG-65637
Pick-to: 6.2 6.1 5.15 5.12
Change-Id: Iac5fb483433441173d770f4fcfb9e93440b9fe6a
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Option-clicking the close-button on a window in macOS is a request to
batch-close all the windows. When this happens we get an event in through
the window that was clicked, which in turn results in AppKit calling
windowShouldClose: for each window. We respond to that by explicitly
closing each window (instead of just responding YES or NO), which
results in the window being released and deallocated. This causes
a crash when AppKit then follows up by closing each window after we
responded YES to windowShouldClose.
We work around this by keeping the window alive in the closest auto
release pool, which is typically at the level of the runloop. This
ensures that the window is alive for the duration of the logic that
AppKit has for batch-closing windows.
Fixes: QTBUG-92232
Pick-to: 6.2 6.1 5.15
Change-Id: I68b6138eb8325af0576b438ffa011137fec27926
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This matches the behavior of other implementations, such as
QEGLPlatformContext: doing a qWarning with the relevant error
from eglGetError() or equivalent is helpful when something
goes wrong and should be common among QOpenGLPlatformContext
implementations.
Pick-to: 6.1
Task-number: QTBUG-85714
Change-Id: Ifa5d22b83808c92ae92efcbcddc8d8c3d3ba34b1
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Make it return bool since the TriState was really only used signify that
the property was unsupported but there is already a separate way to
check if it's supported. More importantly there is no different set of
actions available to a user if they're in the Unknown or False state.
Because of the change to bool, we also rename the property to have an
'is'-prefix.
Change-Id: Iaaaad5ac31e663c36e00223bf5b0e719f412fc69
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
Change-Id: I7e0fce63883019278ceb9dc757bf10ee9d77097c
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
On Windows, building Qt with Ninja Multi-Config failed in a reasonably
named build dir. The reason for that is our deep build dir hierarchy,
created with add_subdirectory, and long target names like
QNetworkListManagerNetworkInformationBackend within that hierarchy.
Rename said target to QNLMNIBackend, and set AUTOGEN_BUILD_DIR to a
short string that does not repeat the target name.
Pick-to: 6.1
Fixes: QTBUG-94211
Change-Id: Iea6fee012fa46d9bfb01142b60e6cb5273893c50
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The generated CMake file Config.cmake in cmake/Qt6Network/
is generated from the target name. If it doesn't end with "Plugin",
then it won't be found by Qt6NetworkPlugins.cmake, which is looking
for Qt6*PluginConfig.cmake files.
Fixes: QTBUG-94108
Change-Id: I43f7056b2caede14509c9ec66b10e2037033762b
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
On macOS, shortcuts should be rendered along the imaginary line between
the modifiers, and the key. The modifiers are right-aligned on the left
side of that line, the key left aligned on the right side.
Make an exception for multi-chord sequences, render those always left
aligned.
Pick-to: 6.1
Fixes: QTBUG-73990
Change-Id: Ie03f3f40278700bdfafbfca7aa52075825e20234
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Android file dialog uses setType() to set the main mimetype for the
dialog if no mimetype or multiple (+1) mimetypes are provided, then
the additional mimetypes can be provided via EXTRA_MIME_TYPES flag.
The problem was that the mimetypes deduction from the namefilters was:
* the namefilter used was empty, now we take the first item
nameFilters(), because mimetypes cannot be changed once the dialog is
open anyway.
* The regex extraction was getting a namefilter ending with an empty
char and that was giving a mimetype of any format thus making it show
all possible files.
Pick-to: 6.1 5.15
Fixes: QTBUG-83089
Change-Id: Ifaef40c2186732ad3a604d28e086409c35dafacf
Reviewed-by: Rami Potinkara <rami.potinkara@qt.io>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
This value is sent by Apple keyboards, and was missing
from the table.
Change-Id: I49ad7ea74f2571c60ee9f2547468777b1f4585d8
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>
This is only a theoretical bug fix because systems that don't support XI
2.2 are ten years old by now.
Change-Id: Ia3770a6ca626b60d0f1e4cd2fdc611adf99cbf25
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
utf16() reallocates a QString if it was created from raw data, in order
to ensure NUL termination. But here we don't need NUL termination
anyways because we also pass the string size, so just use unicode()
instead.
Change-Id: I4a01ab9f4e53b94b80d3d00272cb0f0e35e30959
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Passing SQLITE_STATIC to sqlite3_bind_*() means that ownership
of the data stays in the caller, i.e. SQLite itself doesn't make a copy;
such data must be therefore be kept valid until sqlite3_step() is called.
The code in the SQLite driver uses that option to avoid copying byte
array or string data. But, unlike what the comments in the code say, we
do NOT keep the QByteArray/QString alive long enough: they're contained
by a temporary QVariant object which gets destroyed at the end of the
loop that binds each argument.
Luckily the fix is simple: since that QVariant is just a copy of the
QVariants used as bound parameters, and these are held in a container
(which lives long enough), simply create a reference to the container's
elements rather than a copy. This ensures that the data is alive by
the time sqlite3_step() is called.
This problem doesn't normally appear because of implicit sharing of
QByteArray/QString. When the QVariant is copied, the inner element
is just a shallow copy. Getting the pointer to the data, and destroying
the QVariant, does not destroy the data (it's kept alive by the
QByteArray/QString inside the *copied-from* QVariant).
Of course there's a catch: if the *copied-from* QVariant contains a
QString created via fromRawData, then everything blows up. In this case,
1. the copied QVariant is created (which bumps the QString refcount)¹
2. the QString inside of it is accessed directly (via
QVariant::constData)
3. utf16() is called on that string, which detaches it (!)
4. the result of utf16() is passed to SQLite, with SQLITE_STATIC
5. the copied QVariant is destroyed; this destroys the inner QString,
which, being detached, deallocates the data too early.
6. sqlite3_step() is called, kaboom.
(The copied-from QVariant still has the string created by fromRawData.)
¹ Note that QString uses the Small QVariant Optimization, so the QString
object itself into the QVariant is copied, it's not just a *QVariant*
refcount increase.
Change-Id: Idcdb192809f1f8f79b4a901e1247f933eb06e854
Pick-to: 6.1 5.15 5.12
Fixes: QTBUG-94070
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Rename internal modules to adjust their names to the internal module
policy. Also modify mappings of the qmake file converters.
Change-Id: I69aee1e8136c2379608d9d22d718f8c8a5f73124
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
It was a necessary workaround at the time, but this has since been fixed
Task-number: QTBUG-90812
Change-Id: I8c0d5bf4c61d379736cd07a28c089c55d6befdee
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Give QCocoaMenu access to the item's visibility property by adding a
public getter, and respect that state when syncing the separator to the
corresponding QCocoaNSMenuItems.
Pick-to: 6.1 5.15
Fixes: QTBUG-88651
Change-Id: I7e238e5d3be141ec8f4e8f6f3ab22d761094d2d2
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Fixes Integrity compiler issue where it cannot find implicitly defined
Change-Id: I126333dd4b02b9b653b8e1ac5da07679b343c91f
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reviewed-by: Kimmo Ollila <kimmo.ollila@qt.io>
When we draw menu item text with CoreText instead of QPainter, then we
need to translate QFont::underline() explicitly to an entry in the
attribute dictionary.
Task-number: QTBUG-73990
Pick-to: 6.1 5.15
Change-Id: I7e0c258f7cf80a89b29517fa8122576342654de1
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Silences warnings
warning C4996: 'ID2D1Factory::GetDesktopDpi': Deprecated. Use DisplayInformation::LogicalDpi for Windows Store Apps or GetDpiForWindow for desktop apps.
Using GetDpiForWindow sounds like the right approach, but it's not obvious
where to get the window handle from here.
Task-number: QTBUG-94043
Pick-to: 6.1
Change-Id: I2fc005e0c970c8338184f4500b876774ec9d976b
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
We don't need all of QtDBus + QtCore dragged in.
Change-Id: I99dd6b1a50587d8d520ec5240c5129b406e4f40d
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
In some cases, if X11 is configured with CapsLock as
keyboard language switch key, and CapsLock is toggled
via Shift+CapsLock key combination,
toggled CapsLock is falsely detected as Qt::GroupSwitchModifier
for subsequent key events.
This change fixes this false detect,
but doesn't fix detection of Qt::GroupSwitchModifier
which is likely still broken.
Fixes: QTBUG-49771
Pick-to: 5.15 6.0 6.1
Change-Id: I485e2d4f3c654707c62adaba367c1b8afb3fc36c
Reviewed-by: Liang Qi <liang.qi@qt.io>
Get select handles {Left, Right}Point from a mapToGlobal with a cursorRectangleof anchorRectangle
of the selected word/text.
Task-number: QTBUG-90799
Change-Id: I4fed19a1d3bb2247656b052306307e6980074405
Reviewed-by: Rami Potinkara <rami.potinkara@qt.io>
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
When tool tip was trying to show its window became activated causing
tool tip to hide instantly due to a redundant window activation bug on
show/hide/raise/lower.
Fixes: QTBUG-91095
Change-Id: Id215400a79cdc4ed1961731949ca139b575eec88
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
The logic was removed in 1fc7ca091b when
we removed support for surface-backed views, based on the documentation
for wantsBestResolutionOpenGLSurface noting that layer-backed views
ignore the property.
But in some cases, when plugin views are embedded into host views, and
the host was built with a pre-10.14 SDK, AppKit seems to get confused,
and we still need to set this property.
Fixes: QTBUG-93770
Pick-to: 6.1
Change-Id: Id67f2fcb8c090c0e01c05d00792eaef311146df1
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Edward Welbourne
Morten Sørvig
Lorn Potter
Liang Qi
Volker Hilsheimer
Joerg Bornemann
Richard Moe Gustavsen
Mårten Nordheim
Giuseppe D'Angelo
Tor Arne Vestbø
Laszlo Agocs
Kai Köhne
Assam Boudjelthia
Povilas Kanapickas
Alexey Edelev
Janne Koskinen
Timur Pocheptsov
Aleksei Nikiforov
Pekka Gehör