| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
The patch provided by the customer just jumps to reading the next event fom the
kqueue when an unexpected condition happens.
Task-number: QTBUG-12533
Reviewed-by: Bradley T. Hughes
(cherry picked from commit b7256b1f8c9bb43c5234d24453aaa8b1c079cf06)
|
|
|
|
|
|
|
|
|
|
|
| |
... according to the online MSDN documentation.
Hid the cachedFd member in private data under WinCE, since it's never
used there.
Task-number: QTBUG-9085
Reviewed-by: Zeno Albisser
(cherry picked from commit 7986ab58b9a5d0828291c857d3ce86bfa1af4e6e)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Starting in 1216161584b730576c24fb128131838be1826b37, we started
processing NTFS junctions as symbolic link.
This reverts isSymlink() behavior back to former behavior and processes
only symbolic junctions. The code path there worked just fine for
directory symbolic links but lead to multiple issues for volume mount
point junctions due to incompleteness.
Task-number: QTBUG-7036, QTBUG-7384
Merge-request: 493
Reviewed-by: João Abecasis <joao@trolltech.com>
(cherry picked from commit b4996ab8e3a26d45ee55593da281e4b798c04dd2)
|
|
|
|
|
|
|
|
|
|
|
| |
Ideally, we should write a BOM after seek(0) if setGenerateByteOrderMark
has been called, but we don't store that information. This commit is an
improvement for the default case when generateByteOrderMark is false, and
also when seeking to any other position than 0.
Task-number: QTBUG-6295
Reviewed-by: Olivier
(cherry picked from commit 7837de27309c958dc5a5985afc55492ec3054ee0)
|
|
|
|
| |
Reviewed-by: Trust Me
|
|
|
|
|
|
|
| |
If the host contains invalid caracter, QUrl::isValid should return false
Task-number: QTBUG-10355
Reviewed-by: thiago
|
|
|
|
|
|
|
|
|
| |
This is not supposed to be allowed, but it work with other browsers
Rask-number: QTBUG-7434
Reviewed-by: Thiago
Reviewed-by: Markus Goetz
(cherry picked from commit a8065da96b96fcc4baeca7615c2a4195c05cbc03)
|
|\
| |
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/oslo-staging-1:
QUrl: fix parsing of IRIs with more than one IDN label
QUrl: update the whitelist of IDN domains
|
| |
| |
| |
| |
| | |
Task-number: QTBUG-10511
Reviewed-by: Trust Me
|
| |
| |
| |
| | |
The list is taken from the Mozilla page.
|
|/
|
|
|
|
|
|
|
|
|
|
| |
Change 02532ec80375c686503c4250c6ad6bb211515ec8 removed the early-exit
for 0 byte reads, causing us to hit code that assumed the buffer was
empty since nothing was read. It would thus read more into the end of
the buffer, causing the buffer to grow bigger than QIODEVICE_BUFFERSIZE.
Next, if the actual number of bytes we wanted to read was bigger than
the original buffer size we'd read the same data twice.
Reviewed-by: João Abecasis
Reviewed-by: Thiago Macieira
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The optimizations in d0645d1792e1cbdf417a923ea071975e4390fccd did not
take into account 0-length reads, used in network programming to check
connected state.
By not returning early in the case where lastReadChunkSize == maxSize
and maxSize == 0, we allow the underlying device to report error
conditions on 0-length reads.
This reverts commit aea71e2e02fa966842b094244bc3f5fc88f50f41, while
still ensuring that QIODevice::read() returns -1 on a closed device.
Reviewed-by: mread
Reviewed-by: Thiago Macieira
|
| |
|
|
|
|
|
|
|
|
|
| |
Work done jointly by Gareth and me. Yields no qdoc errors.
Task-number: QTBUG-9342
Task-number: QTBUG-9120
Reviewed-by: Gareth Stockwell
Reviewed-by: David Boddie
|
|
|
|
|
|
|
| |
When a file copy failed on symbian, the file engine error was not set.
Now translates the symbian os error code to set the file engine error.
Reviewed-by: Marius Storm-Olsen
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
atEnd improvements, since atEnd is used in published example Qt read
loops. The two main ideas are to push the buffer test forward, which
optimises the normal buffered case, and to cache the file size till you
get to the end, which otherwise has to get the file size every 16K of
data.
read buffer improvements: The ring buffer structure was causing
significant performance overheads. In practice QIODevice has much
simpler requirements on its read buffer, and a linear buffer can be
used instead. This now uses a buffer optimised to QIODevice's use of
it.
QIODevice read function improvements. There are a number of sub-themes
here, which all are aimed at getting the normal buffered path through
the code done as fast as possible. This gives greatest improvements for
small reads, but it is these small reads that have the biggest
problems.
- removing the isSequential test, by setting a pointer to the qint64
to update and using a dummy qint64 target when isSequential, then
writing through the pointer.
- doing the readability check on the first read only, out of the fast
path
- doing the maxSize<0 test after the getchar fast path
- removing the buffer isEmpty test and giving a fast exit test
instead
- moving arithmetic out of the fast path, so "data" and "maxSize"
are now dynamically updating
- for RCVT builds, ARM mode is used for the read functions because
the 64-bit operations are much slower with this compiler in Thumb
mode
There are some other changes to read() which improve clarity:
- bytesToBuffer is now always set to QIODEVICE_BUFFERSIZE. This has
always been the case, its just that the logic was not clear before.
- moreToRead is set to false now in the case where the request was
met by a direct device read. Leaving it true in this case was
pointless, and setting it true in the converse case seems dangerous
because the function might iterate for a very long time, although
it might meet the API semantics better and would be a change in
behavior because the function used to not read more when it claimed
it would.
Reviewed-by: Joao
Reviewed-by: Aleksandar Babic
|
|
|
|
|
| |
Task-number: QTBUG-8906
Reviewed-by: Janne Koskinen
|
|
|
|
|
|
|
|
|
| |
On Symbian, we incorrectly assumed that filenames are encoded in
local8bit. OpenC's implementation of wcstombs, mbstowcs, etc. are all
UTF-8, regardless of the locale.
Task-number: QTBUG-7175
Reviewed-by: Shane Kearns
|
|
|
|
|
|
|
| |
In fact QIODevice still _does_ buffer access through QBuffer currently.
Reviewed-by: Markus Goetz
Reviewed-by: Joao
|
|
|
|
|
|
|
|
|
| |
QProcessManagerMediator and QProcessActive were not properly canceled
in their destructors, causing crash when global static QProcessManager
was deleted.
Task-number: QTBUG-7735
Reviewed-by: Janne Anttila
|
|
|
|
|
| |
Task-number: QTBUG-8365
Reviewed-by: denis
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Symbian implementation of fileFlags calls both doStat and isSymlink
resulting in calls to both stat and lstat. However most of the time,
ie when accessing a normal file rather than a symlink, lstat gives all
the information we require. So this change uses lstat where possible,
and caches its result for both doStat and isSymlink.
During start of DesktopServices app, this cuts calls to stat/lstat by
45%.
Reviewed-by: Shane Kearns
|
| |
|
|
|
|
|
|
|
| |
Removing trailing whitespace from qurl.cpp
Removing a debug() output from the test.
Reviewed-by: Benjamin Poulain <benjamin.poulain@nokia.com>
|
|
|
|
|
|
|
| |
Patch to QTBUG-3168 and update tst_qurl to
enable the test case that was skipped.
Reviewed-by: Benjamin Poulain <benjamin.poulain@nokia.com>
|
|
|
|
| |
Task: QTBUG-8450
|
|
|
|
| |
Task: QTBUG-8083
|
|
|
|
| |
Task-number: QTBUG-8070
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When adding and removing lots of files, the kqueue implementation's
pipe between the worker thread and the main interface would fill up,
causing the thread adding/removing files to block even though the
worker thread was no longer running.
The kqueue interface gives us information about how much data is
available on the pipe (the data member of the kevent struct), so we
should use this to make sure we always empty the pipe. We then iterate
over all the command bytes and make sure we interpret them correctly.
Reviewed-by: Prasanth Ullattil
|
|
|
|
| |
Task: QTBUG-8070
|
|
|
|
| |
Task: QTBUG-7997
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/berlin-staging-1:
Assistant examples: Fix missing QLatin1String.
Added Czech translations.
Webkit: Fix missing namespace.
qmake: subdirs template: make distclean target use QMAKE_DISTCLEAN
qmake: fix warnings
qmake: Remove macro and simplify string expression previously using it
qmake: use isActiveConfig() instead of hand-crafting comparisons
Fix missing namespace.
Assistant: Fix spelling mistakes.
Assistant: Use const references in foreach loops.
optimization: get rid of QString::fromUtf16() usage
document QString::fromUtf16() slowness
remove duplicated calculation of length
remove pointless conditionals
micro-optimization
optimize qhash()
Designer/uic/related examples: Fix source code scanning issues II.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
QString::fromUtf16() is slow - it does a BOM check and optionally byte
swapping, which is utterly pointless when converting internal data
structures which are raw utf16 in host byte order anyway. so replace
it with QString::fromRawData() (for short-lived strings) or
QString(const QChar *, int) (otherwise) if possible.
Reviewed-by: axis
Reviewed-by: mariusSO
Reviewed-by: Bill King
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
QProcess::waitForFinished was panicking in Symbian when timeout of -1
was supplied. Fixed it to disable timeout and block until process
exit, as docs indicate it should.
Task-number: QTBUG-7667
Reviewed-by: Janne Anttila
|
|/
|
|
|
|
|
|
|
| |
Catching by value causes GCC to use the compiler support function
__cxa_get_exception_ptr. This is not available in the compiler support
libraries on Symbian OS prior to Symbian^3. Catching by reference
avoids this problem.
Reviewed-by: mread
|
|
|
|
| |
Reviewed-by: Peter Hartmann
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/oslo-staging-1: (70 commits)
doc: Fixed the last qdoc errors.
doc: Document the "Type" enum value as a const in variable.
Designer: Add lower/raise to context menu.
Compile with no-webkit - add missing semi-colons.
Fixes visibility update missing when doing setParentItem on graphicsitem
Stabilize tst_QGraphicsScene::polishItems2 (new test)
Fixed an infinite loop that could occur when reading invalid BMP images.
Updated docs regarding QGLWidget::renderText() limitations.
Added optimization flag to QGraphicsItemPrivate.
Fixed child items with graphics effects not inheriting opacity.
Made the trace replayer handle limited resolution cases better.
Small optimization in raster paint engine.
Another ASSERT while deleting spans
Implement QDirectFBPixmapData::scroll
Potential crash when adding items from QGraphicsWidget::polishEvent().
QGraphicsWidget is painted twice on the inital show.
Fix QPainter::redirection() to pass autotest.
Re-added the Close button in QPrintPreviewDialog for Mac/Carbon.
revert parts of 10392eef4fd4f9
Fix y-inverted pixmaps properly.
...
|
| |
| |
| |
| | |
Also, since domain is never empty, the idx > 0 test is unnecessary.
|
| |
| |
| |
| |
| | |
Merge-request: 436
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
This only coincidently worked on Linux, because it creates a QSettings
object to read system settings before any user data was executed, thus
initializing the static default path data. However, if only the
SystemPath was modified, the UserPath never got initialized on
non-Linux platforms, causing settings for the users below $PWD.
Also, it removes unncessary checks and mutex locks, which should
give a slight speed up.
Reviewed-By: dt
|
|
|
|
|
|
| |
as agreed with Warwick.
This reverts commit 089ee7094eabb2058b478f5d2f306a69f6c0b3bf.
|
|
|
|
| |
Reviewed-by: Markus Goetz
|
|
|
|
|
|
|
| |
Don't try to allocate one for QFile::handle and QFile::flush since
they'll be useless there.
Reviewed-by: Markus Goetz
|
|
|
|
|
|
|
| |
Don't repeatedly call the fileEngine virtual function. Instead re-use
the private data member if we know it has been filled.
Reviewed-by: Markus Goetz
|
|
|
|
|
|
|
|
| |
remove was checking for errors from close, but without clearing the
error state beforehand it ended picking unrelated errors.
Task-number: QTBUG-7285
Reviewed-by: qCaro
|
|
|
|
|
|
| |
QFileInfo().absolutePath() is legal and shouldn't produce a warning.
Reviewed-by: João Abecasis
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is private API for now. For 4.7, decide whether to have a public
method for this.
Also investigate whether to provide a qHash function. Adding one to Qt
would break source- and binary-compatibility with code that already
has that (like Soprano).
Patch-By: Warwick Allison
Reviewed-by: Thiago Macieira
|
|
|
|
|
|
|
|
|
|
|
|
| |
Try to retrieve user/group name in a single call to LookupAccountSid, by
using pre-allocated buffers. Only call a second time if the buffer size
is insufficient. 64 bytes should be enough in common case.
Previously, we would always call LookupAccountSid twice, even if
ERROR_NONE_MAPPED occured first time.
Merge-request: 815
Reviewed-by: João Abecasis <joao@trolltech.com>
|
|
|
|
|
| |
Merge-request: 815
Reviewed-by: João Abecasis <joao@trolltech.com>
|
|
|
|
|
|
|
|
|
|
| |
* tabs -> spaces;
* if( -> if (
* some strings rearranged.
no code changes!
Merge-request: 815
Reviewed-by: João Abecasis <joao@trolltech.com>
|