This test runs fine almost all of the time on systems with 1
processor, which were the norm when the test was written and are still
the way that the Qt Continuous Integration system works as of
today. But it falls flatly on multi-processor systems.
The root of the problem is that QSystemSemaphore recreates the
semaphore if it disappears underneath it. However, the recreation
process is not thread-safe at all: if two threads race to recreate it,
weird things might happen. strace on Linux shows that a thread got
stuck trying to acquire the semaphore:
<... nanosleep resumed> NULL) = 0
stat("/tmp/qipc_systemsem_market5c9f73af73334ffe350c60ec076e5744db0ecda3", {st_mode=S_IFREG|0640, st_size=0, ...}) = 0
stat("/tmp/qipc_systemsem_market5c9f73af73334ffe350c60ec076e5744db0ecda3", {st_mode=S_IFREG|0640, st_size=0, ...}) = 0
semget(0x51001388, 1, IPC_CREAT|IPC_EXCL|0600) = -1 EEXIST (File exists)
semget(0x51001388, 1, IPC_CREAT|0600) = 114786308
semop(114786308, {{0, -1, SEM_UNDO}}, 1 <unfinished ...>
This problem does not happen if the creation and destruction of the
QSharedMemory (which uses QSystemSemaphore) does not race with other
threads or processes attaching and detaching. For the threads test
it's easy. For the processes, we use stdin and stdout as a
communication channel.
Change-Id: Ie11b135431d4abfc59234654848b67f622eb03c9
Reviewed-by: Richard J. Moore <rich@kde.org>
On Mac, GCC and Clang print "(framework directory)" at the end of the
directory listing for includes if that's a framework dir. The directory
with "(framework directory)" at the end does not exist, so it doesn't
affect anything by being there.
But we don't have to keep the list longer just for that.
Change-Id: I3d031d3d15c75801ec0d6112b2c913bd63e5def3
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
These headers are technically private API, since they include private
headers. They should be _p.h actually, but that change I'll leave for
5.1.
Change-Id: I2dec222854e147da0fc166de311012472954012e
Reviewed-by: Mark Brand <mabrand@mabrand.nl>
Task-number: QTBUG-29056
QOpenGLContextGroup object is designed to be destroyed by deleteLater(),
but this method will not always work due to the fact that in many cases
event loop will exit before the deferred deletion of the
QOpenGLContextGroup object is queued. Think about the following case:
int main(int argc, char *argv[])
{
QApplication a(argc, argv);
QGLWidget w;
w.show();
return a.exec();
}
In the above program, the event loop will exit before QGLWidget object's
destruction. This will cause the QOpenGLContextGroup object hold by
QGLWidget object never been deleted.
This patch will delete QOpenGLContextGroup object directly with delete
operator if the current thread is the same as the thread which the
QOpenGLContextGroup lives in.
Change-Id: If835d7482474f4a668763fc7c21b293a27f075fd
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
On Mac if AssertMacros.h has already been included then it defines
verify which conflicts with the verify static function. Therefore we
just undef this if is already defined.
Task-number: QTBUG-27316
Change-Id: I5960e504c4efa4fc4ff65ba66bbd7decb33ffc62
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
Introduce a function named deleteClonedAnimationStyleOption() in
qwindowsvistastyle.cpp to delete style option cloned by
clonedAnimationStyleOption() to prevent object leak which is hold by the
actual style option object.
Change-Id: I1afd95ddab237059ce3460ac0b52a65de102eaa5
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
One of the IPv6 autotests was always disabled instead of being disabled
only when the system has no IPv6 support.
Change-Id: I34dffbeae6ba85a706bfeb0cc4750a4514b73a65
Reviewed-by: Peter Hartmann <phartmann@rim.com>
The test for a magic reason was not build by default.
Change-Id: I21c7fc959d76d6faac0091495f965f3da6d415b1
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
This was another bug resulting from modularization.
qdoc is now run twice for each module in Qt 5.
First, qdoc is run with the -prepare flag for each
module. The only thing qdoc generates is the .index
file for each module. Then qdoc is run with the
-generate flag for each module. Here, for each
module, qdoc first reads the .index files for the
modules on which the current module depends. Then
qdoc generates the docs for the module.
qdoc was not reading the index files for the
prerequisite modules, when it was run in the
-prepare phase. This has now been corrected.
qdoc now reads the prerequisite .index files
in both the -prepare phase and the -generate
phase.
Note that this requires that the order qdoc
runs in the -prepare phase must be the same
as the order of building modules when building
Qt 5.
This change also tells qdoc to ignore nodes,
when traversing its main data structure to
output docs, if the nodes came from reading
a .index file, because the docs for these
nodes are generated in the -generate phase
for their respective modules.
Task-number: QTBUG-28508
Change-Id: Id73652ae1c4022e4c9e4f199caab12a854e5f9b4
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
\obsolete for one value was marking the whole enum obsolete.
Change-Id: If022b4af8365e3448556ca11a538203ab5c63c78
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
the KDGETMODE ioctl should use a pointer to int as param,
not a pointer to pointer to int, otherwise it may crash
Change-Id: Ie255d240f3b6ca1ff5398f972308116135374ae3
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>
The mingw headers lack the IPV6_V6ONLY define, depending where you get
them from. Currently the headers provided by mingw-builds are more
complete than those from mingw.org itself.
I have removed the compile time check, defined the macro if it is
undefined, and it should be just a runtime check for if you are running
on windows XP you get no dual stack binding.
Task-number: QTBUG-28787
Task-number: QTBUG-28971
Task-number: QTBUG-28972
Change-Id: Iafadbb55d367c44ba9f812a24115e65591701b54
Reviewed-by: Ilya Sidorov
Reviewed-by: Peter Hartmann <phartmann@rim.com>
Nobody should break ODR, but we were living with apps breaking it
for at least 8 years. Sure it is an undefined behavior but in many
cases it works. I do not think that Qt should enforce usage of
a proper C++, it is role of a tool chain. Qt can only indicate that
something is going terribly wrong.
Make message a bit more verbose. Sometimes the ODR violation is not
that easy to fix, therefore a hint is a nice addition.
Update documentation of qRegisterMetaType.
Change-Id: I61dcccc840eec80a4ed5b8a212a912807d239d8c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Both Q_UNREACHABLE and Q_ASSUME with an invalid condition can produce
really weird side effects and crashes.
Change-Id: I4d808c705ae98388ef5853e6539b70bd5e5ad34b
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The value -1 is used to indicate no stencil buffer, so we have to check
for sizes <= 0 when deciding whether or not to triangulate the path in
QOpenGL2PaintEngineExPrivate::fill().
This fixes an issue where filling a path would end up filling the whole
outline of the path, which was very visible with fonts over a certain
size (when we go from using the glyph cache to drawing filled paths
for each glyph).
Change-Id: Iafa96124481936db1e5109bba6166a6038c7ca83
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@digia.com>
A new qmake's variable RC_ICONS is introduced to specify the icons.
The first one will be used as the application's icon.
Change-Id: I4218db795837d470087dff8526eb0e4cb81ce5ed
Reviewed-by: Orgad Shaneh <orgads@gmail.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
If the widget is embedded in a native window then pos() should be used
instead of mapToGlobal() so that the right position is used. This was
reproduced with the qtwinmigrate solution as the dialogs were not
centered correctly.
Change-Id: I2ce7771f8c1a73aa74ab11faf4f9c57b922eefab
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
Fixes breakage introduced by cdc436ebe6 .
Top level widgets with translucent backgrounds were invisible since
opacity was set to 0.
Task-number: QTBUG-28531
Change-Id: I97058ac1b971422f3bda3a5ffed479ec55bfe5d4
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
CGRect == NSRect only in 64-bit mode.
Change-Id: I069b5b050ccf02654a65375c3ab0f58f7d5cc659
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>
Don't autorelease in the elementWithInterface function,
rename it to createElementWithInterface. The element
can then be released immediately or autoreleased ("delete
later").
Change-Id: I155a85404c34d756c1752eb7c24a7fb0f3cf2e77
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
'timex' is not a Qt class, so there is no need to check it.
Change-Id: Ic77b3518e5a7eaf2c2bc7dcd98d1f9aebf4b655d
Reviewed-by: Caroline Chao <caroline.chao@digia.com>
Instead of testing the "windows" and "motif" styles, that test should
actually test the scrollBarRanges for all available styles. To be able
to do that, the magic numbers 16 (width/height of scrollbars) and 4
(spacing for the faux motif style) were replaced and instead of setting
the explicit values in the data the "number of scrollbars/spacings to
add/remove" is saved in a struct and the value of these (depending
on the style) is obtained in the test run.
This change does not also cause the fusion style to also be tested but
also fixes this test for Windows 7 and 8 (Aero) where the scrollbar
width/height is not 16 but 17.
Task-number: QTBUG-28611
Task-number: QTBUG-29002
Change-Id: I5d103018fde81cee6e6e89cd414426768b2dc8e7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
It mustn't be installed and it mustn't be available in the master
include. It's an assembler header anyway.
Also, take the opportunity to add the "We mean it".
Change-Id: Id4233e5bda458714a0d329549afe499d861acc56
Reviewed-by: Adrian Perez de Castro <aperez@igalia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>
This definition was missing. It's clear MSVC supports this feature
because it has been in use in qtconcurrentrun.h under an #ifdef
Q_COMPILER_DECLTYPE, which is defined for MSVC.
Change-Id: Id76198b2be8bba13fd00c65d6d4f73d3ef601449
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
QQnxCliboard::setMimeData was not cleaning the system clipboard aproppriately
when m_mimeData != 0 but m_mimeData->userMimeData == 0, as it would return
early before the empty_clipboard() statement.
Also, we need to emit the dataChanged() signal after the clipboard has been
cleared as well.
Change-Id: Ie06609e6e3a0a0135f1cdb6cf4c90b490ef2fc28
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Fully decorated windows cannot be smaller than 160x30 (Large fonts).
Enlarge Windows or remove Window frame to get rid of decorations.
Task-number: QTBUG-28611
Change-Id: Idb6ee94fb8d0760d5f97042b3084557f11e9fdf9
Reviewed-by: Oliver Wolff <oliver.wolff@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Adds an auto test that verifies that the replies to POST and PUT
requests are not cached even though they contain a max-age header and
that subsequent GET requests are reloaded from the server.
Change-Id: I188ae1200cb5551e164722c0f479719be8d11bfb
Reviewed-by: Peter Hartmann <phartmann@rim.com>
these variables are somewhat magic: they are relative to DESTDIR, and
they also specify the installation location relative to target.path.
the actual output directories are created by other code.
Change-Id: Iead3006057516f5a49b6fd4bd8996a062c984fce
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Both qdoc and Q_QDOC are used in source code, which looks not good.
Change-Id: I4f3a71670278b0758d92bfa5db086a07e1b1acfd
Reviewed-by: hjk <qthjk@ovi.com>
Reviewed-by: Paul Olav Tvete <paul.tvete@digia.com>
qApp->topLevelAt() returns the QShapedPixmapWindow that the DnD
operation created, but what we want is the QWidgetWindow.
QWidgetWindow has the code to handle QDragMove events.
In Qt4, QShapedPixmapWidget had a parent, so was never returned
by qApp->topLevelWidgets().
In Qt5 we must filter it out. Bug is visible in the QNX plugin,
which is the only user of QSimpleDrag.
Change-Id: I920da86f3a1a92ce8e087f5948292fa4c68d4d81
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@digia.com>
A regression caused by 79a389 - it should only draw a styled frame for
text input widgets.
Task-number: QTBUG-28876
Change-Id: I5cda920b7db55f35e0ab6cee460c0e27951d29b9
Reviewed-by: Jonathan Liu <net147@gmail.com>
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>
Cross bulding on unix for mingw exploits the fact that makefiles
generated for mingw work with plain make. There is no mingw32-make, so
this is the only option.
Arguably, plain make could also be used in an MSYS environment,
perhaps detected by MINGW_IN_SHELL, but there might be good reasons I
don't know about not to do this.
Change-Id: I694c74046a307c2887af1c30cca36f95e242adc1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The operator<= and operator>= overloads for QPoint are used only with
Q_ASSERT. Clang was correct that they are not used.
Even though they are inline, Clang reports the functions as unused
because they were inside an anonymous namespace. So take the off the
namespace. They are correct for external use, should they be exported
by accident (which they aren't now).
painting/qpathsimplifier.cpp:76:13: error: unused function 'operator<='
[-Werror,-Wunused-function]
inline bool operator <= (const QPoint &a, const QPoint &b)
Change-Id: I67415621e777fd0d59cbdede26c34d8bb13f7346
Reviewed-by: Gunnar Sletta <gunnar.sletta@digia.com>
- Add initial set of amd64 data
- Add QtWidgets to global.cfg
Task-number: QTQAINFRA-321
Change-Id: I9bffaa00ac01976546629988ac69965383eb2efd
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Add additional tests for the updates in the Qt4.8 qlocale data (CLDRv2.0) already existing in Qt5 locale data (CLDRv22.1):
- The NumberingSystem for some Indic and Slovak locales.
- The Month/Day name in Irish/Gaelic locale.
- The AM/PM Text in Turkish locale.
Change-Id: Iaea4f13ec79f94ab937b97f8ae60eb8d8f217c4b
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
