| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
The path /lib/ld-linux.so.2 is correct for i386 only, possibly some
other archs. But on x86-64, it's usually ld-linux-x86-64.so and it can
be in either /lib or /lib64.
|
|
|
|
|
|
|
|
|
|
| |
Use the fully qualified classname at relevant places in the
moc-generated code. Also, QMetaObject::newInstance() needs to
strip the namespace part, since the constructor signatures
don't contain the fully qualified name.
Task-number: 246064
Reviewed-by: Simon Hausmann
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
|
|
|
|
|
| |
enabled.
This allows mixing of debug and non-debug code (possible on Unix
systems) without causing assertion failures.
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
|
|
|
|
|
|
|
| |
Two strings are only equal if they have the same size.
Task-number: 256763
Reviewed-by: TrustMe
|
|
|
|
|
|
|
|
|
| |
This doesn't work, the paths are wrong. It's not properly checking the
arch, and none of our compilers use it. Unsupported ones might, but I'd
rather leave it blank and let them figure out how to do it correctly
then lead them down a blind alley.
Discussed with Morten Sørvig
|
|
|
|
|
|
|
|
|
|
|
|
| |
Snow Leopard defaults to 64-bit if you don't specify an architecture,
whereas in the past it defaulted to 32-bit. This isn't a problem for Qt
per-se. It *is* a problem if you just build an application that uses Qt
but isn't using qmake stuff. To help in those situations, we should
error out to let the person know that they need to change their
configuration (in any case, the headers are going to complain and they
get a much more cryptic message).
Reviewed by: Morten Sørvig
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In 4.4 QRectF handled flat rectangles in the same fashion as QRect does,
but that changed with Lars' and Jo's optmizations done in the falcon
branch. The difference is that the optimized version only checks whether
the width or height is 0, whereas in 4.4 both had to be 0 (isNull()) before
we bailed out.
This regression also introduced a regression in
QGraphicsItem::childrenBoundingRect().
Auto-test included.
Task-number: 254995
Reviewed-by: Lars
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
|
|
| |
Changed to QIODevice::isSequential()
Reviewed-by: TrustMe
|
|
|
|
|
|
|
|
| |
This patch prevents the eventloop from waking up needlessly.
Without this patch the event loop will not sleep at all if a 0-timer is already 'inTimerEvent'
Merge-request: 550
Reviewed-by: Olivier Goffart <ogoffart@trolltech.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
The BOM was created correctly, but half of the BOM was then overwritten
by the converted data afterwards.
Also made the autotest also do reverse encoding tests where possible.
Task-number: 255095
Reviewed-by: lars
|
|
|
|
| |
Reviewed-by: Kavindra Palaraja
|
|
|
|
| |
Reviewed-by: David Boddie
|
|
|
|
|
|
|
|
|
|
| |
in atomic operations, we declare Interlock... functions in the namespace
That can confuse the compiler because they are also declared in another
header outside the namespace.
Same problem in clucene where we include windows.h from within the NS.
Task-number: 254214
Reviewed-by: ogoffart
|
|
|
|
|
|
|
|
|
|
|
| |
Use a union instead of an unsafe cast when swapping the bytes in
the QDataStream streaming operators. The old seems to cause problems
with Link Time Code Generation optimizations with the MSVC compilers.
Task-number: 247708
Reviewed-by: Samuel
Reviewed-by: Thiago
BT: yes
|
|
|
|
| |
Task-number: 245501
|
|\ |
|
| |
| |
| |
| | |
Reviewed-by: Trust Me
|
| |\ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Adding details to the documentation of custom storage format and related functions.
Task-number: 207865
Rev-by: David Boddie
Rev-by: Marius Storm-Olsen
|
| |/
| |
| |
| | |
Reviewed-by: Trust Me
|
|/
|
|
| |
Task-number: 246789
|
|
|
|
| |
Task-number: 252491
|
|
|
|
| |
Reviewed-By: Oswald Buddenhagen
|
|\ |
|
| |
| |
| |
| |
| | |
Task-number: 254333
Reviewed-by: Andy Shaw <qt-info@nokia.com>
|
| |
| |
| |
| |
| |
| |
| | |
When connecting to an open file descriptor, set the openMode in the file
system engine, as is done for file handles.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This gets temporary file renaming working on Windows, without requiring
block-copying.
While we could #ifdef this behavior for Windows, it's preferrable to
maintain consistency in the exposed interface.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
With recent changes to QTemporaryFile, allowing the file to be closed,
the engine has to keep track of whether a fileName has already been
generated, so we don't generate new files after the first one.
If the file is closed but we already have a name for it, then just
forward the call to the base file engine.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| | |
Lifetime of the engine is already handled by the native engine.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
In some circumstances, the file descriptor in QTemporaryFile is actually
closed and setOpenMode alone won't give us reOpen semantics.
Added function to QTemporaryFileEngine that checks if we have open file
handles. On open, if we currently hold no handles, re-open the file.
Trying to open a new file while we hold open handles would lead to
leaks, so added an assert there, to be on the safe side.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Although QTemporaryFile hides QFile::open(OpenMode), this function is
still available when accessing instance methods through the base class.
Unconditionally setting ReadWrite allows the temporary file to be
re-opened with different flags.
Task-number: 248223
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| | |
For now, this only happens if Append mode is requested and we're unable
to seek to the end of the file. Theoretically, this could change in the
future so it's better to err on the safe side.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| | |
We souldn't be returning an empty string for the fileName, just because
the file is closed. E.g., after a rename, the file will be closed, but
should still have a name.
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| | |
Note: this showed even if the file descriptor was kept open.
Reviewed-by: Peter Hartmann
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| | |
Reviewed-by: MariusSO
Reviewed-by: Peter Hartmann
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fallback implementation for rename operation should try to copy
permissions from the original file to the destination file. Note that
failures at this point are not treated as errors.
Errors previously set by the native fileEngine are also reset before
returning.
Reviewed-by: Peter Hartmann
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| | |
Also added check in test case for rename fallback.
Task-number: 165920
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Changed the fallback implementation to use 'this' instead of a new
QFile. This allows a QTemporaryFile to be block-copied to the
destination and the source to be removed (QTemporaryFile is special
because it isn't really closed).
Reviewed-by: Peter Hartmann
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Block copying a sequential file is potentially a destructive operation,
because there is no guarantee copy+remove will succeed. In these cases
the fallback should not be tried.
The user is better equipped to decide how to handle such failures and to
ensure no data losses occur, e.g., copy without removing the original
file.
Reviewed-by: MariusSO
Reviewed-by: Peter Hartmann
Reviewed-by: Thiago
|
| |
| |
| |
| |
| |
| | |
Asked by Oswald.
Reviewed-by: Oswald Buddenhagen
|
| |
| |
| |
| |
| |
| |
| | |
I tested it with 32 bits compilation and there is no warning any more.
Task-number: 247325
Reviewed-by: ogoffart
|
|/
|
|
| |
Task-number: 252494
|
|
|
|
|
|
| |
Reviewed-by: Olivier Goffart
Reviewed-by: Peter Hartmann
Reviewed-by: João Abecasis
|
|
|
|
|
|
|
|
|
|
| |
currentFileInfo is only used for returning from the public functions
since the file info used in the algorithm is one step ahead.
nextFileInfo is the one actually used in the algorithm. The bug was
introduced in a compile fix for Windows and broke the stopLinkLoop test
for QDirIterator.
Reviewed-by: Olivier
|
|
|
|
| |
Reviewed-by: David Boddie
|
|
|
|
|
|
|
|
| |
Adding more details on QUrl::addQueryItem()
Task-number: 234125
Rev-by: Thiago Macieira
|
|
|
|
| |
Reviewed-by: Marius Storm-Olsen
|