| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Destroying a QObject which is connected to many signals was slow.
We had to loop over all the ConnectionsList of all the connected object
to remove the connections. (in QObjectPrivate::removeReceiver)
The idea is to now use lists of pointer to Connection both in the
connection list, and in the list of senders. So clearing the connection
should be faster.
Task-number: 251097
Reviewed-by: Brad
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I'm tired of these "hidden" functions. We have an
AA_MacPluginApplication, but sometimes you may have a legitimate reason
for setting this outside of "plugin applications." In the footsteps of
the menu icon attribute, the attribute is the main leader, but menubars
can disable/enable this locally the new QMenuBar::setNativeMenuBar()
property. Otherwise, the menubars take their que from the application
attribute.
This also works for Windows CE. So, there is a bit on convergence as
well.
Task-number: 236757
|
| |
|
|
|
|
|
|
| |
- Windows Mobile does the same, thus we will not have double definitions or such
- COINIT_MULTITHREADED is the default, meaning that 0 usually expands to it. Just be on the safe side and be more precise.
Task-number: 237029
Reviewed-by: joerg
|
| |
|
|
|
|
|
|
| |
These were caused by a comparison between QVariant::Type and
QMetaType::Type. We're using QMetaType::Float because QVariant have no
such Float value
Reviewed-by: denis
|
| |
|
|
| |
also added autotests
|
| | |
|
| |\ |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
constructed
Fix a regression from 4.3 to 4.4 that prevented QThread from starting
timers and event loops before QCoreApplication was instantiated. Even
though this is documented not to work, it seems that people have been
relying on it.
This reverts commit e52e5acdfa198cd079bbfe3a9302debf46c7cadd which
attempted to work around not calling g_thread_init() multiple
times. The proper fix is to serialize the g_thread_supported() checks
in the Glib event dispatcher.
Task-number: 248191
Reviewed-by: denis
|
| |\ \
| |/
| |
| |
| |
| |
| | |
Conflicts:
src/corelib/global/qfeatures.h
src/gui/painting/qtransform.cpp
util/scripts/make_qfeatures_dot_h
|
| | |
| |
| |
| | |
Reviewed-by: Thiago
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The NSOpenGLContext seems to be tied to the window. So if the view
changes from one window to another, the OpenGL context needs to be
cleared. We can do this by hooking into the viewWillChangeWindow and
viewDidChangeWindow events and clear and reset the drawable
respectively. We also found out that QCocoaOpenGLView was not being used
at all, so just remove it to get rid of any confusion.
Task-number: 250066
Reviewed-by: Trond
|
| |\ \ |
|
| | | |
| | |
| | |
| | | |
There is one less test thanks to usage of reference instead of pointers
|
| | | |
| | |
| | |
| | | |
performance by 20%)
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
This is possible if the type is the same of type < Char (simple
types)
|
| | | |
| | |
| | |
| | | |
constDataHelper dones't need to know about the type if the variant
|
| | | | |
|
| | | |
| | |
| | |
| | | |
Instead, it calls directly the v_construct template function.
|
| | / |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The NSOpenGLContext seems to be tied to the window. So if the view
changes from one window to another, the OpenGL context needs to be
cleared. We can do this by hooking into the viewWillChangeWindow and
viewDidChangeWindow events and clear and reset the drawable
respectively. We also found out that QCocoaOpenGLView was not being used
at all, so just remove it to get rid of any confusion.
Task-number: 250066
Reviewed-by: Trond
|
| |\ \
| |/
| |
| |
| | |
Conflicts:
src/gui/graphicsview/qgraphicsitem.cpp
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Zero timers on Windows would continue to fire even after being stopped
as long as a new timer was started that reused the pointer address of
the zero timer. Fix this by only re-firing zero timers if the zero
timer hadn't been stopped (we can check this by looking at the
inTimerEvent flag, which is set to false by registerTimer()).
Task-number: 247401
Reviewed-by: Denis Dzyubenko
Reviewed-by: Prasanth Ullattil
|
| |/
|
|
| |
Reviewed-By: Harald Fernengel
|
| |
|
|
|
|
|
|
|
| |
details about BlockingQueuedConnection in QMetaObject::invokeMethod().
Task-number: 187869
Task-number: 216742
Reviewed-by: Thiago Macieira
Reviewed-by: Morten Sørvig
|
| |
|
|
|
|
|
|
| |
Details: Added a note to the description about using POSIX functions for converting between data types such as floats and strings
Task-number: 244600
Reviewed-by: Denis Dzyubenko
Reviewed-by: Geir Vattekar
|
| |
|
|
| |
Reviewed-by: TrustMe
|
| |
|
|
|
| |
the exact length of the template is known, so there is no point in using
the more complex (and thus potentially slower) strncmp instead of memcmp.
|
| |
|
|
| |
this function is always called with found != 0 nowadays
|
| | |
|
| |
|
