Fixes a crash when invoking various QRegExp methods on an object
*before* doing any match. For instance fixes:
QRegExp re;
re.matchedLength(); // crash
Task-number: QTBUG-23352
Change-Id: I9c239ff790a139c7820ef1aeced89d31320ae6b0
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
this feature was introduced in feb 2005 by Sam with this comment:
> I have implemented a versioning into the keys and do environment
> expansion there as well, these aren't immediately usefull but Marius
> and I agreed that distributors will probably want such features at
> times. The versioning fallbacks will be usefull to us over time no
> doubt.
imo the versioning is a contestant for the most useless feature ever:
- (linux) distributors couldn't care less - they simply configure qtcore
correctly. additionally, the packaging policies state that no config
files should live in the binary dir at all, so no qt.conf for them.
- ISVs don't care, because they ship their software with a particular qt
version anyway.
- SDK distributors don't care, because it doesn't solve any real problem
for them: a) they will isolate the (qmake) versions and b) a
distinction based on version number (as opposed to build
configuration) is utterly useless in the first place.
i left in the variable expansion, as it could at least theoretically be
useful for creating relocatable packages. debatable - the file it easy
enough to modify at installation time.
Change-Id: Ida8a50b16d55d8d8613d1a98a51df56753f6a6e3
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
- AccessibleWidgets
- Windows printer support
- Examples
Change-Id: Icc162bd7fc284b3c76d9966210f983728085c743
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
We do not need to distinguish between different types in the switcher.
Before this patch it was not possible to overload
DelegateObject::delegate with a pointer type. Now it is fixed.
Change-Id: Icd73a53e73e5e66b1b6f6407ba4e0f79e584d930
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Use a union and a type enum instead of three pointers.
Change-Id: I02b11733a4f2e95099064fa9325497d4e04ac615
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Reusing a template is much better then creating a new one, even if it
should inline the same code.
For some reason replacing T* by void* force gcc to remove a few bytes
per template instantiation too, it is not really significant, but it
alows us to simplify the code.
Benchmarks don't show any regressions.
Change-Id: I4fdf1e4dc311b23021eb5758605602937d05b183
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QVariant::Type is marked as obsolete. It is not possible to get rid of
it completely, in a source compatible way, but at least we can remove it
safely from a method arguments list.
Change-Id: I26b58099bfa6d32f3a583a8ae0047f0bb36bcd0d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The user-defined copy constructor and
copy-assignment operators were 100% equivalent
to the ones the compiler would generate, so
let the compiler generate them (so we reap
move constructors, too, even though they're
not needed on this class).
Change-Id: Iecdd579fa5a819d083ec9b2f25734ddba85515e6
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
we already initialize it before parsing a project. if a project is daft
enough to clear TARGET, it does not deserve differently than breaking.
Change-Id: I6c727bc27d72a00e84b676ae3c169024bdb2d929
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
This does not fix anything, because AreArgumentsCompatible already
do all the type checks.
But it make RemoveRef consistant with std::remove_reference
Change-Id: Ic42c872356172d7f5ea10de050254b5d10e50a6e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Make most QFixed{,Size,Point} members constexpr.
See f3141c58ba for rationale.
Change-Id: I0b44f59bb768e1f79c57e1c9a0dc67afb515e03f
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
The main change is the addition of a new constructor
that passes its argument into 'val' verbatim. In order
to disambiguate it from the existing QFixed(int)
constructor, it takes a second 'int' argument. This
is too ugly for public API, so it's private, and only
used by static QFixed fromFixed(int), which is the
existing named constructor with the same semantics.
The rest of the changes simply reformulate their
operations in terms of fromFixed().
This makes them ready to be constexpr'ed.
Change-Id: I2a3813d62bd4124064755de6b00526a60fc82c1d
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
QRegExp is not used, so they're unnecessary.
Change-Id: I3480bcbe013a0bf15e2ee4fa30862fe035820eea
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
In sub-attack an interview, one can't make two implicit conversions at
once, so explicitly convert to the right type.
The change in the torrent example is required because of
https://codereview.qt-project.org/16168 (commit 9491272)
But in that case, using a QByteArray is better anyway.
Change-Id: Ieed22ac7f0d700d5ba5d1e70af3db4dd6c139c8f
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
This would break a namespaced Qt when QT_NO_DEBUG_STREAM
was in effect. Unlikely to hit, but nevertheless fixworthy.
Change-Id: Ie2a4cf4334a6a610c84233ab1ca89b928386c91a
Reviewed-by: Richard J. Moore <rich@kde.org>
Makefile.mingw was using the test command which requires sh.exe. This
adds support for bootstrap using cmd.exe instead.
Change-Id: Ieb67843170d2745ce119cc8feaa5433aa82fd7d7
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
x86 processes get redirected to the 32 bit area of the registry,
while 64 bit processes access the registry in raw form.
Added the registry paths for 64 bit processes to see the 32 bit
registry keys for MS visual studio. (Wow6432Node)
This problem was revealed when we stopped including configure.exe
as a binary checkin to git. Running configure in an x64 compiler
environment results in creating an x64 configure.exe
Change-Id: I5e3e51ddbf20ccc65abf1833bf23ee5670bd973e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Use the correct calling convention. Discovered by the mimetype
test in Qt 5.
Change-Id: I79b97325dd69466885a64c238935107bf14e9a0d
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
(cherry picked from commit bc6e35d94325f4b18caec7ce0824c5beab95cdd7)
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Shared resources may have been used when initializing NSOpenGLContext
but QCocoaGLContext did not implement isSharing(). This caused
default implementation to return false always and therefore shared
resource test case failed.
Implemented also another missing method, isValid().
Task-number: QTBUG-23061
Change-Id: Ia912450035b584ea90a02a7d88d6ae531c3cbadf
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Don't register the thread handle and QThread object pointer to watch list
for adopted thread watcher thread in Windows. Otherwise the watcher thread
will never exit and can't clean up its own QThreadData and QAdoptedThread
object.
Task-number: QTBUG-23961
Change-Id: Ia84326cf3cfd978d2b003ccc1bb6861db950e899
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Fixes remnants of an automated QtGui -> QtWidgets port.
Change-Id: I31f63bdd6ae71aee8e70b20e24def30e0eafb725
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Declare a char array instead of a pointer variable.
Change-Id: I2beff815d05b6dc9c35bb0a55d7294189afbf17e
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
- Fix key handling in QWidgetLineControl according to
the keyboard scheme returned by the QPlatformTheme,
remove #ifdefs.
- Do the same in the test.
Task-number: QTBUG-21402
Change-Id: I36d836584e7122309061af72819a4147cadd0a74
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
The libName that is used in pkgConfig files should include the
TARGET_VERSION_EXT.
This is needed because in Win32MakefileGenerator::fixTargetExt(),
the TARGET_VERSION_EXT is added to the library name.
In Win32MakefileGenerator::processVars(), if TARGET_VERSION_EXT is empty,
it is set to VER_MAJ.
On platforms != Windows, TARGET_VERSION_EXT does not seem to be used.
We probably got away with this so far because pkgconfig files generation
for win32 was just added in 4.8 and nobody uses them yet, and because on
platforms != windows the TARGET_VERSION_EXT is not used.
Change-Id: I56f239e389f0ef926030e4c2376cadd92c4f673c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Change-Id: Ia68fb00afb81a0f521f9fbc3f0dbfc7e43075bbe
Reviewed-by: laurent montel <laurent.montel@kdab.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
It might be silly to have sections with size 0 without using hideSection.
Nevertheless we should still use the principle of least surprise.
It does not make sense that hiding a row 'far' away should affect
the semantics of setOffsetToSectionPosition on lower indexes.
Change-Id: Iaf847eba2ea4d28fc7bcfe3a27d62f432f6f61e0
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Support pixel-based deltas as well as sending
dx and dy values in the same event. Keep source and
behavior compatibility with Qt 4.
New API:
QPoint pixelDelta() const
QPoint angleDelta() const
Deprecate delta() and orientation().
Both pixel-based deltas and combined updates are
necessary for smooth trackpad-based scrolling on
OS X.
Qt 4 compatible behavior is achieved by sending an
extra wheel event in cases where the initial event
has a combined dx and dy update. This extra event
sends dx in delta() and orientation(), with pixelDelta()
and angleDelta() set to null.
Modify the Cocoa implementation to provide pixel
deltas. It is expected that not all platforms can
provide these. Angle deltas will always be available.
Change-Id: I20c10f0df338ddcd6a3f7a4d40949ed5ae3b4795
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
This removes an if that should always be false. If it isn't
false we had more logical indexes than sections.
Unless visualIndices and logicalIndices are both 0-sized
(meaning that visual and logical indexes are the same)
then they better have exactly as many indexes as sectionCount.
Otherwise there must be inconsistent data - some logicalindex
would refer to non-existing visual sections or some
visualindeces wouldn't be hit by logicalIndices.
Change-Id: I2dc92f40895438d912bba20612131baf63df4fbe
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
* Document AllRelations
* Assigned new values (starting from 1) for the remaining members of
the enum.
Change-Id: I55ec197f24ff7d43ff3c0d101edd33b9615f9287
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
