When the window flags are changed for the widget then it is possible to
minimize the dockwidget via the titlebar. This will cause it to be ready
to start a drag internally but since the dockwidget never gets a mouse
release event it doesn't end it. Therefore it ends up being in an invalid
state later on when restored, so the endDrag() needs to happen in the hide
event to ensure this is not an issue later on.
Change-Id: Ia84bee96b9eed49896869e6a15d4de6d01964264
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Normally setting the mode again and again on each application startup
is not desirable as it takes time and considerably affects the application
startup experience.
However, in some cases (like shutting down the X server) the output
gets powered down and there seems to be no way to detect this. Here
setting the mode again becomes necessary. Until there is a better
solution, add an environment variable QT_QPA_EGLFS_ALWAYS_SET_MODE
which, upon a non-zero value, leads to always calling drmModeSetCrtc.
The issue is not specific to Qt, the same is visible with the NV examples
as they try to minimize redundant mode set calls in the same manner.
Change-Id: If2b5e3fda74e077a747dd03cee65c936560a5026
Reviewed-by: Louai Al-Khanji <louai.al-khanji@theqtcompany.com>
Similar to the existing methods for QDBusPlatformMenuItem. These methods
will be needed for implementing the D-Bus platform menu bar.
Change-Id: Iae3a5892b71fa9a7c72c323ee4b3ab2afd7d4d76
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
Add support for ABS_MT_ORIENTATION value. Linux kernel multi-touch
protocol allows it to be in a wide range:
https://www.kernel.org/doc/Documentation/input/multi-touch-protocol.txt,
but we need only to know whether the touch point rect is rotated
by 90 degress. So adjust the angle to the closest axis.
Change-Id: Ie20725dc4bef509e2f9b19571efc69502d00c019
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
It's only supposed to be included by clients of the library, and
confuses qmake's dependency resolution into thinking the header-only
features such as qtestkeyboard.h and qtestmouse.h are dependencies
of this source file.
Change-Id: I5fe73b752c1c491ee0088bced77f997696d96c62
Reviewed-by: Edward Welbourne <edward.welbourne@theqtcompany.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
To calculate the real count we need to use the actual fixed point
increment and can not use the floating point value increment wass based
on since it might round differently.
Includes auto-test by Gabriel de Dietrich.
Task-number: QTBUG-50153
Change-Id: Ia973088f361c90370fa20bac14a4b8f373b5d234
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
The deprecation was introduced in 5.6
Change-Id: Ief6b749b40ec75c3c9f904caed8447bfb5ef5439
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Modern FreeBSD doesn't come with GCC by default anymore and doesn't even
provide the "gcc" or "g++" falback that OS X does. So there's no point
in keeping the freebsd-clang mkspec in unsupported/ since it's the only
one that works, or keeping the freebsd-g++* ones outside, as they won't
compile.
I'm not removing the GCC mkspecs because you can still install GCC from
the ports tree.
[ChangeLog][FreeBSD] The "freebsd-clang" mkspec is no longer in the
unsupported/ subdir. If you have scripts you use to build Qt, you'll need to
update them to say -platform freebsd-clang or remove the -platform argument.
Change-Id: I7a9e11d7b64a4cc78e24ffff142dfc11d3aabb1e
Reviewed-by: Raphael Kubo da Costa <rakuco@FreeBSD.org>
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
On FreeBSD, this variable is defined as a common symbol in the object file
/usr/lib/crt1.o, which is injected into the final application by the
compiler. This means when linking any library, 'environ' cannot be found
and we can't use -Wl,-no-undefined on FreeBSD.
I don't know why this wasn't caught before. Most likely, we failed to
pass the linker flag until some recent change to the buildsystem.
qprocess_unix.cpp:279: undefined reference to `environ'
Change-Id: I7a9e11d7b64a4cc78e24ffff142e02dbf188bca5
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
The qversiontagging.h inline assembly expands to:
.long qt_version_tag@GOTPCREL
Which, with GCC, Clang >= 3.7 and with the option -no-integrated-as in
previous versions, produces the proper relocation (a R_X86_64_GOTPCREL).
With Clang < 3.7, it instead produces a R_X86_64_32, which is unsuitable
for use in shared libraries: 32-bit displacement is insufficiently wide
and would produce linker errors like
obj/qftp.o: requires dynamic R_X86_64_32 reloc against 'qt_version_tag' which may overflow at runtime; recompile with -fPIC
Instead, force a 64-bit relocation (an R_X86_64_GOT64), which like the
32-bit version is simply an offset into the GOT of where the address of
the symbol is stored.
Task-number: QTBUG-50749
Change-Id: I7a9e11d7b64a4cc78e24ffff142e039c172b802c
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
This only happened in debug mode, though, because in release mode the
warning wasn't printed and the socket notifier was removed. In debug
mode, this loop in closingDown() never exited:
while (!d->sn_read.isEmpty())
unregisterSocketNotifier((*(d->sn_read.begin()))->obj);
[ChangeLog][QtCore][QThread] Fixed a bug that would cause debug-mode
applications to live lock on exit if they had a global static containing
a QThread that wasn't properly exited.
Task-number: QTBUG-49870
Change-Id: I7a9e11d7b64a4cc78e24ffff142e457a4540d6b6
Reviewed-by: Mat Sutcliffe <oktal3700@gmail.com>
Reviewed-by: Roland Winklmeier <Roland.M.Winklmeier@gmail.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Being called from an inline function we were breaking BIC (however
insignificant) by renaming the symbol in c7e5e1d9e0.
Change-Id: I683bfd53a5ad0de7db0fae6d9aa7d175e00f96ed
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
The name, format and maxChange of a Section depend only on the
section, not on the details of the currently parsed text it matches;
so we don't need the parser object's list of all sections to work them
out. Move these methods to the SectionNode and act directly on that
instead of going via the section list.
Make the name take a Section enum instead of an int.
Likewise, make stateName take a State enum instead of an int.
Change-Id: Ie340d042ab95aec517013c4dcc30901d40305c78
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
The actual section members grew up to the internal value, leading to
interleaving of subsequently-added values. Change existing enum
values to make this go away (class is private, so no BC constraint
forbids this).
Change-Id: If266b557817eae19d375519bf38268f0f2a0d372
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
One #if...#endif started in one function's body and ended in
another's, which is worth remarking on in comments at both ends.
A later #if...#else... opened blocks in both halves, that were closed
just after their #endif, leading to simplistic brace-matching failure;
so move that closing brace to inside both halves, so each half is
brace-balanced and the function, as a whole, brace-balances simply.
Commented on each #endif that was distant from its #if.
Change-Id: I74d97b6dc4b291ec05053d299077a61cecf41ca1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Generate the qundo*.qm files for tst_QUndo*::commandTextFormat() in
the current directory (out of source, for a shadow build) rather than
alongside the qundo*.ts from which they're derived (in the source
tree); and remove them once loaded, if that succeeds. (On failure, we
might even want to look at them.)
Task-number: QTBUG-49081
Change-Id: I666985fa3ceb8c25c917b617d6d39141eddebb76
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@theqtcompany.com>
To do that, split out and export a QKeySequencePrivate::keyName() function
that returns the key text representation without the modifiers.
The implementation is compatible with libdbusmenu-qt.
Change-Id: Ibc3190a3949813a1610728cf89574027e5d2778d
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
Rename TestWidget to PaintTestWidget, because this class tests only
whether paint event occurs.
Remove unused variables.
Remove reset() method, because there is no need to reset only one
variable in class method.
Change-Id: I8f442c73598e3431baf2b0433fefc157d3f10ea6
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
This signal got missed when 3fb0d851a was submitted so this compliments
that fix.
Change-Id: I91d20b709fce2c78d41166779954a3bb618feb37
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@theqtcompany.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
This warning was a good idea when the plugins were needed to give a hint
where to start debugging. Now the a11y plugins were merged into
qtwidgets/qtquick, so there is no more need for this warning which may
pop up on custom widgets e.g. inheriting QWindow.
Task-number: QTBUG-50215
Change-Id: Ic5efbb3dfbd07a7139884126a2604b54c99005e4
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@theqtcompany.com>
This solves an instance where restoreState() was used when the
dockwidget was already floating and the saved state was also for
a floating dockwidget.
Change-Id: I1fe764ae2a6b0351ae26e33ffec682ad37c944d7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
The QKeySequence test no longer uses private members (since commit
725bdc3fd2), so that is no longer needed.
Also, remove the unused include from the test.
Change-Id: I4d252bb3efd7282f74c44e48444c23ab51d48ea5
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
This reverts commit 4c71db7567.
It's too risky for 5.6, we should let it cook in dev for a while
and backport when ready.
Change-Id: I91e677e65d967f29c84a254cd3dffc8bb847b263
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
These are "special" XCB_NOTIFY_MODE_(UN)GRAB
Enter/Leave events and we do not have handlers for
them in Qt, so lets just ignore events with this mode.
Patch fixes Qt+ArchLinux+Awesome WM issues, where Qt
applications were receiving Enter/Leave events for
mouse clicks in the application window.
This patch does not affect "normal" XCB_NOTIFY_MODE_NORMAL
Enter/Leave event handling.
Task-number: QTBUG-45818
Change-Id: Ib70fdd9ed9200364a9753904f8e63d1ed9e2072f
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
Empty clipRect means "clip away completely", not "no clipping".
Task-number: QTBUG-50719
Change-Id: I6a9dd66130716a921fe9fc245582274e3c9718fe
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
The note about RasterGLSurface leading to calling composeAndFlush()
instead of flush() is simply not true. We have to have visible
render-to-texture widgets in the window to enter the composeAndFlush()
path.
Change-Id: I8331b10c75a3fbefc21009c7e5bb2b4de991bf5a
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
The memory should be allocated only if we operates on a valid type,
It is a regression introduced by 3d575d4845
Change-Id: Ia31bccd5b41fe090c29df1aeaa69efb706cd25bb
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There is no reason why they should be disabled, and also the current behavior
breaks the D-Bus platform menu bar implementation.
Change-Id: I8a8b5b6a66e40f1f1a25394dc87ac2178c7907fa
Reviewed-by: Martin Klapetek <mklapetek@kde.org>
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
Changed setCurrentAction() function so that it reselects parent menu
action only if mouse is over a menu and parent menu is not already
selected.
Task-number: QTBUG-47987
Change-Id: Id077d6ce1de1335ba2709180c8480af5d1720de0
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@theqtcompany.com>
Reviewed-by: Michal Policht <michpolicht@gmail.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@theqtcompany.com>
When we set accessory view sometimes (sic!) a delegate's callback
fires: -panel:directoryDidChange: with an outdated path (probably because
panels are shared?) resetting our current directory; later we open
file dialog with a wrong path as result.
Change-Id: Iffb02e801c44c5d9a62c2cca3acdf9278eaadb26
Task-number: QTBUG-50140
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@theqtcompany.com>
When getting the glyph runs from a QTextLayout with multiple
lines, the glyph runs would be merged if possible, but not their
bounding rects. This was an oversight.
[ChangeLog][Text][QTextLayout] QTextLayout::glyphRuns() now returns
united bounding rects for glyph runs that are merged.
Change-Id: Ibbeaa99ecfc4e82e7965342efdae7c3c2b637343
Task-number: QTBUG-50715
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
qfeatures.txt says that it depends on PROPERTIES, so this condition will
never happen.
Change-Id: I42e7ef1a481840699a8dffff1409e2e63c6857fa
Reviewed-by: Alex Blasche <alexander.blasche@theqtcompany.com>
Resetting focus_window and other internal QGuiApplication variables before
calling setVisible(false) and destroying the platform window means that the
platform window can't reason about whether or not it was the focus window
unless it can resolve that using native APIs. We should let the platform
window take care of resetting the focus window and related states, and
only execute our fallback logic if the plugin doesn't do the right
thing.
We also use QPA to update the state instead of modifying the internal
QGuiApplication variables directly, so that events and signals are
emitted as a result of the reset.
The QLineEdit test gets two added calls to processEvents(), since
assuming that activateWindow() is synchronous is not correct, and
would result in the QMenu resetting the focus window to 0 on destroy.
Task-number: QTBUG-46414
Change-Id: I562788393ed0ffd77d7a4be2279862322f721c1a
Reviewed-by: Błażej Szczygieł <spaz16@wp.pl>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
Can be useful for e.g. testlib for handling native key events.
Change-Id: I6560c6e28799e25eb3bdcaa0f2ca3c17644c62db
Reviewed-by: Jake Petroules <jake.petroules@theqtcompany.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
createPlatformOpenGLContext() used to silently set the widget
compositor's context as the context to share resources with.
This works mostly, but is the wrong level to enforce the resource
sharing. For example, QOpenGLContext::shareGroup() becomes
inconsistent since from QOpenGLContext's view there was no
shareContext specified.
The inability to test via shareGroup() is the reason eglfs started to
show warnings when exiting applications. The resource sharing was in
place on EGL level but QOpenGLContext knew nothing about it.
Therefore, let's switch over to the way other components, f.ex. Web
Engine use: set the internal global share context pointer to the
widget compositor's context. This way everything remains consistent:
the widget compositor's context is stored upon creating the main
QEGLFSWindow, QWidget::shareContext() picks this up then, and as a
result we have sharing set up on QOpenGLContext's level instead of
sneaking it in in the QPlatformOpenGLContext implementation.
Task-number: QTBUG-50707
Change-Id: I5fc1dec58c69c46aa83c7b4cab1eadce6fa633ce
Reviewed-by: Paul Olav Tvete <paul.tvete@theqtcompany.com>
A sequence of state changes fullscreen, maximized and back can leave the
window in a maximized state after setting the top level style. It needs to
be cleared before applying the normal geometry, otherwise, the window ends
up with a maximized button and normal geometry. Amends change
e3288f246b.
Task-number: QTBUG-49709
Change-Id: I0bb4ac1d60693e25d5ee74e763d293405636bb13
Reviewed-by: Błażej Szczygieł <spaz16@wp.pl>
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
The feature is enabled by CONFIG += unsupported/objc_namespace,
but can be easily integrated into other build systems such as
CMake or native Xcode by modifying the LD and LDFLAGS equivalent
for each build system.
This is a less resource-intensive alternative to using multiple
Qt builds with different -qtnamespace settings.
Note: The feature is not supported in any way, and should be
used with care.
Change-Id: Ibb8ba1159db36efd7106c117cc2210c7e2e24784
Reviewed-by: Martin Smith <martin.smith@theqtcompany.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@theqtcompany.com>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
Introduce separate scale factors for hot spot and pixmap and
set the devicePixelRatio of the scaled pixmap to 1 matching that
of the target pixmap which will be converted to a Windows cursor.
Change-Id: I0b0f6c6a79589ec954b5a1a09a86b87c91b5147d
Task-number: QTBUG-46068
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
Reviewed-by: Alexandru Croitor <alexandru.croitor@theqtcompany.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@theqtcompany.com>
