Some time ago this was a blocker that didn't allow to refactor QLocale
implementation due without making binary incompatible changes. Deinlining those
functions for Qt5, it shouldn't be performance critical code path.
Change-Id: I6cb19e32188a2df223d04be0c613a6176ad8d118
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
QCOMPARE and friends should only be called in a test function. Instead
of calling QCOMPARE elsewhere, keep a count of the number of errors and
QCOMPARE that count with zero in the test function.
Change-Id: I9a264e91169a98c30980fdc04a3e45bfb0ca8063
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
There is no need to print out the name of the backend used by each test
run of a test function as every message output by the test function will
have the name of the current data row included.
Change-Id: Ie69881d2ecedce728ea67b5aae1c1196776552a5
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
CallConstructor<T, /* CanUseInternalSpace = */ false> is called when we need to
construct an object that couldn't be fit in qvariantdata, meaning either it is
not a POD type (Q_PRIMITIVE_TYPE), or it is simply too large to fit there.
Change-Id: Ied122b4a6f600e14312a8d515f5b3e91214a94f1
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@nokia.com>
... instead of QEventDispatcherUNIX. The Cocoa dispatcher does not use
any of the facilities of the UNIX dispatcher, and it reimplements every
virtual method already (with the exception of flush(), which just needs
an empty implementation).
Change-Id: I24aefd169888946afac7800192a0f96770787718
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
This change makes Qt load the default openssl config always, not just
when compiled with OPENSSL_LOAD_CONF. This means that facilities like
openssl engines (and their configuration) are usable. An alternative
would be to call OPENSSL_config(NULL) ourselves, but that's exactly
what the OPENSSL_add_all_algorithms_conf does for us.
Task-number: QTBUG-16018
Change-Id: I4cda701f82627e0541b6225009f4e1249aec9d47
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Much like DataLocation = GenericDataLocation + domain + appname,
this makes CacheLocation = GenericCacheLocation + domain + appname.
This way a framework library can have an application-independent cache
(like ksycoca).
Change-Id: I6a8c47ff85b7d5c68b594cc8b071a752d96b029d
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
On Mac, the application's dir is in the bundle, so we need to "escape"
the bundle when looking for the an executable relative to the
application's dir path.
Change-Id: I5c01f7d816ec8cc30f5277202f4eefb0c49a2bc3
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Use the C++ boolean constants true and false instead of the C macros
TRUE and FALSE (which are actually integers), and use QVERIFY instead of
QCOMPARE for verifying simple boolean expressions.
Change-Id: Ie76dfcab6722df6b93b3fa62b0f3437901482932
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Inotify has been available in Linux for some ~6 years now, 7 when Qt 5 will
actually be released, so I'd say it's safe to remove this fallback path now,
particularly as the autotest notes that it's broken.
Change-Id: I49dbb161d4765d63e92f512a6375323c7d37ccbe
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Similarly to ec9ea7f3e8, the code in unix
(non-Linux/OS X) is actually the same as for integrity, so merge the
conditionals together to save duplication.
This should have the side-effect of unbreaking Qt 5 on integrity wrt
the new QStandardPaths introduction, which was not added to the integrity
block.
Change-Id: Ib512fa781f5ceb240069888ce6958c9af2990d37
Reviewed-by: Rolland Dudemaine <rolland@ghs.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
The value is changed to true. It is a common bug that developers expect
this proxy model to reflect the source model when the source changes.
That requires setDynamicSortFilter(true), so we change the default to
optimize for the common case.
Change-Id: I9bf7efdbda10309fa77aed9391c33054aaae4a29
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Instead of only from QApplication in QtWidgets, as we need these events
for example in QDeclarativeApplication.
Task-number: QTBUG-21573
Task-number: QTBUG-23331
Change-Id: I0c960bd1c7911d306d274a6e9a1838f158235ed0
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@nokia.com>
GCC 4.3 introduced the alloc_size attribute to hint the compiler
that allocated memory will be returned, inform the compiler which
parameter holds the size of the allocation.
Change-Id: I8734868f6bd19e201abdacd0a1b0fb80a27883c0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Pure syntactical sugar, to match up with what the other container
classes offer.
Change-Id: I0f97de011923d9d204cca0fa906b059dc5054a89
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Qt5 modules are supposed to be in a namespace, but if the namespace definition
is hidden in a macro, then moc doesn't know about it and generates invalid
moc_xx.cpp that cannot be compiled due to usage of classes outside of their
namespaces - e.g. in qtjsondb we have QtAddOn::JsonDb::Foo class, but the
moc_foo.cpp expects to find that class in the global namespace instead.
Fixed it in QtJsonDb to define QT_BEGIN_MOC_NAMESPACE="QT_USE_NAMESPACE
QT_ADDON_JSONDB_USE_NAMESPACE", however we need to ensure qglobal.h doesn't
re-define that macro back.
Change-Id: Ic8407f50c11d2d787167ad2f92457aa3ec126d45
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
gray_conic_to, gray_cubic_to and gray_line_to were all single line wrappers
around their equivilent gray_render counterparts, with an additional lie of
error handling that never actually happened.
Since this doesn't really do anything except confuse the reader, let's ... not
do it :)
Change-Id: Id5d86c49174acb92514b628a70bd32d6c6640a5d
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Usually we don't have a plugin for this on linux anyway.
But if we do, we should actually allow it to interface with the
system.
When using AT-SPI the plugin can detect if it should be active.
Other plugins can fall back to using an env var if really needed.
Change-Id: Ic9dcfa305e7cdafbf4a93bcc2dc9a0fcd9b9a7a2
Reviewed-by: Jan-Arve Sæther <jan-arve.saether@nokia.com>
The original intention was to identify a clipping bug, it turns out
that during a ::begin() (and systemChanged) we should forward the
QPaintEnginePrivate state to our proxy engine.
Instead of using the proxy-pattern subclass rasterengine and
specialize the paths we are able to accelerate using the blitter
interface. This will avoid similiar problems in the future. I have
no performance measurement to show which of the two approaches is
faster/slower.
Change-Id: I39bff11b32b1fe20284c7e8df60050de5991bb6e
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
* Group code responsible for tracking the render pipeline state
to check if we can easily accelerate it
* Code that will call into the QBlittable
* Code that will lock the QBlittable before calling into raster
Change-Id: I862e242d59805de5094ed363b486afcdbc23ff78
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
A default QPainter will set a QBrush with Qt::NoBrush BrushStyle, the
current code detects this as a non solid fill and all calls with fillRect
and a color will not go through QBlittable. Check for Solid or NoBrush
style. Manually verified that a p.fillRect(rect, Qt::red) goes through
the accelerated path now.
Change-Id: Ic0d98030e94f5d11abbe61628fbf71d1e08219c2
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
Create updateCompleteState(QPainterState*) so we don't need to
call into the paintengine, move all state methods into the
QBlitterPaintEnginePrivate class.
Change-Id: If30fdcc3f63755e0443bced7d9d9fb993d4ec2b7
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
Cosmetic change to remove four spaces of indention from some of the
deeply nested methods.
Change-Id: I67fdd0ab722b7c7c67c4da7a0a0bd86459751700
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
It starts with being initialized wrongly, the call to buffer() will
lock the data while we think it is not locked, it can also get out of
sync by someone calling buffer() again. Remove the variable and
check with the QBlittable if we need to lock the resource into memory.
Change-Id: I350375011138d1b4c2c48c100b7b30b8ea2ae460
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
Move the bit helper functions into CapabilitiesToStateMask as they
are only used in this class, allocate the class as part of the
QBlitterPaintEnginePrivate, shorten the name as well.
Change-Id: If22ddd117a9789cd98edb08f23fd0ffabb17d5a5
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
Spend some bytes for spaces after comma and keywords. This should
now mostly follow the Qt style guidelines.
Change-Id: I3298c8d41d40ab5b0153a33d44b1b607a2edca8e
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>
CoarseTimers are worst in their first firing, so we prefer a
PreciseTimer for something that happens only once. If the timeout is
too big, we use a CoarseTimer anyway (current threshold is 2000ms).
Author: Thiago Macieira <thiago.macieira@nokia.com>
Change-Id: I30b20acf506e442cd58126abfe3a4d70fc13b075
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The API passes the interval as an int, there's no reason to convert
it to a timeval struct. This also prepares for changing the UNIX
timer code to support the different timer types.
Author: Thiago Macieira <thiago.macieira@nokia.com>
Change-Id: Ie3cc1ae8f1be6a9ad3f1766051642cbf3e614418
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
... and deprecate the old registerTimer() functions. The new pure-
virtual registerTimer() breaks source-compatibility. Subclasses cannot
be instantiated anymore, since the pure virtual function signature has
changed.
QAbstractEventDispatcher::TimerInfo is no longer a QPair. It is now a
struct with timerId, interval, and timerType members. This is a source
incompatibility that should only affect subclasses of
QAbstractEventDispatcher, which will need to pass 3 arguments to the
TimerInfo constructor instead of 2. If the subclass used QPair<int,int>
instead of the TimerInfo typedef, the QPair<int,int> declarations will
need to be replaced with TimerInfo.
Call the new registerTimer() function with the type from
QObject::startTimer(). Change all subclasses of QAbstractEventDispatcher
to reimplement the new virtual function. The type argument is unused at
the momemnt, except to ensure that registeredTimers() returns the type
each timer was registered with. Implementations for the various
dispatchers will be done in separate commits.
Author: Thiago Macieira <thiago.macieira@nokia.com>
Change-Id: Ia22697e0ab0847810c5d162ef473e0e5a17a904b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The timer type will control the accuracy of the timer. By default, all
timers are CoarseTimers, which allows for +/- 5% interval adjustment.
PreciseTimers will not have any interval adjustments, VeryCoarseTimers
will have intervals adjusted to full second resolution.
Use QTimer::setTimerType() or the QTimer::singleShot() overload to
specify the type.
QObject::startTimer() now takes a Qt::TimerType argument which defaults
to Qt::CoarseTimer. QBasicTimer::startTimer() gets an overload that
takes a Qt::TimerType argument. The argument is unused for now, since
the QAbstractEventDispatcher interface needs to change (done in a
separate commit).
Author: Thiago Macieira <thiago.macieira@nokia.com>
Change-Id: I3100da5aa1fe17ec30b8644897d0fe6ec4a07f52
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Previously they would fall into the unicode handling and return very
strange values.
Change-Id: I62a53894c0983bf53fd79f924b40a6fd3ba02993
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
These tests used requires(contains(QT_CONFIG,private_tests)) in their
.pro file, but did not subtract themselves from their parent project
SUBDIRS when private_tests weren't enabled.
In the best case, this wastes a little time as qmake iterates over these
projects which won't be built. In some worse esoteric cases, this may
break compilation or packaging.
Change-Id: If36b1b8f69c3509128786fec67899ae18ffaa2bc
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This testcase fails on the Ubuntu 11.10 x64 platform, add flag to .pro
file so test is ignored on this platform.
Task-number: QTBUG-23380
Change-Id: I51831df8c8e9bfcf63d3689e37552ca1a62691cd
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Several of the test functions in the QDebug autotest call
qInstallMsgHandler() to temporarily use a custom message handler.
Unfortunately, these test functions were then resetting the message
handler back to Qt's default handler at the end of the test.
QTestLib also calls qInstallMsgHandler() to set a message handler that
redirects debug/warning/fatal messages into the test log. When the test
resets the message handler back to Qt's default handler, testlib's
message handler is bypassed for the rest of the test, preventing any
subsequent debug/warning/fatal messages from being visible in the test
log or subject to testlib's ignoreMessage() function.
This error also caused several of the test functions to fail if they
were run manually. The "defaultMessagehandler" test would fail if it
was run before any other test function and the "assignment" test would
fail if it was run after any other test function.
This commit fixes these failures by using a helper class to ensure that
the previously active message handler is restored at the end of each
test function, even if the test function fails or throws an exception.
Change-Id: I51376724d164c8ad126e5b9be76890bf3e6a9fb0
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
