| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
Symbian generates a directory change event when a file is overwritten in
that directory.
Autotest was asserting that no change event is generated in this
situation (documentation allows it in the signal description)
So the test now allows 0 or 1 dir changed events, to allow existing
platforms and symbian to both pass.
Reviewed-by: brad
|
|\ |
|
| |
| |
| |
| | |
That way we will avoid duplicated functionality.
|
| |
| |
| |
| | |
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
|
| |
| |
| |
| |
| | |
RevBy: Trust me
AutoTest: It built...
|
| |\ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This was caused by the widget not calling update on the input context
after changes to the cursor position.
This is a reimplementation of 8d1691e9395, after the QLineControl
refactoring.
RevBy: Trust me
AutoTest: Passed
|
| | | |
|
| | |
| | |
| | |
| | | |
autotest repo.
|
| |\ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
The change is actually correct, as we only need the commented out
codepath for debugging.
|
| |/ /
| | |
| | |
| | |
| | |
| | | |
Conflicts:
src/gui/styles/qwindowsstyle.cpp
|
| | |
| | |
| | |
| | |
| | |
| | | |
In some alternate universes it might make sense to use a pointer before
checking if it is valid, but since this isn't one we shoudn't do
it. Also extend the autotest a little to make sure we catch this case.
|
| |\ \ |
|
| | | |
| | | |
| | | |
| | | | |
attribute in Symbian in a meaningful way.
|
| | |\ \ |
|
| | | | | |
|
| | |/ /
| |/| | |
|
| | | | |
|
| | | | |
|
| |\ \ \ |
|
| | |/ /
| | | |
| | | |
| | | | |
forwards the event and doesn't just send a fixed one.
|
| | | | |
|
| |/ / |
|
| | |
| | |
| | |
| | |
| | | |
Turns out it's way too much work to make all the Pixmap backends
strongly exception safe. Bumped to "future Qt version" :)
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The wrong version of cancel was used, resulting in all filesystem
notifications being cancelled when one path was removed from the engine.
For directory notifications, changed ENotifyAll to ENotifyEntry to
reduce the spurious directory changed notifications which come from
file modifications within that directory.
On Symbian OS, opening a file for overwrite (truncate mode in Qt terms)
will cause a directory change notification, the OS considers it to be
the same type of event as creating a new file for the first time.
Other platforms presumably do not have this behaviour otherwise the auto
test would fail there as well.
Reviewed-by: miikka heikkinen
|
|\ \ |
|
| | |
| | |
| | |
| | | |
this was only accessible from Nokia network.
|
| | |
| | |
| | |
| | |
| | |
| | | |
different server machines.
On Symbian we read replies from external file,
for others we put data into network-settings.h
|
| | |
| | |
| | |
| | | |
This commit fixed QLabel::task226479_movieResize autotest.
|
| |\ \ |
|
| | |\ \ |
|
| | | | | |
|
| | | | |
| | | | |
| | | | |
| | | | | |
Makes render test case to pass at least on emulator.
|
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | | |
The scrollBarRanges test case is style specific, but we were using
s60 style to run the test. Added Q_OS_SYMBIAN ifdef to enable windows
style for this particular test case.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Started to use QApplication::closeAllWindows instead of unexported
QThreadData::current() and friends.
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The implementation is likely not complete, but makes autotests using
QApplication::activeModalWidget to pass. At least the following test now
pass successfully:
qcolordialog
qmessagebox
Might have positive impact to other autotests as well.
|
| | |
| | |
| | |
| | | |
tst_qobject deleteQObjectWhenDeletingEvent needs QtGui.
|
| | |
| | |
| | |
| | | |
due to http://qt-reviews.europe.nokia.com/r/67/
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
It happens during the configure phase on Non-Windows. When making
Qt for S60 build on Linux, we may want to revert the change or have
a different fix.
Reviewed-By: Jason Barron
|
| |\ \ |
|
| | | | |
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Add a bracket to the right place and re-arrange the code a little.
The indenting was wrong here in the non-Symbian case since the
if (exists) did not have curly bracket. Also copy the union of
the Hidden flag into the Symbian #ifdef to make this code slightly
more readable.
|
| | |
| | |
| | |
| | | |
As per Marius' S-O's request :)
|
| | | |
|
| | |
| | |
| | |
| | |
| | | |
Ensures that we don't leak when creating a new Pixmap while running out
of memory
|
| | |
| | |
| | |
| | | |
First step to get rid of QCustomScopedPointer class
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
as we cant expect that order of the event
hansling will be same on all platforms.
For example, on Symbian quit will happen
first, on Linux not.
It is better to check if we processed 2
events in total.
Reviewed by: Axis
|
| | | |
|