Remove the forced bail out when isES() returns true. This is not
necessary.
Change-Id: I5ee21fe1e66163e2391bd11b647827b3c0a020c1
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Windows differs from the other platforms in that transient
parent relationship is not just a window property but also
implies ownership; windows will destroy their transient
children in DestroyWindow(), which interferes with Qt.
Explicitly clear the relationship in QWindowsWindow::destroy()
to prevent this.
Task-number: QTBUG-36666
Task-number: QTBUG-35499
Change-Id: I5e72524ef57422831f60484993f6c8d7c80c8601
Reviewed-by: J-P Nurmi <jpnurmi@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
Instead of QSharedPointer<T>(new T), use QSharedPointer<T>::create(), since the latter
co-locates the refcount and the T instance in a single memory allocation, unlike the
first form, which uses separate memory allocations.
Change-Id: I5095ac43448aad9a7e3ec07ed4dcdca869bcd9e8
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
Two changes broke compilation on armv5, where we're currently
not CI-testing:
634f82f1f1 changed the signature
in a function definition without changing its declaration, while
it was actually intending to add this as a new overload.
bfe0db6fbe added an #error
condition without fixing compilation on armv5. I don't know
if the fix is correct, but at least it compiles.
Task-number: QTBUG-37034
Change-Id: If99142fafb9bd55afc20b17f8b3cce5ee0ffec13
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The old code creates a default-constructed QByteArrayList, then performed two
list-appends, the first one of which just performs assignment.
Optimize by replacing the default construction and assignment with a copy
constructor call.
Change-Id: I6d5bd14172798c925b05bd3602e6d1d037d90796
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
This reportedly does fix crashes when running a Qt Quick 2 application
over remote X.
Task-number: QTCREATORBUG-11207
Change-Id: I6fa82420f9d12e56e52fa8efd263bf18d868d7d8
Reviewed-by: Ville Nummela <ville.nummela@parker.com>
Reviewed-by: Jørgen Lind <jorgen.lind@digia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
This prevents a tool button menu from being opened by a replayed
click. This partially reverts
8301c0002280c10970cce1e17f634e74c61f2f5d .
Task-number: QTBUG-36863
Change-Id: I396e3694de8b3d4ca916457c2b2df39798502530
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>
If there's no qApp, then QCoreApplication::arguments() returns an empty
list (after printing a warning). That means the callgrind runner can't
get the arguments it needs in order to rerun the benchmark.
The crash happens because it always uses .at(0) to try and get the
executable's path. Even if we get the path from somewhere else, we still
need the arguments.
Change-Id: I5c74af4d96fc5824b2b7fd7a89648d78393016e2
Reviewed-by: Sergio Ahumada <sahumada@blackberry.com>
Needed for QOpenGLShaderProgram to be usable with
GL_OES_get_program_binary and potentially other extensions.
[Changelog][QtGui] The function QOpenGLShaderProgram::create()
has been added. It is can be used to force immediate allocation
of the program's id.
Change-Id: I36b3f45b00e7a439df12c54af7dc06c0ba913587
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
If raster only apps set QT_ANDROID_RASTER_IMAGE_DEPTH to 16 (RGB16), we
should create also RGB16 native surface.
Change-Id: I82692ff34b0e604e627d1d86a437272e3700daf8
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
The previously used GetActiveWindow() returns the application's
window also if it is minimized.
Task-number: QTBUG-36806
Change-Id: I8ede3ea30e7b714aa1af85ed67e510e1692ebb8f
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Print "QMap<QString,QString>" for "a{ss}" and print a nicer expansion of
other types.
Change-Id: I0a7a2ecf8f0a62bd97931f3c129cd4cb4f471ef1
Reviewed-by: Lorn Potter <lorn.potter@jollamobile.com>
The bit scan function returns the index of the last non-ASCII
character. The next ASCII is the one after this. This means all the
benchmarks were made while reentering the SIMD loop uselessly...
Change-Id: If7de485a63428bfa36d413049d9239ddda1986aa
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Profiling showed that Qt Creator spent 2% of its load time normalizing
signals and slots. By pre-normalizing everything, we ensure that there
is no runtime cost. Profiling after this commit and the others in this
series shows that the cost dropped down to zero.
Change-Id: Ifc5a2c2552e245fb9a5f31514e9dd683c5c55327
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Disassembly shows the Intel compiler does expand to SIMD, but a much
worse code than ours. Clang 3.4 does generate a compact SIMD version,
probably of the same quality as our hand-written code. And GCC 4.7
through 4.9 don't generate SIMD at all.
So let's use the most efficient version.
Change-Id: I418e201a774ac0df1fb2b7a7d9589df7c9b655db
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Unit-test this by making the QList, QVector, QHash and QMap unit tests
be duplicated under strict-iterator mode. There's no test for
QLinkedList.
The tst_Collections test does not compile under strict-iterator
mode. It generated over 15000 errors when I tried.
The strict iterators required a small change: the difference_type
typedef needs to match the operators that get distances
(operator-(iterator)) and move the iterator around (+, -, +=, -=, etc.).
Task-number: QTBUG-29608
Change-Id: I834873934c51d0f139a994cd395818da4ec997e2
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Reviewed-by: Jason McDonald <macadder1@gmail.com>
Mouse selection does not work well with Android and text will be
randomly copied, pasted or deleted. This behavior is especially bad
when predictive text is enabled.
Task-number: QTBUG-34616
Change-Id: I732ad7db52169bfb5735c237cf24597a3d6d64ba
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
Text selection does not work correctly and is currently causing
selected text to be randomly copied, pasted or deleted.
Task-id: QTBUG-34616
Change-Id: I98678b7575034325dd8a4fa181ee4cb182783a3b
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
A QSslCipher is not an equivalent representation of a QString, so
the constructor that takes a QString should be explicit.
Change-Id: I4c1329d1eebf91b212616eb5200450c0861d900f
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
A QPainterPathStroker is not an equivalent representation of a QPen, so
the constructor that takes a QPen should be explicit.
Arguably, the named constructor idiom would be even better here:
static QPainterPathStroker QPainterPathStroker::fromPen(const QPen &pen);
But QPainterPathStroker is non-copyable.
Change-Id: I3148dc0ee336026781d8bc1baf21c113c7b41ce8
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
This fixes a problem that QScanConverter::mergeLine didn't recognize
lins as being the same (when they where), causing aliasing effects
Task-number: QTBUG-36354
Change-Id: I29d92ddb4e867025541bdc6b294cfaca55c0d3e1
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Add a new QMarginsF class to complement QMargins in the style of
QSize/QSizeF and QRect/QRectF.
[ChangeLog][QtCore] Added class QMarginsF to support handling margins
with floating-point values.
Change-Id: Iaaa95ec85f5d126d9d864fc4b607241a8c8a8f3a
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
On Windows, trailing spaces in a filename are silently ignored, so we
need to strip it before trying to open a file with it. Otherwise it ends
up being stripped later and in a case like " ." it will end up causing
Qt to think that a folder exists when it does not.
[ChangeLog][Platform Specific Changes][Windows][QtWidgets][QFileDialog]
Handled the case of having trailing spaces in a filename correctly so if
the filename ends up being empty that the parent path is used instead.
Change-Id: I6500cc3a44746bf4a65e73bcfb63265a0a97c8a3
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Trigger QOffscreenSurface's fallback mode (hidden QWindow and a regular window
surface) instead. queryDummyContext() already works like this but the same must
be done for any QOffscreenSurface.
Task-number: QTBUG-36900
Change-Id: I64176ac6704e9d6ed768fa3d456c40c8818be6dc
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@digia.com>
Allow the developers to define a splash screen which will be visible
until the first window is created.
[ChangeLog][Android] Allow the developers to define a splash
screen which will be visible until the first window is created.
Task-number: QTBUG-30652
Change-Id: I5da80be417ffffb03e66009f45745d4b387d2912
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
Android is using double buffering, so, we need to repaint the bounding
rect of the repaint region, otherwise black holes will appear.
Change-Id: I21f36a6f5f1a6c64b605c0fef3af10dfdc5ec6e2
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
The QTextLayout::FormatRange is larger than void* and thus should not be held in
QList. Use a QVector instead.
This is public, but as of yet unreleased API.
Change-Id: Ie04a561b43c91c3b2befb3cac2981821f84d5f77
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Otherwise, OleFlushClipboard() might query the data again which causes
problems for QMimeData-derived classes using QPixmap/QImage.
Task-number: QTBUG-36958
Change-Id: I89e58eeb64bd3481e89ad789f310f19ddb4604a2
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
So far we did bind the definition of Q_COMPILER_VARIADIC_MACROS to
C++11 (so gcc, clang will not define it in default gnu++98 standard).
However, variadic macros are a feature of the gcc preprocessor since
version 2.97, and are enabled in the default configurations on gcc, clang,
icc.
This might cause warnings and errors though if one enables additional
warnings in gcc, clang (e.g. by -pedantic). Anyhow, as a precedent
qglobal.h already relies on 'long long' ... The warning can be disabled
by adding '-Wno-variadic-macros'.
[ChangeLog][Compiler Specific Changes] Variadic macros are now enabled
more liberally for gcc, clang, icc. If you have warnings (because you e.g.
compile with -pedantic), disable them by -Wno-variadic-macros.
Change-Id: Ie979b85809508ad70cab75e6981f20496429f463
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Commit f4dd534 introduced a regression, so that QFontMetrics reported
a wrong size (to be more specific width) for FreeType fonts.
The calculation of glyph advances has to to reflect (rounded)
integral number of pixels. This was only done when the glyph was cached.
So in some cases the first call to QFontMetrics::size gave a different
result than the second.
This patch reverts f4dd5344fb.
The tst_QFontMetrics::same auto test only happened to work on some
platforms, on BlackBerry for instance it did not. Extended the test
case to make sure it works for different font sizes.
Change-Id: Ia5bb9abd3ff98193c9bba048b85207672ed8d9c3
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
-Qt Examples and Tutorials page does not need the table of contents.
Task-number: QTBUG-36838
Change-Id: Id51ebc7cba7831a24cd9d8e8e6bde7f96bece326
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Topi Reiniö <topi.reinio@digia.com>
After 37ca2224ec it's necessary to
read the saved QSettings again after creating widgets, in order to
deal with the settings which only affect widgets (such as saved
bookmarks). It's also necessary to read them if widgets are not used
though, because some of the settings affect native dialog options.
Task-number: QTBUG-36888
Change-Id: I8cf53db864b173c50a876a1d5ce29c1e073fcaa6
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
By ensuring that the widget is already polished before doing a repolish it
means that if you reset the stylesheet then it unsets any changes that
the stylesheet had applied to the widget.
Task-number: QTBUG-18958
Change-Id: Ie0aeda0dac9f2211b7feca138c115cf2b48aac80
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
When the emulation paint engine was no longer needed then it would still
end up using it because the flags would prevent it from being switched
back. This ensures that it has the right engine when something triggers
it to be switched.
Change-Id: I7571923d16cbebd9fdd34560631b561c07a724f7
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
Fall back to /var/tmp instead of /tmp if neither TMPDIR nor TEMP are set.
/tmp is not a true filesystem on BB10 but rather a symbolic link to /dev/shmem
For more info see
http://www.qnx.com/developers/docs/6.3.0SP3/neutrino/user_guide/fsystems.html#RAM
Change-Id: Ie690ed74ffd81b52ef4623458c3ff88629aee00a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QHash uses an array representing the difference between 2^i and
the next prime; when growing, it calculates 2^x + array[x] (with
`x' representing the "hash table size in bits").
For some reason lost in history the differences are actually wrong
and the calculation above leads to using composite numbers.
Hence: use the right sequence and always produce primes. The right
sequence is actually A092131 from OEIS: http://oeis.org/A092131
Note that the sequence starts at A(1), but we need A(0) too.
Also we truncate the sequence to when growing too much, just like
the old code did, and use powers of two in that case instead.
Task-number: QTBUG-36866
Change-Id: Id2e3fc9cb567c0fdca305dee38f480e17639ca04
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Currently the only supported SPDY version is 3.0.
The feature needs to be enabled explicitly via
QNetworkRequest::SpdyAllowedAttribute. Whether SPDY actually was used
can be determined via QNetworkRequest::SpdyWasUsedAttribute from a
QNetworkReply once it has been started (i.e. after the encrypted()
signal has been received). Whether SPDY can be used will be
determined during the SSL handshake through the TLS NPN extension
(see separate commit).
The following things from SPDY have not been enabled currently:
* server push is not implemented, it has never been seen in the wild;
in that case we just reject a stream pushed by the server, which is
legit.
* settings are not persisted across SPDY sessions. In practice this
means that the server sends a small message upon session start
telling us e.g. the number of concurrent connections.
* SSL client certificates are not supported.
Task-number: QTBUG-18714
[ChangeLog][QtNetwork] Added support for the SPDY protocol (version
3.0).
Change-Id: I81bbe0495c24ed84e9cf8af3a9dbd63ca1e93d0d
Reviewed-by: Richard J. Moore <rich@kde.org>
Padding the bounding rect was not enough, as we failed to shift the glyph
accordingly so that it would end up in the center of the bounding rect.
We also didn't take subpixel-positioning into account, which may shift the
position of the glyph too far to the right to be within the image size that
we reserve.
There are still cases where the glyphs seem clipped compared to the same
text rendered with CoreText, but that's because we end up shaping the
text slightly differently, resulting in different subpixel positions
than what CoreText chooses.
Change-Id: Icb88c829f86457b16bdecbc4c24b3f1c23448261
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Instead of padding the image size manually, we rely on alphaMapBoundingBox
to give use the right glyph metrics.
For clarity, a few function arguments were renamed in the affected code.
Change-Id: I84c31e613a1048ea839a390af70342e5388ed0cb
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Instead of the glyph cache having its own cache type that always mapped
one to one to a font engine glyph format, causing confusion and needless
conversions, the glyph caches now use QFontEngine's glyph format enum.
This also removes the iffy use of an int for the glyphFormat in the font
engines.
Change-Id: I529bad5c179e004f63e152f7dcc311d298c3db98
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
GNU/Linux distributions like Fedora update the __GLIBCXX__ date (e.g.
because of applied extra patches on top) and thus make the current
workaround useless. As long as no official gcc/libstdc++ version is
released, we can increment the "upper bound" date to cover such cases.
See also comment #1 at QTCREATORBUG-11129.
Change-Id: I1fdf6f312664163f6a99a8eddf490646ff25a87d
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Change 4b2c73b476 triggered a
regression on Android, because the DroidSansFallback.ttf font
has codepage ranges that list Arabic as a supported codepage
despite the fact that the font has no glyphs for Arabic.
If a font has valid unicode ranges which does not specify
support for e.g. Arabic, then the code page ranges for the
same script should be ignored.
[ChangeLog][Android][Fonts] Fixed support for Arabic text.
Task-number: QTBUG-36789
Change-Id: I7c5a0e303fd9ed2733275814fe752ae0fb54a207
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
ICC defaults to "fast math" mode, which allows it to do non-compliant
operations that may or may not result in unexpected values. Generally, it's
ok, but apparently the code in qlocale_tools.cpp is too complex, so we're not
taking chances.
I can't rule out an issue in the code, though.
Task-number: QTBUG-36795
Change-Id: Ica5cb77fb3a65d22ae8ad22e13b4ba78f1b5dadf
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
It is in nearly all cases much faster (in our benchmark it is
about a factor 8) - and it also makes the code more readable.
Change-Id: I075e1d1a96756e5e9a2bf53b5479094f7d87883d
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
updateBlock is part of the interface of QAbstractTextDocumentLayout
however QTextDocumentLayout implementation is not invoking it.
This commit adds similar behavior to QTextDocumentLayout as the
QPlainTextDocumentLayout (QtWidgets) has by implementing the missing
functionality.
[ChangeLog][QtGui][QTextDocumentLayout]
Emit updateBlock signal in QTextDocumentLayout.
Task-number: QTBUG-36743
Change-Id: I7f812d4d5d7c2148393ee672fb0db6c0b2e73783
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
This fixes compilation of Qt for Android when configured with
"-no-widgets". One good use-case for that configure switch is
that QtSvg loses the dependency to QtWidgets.
Change-Id: I8a645d9fee6ce0467ff0b5d1d62531b7c6272aab
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
Reviewed-by: Daniel Teske <daniel.teske@digia.com>
This partially reverts 5764f5e6ea and
fixes the problem differently.
After this commit, "file:///foo" is still equal to "file:/foo", but
"foo:///foo" becomes different from "foo:/foo", as it should be.
Task-number: QTBUG-36151
Change-Id: Ia38638b0f30a7dcf110aa89aa427254c007fc107
Reviewed-by: David Faure <david.faure@kdab.com>
The drawing code currently only optimizes the bilinear interpolation
under specific conditions that allows the optimizations used there.
The patch adds a SSE2 version of the fallback 4 pixel interpolation.
Change-Id: I4e8a2ba6cb44647105a9b24e38b3ab755a435050
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
In case the QPlatformTheme does not provide a QPlatformMenu the
QPlatformSystemTrayIcon is not able to forward the menu because the
QPlatformMenu pointer passed to updateMenu is always null. Providing a
QPlatformMenu in the QPlatformTheme implementation should not be a
requirement for having the menu in the system tray icon. There are
cases where no QPlatformMenu should be created by the theme, e.g. if
the X11 implementation of QSystemTrayIcon gets provided by the xcb
plugin.
The change adds a virtual method to QPlatformSystemTrayIcon to create
a QPlatformMenu. This method is called from the QPA implementation of
QSystemTrayIcon if the QMenu's platformMenu is not present. Thus the
system tray icon is able to provide a custom implementation of the
menu. This gets installed through a new internal method in QMenu to set
the platform menu. It creates the required connections and sync the
state to the QMenu to the newly created QPlatformMenu.
Last but not least QPlatformMenu is extended by a method to create a
QPlatformMenuItem. The default implementation delegates to the platform
theme. This allows to provide the menu item implementation for the
system tray icon without providing the QPlatformMenuItem through the
platform theme.
Change-Id: I17234bd8bcf8c05f8bd786feff0cf8f860430e82
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
The setting of the default format of Raster_A8 was mistakenly
moved down to below the override of Raster_RGBMask.
Change-Id: I52ee19180471c6f5c8d6824ee35f8d39632c9d94
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
This fixes regression introduced in f864bdaf59
(in resolveAdditionalFormats(), formatIndex(si) returned an invalid
block format index due to non-empty specialData->resolvedFormats)
Task-number: QTBUG-36759
Change-Id: I4da13d2edaf1987bcd85b2ef326e9d22345eb8f3
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Turns out that \obsolete means "Qt3Support", while the correct tag
is \deprecated.
Change-Id: Id9896893f3078a0d516bd8751bce0b2df441509d
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
0x80000000 is too big for a signed 32-bit int, so the compiler ends
up choosing a 64-bit number. Then it will not be the same size as
the parallel enum QDialogButtonBox::ButtonRole, which does not have
these extra flags. By making EOL be the same as InvalidRole and
changing Reverse to 0x40000000, we don't need to use 0x80000000 at
all, so again the compiler can choose a 32-bit representation
for both if appropriate. EOL is used as a terminator in the
static const int buttonRoleLayouts table and therefore as a terminator
in arrays returned from QPlatformDialogHelper::buttonLayout().
It's internal API so it's OK to change in a minor release, but
QtQuick.Dialogs is also looking for the EOL now, so we should avoid
changing it again if possible. This is mainly for efficiency and
for avoiding surprise data type conversions when casting from one
enum to the other.
Change-Id: Ia7b7ce43c8f929d09dd999769e9b7114f695e9a4
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
When the focus changes, screen sends a focus deactivated change for the focus window.
Depending on the order, this might deactivate the window focus completely. This
for example leads to problems with QComboBox.
Change-Id: Ia57f061a765c1f971d86d941b17f573ce9221ae1
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Bernd Weimer <bweimer@blackberry.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
The navigator swipe down event is not mapped to a platform panel event any more.
Instead the NavigatorEventHandler is exposed through the QPlatformNativeInterface.
Change-Id: I6d29bba011849da5210f6f4d595e3c2e0c021449
Reviewed-by: Bernd Weimer <bweimer@blackberry.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Modern websites use around 15 hosts on average so the limit of 5 DNS
looks meant that most non-trivial sites have to wait for at least one
unneeded round trip.
Task-number: QTBUG-30866
Change-Id: I584dd46c247e0413b2ed88dd66f111cff08ff987
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Implemented the use of the new QDnsLookup property "nameserver".
On the Linux platform, we can specify both IPv4 and IPv6 addresses
for the nameserver.
On Windows since we are using DnsQuery_W(), which does not have a
way of accepting IPv6 addresses, passing IPv6 nameserver address
is not supported.
On OSX/BSD platforms, specifying IPv6 addresses for nameserver
require access to the __res_state_ext structure which is in a
private header of libresolv (this header is different for BSDs and
OSX). If this feature has to be enabled in the future, we have to
figure out a way to access this struct by either accessing the
private header or by specifying one of our own. Currently, I'm
disabling it till such a solution is arrived at.
Nameserver support on different platforms:
Platform | IPv4 | IPv6
-------------+---------------+---------------
Linux/X11 | supported | supported
-------------+---------------+---------------
Windows | supported | not supported
-------------+---------------+---------------
OSX | supported | not supported
-------------+---------------+---------------
WinRT | not supported | not supported
-------------+---------------+---------------
Others | supported | not supported
| (not tested) |
-------------+---------------+---------------
Task-number: QTBUG-30166
Change-Id: Iedbddf15b9a62738ce4c2cfa0fce051514d64766
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This fixes the issue that the blitter required sometimes the
texture wrapping to be repeat
Change-Id: I86150d008422facf9040873b0983b0e44be9ad24
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
The Origin for Target rect was deemed a confusing concept. The current
implementation would translate the target rect to the coordinate system
specified. However, the order and "direction" of the vertices would
always be the same. So drawing a texture in for one target rect defined in
one coordinate system would paint the texture the same way as it would
when a texture was drawn for a target rect drawn in the "opposite"
coordinate system. The point with this was that if you wanted to "flip"
the texture you would specify that with the source coordinate system.
However, this approach breaks on different levels, such as QRect has functions
which expects a top left coordinate system (ie. top() and bottom()).
In the end Qt uses a top left coordinate system, hence QWindow specifies
a top left coordinate system, and hence the api becomes easier if it is
not possible to define the coordinate system of the target viewport.
Change-Id: I7dd59b3718380876e87a4bff88381d7a1c7d58c1
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
This also has the advantage of not requiring the use of the locale
codec. Quite an advantage if you're debugging the locale codec. But it's
mostly so that we don't get question marks that hide the difference we
were trying to locate.
[ChangeLog][QtTest] QtTest now prints an escaped version of QStrings
that failed to compare with QCOMPARE. That is, instead of converting
non-printable characters to question marks, QtTest will print the
Unicode representation of the character in question.
Change-Id: I44c1ef3246b188c913dacd3ca4df02581356ea41
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
[ChangeLog][QtCore][Atomic support]Added more operations to the atomic
classes, including operator T(), operator=(T), operator++,
operator--. For the QAtomicInteger, bit-manipulation operations are
also provided, both in operator and in fetchAndXxxYyyyyy modes.
Change-Id: I39c07be74e15e0a48f9e931f4342b182004dee1a
Reviewed-by: David Faure <david.faure@kdab.com>
This is extremely useful, since the most common action after a failed
compare-and-swap is to loop around, trying again with the current
value as found in memory.
Code currently written as:
do {
Type value = atomic.load();
...
} while (!atomic.testAndSetRelaxed(value, desired));
Becomes:
Type value = atomic.load();
do {
...
} while (!atomic.testAndSetRelaxed(value, desired, value));
In most CPU architectures, the value that was found in memory is known
to the compare-and-swap code, so this is more efficient than the
previous code. In architectures where the value is not known, the new
code is no worse than before.
The implementation sometimes modified an existing function, sometimes
it added a new one, depending on whether more registers were needed in
the assembly (like ARMv6-7), the code became more complex (ARMv5), the
optimizer failed (C++11), or it was just plain equivalent (MIPS).
Change-Id: I7d6d200ea9746ec8978a0c1e1969dbc3580b9285
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
ICC 8 and 9 are positively ancient. I doubt anyone is using them for
Qt, let alone Qt 5. ICC 11 through 13 haven't supported OS X.
ICC now masquerades as Clang, so we need to let qmake and
qcompilerdetection.h know about it.
Change-Id: If0d2bd8b6a4a45250c15c9472c062effc76f17de
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Apparently it doesn't like the text.
Change-Id: If8e14df84f0d9915018eac94df16bf1b679155e0
Reviewed-by: Keith Gardner <kreios4004@gmail.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The QByteArray version was missing the overflow check that the other
versions had.
Change-Id: I03cd92e5e5a84c038bee1f1ee217e93e9d9a675a
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
This implementation is based on GCC's implementation of std::hash<FP>, but only
to the extent of checking for zero before hashing the bits. The bit hasher is
the Qt one; I didn't even look what GCC uses.
The check against 0.0 is mandated by the requirement to have
\forall x,y: x == y => qHash(x) == qHash(y)
which would be violated for x = 0.0 and y = -0.0 if we only hashed the bits.
Implemented out-of-line to avoid potential FP-comparison warnings, as well
as to be able to use the file-static hash() functions, which gets inlined
unlike qHashBits(), which cannot be.
[ChangeLog][QtCore][QHash/QSet] Allowed to use float, double and long double
as QHash/QSet keys.
Change-Id: I38cec4afb860f17e9f8be7b67544e58b330f8fff
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
The properties that make up the hash value are chosen to be the same as those
that make up QFontDef's op<() and op==(). Indeed, the implementation for QFont
simply delegates to the one of QFontDef, which has been added for this purpose,
but may prove useful in its own right down the line.
The code would greatly benefit from a qHash(qreal) implementation. Lacking
this, the patch uses multiplication with 10000 and qRound64() to convert
the one floating-point property used in the hash to an integer. This is
probably the right thing to do anyway, to avoid epsilon problems.
[ChangeLog][QtGui][QFont] Added qHash overload for this class.
[ChangeLog][QtCore][QHash] Allowed QFont to be used as a key in QHash/QSet.
Change-Id: I2c1cb5d9da53e26cb2c0f1a7c357731e73eea78e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When you launch an activity through an intent,
data can be provided back from the activity when it has finished
using onActivityResult() in the activity which launched it.
This is okay for applications, since they can easily create their
own activities, but does not work for libraries that need to
use intents. There is no listener API for activity results which
allow external classes to eavesdrop.
In order to support launching intents from third-party or add-on
libraries, we provide a low-level way to hook into the activity
result event. The corresponding public API will be added to
QtAndroidExtras.
Change-Id: I89417f485e2c0e69028dcccc7c155788346a7417
Reviewed-by: Christian Stromme <christian.stromme@digia.com>
Change 85e57653 caused a compile error for code that does
Q_DECLARE_LOGGING_CATEGORY(cat);
//..
qCDebug(cat()) << // ...
error: C3848: expression having type 'const QLoggingCategory' would lose
some const-volatile qualifiers in order to call 'QLoggingCategory
&QLoggingCategory::operator ()(void)'
This is a regression from Qt 5.2. Fix the error by adding a const version
of operator()().
Change-Id: I2fb04f2e155962adee0f98089fc5a159000bef56
Reviewed-by: Alex Blasche <alexander.blasche@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The use of if {} else {} in the macro causes compiler warnings
about "ambiguous 'else'" if qCDebug is used in an if / else without brackets.
Revert to the for loop, but make the variable name less likely
to clash.
This reverts commit bab5f5873680aa3c5a22d94da112aafd2b0b53d3.
Task-number: QTBUG-36605
Change-Id: Ie4b075b63b83b7f8a2ad61437b7bf3e6a6c0177a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
Reviewed-by: Alex Blasche <alexander.blasche@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Added basic functionality to socket for WinRT. Even though not
all auto tests pass yet, this patch can be seen as a foundation
for upcoming work in this area. Reading from and writing to TCP
socket works and one can listen for tcp connections.
Change-Id: Id4c25ba1c7187ed92b6368c785c4f62837faded7
Reviewed-by: Andrew Knight <andrew.knight@digia.com>
Moving them into QPlatformDialogHelper for the convenience of both
widgets and QtQuick.Dialogs. The main reason is to ensure that
QtQuick.Dialogs does not need to depend on the widgets module, in
order to re-implement the button box concept in a generic dialog.
Change-Id: If400d215338d7cb6dade39d9de60e50b5e7515ef
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Currently, inputMethodQuery() only provides information about the
current paragraph. On some platforms, such as Android, the input method
needs information about the global cursor position, and more of the
surrounding text. Some queries need to pass parameters.
The current inputmethodQuery() implementation does not allow parameters to
be passed. Changing this would require new or modified virtual functions, which
is not possible until Qt 6. Therefore, a completely new mechanism is needed.
Change-Id: Ic64fd90198ade70aa0fa6fa5ad3867dfa7ed763c
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
category() returns a const QLoggingCategory so our code does not
compile anymore. Do it differently.
Fix also the Windows error string formatting a bit.
Change-Id: Ie0b6b02947d94b7ccf4a4a57da487dfa8a15709d
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
I find the current description of primaryOrientation and orientation
a bit confusing.
Change-Id: I25d77cff2c27c481607903bc1aeb54eacf616718
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
The Qt documentation says that PortraitOrientation is rotated 90 degrees
clockwise relative to LandscapeOrientation. This means that the home
button should be on the right when held in LandscapeOrientation,
therefore, Qt::LandscapeOrientation == UIDeviceOrientationLandscapeLeft.
Without this patch, all QScreen mapping functions are broken.
Change-Id: I2c570cd0307b7fbd59c749d6574dcb258790cfbc
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
The code was not creating all of the storage necessary for cubemaps
as well as attempting to bind to the cubemap face targets which is
invalid when using mutable storage - typically on OS X where
EXT_direct_state_access is not available and immutable storage is only
available at all if using an OpenGL 4.1 context.
Change-Id: I4cf84f1b88c90e8359366392b3ccda65669ebfa7
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Pasi Keränen <pasi.keranen@digia.com>
moc would skip the 'operator' keyword as unknown and try to parse a type again
but as it sees the '<' it looks for the corresponding '>' which does not exist
types can't start with '<' anyway, so return an invalid type and continue
parsing as usual
Task-number: QTBUG-36834
Change-Id: If3d27076ef9947abf8c57c594713eece9334d0b0
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
I could trace the dangling 'However, ' back to Qt 4.2. We probably will
never knew what the rest of the sentence was, exactly ... since the
previous paragraph is anyhow covering the same grounds we can just
remove the paragraph.
Change-Id: I1dda56e1ad17932205b8f40a527ca90a7592ad40
Reviewed-by: Ulf Hermann <ulf.hermann@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use FSEvents to monitor changes in the filesystem instead of the kqueue
implementation. This removes the limit of wathed files: kqueue uses a
file descriptor for each file monitored, for which the ulimit was set by
default to 256. Now the OSX implementation on par with the other major
desktop platforms.
Change-Id: I2d46cca811978621989fd35201138df88a37c0fb
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
The patch introduces a new build configuration on Windows which
can be requested by passing -opengl dynamic to configure.
Platforms other than Windows (including WinRT) are not affected.
The existing Angle and desktop configurations are not affected.
These continue to function as before and Angle remains the default.
In the future, when all modules have added support for the dynamic
path, as described below, the default configuration could be changed
to be the dynamic one. This would allow providing a single set of
binaries in the official builds instead of the current two.
When requesting dynamic GL, Angle is built but QT_OPENGL_ES[_2] are
never defined. Instead, the code path that has traditionally been
desktop GL only becomes the dynamic path that has to do runtime
checks. Qt modules and applications are not linked to opengl32.dll or
libegl/glesv2.dll in this case. Instead, QtGui exports all necessary
egl/egl/gl functions which will, under the hood, forward all requests
to a dynamically loaded EGL/WGL/GL implementation.
Porting guide (better said, changes needed to prepare your code to
work with dynamic GL builds when the fallback to Angle is utilized):
1. In !QT_OPENGL_ES[_2] code branches use QOpenGLFunctions::isES() to
differentiate between desktop and ES where needed. Keep in mind that
it is the desktop GL header (plus qopenglext.h) that is included,
not the GLES one.
QtGui's proxy will handle some differences, for example calling
glClearDepth will route to glClearDepthf when needed. The built-in
eglGetProcAddress is able to retrieve pointers for standard GLES2
functions too so code resolving OpenGL 2 functions will function
in any case.
2. QT_CONFIG will contain "opengl" and "dynamicgl" in dynamic builds,
but never "angle" or "opengles2".
3. The preprocessor define QT_OPENGL_DYNAMIC is also available in
dynamic builds. The usage of this is strongly discouraged and should
not be needed anywhere except for QtGui and the platform plugin.
4. Code in need of the library handle can use
QOpenGLFunctions::platformGLHandle().
The decision on which library to load is currently based on a simple
test that creates a dummy window/context and tries to resolve an
OpenGL 2 function. If this fails, it goes for Angle. This seems to work
well on Win7 PCs for example that do not have proper graphics drivers
providing OpenGL installed but are D3D9 capable using the default drivers.
Setting QT_OPENGL to desktop or angle skips the test and forces
usage of the given GL. There are also two new application attributes
that could be used for the same purpose.
If Angle is requested but the libraries are not present, desktop is
tried. If desktop is requested, or if angle is requested but nothing
works, the EGL/WGL functions will still be callable but will return 0.
This conveniently means that eglInitialize() and such will report a failure.
Debug messages can be enabled by setting QT_OPENGLPROXY_DEBUG. This will
tell which implementation is chosen.
The textures example application is ported to OpenGL 2, the GL 1
code path is removed.
[ChangeLog][QtGui] Qt builds on Windows can now be configured for
dynamic loading of the OpenGL implementation. This can be requested
by passing -opengl dynamic to configure. In this mode no modules will
link to opengl32.dll or Angle's libegl/libglesv2. Instead, QtGui will
dynamically choose between desktop and Angle during the first GL/EGL/WGL
call. This allows deploying applications with a single set of Qt libraries
with the ability of transparently falling back to Angle in case the
opengl32.dll is not suitable, due to missing graphics drivers for example.
Task-number: QTBUG-36483
Change-Id: I716fdebbf60b355b7d9ef57d1e069eef366b4ab9
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Jørgen Lind <jorgen.lind@digia.com>
Added configure test, whether lgmon (liquid graphics performance monitor)
is available. The test is supposed to be positive only for internal
BlackBerry NDKs currently.
Added calls to initialize lgmon and to indicate when an app is ready for
user input.
Change-Id: I5cbc29fb38a86585dcebd14d462436deaa1998aa
Reviewed-by: Wolfgang Bremer <wbremer@blackberry.com>
Reviewed-by: Fabian Bumberger <fbumberger@rim.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Update drawColorLessButton() and qt_mac_cg_context()
to handle paint devices with a devicePixelRatio greater
than one.
Task-number: QTBUG-36792
Change-Id: I9d642846b299fc7048bb8e08765b9e1d7ee631fc
Reviewed-by: Denis Dzyubenko <denis@ddenis.info>
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>
At first, my goal was just to fix Moc::until() to parse properly
template arguments containing expressions containing > or >>
such as Foo<(8>>2)>
But with the test, I realized that normalizeType also requires change
not to split the > > too much.
And QMetaObjectPrivate::decodeMethodSignature should not interpret
the ) within the template parameter as the end of the function.
Change-Id: Ia9d3a2a786368aeda1edcf66280d70f64cf05070
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
In general QLoggingCategory should be treated as a const object that's
configured by the backend. The only legitimate place where user code
should call setEnabled is in a CategoryFilter ...
[ChangeLog][QtCore][Logging] Make Q_LOGGING_CATEGORY
and Q_DECLARE_LOGGING_CATEGORY return a const object.
Change-Id: I6140442ab48286e05cd3b55064a502bbe6dfb16a
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Including qnumeric.h causes QtGlobal to have all of the functions that
the documentation indicates that it brings.
Task-number: QTBUG-36715
Change-Id: Ib08ccc18a85dfe4ecd18ef1dfbfec3e90ab8ec2c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Enums should be named LikeThis, not likeThis, even in private API.
Change-Id: I197f9f888204a7c495364bd09357bfca24146560
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The arguments are all strings, so multi-arg is available without
other changes to the code. Even though insertion of a format placeholder
can be ruled out in the present case, multi-arg should be faster than
a 3-chain of arg() calls.
Change-Id: I8d030227e1bd30c56f1062a0c9dbbaae0143885f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Instead of using Q_STRINGTABLE, use a switch/if construct
to be able to share the string data for "write" and "readwrite".
Change-Id: Ia1c7b8a0f13a809372de2e5a956978dc2d569e92
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It was not possible to get the actual process ID (in a cross-platform
manner) from QProcess, as the user would need to handle the returned
typedef (Q_PID) differently on Unix and Windows.
On Unix Q_PID is the actual process ID, but on Windows it's a pointer
to a PROCESS_INFORMATION structure, which among other fields contains
the process ID. Instead of returning a pointer on Windows,
QProcess::processId() will return the actual process ID on both Windows
and Unix.
[ChangeLog][QtCore][QProcess] Added processId() to QProcess. This
function will, unlike pid(), return the actual process identifier on
both Window and Unix.
Task-number: QTBUG-26136
Change-Id: I853ab721297e2dd9cda006666144179a9e25b73d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The docs said it supports the feature, but now that we've tried to use
it, we can't. It might have been referring to the non-C++11 extension
from GCC (__thread).
qlogging.cpp(1253): error #303: explicit type is missing ("int" assumed)
static thread_local bool msgHandlerGrabbed = false;
^
Change-Id: I9343cf61bd3b2eacac686e602cc0ffea2d4a7a22
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
Reviewed-by: Mehdi Fekari <mfekari@blackberry.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
By way of templates. This makes the code a lot cleaner.
Change-Id: Ie369561c7631b0d34d76a6852883716cc0aa89d4
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
This version will parse the string only once and will not do any
memmove. This is more efficient.
Change-Id: I59026ad0fa61cc3f16146bdcd622fc54cbd8a321
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Along with some more helper functions.
There are two more functions used in QIntValidator
Change-Id: I469ef40426cbb73ab515454bd5ecb12d944f5c0a
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Those functions do not need any of extra QLocale settings in
QLocalePrivate, so we can move them easily, along with their flags.
It's also very convenient that we can now bypass completely QLocale
when formatting numbers to strings.
Change-Id: I8cae64e8e2056a6b2d716758e4be79f746644732
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The difference between them was simply whether they operated on QString
or QStringRef. So drop down to what's common between them: a pointer and
a length (or two pointers, but numberToCLocale already operates on a
pointer and a length).
Change-Id: Ie7c8955ac13d6023761e6d3bafe7ab04bd6984e1
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
No need to do it in QLocalePrivate::xxxToDouble() because
numberToCLocale() already does it for us. Centralized code = better
code.
Change-Id: Ifecf9119556d4465582212b5be773c18edd13563
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
From before we would activate all QWindows that the user
tapped on, or setVisible were called on. This is wrong
since a QWindow does not have to be a top-level window.
For a non-alien widget application this would mean that
we would send activation events for all widgets that the
user tapped on all the time.
With this patch we do some extra checking before we
tell a QWindow to activate.
Change-Id: I1afe97e5384c36c67fee0bbd070d880bba7528a1
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
We need to activate a window on touchesBegan instead of
touchesEnded. The reason we used to do this on touchesEnded was
to delay activating a window in case the user started e.g a
flick. But delaying the activation can cause problems if the app
activates a different window on press. We will then cancel
this out on release since we then raise the pressed window instead.
This is e.g typical when opening popups, and will cause focus to
not be restored properly when later closing the popup again.
Change-Id: I709b2f2e2633c9dc85c2761b0b176cd23c2f6b36
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
Sometimes focus object is updated after we get a
callback that the cursor rectangle has changed. And
there is no reason to keep a local reference to it.
Since we also send events to the qApp->focusObject from
UIView_textInput, we now end up more consistent.
Change-Id: I3976175aae4e3f346be9bc5b771ac0fdefc03ae6
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@digia.com>
This is only useful for logging benchmarks, since it won't print test
passes, failures, etc. It's useful for importing to spreadsheets to do
number-crunching.
[ChangeLog][QtTest]Added a CSV logging mode that is suitable for
importing benchmark results into spreadsheets. This can be enabled by
the -csv option on the command-line. The CSV logging mode will not print
test failures, debug messages, warnings, etc.
Change-Id: I245d6f86bb380645c9bc0d748cf474b3ed42cab8
Reviewed-by: Sergio Ahumada <sahumada@blackberry.com>
Reviewed-by: Jason McDonald <macadder1@gmail.com>
Introduce new virtual QGuiApplicationPrivate::tryCloseAllWindows()
which allows overriding the behavior in QApplication to properly close
the widgets first.
Without this, QGuiApplication closes the widget windows leaving a stale
window handle behind in the associated QWidget which then causes the
application not to terminate since QApplication::shouldQuit() stills
finds the affected widgets to be visible.
Task-number: QTBUG-35986
Change-Id: I19ac4b5a19250ee68d09e461c03dbace458c98e4
Reviewed-by: David Faure <david.faure@kdab.com>
Integrate with QOpenGLTextureBlitter, QOpenGLWidget and friends.
Change-Id: Ic2867b713a21a3d2820d546174fc9164b3dd220c
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
This can be quite useful on some embedded systems to free up
graphics memory when windows are not used. QEglFSWindow already
implements the function.
Change-Id: I79b08efbd3c67d7be34df6a0e12dd184a92d48c5
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
Silences warnings in tests that instantiate QGuiApplication multiple
times.
QSYSTEM: tst_QGuiApplication::removePostedEvents() QApplication::regClass: Registering window class 'TabletDummyWindow' failed. (Class already exists.)
QSYSTEM: tst_QGuiApplication::removePostedEvents() UnregisterClass failed for 'TabletDummyWindow' (Class still has open windows.)
Change-Id: I6af2d38a2debd35f4dc0d48c09244dff022bd6b8
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
84be1bd4d3 has changed the stored data type
from QFontEngine to QFontEngine::FaceData. Update the implementation
and revert changing the y_scale sign on non-Mac
(aka fix-up 2d576f79f7).
Change-Id: I4180257bc8f610fb014fd2a2ad6f8fdceece2f13
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Make sure that stacks are properly aligned by replacing CALLBACK with
QT_WIN_CALLBACK so that we don't crash in SSE2 code.
Task-number: QTBUG-36807
Change-Id: I6952d0f252c7b8e481c48521ed1377b7d7510e15
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
... to defer the decision which protocol will be used on a specific
channel. This is to allow using the SPDY protocol instead of HTTP (to
be implemented in a later commit); which protocol will be used can
only be decided after the SSL handshake.
Change-Id: I6b538320668fe4994438f0095ecdc445677cf0a6
Reviewed-by: Peter Hartmann <phartmann@blackberry.com>
... from the private to the public class, because we need to access
these methods from other classes.
Change-Id: I2c5ea84e0f5d3641c1dc02342348f1022d886249
Reviewed-by: Richard J. Moore <rich@kde.org>
This is additional fix for bug 23569.
Previous fix (SHA1 7fcf1cf674) is not enough.
QGtkStyle was still crashing with old gtk (< 2.14)
in drawComplexControl () function.
Bug was reproducible on CentOS 5.x and Red Hat 5.x.
Current patch makes the same check as in commit mentioned but in another
line of code.
Task-number: QTBUG-23569
Change-Id: I261b61bc93ccaada879ed02ad4d0bef62935335b
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Harri Porten <porten@froglogic.com>
Initial submission of a new class QByteArrayList with the
purpose of aggregating and then joining QByteArray instances.
[ChangeLog][QtCore] Added new QByteArrayList class.
Change-Id: I2a9dc71ff7aadb19ebc129a0d47ac8cd33895924
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The native event dispatcher is responsible for delivering callbacks
to non-GUI handlers, such as network socket listeners. So, the non-GUI
logic is moved into the core dispatcher so that the event loop works
better for apps (and test cases) which use QCoreApplication.
Change-Id: Ic5f7d939cf164198fd39aa5880e265ae560b39b4
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
Make sure that stacks are properly aligned so that we don't crash in
SSE2 code (as used in QString).
Task-number: QTBUG-36807
Change-Id: I765c1972fbb0a4b0c2dc5e4465e1d19e00b3322b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
This changes the code added by 0e4d94edd0
by removing the Q_INTEGRAL_TYPE enum value and instead relying on our
private copy of typetraits.
Change-Id: Icb6c814a36125de0d712421b97e11e8abc0f7416
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Currently QAndroidPlatformTheme::palette has been returning null so
the default colors came from elsewhere. Now they will default to
colors taken from the Holo Light theme to the extent possible.
Change-Id: I022c5ef679d9d0dcb76721a4d520492bc8a6f4cf
Reviewed-by: BogDan Vatra <bogdan@kde.org>
Benchmarking shows it took up to 3.5% of Qt Creator's initialization
cost. Optimize by modifying only one variable per loop: instead of
updating n and dst128, we only update one variable at a time.
Removing the Duff's Device also improves the code, since the compiler
won't try to update dst128 four times per loop, only once.
The moving of the epilogue close to the prologue was just to make the
code a little cleaner.
Change-Id: I5b74e27d520ca821f380aef0533c244805f003b7
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
This function has a flawed design. It was flawed when it was added in
Qt 3.0.
A "false" return value is racy: any other thread running may post events
to the current thread, thus making the result stale. That includes Qt
starts for its own purposes when it comes to the main thread, like the
Scene Graph thread, the QProcessManager thread, the Windows
QAdoptedThread watcher thread, the Windows pipe writer thread, etc.
A "true" return is stable only if the selected thread is stopped, which
includes selecting the current thread (the case of QCoreApplication).
For that reason, this method should not be public, but a protected one
so that a public static could call it. But even that would not solve the
race condition from the previous paragraph (hence why
QCoreApplication::hasPendingEvents being deprecated too).
And, to top all of that off, all but one of the implementations access
the GUI thread's event loop counter in a non-thread-safe manner. I've
changed the documentation to restrict to the only currently-working use-
application.
[ChangeLog][QtCore][Event loop] QCoreApplication::hasPendingEvents and
QAbstractEventDispatcher::hasPendingEvents are now deprecated. Please
refer to the documentation for more information.
Task-number: QTBUG-36611
Change-Id: Iac61f307e9672839944ae2f75abb1aea30c419f6
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
By reimplementing canBeNativeDialog() we ensure it is only shown
natively if the circumstances are correct. The change was modelled on
how QFileDialog implements the same checks.
Change-Id: If58e33d6d0547e95975d7c0453901fa592eb96a0
[ChangeLog][QtWidgets][QColorDialog] Ensured
QColorDialog::DontUseNativeDialog is respected when showing the dialog.
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
[ChangeLog][QtGui][QWindow]QWindow::icon() now defaults to the application
icon, which can be set with QGuiApplication::setWindowIcon().
Change-Id: Id1974e5cda81775e515c14b294f67fb99351c6c9
Reviewed-by: Albert Astals Cid <albert.astals@canonical.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
-title disappeared between Qt4 and Qt5, due to all the refactorings around QPA.
Making the caption of the mainwindow configurable allows custom setups
for specific users or use cases.
[ChangeLog][QtGui][QGuiApplication] Restore support for -title command line
argument on X11, add -qwindowtitle on all platforms.
Change-Id: I73e6bf21248f3419178eba583b257172a175e74e
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Albert Astals Cid <albert.astals@canonical.com>
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
Enable child widgets (without a native window) that render to an FBO
and are composed with the raster backingstore by the platform plugin.
A preliminary version of QOpenGLWidget is included as private API.
Change-Id: I8f984a4d7db285069ce3d6564707942c823d890d
Reviewed-by: Jørgen Lind <jorgen.lind@digia.com>
When the Cocoa menubar is updated then it ensures that the merged native
menu items are visible if appropriate. However when the old menubar is
deleted then it causes the merged native items to be hidden. Therefore
we ensure the children are deleted first which causes the native items
to be hidden and then update the menubar so that they can be correctly
visible if they should be.
Change-Id: I426864a5d2ec1f34c03290ac66371b12bf77cc00
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>
Useful for bug reports and test logs.
[ChangeLog][QtCore] QLibraryInfo provides information on how
Qt was built.
Change-Id: I867197fd7d2e10bcdf01a8eb47c9c1e03647e2c1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Qt::WA_Mapped corresponds to 'exposed' in Qt 5, hence a parent window
fully obscured by its children may not have the attribute set.
In that case, go by the attribute of the child widget.
Task-number: QTBUG-33520
Change-Id: Ia08f283172daaa58f58e5636004461b2c48784f7
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
While inheriting from NSPanel proved to be robust enough, it is
not really future proof as we're at the mercy of Apple changing
NSPanel's behavior. On the other hand, we can't inherit exclu-
sively from NSWindow as the tool window case, where the QWindow
should look like an NSPanel, can't be emulated perfectly without
using private APIs.
This reverts commits 79fb39a87c and df86721bb4.
Change-Id: I9021193e3614633a943578df9e2794b00094a1f7
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
Added a basic theme class, which creates accessors for the
QPlatformMessageDialogHelper. This handles creation of the
native dialog, but also spawning an event loop around it.
There are some limitations shared with the Android implementation.
First you cannot create custom labelled buttons and also selecting
the default button is not possible. This will be added at a later
point.
Note that Windows Phone is not supported.
Change-Id: I2dec83dcded7919835a6dcdf8dc2a56282024a7e
Reviewed-by: Andrew Knight <andrew.knight@digia.com>
These functions now return a QStringList to reflect the possibility of
there being more than one entry of a given type, but the documentation
did not reflect this.
Task-Number: QTBUG-36304
Change-Id: Iba2eda5e2c3174c8dcea640b5aed9cdc9a432392
Reviewed-by: Peter Hartmann <phartmann@blackberry.com>
Avoid dereferencing the q-pointer if the button has been
destructed meanwhile popup was open.
Task-number: QTBUG-26956
Change-Id: I68190e9fe84c669229ae0ce4d573ee7a02a8a141
Reviewed-by: Ivan Komissarov <ABBAPOH@gmail.com>
Reviewed-by: Thorbjørn Lund Martsum <tmartsum@gmail.com>
Partially revert 54865d47cfd859a0e84ba8e1bbff3b56c93d4e0d and
bring back parts from abba37cd0eff959edf2d011842cbf2b36cce8467 which
failed some tests on other platforms and never made it in.
Avoid setting a geometry on maximized windows since that leaves
them in a broken state. Instead, if there is a need to restore
maximized windows to a different screen, clear the maximized flag
before applying the geometry.
[ChangeLog][QWidget][Windows] QWidget::restoreGeometry() now restores
maximized/full screen widgets to the correct screen.
Task-number: QTBUG-21371
Task-number: QTBUG-4397
Change-Id: Ibd5b736e921f20d4b7365c06c6f9c5dd4469c781
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Allow configuration of logging rules from outside of the application,
either through a configuration file (.config/QtProject/qtlogging.ini),
or through a file specified by a QT_LOGGING_CONF environment
variable.
The logging rules from the different sources are concatenated: First
the rules from QtProject/qtlogging.ini are applied, then
QLoggingCategory::setLoggingRules(), finally from the environment.
This allows an application to overwrite/augment the system wide rules,
and in turn that can be tailored for a specific run by setting a
configuration in the environment variable.
[ChangeLog][QtCore][Logging] The logging framework can now be configured
with an .ini file.
Change-Id: I442efde1b7e0a2ebe135c6f6e0a4b656483fe4b1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QWidgetPrivate::updateWidgetTransform will only update the
transform if the widget is the current focus object.
But if someone calls QApplication::setActiveWindow, Qt will
call QWindow::requestActivate and then send focus
in/out events, all in the same event loop recursion.
The problem now is that requestActivate is not guaranteed to be
synchronous (it's not on iOS). So the window activation
(together with updating the focus object) will still be pending
when the widget receives the focus-in event. As such, the
transform update will also fail.
This patch will give the event as input to the function, so
that we don't depend on window activation being synchronous.
This will fix IM spell checking popups on iOS to show at the
correct place, also after closing font popups etc.
Change-Id: If0ee70f55692bbd613821b126923364e39ed1199
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Makes it possible for QFusionStyle to utilize the same animation.
Change-Id: Ifac9eaa3138cf1068439d5b0271a2acce54a3c16
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>
This patch cleans up the coding style and includes
Change-Id: I710d4a60795e9975d4f8ee79599018e05d85debe
Reviewed-by: Jørgen Lind <jorgen.lind@digia.com>
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
The fixme comment is invalid since QAccessibleWidget actually checks for
isWindow and sets the active state.
This is messed up because in Qt 4 there was a work-around to set active
for windows in a different code path.
[ChangeLog][QtGui] Accessibility on Linux now reports the active state
correctly.
Change-Id: I9e2cf436b3ffa7ef28286ee49d6e582f179930c6
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>
... which is needed to negotiate the SPDY protocol.
[ChangeLog][QtNetwork][QSslConfiguration] Added support for the Next
Protocol Negotiation (NPN) TLS extension.
Task-number: QTBUG-33208
Change-Id: I3c945f9b7e2d2ffb0814bfdd3e87de1dae6c20ef
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
Right now socket notifiers are attempted to be created before having
the event dispatcher up. This is wrong.
Change-Id: Ica3034b9fa790c037f28982db23c71342f6597d7
Reviewed-by: Andy Nichols <andy.nichols@digia.com>
Use correct parameters for \enum commands, and add documentation
for QOpenGLTexture::Filter enumeration.
Task-number: QTBUG-35576
Change-Id: If7099da0b2b570c28e683126f0ba3a885d80f741
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Passing a constructor as second argument to foreach didn't work when
building with gcc.
For MSVC this already worked as a different foreach implementation is
used.
Change-Id: Id98444c699b4cebc14ea62076c5f7cba33ffb824
Reviewed-by: hjk <hjk121@nokiamail.com>
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@digia.com>
Better late than never :)
Task-number: QTBUG-33749
Change-Id: I5035255e66a56754b609441f5b81ab119565a7cb
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Remember the engine index for each sub-item and avoid moveGlyphData()
where possible (ie. when there are no glyph indexes to care about).
Also don't memmove data we didn't ever initialize.
Change-Id: Ib8e5fd937a10e4e3c8c0e18961a2e2c1a4167924
Reviewed-by: Ahmed Saidi <justroftest@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The old code wasn't good enough to catch all the glyph (de)composition cases,
thus leading to an assertion in QTextLayout's addNextCluster() helper.
The new code catches all the corner cases and introduces somewhat
better performance to the HB-NG shaper backend.
Change-Id: I5b6c673395a4a039dc55b200abbf74b0ba5d0829
Reviewed-by: Ahmed Saidi <justroftest@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
The RGBA8888 formats was incorrectly using the qt_gradient_quint32
which is argb specific. This caused vertical gradients but only
vertical gradients to be drawn incorrectly.
This changes the RGBA8888 formats formats to use the generic gradient
method and renames qt_gradient_quint32 to qt_gradient_argb32 to
indicate its limitation.
Change-Id: Ia1cd48ca7f4f78b64f31d6263e81cd8ac3b0954e
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
LRESULT should be non-zero if the application can quit,
and it is always handled.
Task-number: QTBUG-35986
Change-Id: Ic8b0ecc731d3efaf94e2a62491918d379fe59f5c
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
The debug output of all categories will be visible by default,
except from the "qt.*" categories. "qt.*" categories are private
and their default debug output will be hidden.
[ChangeLog][QtCore][Logging] Enable qCDebug's for all categories except qt one's
Change-Id: Ibe147c8bbe0835a63b3de782288b9c3251321d8f
Reviewed-by: Kai Koehne <kai.koehne@digia.com>
RGBx8888 to ARGB32_PM is incorrectly using the RGBA8888 to ARGB32_PM
which asserts the input format is RGBA8888. Since the routine also
performs an unnecessy premul, we should be using a the generic rgba2argb
routine.
Change-Id: I7b67328f804f5f2a9664a35c04836679e8c8b8e5
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
The XDG specification for file URIs requires us to use triple slashes in
file:/// for URLs with absolute paths. I don't like special-casing any
particular scheme, but we've done it for file for many years now.
Since we need to test this situation in a couple of places, it's easier
to just cache the result once, in setScheme (both functions).
Change-Id: I078b45b5b6c861f4caee082b4730fd6f67684ae4
Reviewed-by: David Faure <david.faure@kdab.com>
Calling this function associates the given native
toolbar with the QWindow. QWindow will then set
it on the native NSWindow at the appropriate time
during window creation.
Change-Id: I2a50f79b2a0453cc739f8d68e965e37b95998083
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
QNSWindow inherits QNSPanel regardless of the kind of QWindow. With
NSPanels, pressing escape is equivalent to dismissing the panel.
That is not the behavior we aim at for normal toplevel QWindows.
Task-number: QTBUG-36694
Change-Id: I1a1c138183cc43c968249442021f334cd37f7ce0
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>
This patch optimizes the unpremultiply method further by using a
lookup table to avoid any divisions at all.
The opportunity is taken to export both premultiply and unpremultiply
since they are commonly used methods relevant to the exported QRgb type
that can be both premultiplied and unpremultipled ARGB.
[ChangeLog][QtGui][QColor] Exported highly optimized methods for
premultiply and unpremultiply of QRgb values.
Change-Id: I658bcf57b0bc73c34c1765b64617d43b63ae820b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Gunnar Sletta <gunnar.sletta@jollamobile.com>
Don't delete the thread object without detaching it, use detach()
instead of CloseHandle(), and avoid a double-delete.
Change-Id: Ia169a96fb32805e06abe099c3c35e97ce485f088
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
private api, but useful for 2d gui code that suddenly finds itself
needing to get a set of textures onto some fbo
I didn't want to include ARB_copy_image since it looks like its from
texture/renderbuffer -> texture/renderbuffer while this class implies
texture -> write fbo. We could wrap ARB_copy_image in QOpenGLTexture or
some other class or we can add it later.
I have not added any QOpenGLTexture functions since this class opperates
on the GLuint identifier. We can add overloads later.
Change-Id: I3e565b33466c1c183a249a33c3e82c6786debd55
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
NVDA for example reads name and value, so that most of our
accessibles would lead to the same text being read twice in a row.
Instead use Name as that's best supported on all platforms.
[ChangedLog][QtWidgets] Fixed accessibility issues that would lead to
screen readers reading the same text twice.
Task-number: QTBUG-36678
Change-Id: I6c5c9cdcf5392c7135a65bd30f87a590c3c07fb4
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>
The basedir xdg spec says:
"All paths set in these environment variables must be absolute.
If an implementation encounters a relative path in any of these variables it
should consider the path invalid and ignore it."
Therefore we ignore relative paths including the empty string.
Change-Id: I8f779b78981018051b16de23b2514f2e62b7ab39
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
tst_QDir::tempPath already checks that there is no trailing slash.
Except of course when the path is "/" or "C:/", but we can't do much
about that unlikely corner case.
Change-Id: If71d5de1aeebc6720348cecbf659b7fceb83fb0e
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
This is a no-op because QDir::cleanPath() already takes care of removing
the trailing slash.
Change-Id: Ic19d9a9dd7e377e04447c3ebc776b025f5f0c43a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Instead of "blacklisting" Unix platforms that don't have Linux headers,
do the opposite and check for Q_OS_LINUX in the framebuffer code that
has Linux-specific code.
This should help fix the build on other Unices, such as the BSDs.
Change-Id: Icb6edf34bb20c9a98843b6dc0e2a87ee71fcd046
Reviewed-by: Laszlo Agocs <laszlo.agocs@digia.com>
This change does not modify the actual algorithm implemented. It only
updates the source code so that the code generation is more optimal:
- change only one variable per loop (the "offset" variable)
- unroll the tail expansion of the last 15 characters
The Neon code for the toLatin1 codec most likely benefits from the
unrolling of the tail too, but I can't verify that I haven't broken
anything.
Change-Id: I8a92fd3c1aa700e6f8b0c8ebdb1978ade394757f
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Laszlo Agocs
Frederik Gladhorn
Friedemann Kleint
Marc Mutz
Eskil Abrahamsen Blomfeldt
Paul Olav Tvete
Kai Koehne
Tor Arne Vestbø
Sze Howe Koh
Thiago Macieira
Gunnar Sletta
BogDan Vatra
Sergio Ahumada
Jason McDonald
Christian Strømme
Christian Strømme
Jorgen Lind
John Layt
Andy Shaw
Bernd Weimer
Jerome Pasion
Shawn Rutledge
Giuseppe D'Angelo
Peter Hartmann
Tor Arne Vestbø
Oswald Buddenhagen
Nikolai Kosjar
Eike Ziller
hjk
Thorbjørn Martsum
Sarunas Valaskevicius
Leena Miettinen
Jani Heikkinen
Alessandro Portale
Allan Sandfeld Jensen
Martin Gräßlin
Konstantin Ritt
Fabian Bumberger
Richard J. Moore
Lars Knoll
Mandeep Sandhu
Oliver Wolff
Yoann Lopes
Sean Harmer
Olivier Goffart
Erik Verbruggen
Jan Arve Saether
Morten Johan Sørvig
Glen Mabey
Richard Moe Gustavsen
Andrey Volkov
Andrew Knight
David Faure
Gabriel de Dietrich
Maurice Kalinowski
Sérgio Martins
J-P Nurmi
Samuel Gaist
Topi Reinio
Joerg Bornemann
Giorgos Tsiapaliokas
Jorgen Lind
Raphael Kubo da Costa