| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
| |
the temporary cache rather than QDesktopServices::DataLocation
Merge-request: 1124
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes a regression introduced in 4.5.2 where QTemporaryFile would
no longer attempt to generate a random name after a failed open. Under
certain situations, this led to a non-random file being left behind in
QDir::tempPath when using the fallback implementation of QFile::copy.
Avoid calling QFSFileEngine::setFileName() on a template, so as not to
process it as file name.
By consistently not calling setFileTemplate in the constructor, we also
delay allocation of the fileEngine. Changes made to that function also
keep it from unnecessarily allocating the fileEngine.
Task-number: 260165
Reviewed-by: Thiago Macieira
|
| |
|
|
|
|
|
| |
ietf.org changed their server to use gzip, therefore our check for the
content-length reply header was bogus.
Rev-By: Peter Hartmann
|
| |
|
|
|
|
|
|
| |
This implicitly adds "grayscale" support for indexed 8, but
only for the conversion. The alternative would be leave the pixels
uninitialized which would be less nice...
Reviewed-by: Samuel
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| | |
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
|
|
|
|
| |
Simple misuse of QLatin1String. Use QString::fromLatin1 instead and
avoid the QByteArray temporary.
Reviewed-by: Andreas Aardal Hanssen
Tracking: CVE-2009-2700
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| | |
|
| |
|
|
|
|
|
|
| |
The subControlRect of the arrow and the listBoxPopup where assuming
the widget rect is at the origin.
Reviewed-by: Richard Moe Gustavsen
Reviewed-by: Pierre Rossi
|
| |
|
|
|
|
|
|
| |
We integrate QMenu into the native menu system. Thus it doesn't make
sense to click at position (5,5) or so. We have the windowsmobile
auto test for this.
Reviewed-by: thartman
|
| |
|
|
| |
Reviewed-by: thartman
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes a regression when computing the clip path for a
QGraphicsWidget that clip its children to its shape. The clip path cache
use the bounding rect which has a null width and height the very first
time on a QGraphicsWidget. This might happen as well for any item that
have a null width and height (e.g. QGraphicsRectItem before you set the
rect). We should do better in mainline by refactoring this clip path
cache and calling it in prepareGeometryChange which should be called
when you update the shape and the boundingRect of the item. Do we
really need also in 4.6/mainline the clip path as it is now? This should
be investigate also. The auto-test cover the problem so we can use it
after for refactoring the cached clip path.
Task-number: 257232
Reviewed-by: andreas
Reviewed-by: bnilsen
|
| |
|
|
|
|
| |
Fixes https://bugs.webkit.org/show_bug.cgi?id=28016
Reviewed-by: andreas
|
| |
|
|
|
|
| |
test was broken
Reviewed-by:Bill King
|
| |
|
|
|
|
| |
We run out of memory on the test system in this test.
Reviewed-by: mauricek
|
| |
|
|
|
|
|
| |
Nowadays every Windows mobile application has a menu (for correct soft
key support). That's why we need a new screen shot.
Reviewed-by: thartman
|
| | |
|
| |
|
|
|
|
| |
The screenshots in the QRC were taken on the WM5 emulator.
Reviewed-by: thartman
|
| |
|
|
| |
Reviewed-by: thartman
|
| |
|
|
|
|
|
|
|
|
|
|
| |
The style was assuming that the combo box is painted at (0,0). This is
not the case when the painting is done in the delegate of an item view.
The offset of the rect is now taken into account to paint the style.
HIRect has been replaced by QRect when it make sense.
Task-number: 00026815
Reviewed-by: Richard Moe Gustavsen
|
| |
|
|
| |
Reviewed-by: Leo
|
| |
|
|
|
| |
Reviewed-By: mauricek
Reviewed-By: thartman
|
| |
|
|
|
|
|
|
| |
This test created a widget at position (0,0) and grabbed the screen.
That's a bad idea on Windows mobile because the upper task bar will
cover the widget.
Reviewed-by: thartman
|
| |
|
|
|
|
|
|
| |
The widget must be created before calling QInputContext::setFocusWidget.
Otherwise we run into an assertion. Yes, this only occurs in debug
configuration but its still annoying...
Reviewed-by: thartman
|
| |
|
|
| |
Reviewed-by: TrustMe
|
| |
|
|
|
| |
Make sure the testcase and directory name are the same
(excluding `tst_').
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
| |
Reviewed-by: Trust Me
|
| |
|
|
|
|
| |
Kramer is no longer accessible from the pulse test system
Reviewed-by: jesperht
|
| |
|
|
|
|
| |
This is a vulnerability in some implementations. Qt isn't affected
because... well, we never implemented the decoding of escape sequences
:-)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
QTemporaryFile on Windows doesn't open the file as a sharable, and
doens't close the file when you call .close(). So the testcase
fails on Windows with a Sharing Violation when the compiler tries
to compile the file.
By switching to QFile we can at least close the file before letting
the compiler chew on it, and remove it at the end when the testcase
is done. Open the file with Truncate, in case the testcase fails to
remove the file.
Reviewed-by: trustme
|
| | |
|
| |
|
|
|
|
|
|
| |
On mac, we prefer testing with mp3 rather than ogg, Because in the latter
case, you must install perian to make it work. And the test machines
dont.
Reviewed-by: brad
|
| | |
|
| |
|
|
|
|
|
|
| |
The test was failing because the goal of the test expect to get only
directories displayed. But if you don't have some in the current dir
then it fails. home dir should be enough.
Reviewed-by:TrustMe
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
When an autotest fails to compile, there's no way to determine which
testcase(s) have been prevented from running.
Our results parsing tools have been guessing, under the assumption that
a directory called `qdogwalker' always contains a testcase named
`tst_qdogwalker'. That wasn't true for all our tests, so let's make it
true.
Also changed the platform-specific tests so that qmake will simply skip
those tests on unsupported platforms, instead of wasting time compiling
a useless QTEST_NOOP_MAIN test.
|
| |
|
|
|
|
|
|
|
| |
If there is severals matching items but all of them are disabled, it
would have produced an infinite loop.
Block the loop once we reach an already matched item.
Task-number: 258949
Reviewed-by: Thierry
|
| |
|
|
|
|
|
|
|
|
| |
We never told Cocoa that it needed to redraw the window view
when a window was shown. This is implicit if the window is
shown for the first time, but needs to be done explicit
if you hide and show it again.
Task-number: 254672
Reviewed-by: bnilsen
|
| | |
|
| |
|
|
|
|
|
|
|
| |
updateAncestorFlags was not reseting the flags if you change the parent
that have for instance itemsClipChildrenToShape to a new one that
doesn't have that flag.
Task-number:258956
Reviewed-by:bnilsen
|
