[docs] QApplication: fix keyboardInputDirection() docs

There were two documentation blocks: the first had the wrong class name,
the second was duplicate.

Change-Id: I241b3e6567376a46a6270837cce4aa380a6c8c49
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

[docs] QInputDialog: document deprecated member getInteger()

Change-Id: I2ac24acbecd67db74f9474a7a2a1ee94b03282d4
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

test:Remove QSKIP from tst_QPainter

Removed QSKIP ("currently broken... ") from tst_qpainter.cpp.

Change-Id: I2a5af009198d26015779ca368ef82188ddf1b060
Reviewed-by: Caroline Chao <caroline.chao@digia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

xcb: better error handling if no randr or screen resources query fails

Marc Mutz already reported problems with ssh -X fowarding.  Now all
such errors are treated the same: if we can't get screen output
attributes, just assume there is only one.

Change-Id: I96802fc90072c623de3370ed2898893daf58198a
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

[docs] QFlags: fix a whitespace error

Change-Id: I316ceb3d630d6697466a46fa14bdbd8379079875
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QFileSystemModel: fix permission mangling

The old code masked out write flags before returning permissions
from permissions() or data(FilePermissions) in order to force
QFileDialog to disable the rename and delete actions. This was to
fix Task 143519, but introduced QTBUG-20503.

Instead, revert to the pre-143519-bugfix code and do the necessary
check in QFileDialog directly.

Also add a testcase for 143519.

Reported-by: Gilles Pascual
Task-number: QTBUG-20503
Task-number: 143519

Change-Id: I140109341c0ed40722e3aac4327c2a740fb014c2
Reviewed-by: Jan Arve Sæther <jan-arve.saether@digia.com>

Remove duplicate words and punctuation from documentation.

Change-Id: I5550c62d412510bc2c5acceb2cae7d2f2ef6a8d3
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>

QCharAttributes: add wordStart/wordEnd flags

A simple heuristic is used to detect the word beginning and ending by
looking at the word break property value of surrounding characters.
This behaves better than the white-spaces based implementation used before
and makes it possible to tailor the default algorithm for complex scripts.

BIG FAT WARNING: The QCharAttributes buffer now has to have a length
                 of string length + 1 for the flags at end of text.

Task-Id: QTBUG-6498

Change-Id: I5589b191ffde6a50d2af0c14a00430d3852c67b4
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>

test: Mark tst_QMdiSubWindow failures as XFAIL

The test has lots of failures, so mark these with QEXPECT_FAIL and
remove the QSKIP.

Task-number: QTBUG-22310
Task-number: QTBUG-27274
Change-Id: I0d38cedb581741b6edae5d1c3f4410714099a7fb
Reviewed-by: Caroline Chao <caroline.chao@digia.com>

fix incorrect uses of qmake in documentation

Assigning backslashes to qmake variables yields a warning
these days. Use forward slashes as they get automatically converted.

Task-number: QTBUG-27325
Change-Id: I804fa641064ce183e9794026ad1dbc91725d334d
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

Test: Remove QSKIP from tst_QWidget::raise

Remove QSKIP("Not yet sure why this fails."), the test is passing
on Mac.

Task-number: QTBUG-22321

Change-Id: I5f09d067b1cc837c5e3ada5bbd34091fe1fd723d
Reviewed-by: J-P Nurmi <jpnurmi@digia.com>

Test: tst_QFile::writeLargeDataBlock() unstable failure as XFAIL

The current check is not enough to catch the unstable failure.
Update the test to catch the failure in all cases.

Task-number: QTBUG-26906

Change-Id: I2e37a1f6513df768cd410df7c91a9fd843150e57
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

Test: Remove QSKIP from tst_QLineEdit::taskQTBUG_7395_readOnlyShortcut

The test is passing on Ubuntu 11.10.

Adding qWaitForWindowExposed after the line edit show() call to ensure
the line edit is actually shown on the screen.

Task-number: QTBUG-24518
Change-Id: I2af65bef76d171b36032120738dfbd7cfff51d7f
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

qmake: support for Visual Studio 2012 project files added

The differences to VS 2010 project files are the
version number (surprise!) and the PlatformToolSet tag which
sets the version of the toolchain.

Change-Id: If26f08fad1a69d7e6cd28cc5e860ff964f19b264
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

QFileInfoGatherer: general cleanups

- fix typo in comment
- make translateDriveName file-static
- use QList::reserve()
- explicit ctor
- annotate functions called by owning thread/run()
- iconProvider() and resolveSymlinks() shouldn't be slots
- private instead of protected
- run() marked Q_DECL_OVERRIDE
- annotate which members are protected by 'mutex'

Change-Id: Ic176a030b0597c98f48185efd17260fd62532460
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Fix default-constructed QFileSystemEntry

Member variables for lastSeparator, first and lastDotInFileName are now
initialized to -1 (non-existing), where the previous value of zero would
mean a separator/dot at that position and resulted in path() returning
'/', instead of '.'.

Tests were expanded for better coverage of empty state and
default-constructed instances.

Change-Id: Ie27547886b52224d38b5be0b4f920c9927fd440f
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

Ignore the QT_NAMESPACE for moc runs

Moc should not see a defined QT_NAMESPACE, to
keep the names of all symbols consistent with
an un-namespaced Qt.

Change-Id: Ia72afabaa0820b02a0863148510122ce942a5082
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Test case for moc handling defines

Added some test cases that check that moc
correctly expands #defines

Change-Id: I7fe6eed129d46ca9281d73064571cae43b32410d
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Moc: Correctly expand macros with arguments

Moc now supports full expansion of macros with arguments
with the exception of some keywords such as Q_OBJECT
and others that are used internally

Change-Id: I283d47152f75de81ec68a3df1f0f2273f11c5149
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Moc: Make # and ## valid tokens in cpp parsing mode

This is required so preprocessing macros with arguments
can work correctly.

Change-Id: Ia30ede6a3579be13800bf1255d313748e2d696f9
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Correctly parse function macros

Parse function macros and add it's list of arguments
to the Macro definition.

Change-Id: Id22f5cf4a1c098f7b4f5b72f002900cd40d03e0f
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Correctly expand macros without arguments in moc

This helps e.g. cases where a namespace is defined
through a macro and moc doesn't see it at all.

Expanding macros with arguments is significantly more
work, and should happen in a separate commit.

Change-Id: Ic8d0443d06fab2ed343115d8c43022f2c67ec3cd
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

qdbusmetatype_p.h: add missing QT_NO_DBUS guards

Change-Id: Idb459fe6300f710b959247cd9c1997a4d5774b2d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QSqlTableMode::selectRow(): lack of WHERE clause is an error

Since we only want one row, never ever run a SELECT without a WHERE
clause.

Change-Id: I40a78935f5573111faa3922eae97e6d5961be5f2
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

QSqlTableModel::selectRow(): reduce scope of QSqlQuery

It's good to clean up the query before emitting signals about the
updated row. It's possible that connected slots will call selectRow()
again for other rows.

Change-Id: I482fe2dd58218f53567ce8725ee591ce2eeda348
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

QSqlTableModel::selectRow(): fix failure on uncached rows

This method was originally intended for refreshing rows after
submitting changes. It should also work for refreshing rows
that are unchanged (i.e., not cached), but did not because
constructing the primary values depended on the cache. As a
consequence, the WHERE clause for the query was not created.

Fixed by deriving primary values for uncached rows from the
query record. Note that the cache is still authoritative for rows
it holds. This is important because the prmary values there may
differ from the original query record due to changes to columns
of the primary key.

Includes new test.

Change-Id: I41cca2cbf26019d4b495ffa6d876e2b55ec57803
Reviewed-by: Andy Shaw <andy.shaw@digia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

Fixed crash in moc on big endian platforms.

Change-Id: Icaa38eb4b404e5f52248fdeaf9180a5d70d0f5f2
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Fixed QJsonDocument::fromBinaryData on big endian platforms.

Change-Id: I1786b6222867c8780f6768e5220e7ddff952b28e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Fixed ifdef condition in QJson resulting in big endian breakage.

Change-Id: I3d36d75ff95ad2fe2fcbbe262f9782f0709d7041
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Add a manual test for QMacCocoaViewContainer

Added a test for QMacCocoaViewContainer which will enable the testing
of the fixes coming

Change-Id: I73e9540bc58411634f0da7d398eca4b7ffafba8e
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>

Test: Enabled tst_QFile::invalidFile

The test is not hanging on Windows anymore.

On Windows, add an expected failure for the failing case.

Task-number: QTBUG-22801
Task-number: QTBUG-27306

Change-Id: Iede95766504f3e8a278a4554a5967ca333aae3bf
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

Fixed instances of "to to" in qtbase.

Really it should just be "to".

Change-Id: I7d0fff334bac3f0ac78adb8131e3bf3df1f1b242
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>

Add some tests for compatible and incompatible custom deleters

Change-Id: I2bb7f597aab0f00a250881602e6a6bb7bfe42e18
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Move the fetch and store pixel functions to qdrawhelper.cpp

These functions have begun showing in the tst_symbols unit test as
"symbol does not start with q". Since they're never called directly,
they're never inlined. Since they were inline, the compiler was probably
deciding to not export them.

Something changed and it could be anything (new compiler version, new
options, etc.). So mark them static.

Change-Id: I838dfc94edd7f09c202743bff0daf9d20c10c3a6
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Introduce an option to disable stripping during installation

The default is the current behaviour: strip on installing release, no
strip on installing debug. This option does not change the
installation of debug builds because qmake does not support that.

Change-Id: Ic208d5ffe860d5f1ee1cafdc944e12001673d33f
Reviewed-by: Davide Pesavento <davidepesa@gmail.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

Centralise handling & ignoring of SIGPIPE in qcore_unix_p.h

We had two instances of this function in the Qt source code, one clearly
a copy of the other, so both had the same thread-safety issue. Instead,
let's have one copy and have both write_nosignal() and sendto() call
them.

Q_NO_POSIX_SIGNALS is also gone. It was only used with Symbian.

Change-Id: I0f1354a8e9df8e6b10a02f86a940e3c6d1222087
Reviewed-by: Peter Hartmann <phartmann@rim.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

fix build failure for MinGW using "system" freetype

The configure script correctly detects freetype and sets
"system-freetype" in QT_CONFIG. However, the project file did not
consider this possibility.

Change-Id: I9ce90e7cd032a12ed7d06d3858b16c5a8d90c073
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

Try to auto-detect the font rendering for WinCE

Native font rendering is broken for QML applications, use free type
rendering for those scenarios. Use native rendering for all other
applications.

Use a workaround to find out if we are running a QML application on the
target.

Related to QTBUG-24205.

Change-Id: I653ea579098db1e58af8176cb2c3f943be0b9602
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

QModelIndex: add constexpr

The functions dealing with the void* internalPointer() can't be
constexpr on GCC 4.6 and Clang 3.2-trunk, even though GCC 4.8-trunk
accepts it, because of the casts required.

Change-Id: Id04105312da3d0c7632f7df06a34bc5a71120b32
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

QModelIndex: store quintptr instead of void*

Rationale:
1. Comparing pointers that don't point the same array is undefined behaviour, IIRC,
   and op== and op< did that.
2. The functions that cast to/from the storage type can't be constexpr. It makes
   more sense to have the quintptr functions be constexpr (they have a fighting
   chance to actually get passed something constant) than it is to have the void*
   functions constexpr. Thus, the storage type should be quintptr.

Also prepare op< to be constexpr-compatible.

Change-Id: I4b2d4a0ec8ca80d619d272bf07c57887cbd11c2f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

When trying to run cmake, don't show stderr output.

It looks like the qmake process failed if cmake is not installed.

Change-Id: I721796a602d8c572144e9d21be5d62b737698b73
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

Fix hover handling in QTreeView

Cached hoverBranch could get invalid if branches were collapsed or
expanded programmatically, leading to a crash in some situations.

Fixed the logic for updating hovered over branch indicators and
also now update hoverBranch when drawing so that it is guaranteed to be
up to date there - this fixes issues like hover indicator not updating
when the view is programmatically scrolled.

Task-number: QTBUG-27158
Change-Id: I5bd1ad76aee512ad78df33959a84ead16886a47c
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

Implement viewOptions() logic in QListViewPrivate.

The private method is called in several places in QListViewPrivate,
but before this patch, the implementation in QAbstractItemView was
 called. This meant that the options were not set properly,
resulting, for example, in icons in icon mode being laid out in
list mode (on the left and small).

This is a regression resulting from
8eab9cbce2014ffc6186dd6b6fb73da85956ee86 and
3578e05b293c1cc53582109001b20f02c7a1eeb7. Other views are not
affected in a similar way.

Change-Id: I753cb99410e367266753eaf2fa43361b9212ab96
Reviewed-by: Pier Luigi Fiorini <pierluigi.fiorini@gmail.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

Specify the --build-config when invoking ctest.

Otherwise we get strange build errors by attempting to build an
project in debug mode against a release-mode Qt:

http://testresults.qt-project.org/ci/QtBase_master_Integration/build_03955/win32-msvc2010_Windows_7/log.txt.gz

(grep for MSVCRTD)

The CMAKE_BUILD_TYPE variable is always specified by the ctest_testcase
driver as either Debug or Release.

Change-Id: Ia5af7f0c31550b92679a3190e07b4e5c4ea638c6
Reviewed-by: Brad King <brad.king@kitware.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

Pass the unmodified directory/project name to build-project.

The result of replacements for special characters should not be
used here. CTest needs the replacements for the test names, but
requires the argument to the project() command to match the
argument to ctest --build-project.

Task-number: QTBUG-27087
Change-Id: I4ff7a7e00ddb65ab0c0f2305901e1d03f46e07d1
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

Don't append _build to the directory name of tests expected to fail.

The name passed to the project() command must be the same as the argument
to ctest --build-project.

Initially I had appended _build to disambiguate the generated project
from the project under test, but that is not needed anyway as the
project under test is self-contained, so it's ok.

Task-number: QTBUG-27087
Change-Id: I648d57271529d4e8d308ff60b81419ade29b2e44
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

wrap QLibrary related code with QT_NO_LIBRARY

To fix a compile error when QT_NO_LIBRARY is defined.

Change-Id: Ie72b60b8204641fa05f4cdbf66e908cb3526217e
Reviewed-by: Jing Bai <jing.bai@digia.com>
Reviewed-by: Richard J. Moore <rich@kde.org>

OpenGL: Remove API for stubbed Geometry shader support in gui/opengl

There is a pending patch to reinstate this for Qt 5.1 (the patch
introduces new API top QOpenGLContext). I will fold the revert of this
patch into:

https://codereview.qt-project.org/#change,31231

Note that QGLShaderProgram still supports Geometry shaders so source
compatibility with Qt 4.8 is maintained.

Change-Id: I53faedd4e4a2a6c2a9765afdc1bee6df5ae2f711
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

Fix potential breakage in QString.

Negation operator ("!") have precedence over bitwise and ("&").

Change-Id: I39e2d99da6eaa4477bbe35a1259f745e05c9841a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Update "About Qt" dialog from Nokia to Digia

Task-number: QTBUG-25205
Change-Id: I9294f8e36523f242b3d05d521f6a24274ba1b0a0
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Janne Anttila <janne.anttila@digia.com>

xcb: dynamic QScreens; primary first; corrected logical DPI

A new QScreen is created when an output is activated (monitor or
projector is added, for example), and destroyed when the output is
turned off.  Ensures that screens and siblings are always in
the right order: primary comes first.
Logical DPI is derived from virtual geom / virtual size,
which will be different than output geom / physical size
if X was started with --dpi override.  This is a good thing:
when X gets wrong EDID info for physical size and you need to
override it to get reasonable font sizes, Qt will heed the
logical DPI for font sizing.

Change-Id: I5e3de34013c1b6b21067243de56f3f1eb72787fa
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

Use QStringList::join(QChar) overload where applicable [tools]

This is an automated change performing the following replacements:
                    join\("(.)"\) -> join('\1')
  join\(QLatin1String\("(.)"\)\)  -> join(QLatin1Char('\1'))
  join\(QStringLiteral\("(.)"\)\) -> join(QLatin1Char('\1'))

Change-Id: Ia087beb886bbaec1a0976cd924440d8904044879
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Use QStringList::join(QChar) overload where applicable [QtCore]

This is an automated change performing the following replacements:
                    join\("(.)"\) -> join('\1')
  join\(QLatin1String\("(.)"\)\)  -> join(QLatin1Char('\1'))
  join\(QStringLiteral\("(.)"\)\) -> join(QLatin1Char('\1'))

Change-Id: I81c378ef6aeeada5e116f1394cc9fc67f901ffd6
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Use QStringList::join(QChar) overload where applicable [examples]

This is an automated change performing the following replacements:
                    join\("(.)"\) -> join('\1')
  join\(QLatin1String\("(.)"\)\)  -> join(QLatin1Char('\1'))
  join\(QStringLiteral\("(.)"\)\) -> join(QLatin1Char('\1'))

Change-Id: I1d0c9782cc1522d937b930531720e32d4c8f7ce8
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

QTestLib: clean up qCompare specialisation/overload mix

qCompare() was both overloaded and specialised, but always as a template.
This lead to the QIcon specialisation actually invoking
   qCompare(QFlags<void*>, ...)
when specifically asking for qCompare<void*>() (detected by adding
underlying-type detection to QFlags).

Fix by preferring overloading and not specialising anything.

Change-Id: Ie001ebb9dfb0847c6c33a3f45177a61579fd61ee
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

configure: Update contact information for commercial users

Task-number: QTBUG-23949
Change-Id: Iaca7f45eae39ab7f6368431daead93358b17fe36
Reviewed-by: Janne Anttila <janne.anttila@digia.com>
Reviewed-by: Sinan Tanilkan <sinan.tanilkan@digia.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

test: Remove QSKIP from tst_QFocusEvent::checkReason_ActiveWindow()

This test actually passes using the "xcb" platform and fails otherwise.

Second, the Jira task is misleading since it refers to
tst_QFocusEvent::checkReason_Shortcut()

Change-Id: Icab91ace8c214d958b534c5cebae900242522372
Reviewed-by: Caroline Chao <caroline.chao@digia.com>

Redirect keyboard/mouse grab to the widget parent window.

Use the native parent's window if the widget in question does not
have one. This should be in line with Qt 4.8 using effectiveWinId().

Remove redundant code in grabMouse(QCursor).

Change-Id: Id6ab192e739221fe89f865f4d2f7a6d4671a190b
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

QStringList: adapt documentation to cover new join(QChar) overload

Change-Id: Ia73501e1e8a562cfab750d1796fb597cd89eab74
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QFileInfoGatherer: remove useless locks

The m_iconProvider and m_resolveSymlinks members are only ever
accessed from getInfo(), which is not called from run(). So the
mutex-lockers aren't needed in the setters.

Change-Id: I3257c95366528c5c66352d20a09e8a43b811f3ba
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

QModelIndex: clean up integer size confusion in the API

QAIM::createIndex() took either int or quint32, but QMI::internalId()
returned qint64.

In the new interface, createIndex() takes, and internalId() provides,
integers of type quintptr.

This matches the storage size of the void* in the model index and
avoids truncation.

Remove the
  createIndex(int, int, quint32) and
  \obsolete createIndex(int,int,int)
overloads.

This makes a literal 0 in the third parameter ambiguous now.
The solutions have been noted in changes-5.0.0.

Change-Id: I0a0ecd8430eaf695129a4d09d14d4e30745485c4
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

Change copyrights from Nokia to Digia

Change copyrights and license headers from Nokia to Digia

Change-Id: If1cc974286d29fd01ec6c19dd4719a67f4c3f00e
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

QFlags: don't provide a constructor from void**

Use a pointer-to-member instead of void** for Zero, the type that is
used to accept only a literal 0, but not other ints, as QFlags ctor
arguments.

This was developed while trying to find the cause for a build failure
in the qCompare<QIcon> specialisation after adding underlying-enum
detection to QFlags, and it didn't help, but I think it's a saver
alternative to void**, in particular since the ctor in question is
implicit.

Change-Id: I71c67b0b50e9404e4e42836d09d62663296f58af
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

QDBusMetaTypeId: replace a volatile bool with an atomic int

Since there is no non-atomic data that is protected by 'initialized'
anymore, the read from, and the store to, 'initialized' may now have
relaxed memory ordering.

Change-Id: I58004e782d9fd93122efb31fa5b30ee160646d99
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QDBusMetaTypeId: don't cache the result of qMetaTypeId<>() in static ints

There's not much point in caching the result of qMetaTypeId<>,
because it's already internally memoised.

In addition, the code that initialised the static int caches wasn't
protected against concurrent access under the assumption that the
operations performed were thread-safe.

That is true for most of them, but not for the stores to the static ints,
which race against each other:

   // Thread A               // Thread B
   r1 = initialized /*=false*/
                             r1 = initialized /*=false*/
   r2 = qMetaTypeId<...>();
                             r2 = qMetaTypeId<...>();
   message = r2;             message = r2; // race, ditto for all other ints

To fix, turn the ints into inline functions that just call the respective
qMetaTypeId<>() function.

Change-Id: I5aa80c624872c3867232abc26ffdcde70cd54022
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Correct QSpiAccessibleBridge::setRootObject definition

QSpiAccessibleBridge::setRootObject was defined to require a
QAccessibleInterface* as parameter instead of QObject*, as a result of
that, QSpiAccessibleBridge::setRootObject did not reimplement
QPlatformAccessibility::setRootObject and the adaptor was not initialized.

After this change, applications should work with screen readers.

Change-Id: Iee3d39f8e0959c1d75b1e682847d266a723c141b
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>

configure: remove CONFIG features while getting PKG_CONFIG

When qmake is used to get PKG_CONFIG, core and gui are not
available. This motivates CONFIG-=qt. In fact, we don't need any
features for this job.

Change-Id: Id247054d43c50f6aeb62db7585c3e90f57aa36a1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

QGtkStyle: remove an unnecessary sanity check for theme name

QGtkStyle was unable to detect the current GTK+ theme.

Change-Id: Ifd816e044bb8294f4ee2e3598ff53eb7cbce6064
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>

A step out from Harfbuzz (reduce dependency)

Introduce QCharAttributes and use it instead of HB_CharAttributes everywhere in Qt
(in Harfbuzz, the HB_CharAttributes is only used in the text segmentation algorithm
which has been moved from HB to Qt (well, most of it)).
Rename some members to better reflect their meaning,
remember to keep HB_CharAttributes in sync with QCharAttributes.
Also replace HB_ScriptItem with a (temporary) QUnicodeTools::ScriptItem struct
that will be replaced with a more efficient/friendly solution a bit later.

The soft hyphen and the mandatory break detection has been factored out
of the default text breaking algorithm to a higher level in order to refactor
the QCharAttributes bitfields and to optimize the implementation for the common case.

Change-Id: Ieb365623ae954430f1c8b2dfcd65c82973143eec
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Fix build under MinGW after 1e3269c0863bfac26be9e9239b6427ea77c6175b

Change-Id: I3c8c28ba1016af6351afa5118893133d373cb4e6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Richard J. Moore <rich@kde.org>

Test: Update QSKIP message of tst_QVector::outOfMemory

This test is crashing. Specify the bug report number in the QSKIP
message.

Task-number: QTBUG-22343
Task-number: QTBUG-27285

Change-Id: I4d4ead4f54944a545103a3d01c5c9d302d7fb1df
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

Test: fix warning for tst_QPluginLoader::loadGarbage

Garbage libs (under elftest) are numbered from 1 to 5 and not from 0 to 4.

Change-Id: Ia0162372bf5cd1fb53a0442543c5a65716880611
Reviewed-by: Sergio Ahumada <sergio.ahumada@digia.com>

Fixed multiple modal windows from all blocking each other.

A modal window in front in the modal window list should never be blocked
by a modal window further back in the list. This was taken care of in
QGuiApplicationPrivate::isWindowBlocked(), we just need to make sure it
gets called when a new modal window gets shown so that its blocked
status is up to date.

Task-number: QTBUG-27206
Change-Id: I590f1715e66067edb178081352636f34fe54a885
Reviewed-by: Tim Jenssen <tim.jenssen@digia.com>

Added manual test for QScreen properties

Shows property values in fields which auto-update when the properties
change.

Change-Id: Ib97566a74cb8d0fff5f85bf97783e89dfb07481f
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

OpenGL: Add runtime check for OS X version and improve logic

Apple recommend adding NSOpenGLProfileVersion3_2Core when asking for
a 3.2 Core profile context and NSOpenGLProfileVersionLegacy in all other
cases.

Also added a missing runtime check for OS X 10.7 or newer. Fixes a
potential crash if Qt was built on 10.7/8 but executed on 10.6.

Change-Id: I4c09d2dbbe8df25a3553cc01b468dabab0f8eaa4
Reviewed-by: Björn Breitmeyer <bjoern.breitmeyer@kdab.com>
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>

Documentation fixes.

Change-Id: I2e234ec4307bbe39359cdfa8b13e86661dad43b6
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@digia.com>

Fix network doc links.

Change-Id: I319644da80d91c5cf6d247410f022fbfd062a1f3
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Martin Smith <martin.smith@digia.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>

Fix doc errors.

Change-Id: I9974d870c081620217582fcb6f8b146d208a471c
Reviewed-by: Peter Hartmann <phartmann@rim.com>
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>

Move examples to proper location.

Change-Id: Ib808f5d99cb8217f8786411b531fa5bc7fa5250a
Reviewed-by: Martin Smith <martin.smith@digia.com>

Move opengl/wid/net example docs to proper folders.

Change-Id: I846439a9cf7ad965ed27a00f98dbc4ff97abe73b
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
Reviewed-by: Martin Smith <martin.smith@digia.com>

Clarify valueChanged() signals' docs in QSpinBox and QDoubleSpinBox.

Explicitly mention that two signals are emitted and that the QString
overload does include prefix() and suffix().

Task-number: QTBUG-26190
Change-Id: I90435779dcb2e4aa494c4b1b6e73e7f69d7998a4
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Jerome Pasion <jerome.pasion@digia.com>
Reviewed-by: Geir Vattekar <geir.vattekar@digia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>

Document that date/time formats without separators are unsupported.

The code is not handling formats like "HHmm ss" correctly, so it needs
to be documented until such support is provided.

Task-number: QTBUG-26067 QTBUG-26596
Change-Id: Ia456d8020e3e0aa9422e6e6987ac984f308facf9
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Fix regression in QAbstractScrollArea

The style hint SH_ScrollView_FrameOnlyAroundContents was currently
being ignored by QAbstractScrollArea. This looks like an accidental
regression following 10c6f015f45092040c281bb90a65179f598a00b1.

This code path does not execute on mac so it should have no impact
on that patch.

Change-Id: I78ca0a6b87dfdd7d426acbb3ef49480390211af2
Reviewed-by: Christoph Schleifenbaum <christoph.schleifenbaum@kdab.com>
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>

Add comments to document the internals of QMutex

Change-Id: Ieb5632017e5e8e09a11dc6b929efa19b4f350086
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Use QStringList::join(QChar) overload where applicable [qmake]

This is an automated change performing the following replacements:
                    join\("(.)"\) -> join('\1')
  join\(QLatin1String\("(.)"\)\)  -> join(QLatin1Char('\1'))
  join\(QStringLiteral\("(.)"\)\) -> join(QLatin1Char('\1'))

Change-Id: I9c9964703dedfdab6e7bfac80be22bd5570e2e49
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

qmake: add ProStringList::join(QChar)

Same reasoning as for 68e04c3ac148bcbe71f2deeb7288563f6cdbcab5 applies.

Adding the overload was easier than to teach a Perl script to distinguish
between QStringList and ProStringList instances...

Change-Id: I6de6ecf21fdad135ac213b5c794927a9bc120a92
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

announce the android platform in the respective specs

Change-Id: I55c8d9fe4160c37eec6073894481e313e4dea191
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

add missing maemo platform announcement to n9 spec

Change-Id: I24855dc3fdfcc19a81f7e030c9dfbbdddd2bb552
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

claim that mingw is a platform

it's mostly win32+gcc, but sometimes there are library differences, so it
makes sense to have an own platform flag for it.

Change-Id: I985f0dd39a47bfc16f10f4b4a9c168874ae2f20d
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

make qmakespecs announce the compiler family

on the way to eliminate scoping based on the spec.

gcc and msvc go as such into CONFIG, the other ones get the vendor
prefixed, as most are mostly unknown and thus likely to clash with
users' flags.

Change-Id: Ie622f53d90e96dbf05ce7d8c638cd355f04fa20c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

don't skip the configure.exe build even when it already exists

the build steps are now fast enough to make no-op rebuilds no problem.

Change-Id: I1018735bf50f3e7a66e22637237f26f98a3baf1d
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

don't skip the syncqt call even when include/ exists

the syncqt run is now fast enough

Change-Id: Ie95a617fe99791ed38d02fe41578a0136fb68ddc
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

don't set MODULE_BASE_OUTDIR unconditionally

webkit actually tries to set it, and after the syncqt invocation changes
we were breaking it.

Change-Id: I1ec740b7e96c972ca173f384529dde1fc7e7cc13
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

Replace QStyleOptionViewItemV? with QStyleOptionViewItem

The former are a typedef for the latter in Qt 5. This only touches
internal implementation, as the API was migrated long ago.

Change-Id: Ided73021ebecc00508e6325c3d988b6c6ad336cd
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Jens Bache-Wiig <jens.bache-wiig@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>

move syncqt call to qt_module_headers.prf

for one, the syncqt invocation is qt module magic and thus simply does
not belong into the generic default_pre file.
second, this way the forwarding header generation is now linked to the
rest of the build magic for a particular module, which is way less
confusing for the unsuspecting developer.

Change-Id: Idc8e420d3faf173d7fff4a41e6e1c59af15c3023
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

make error() abort the qmake run, not just the current file

Change-Id: I82fc55680f9ffb227e25acb39c797596225ba89e
Reviewed-by: Daniel Teske <daniel.teske@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

Fixed outline / fill inconsistencies in raster paint engine.

We did coordinate rounding of the fill in the raster paint engine to
match how drawLine_midpoint_i rendered lines. With the new cosmetic
stroker in 4.8 this rounding is not needed anymore.

Task-number: QTBUG-26013
Change-Id: Ibe86fa809276c8e7988d393a67e772d57b9a4799
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Fixed inconsistent rounding of square cap pens.

A horizontal line should round up at the same time as a vertical line
with square cap, when rendering at subpixel coordinates. Thus, the
special casing in the cosmetic stroker of offsetting by half a pixel
should be for flat caps instead of for square caps.

Task-number: QTBUG-26013
Change-Id: Ic09249337f814c7de95a17976ec9e651561a744b
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

Do not refer to non-existant member in QSurfaceFormat docs

Sample buffers should be enabled automatically under the hood
when samples is > 0 and therefore the getter and setter are
gone.

Change-Id: I9cd6438e89060f901598a6aacacb99362fac4718
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>

Cocoa GL context: do not crash when closing windows

Running hellowindow and closing its windows one by one would result
in crashes without the additional checks for the presence of the
underlying native window.

Change-Id: Id32bdfda9a77936380b2e9a95fb56ceb52028d9d
Reviewed-by: Morten Johan Sørvig <morten.sorvig@digia.com>