fold m_qmakespecFull back into m_qmakespec

the original value is not used any more after the final resolution.

Change-Id: Icadc219f045a1bbfd20506c4c72c53d1fb352969
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

make 'make check' work with non-installed prefix builds

add the appropriate environment to the command lines

Change-Id: Ic9d3c60255be88f431213acd8c5e0ff55e014f5b
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

factor out qtAddTargetEnv()

Change-Id: Ib0517da54da98de4b1f2ee7d80bee22316231091
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

remove pointless conditionals+branches

the surrounding TEMPLATE == vc* scope already implies vcproj, as dsp is
not supported any more.

Change-Id: I68363aca62e21135f42572040ccc7b189dcf32c8
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

remove mentions of borland from project files

Change-Id: I3049b24ae287bee50aca1de06664a2309df69edf
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

don't make the outputs depend on $$QMAKE_DUMPCPP

the variable can contain a complex command which would make a mess of
the makefile.
also, the tool doesn't change all that often, so it's not particularly
useful to depend on the binary.

Change-Id: I965d6555cb22bd6d5306b324f5cad0d747128e99
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

remove now unnecessary indirections

qmake's new parser can deal with this much more readable code.

Change-Id: I772c9da4c9d049e2b332305f41496ed1196f60eb
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>

Fix typo in docs.

Change-Id: Icfafa00062b442903579dd7993c75fffb60187f9
Reviewed-by: Peter Hartmann <phartmann@rim.com>

Make QStatusBar text available via accessibility.

UNTESTED cause the build fails for me (unrelated to this patch).

Change-Id: Ie6836596b6fbaf93baa21d7008509774f1379f5c
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>

Fixed potential use of uninitialized variable in qgl.cpp

If glGetBooleanv fails this variable is left uninitialized.

Task-number: QTBUG-26952
Change-Id: Idb42833f2907bb66fd87c5cdb39753fca04e7438
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>

Split the qt_message_output function in two

This way, I can ensure that the noreturn qFatal function calls a
noreturn function as its last action.

Change-Id: I6db7aa2539af430414645f386d7fce87cdb97186
Reviewed-by: Kai Koehne <kai.koehne@digia.com>

QFileInfoGatherer: remove unused members userId/groupId

None of the code references these, so remove.

Change-Id: Ie48d30b49c644f093df9f42c8d087751c4e980ba
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QFileInfoGatherer: fix race conditions pt.1: abort

Fix a race on the 'abort' variable. While there was a mutex lock around
the code that sets the variable in ~QFileInfoGatherer, there was no
protection in getFileInfos(), where it is read:

   // T:this->thread()  // T:*this
                        // in getFileInfos(), after last mutex.unlock()
   mutex.lock();
   abort = true;        while (!abort...
   // ...               // ...

Fix by making 'abort' an atomic. This means that we can drop the mutex
locker in the destructor, too. We still mostly access 'abort' under
protection of the mutex, because we need to protect other variables that
just happen to be accessed together with 'abort', but we avoid the mutex
lock/unlock on each iteration of the while loop in getFileInfos().

Also cleaned up the logic in run():

- by using the canonical form of condition.wait() (in a loop that
  checks the condition), we can ensure that !path.isEmpty() and avoid
  having to use the updateFiles boolean.
- by checking for abort.load() after we return from
  condition.wait(), we minimise the waiting time for thread
  exit.
- by using different local names, we avoid having to this->qualify members.

Also changed one condition.wakeOne() to wakeAll() for consistency
with fetchExtendedInformation().

Change-Id: If35f338fe774546616ec287c1c37e2c32ed05f1a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: David Faure <faure@kde.org>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Pass correct argument to QShortcutEvent ctor in tst_qabstractbutton.cpp

The QShortcutEvent constructor takes "int shortcutId" as its second
argument, not a bool. Since the default shortcutId is 0, this test
passed, since false == 0.

Change-Id: I43bbae4613f3badb1578dccec76dcdd3c96a3a2f
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

QVarLengthArray: provide STL-compatible member function names

This allows, among other things, to use QVarLengthArray as the target of
a std::back_insert_iterator.

Change-Id: I507f612a23da854bf865780aa0a7e6312f4a896b
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

OpenGL: Make use of the requested version and profile in QGLWidget

When instatiating a QGLWidget in Qt5 at present it ignores the version
and profile information in the QGLFormat meaning that we always end up
with an OpenGL 2.x profile rather than what we asked for. This commit
properly takes the version and profile into account.

Change-Id: I097f10e397c23f5d97c5fcd8d5354667da286896
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

OpenGL: Add flags and logic for newer versions of OpenGL in QGLFormat

Change-Id: Iad57286c36c7fd205f06abc40cca5f568d572e81
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

OpenGL: Request latest GL version when using QGLTemporaryContext

This is necessary so that QGLFormat::openGLVersionFlags() returns the
complete set of versions. Without this patch the temporary context
ends up obtaining an OpenGL 2.x context as the QPA plugins now prefer
to use the ARB {wgl,glx}CreateContextAttribsARB function which creates
contexts of the specified version (2.0 for a default QSurfaceFormat).

Change-Id: I2c3e02c9d7b77d4ee0484f9e8786f41675f8a439
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

Make QTimer::singleShot take a pointer-to-const QObject

The obvious idea is that a connect() happens behind the scenes.
As QObject::connect takes a pointer-to-const, singleShot should
do that as well.

Change-Id: I36433c723441294b2088b23f0c37724ab43d9503
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Remove qUpperBound usages from qtbase

Replace them with std::upper_bound; this allows for deprecation of
qUpperBound.

Change-Id: Idef01d2228b9a70eee3d52931d7aedb5bb6ba902
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>

Remove qLowerBound usages from qtbase

Replace them with std::lower_bound; this allows for deprecation of
qLowerBound.

Change-Id: I536e7338eb85ea6c7c1a5bf23121292767927e0b
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>

configure.exe: Improve help output

Add line breaks where it's looking awkward in cmd.exe , and new lines
where needed. Also adapt some descriptions to the configure shell script
output.

Change-Id: Ie784e715f51f7ff692ee85f7c960dc4583b65b23
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

Fix configure.exe output

Fix help output for -no-fontconfig, introduced in commit 1838a6c2.

Change-Id: I686000d51c56fde56179dcebef31b0783c0dfc29
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

sql: Include own headers first

cpp files should include their own headers first (but below config.h)

Change-Id: I225bc0f09988167ae7f938f7f21a77d05a3d191b
Reviewed-by: Caroline Chao <caroline.chao@digia.com>
Reviewed-by: Mark Brand <mabrand@mabrand.nl>

Doc: Use QStyledItemDelegate and frameless spinboxes in spinbox delegate example

Task-number: QTBUG-18847
Change-Id: I5347a1028b6f45a60e43fb75f6d362efecb6880b
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Marius Bugge Monsen <marius@cutehacks.com>
Reviewed-by: Geir Vattekar <geir.vattekar@digia.com>

Fix tab navigation in windowflags example

Change-Id: Ia0e3087ea60262546cd1a21614591ad3626469d3
Reviewed-by: J-P Nurmi <jpnurmi@digia.com>

Introducing NoDropShadowWindowHint window flag

Added suppport on QCocoaWindow.

Also we deprecate WA_MacNoShadow since it isn't used anywhere, and updated
the 'windowflags' example app.

Change-Id: Id0b453ba15a23b768b0615838597bca139f507ad
Reviewed-by: J-P Nurmi <jpnurmi@digia.com>

fix non-bootstrapped $$system()'s line end conversions on windows

Change-Id: I026da1312bc28619dc7aa5974b7f4c00710517ac
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

Core: Fix gcc warning about signed/unsigned integer comparison

Fix warning: 'comparison between signed and unsigned integer expressions'

Change-Id: I5ce2857115137f9a06eddc74fda55ae49e37d345
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Network: Fix mingw-gcc warning

Fix warning: missing braces around initializer for 'BYTE [6] {aka unsigned char [6]}'

Change-Id: Ic247ff0721955c4562ed027fad7cb3786809a840
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

Network: Fix gcc warning about qDebug format

Fix gcc warnings: 'unknown conversion type character 'l' in format' , and
'too many arguments for format'.

Change-Id: Ic7547a5bd5ba0d4c9f115004f250a0f7d3b378cb
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

QElapsedTimer: Fix typo in restart() documentation

Change-Id: Iac5df9edbe16ec841f783bc5266091a6e34e9cab
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Fix mapping to/from global coordinates for child/embedded windows.

QWidget's mapToGlobal() and mapFromGlobal() functions assumed that
if the widget reports it's a window or if it has no parent widget, it
must be a top level window whose coordinates are in global coordinates.
This is not true for child QWindows or embedded native windows
(QAxWidgets).

Changed the logic for mapping coordinates to use equivalent methods
from QWindow if widget has a window handle, and changed QWindow's
methods to map coordinates using native methods if window is embedded.

Also fixed newly failing accessibility autotest. The geometry related
failures there popped up because now the position of the rect returned
by accessible interface is actually correct while widget geometry still
reports position 0,0 before widget has shown up.

Task-number: QTBUG-26436
Change-Id: I658fafd0ce01eb1604ba255efeeba3073ca0189f
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

X11 (xcb) support for QScreen-per-output and runtime changes

QScreen has notifiers for its properties, but they were not being
emitted when one changes the resolution or arrangement of individual
outputs, e.g. via xrandr.  Also there should be one QScreen per
"output", e.g. laptop LCD + external monitor means 2 QScreens
which will be siblings, rather than just 1 QScreen to represent
the whole desktop.

Change-Id: Ia61bbc5e6a3506f813ab11f87c03d14cf7f4ce85
Reviewed-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Samuel Rødal <samuel.rodal@digia.com>

Disable SSL compression by default.

Disable SSL compression by default since this appears to be the a likely
cause of the currently hyped CRIME attack.

Change-Id: I515fcc46f5199acf938e9e880a4345f2d405b2a3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>

Fix thread-safety of qt_ignore_sigpipe

The testAndSet operation would mean another thread could see the value
of 1 and proceed to write(2)/sendto(2) before SIGPIPE had been ignored.
If the pipe or socket were already closed by then, a SIGPIPE would be
delivered to the application with its default action: terminate.

Change-Id: I62dc8f5fa14c1dd453d13e4053c642bd78fbc468
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>

don't allow overloading of built-in functions

the functions are not versioned or scoped, so user-defined overloads would
mess up qmake's own feature files. it seems safer to break user projects
than to allow the user to break qmake.

Change-Id: I020a2e6416bbb6e2fd2ece339629d848c00c8398
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

remove the entirely insane whitespace compression

it was merely an artifact of using QString::simplified() on the
unparsed (!) project code. there is no reason why anyone should actually
rely on it, so just remove it.

Change-Id: If9b957c4b1263f3990a2331f8851bb1c06154ea8
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

Build tools in release mode if using -debug-and-release option

Change-Id: I1d4b69e6954526b4f999e248f4f2cde6a9da26cd
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

Prevent buffer overflow in QXcbWindow.

Thanks to Janne Kulmala for noticing this and informing about the fix.

Task-number: QTBUG-27123
Change-Id: Idd3cfd74fb7be277b6d805446aea5784b8fdf2f6
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Make qdoc read multiline macros in files with CRLF endings

This caused qdoc for instance to not being able to parse qglobal.h
correctly. (On windows, it stopped to parse anything meaningful after
the line with this macro:)

  #define Q_INIT_RESOURCE_EXTERN(name) \
      extern int QT_MANGLE_NAMESPACE(qInitResources_ ## name) ();

It worked on linux just because on linux a line continuation 'token' is
the sequence "\\\n" (on windows it is "\\\r\n")

So for files with CRLF line endings, it treated *only* the first line
as a macro, potentially causing the subsequent lines to affect the
state of the tokenizer.

Change-Id: If7c80ee7eb317f2d324ace7ff540ced7c31185dc
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Martin Smith <martin.smith@nokia.com>

fix QMAKE_MANIFEST for shadow builds

Change-Id: Iea71f4a4d483e9465b3a7f0ce4560b2e9fa6cd63
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

properly C-quote the manifest file name in the rc file

Change-Id: I8c8eb85d412becc8c4029c2aa393abf6f8e949b6
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

rewrite *_manifest.rc unconditionally

the file has no dependency tracking, so changes to the source would get
missed and cause hard to debug build issues.
and as nothing does dependency tracking on that file, this change
doesn't even cause a noticable performance regression.

Change-Id: I108b490b71a43018e0c7ef5d7c0b11d79a8e726b
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

factor out qtAddToolEnv()

so it's available for other users as well

Change-Id: I2d5a14ae427575c07321ac532b13ee03308b837f
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

remove 3rd parameter from qtPrepareTool()

it's been superseded by the QTTOOL.* module stuff.

Change-Id: I01c9fc3ebbb22111bfb03f82693c7cf08b5fc9d4
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>

define QT_OPEN_LARGEFILE

it must be defined, even if zero

Task-number: QTBUG-27232
Change-Id: I950ef61ede197af9c2d8aedddd712cc6d4cc745b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Add Q_PRINTSUPPORT_EXPORT to qdoc ignores.

Change-Id: I0b8676a14133baf818ec12cda77dcb0670b5ee85
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>

Let qVersion() be extern "C" if this is not a namespace build

This allows one to even dlopen() QtCore and call qVersion, to figure
out which version it is. A bit crazy, but someone might want to do it.

Change-Id: I932460515d07bed3f0e41c8ab2b46fc268ca73ff
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

HTTP header may be damaged - fix, unit test

"HTTP/1.1 100 CONTINUE\r\n"
If the header from a server is splitted between two packets
the first packet contains "HTTP/1.1 100" and the second one
contains " CONTINUE\r\n", one space (0x20) is skipped. After
processing the line looks in this way "HTTP/1.1 100CONTINUE".
QHttpNetworkReplyPrivate::readStatus(QAbstractSocket *socket)
is called twice, if a http header is splitted as above.
The function always removes whitespace from the beginning of a packet,
even if it is the second part of a http header.

QHttpNetworkReply returns QNetworkReply::RemoteHostClosedError
due to damaged http header during processing.

Improvement of unit test.

Task-number: QTBUG-27161

Change-Id: Ifc2949f62473209b4032185effbf5078b4130cda
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>

examples: Mark ctor's as explicit

Make C++ class constructors that can be used with only one
required argument 'explicit' to minimize wrong use of the class.

Change-Id: Ida9f9c2f0c8608c35b0137b2512a6747afd69515
Reviewed-by: Mitch Curtis <mitch.curtis@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

qmake: Include own header first

cpp files should include their own headers first (but below config.h)

Change-Id: I10ef37854843ae6438d68f96ce5ee83eede33db5
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

SIC: QtPrintSupport - Remove QAbstractPageSetupDialog

QAbstractPageSetupDialog is a completely unnecessary base class that
is not really abstract and is used nowhere else.  This changes merges
its methods into the QPageSetupDialog main class.

While technically SIC no-one else uses this so no apps should be
affected.

Change-Id: I59b1739f1c453c34c25d1664d5d042e7918db316
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

use qmake to get PKG_CONFIG in configure script

The value of PKG_CONFIG might depend on device options.
For example, "-device-option PKG_CONFIG" might be used with configure
or a mkspec might prefix PKG_CONFIG with CROSS_COMPILE which is
specified as a device option.

The shell functions of configure for parsing mkspecs do not take
device options into account, but qmake is pretty good at it now.

Change-Id: I1c9558e550c48e8441ebdac34b82066473c2ce3a
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>

test: Mark tst_QTabBar::sizeHints() failure as XFAIL

The test has one stable failure on Mac OS X, so mark this
with QEXPECT_FAIL and remove the QSKIP

Task-number: QTBUG-22320
Task-number: QTBUG-27230
Change-Id: I7660df5770c39788792068a5b68e8236551288c4
Reviewed-by: J-P Nurmi <j-p.nurmi@nokia.com>
Reviewed-by: Caroline Chao <caroline.chao@nokia.com>

Exclude QIcuCodec when QT_NO_TEXTCODEC is defined

Change-Id: Iec0178c427abcc1c79e4fe6ef449d399ac8ca363
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

network: Fix missing or improper include guard in headers

Use an include guard in headers to ensure the header is not included
more than once. Make the header guard match its file name.

Also, cpp files should include their own headers first (but below config.h)

Change-Id: Iecf5da23c0f8e6d457f67657b88ef7557bde9669
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

concurrent: Fix missing or improper include guard in headers

Use an include guard in headers to ensure the header is not included
more than once. Make the header guard match its file name.

Also, cpp files should include their own headers first (but below config.h)

Change-Id: I902c8936382f5c1a8e0de7dbf49e5423f9b72bbe
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

gui/opengl: Fix missing or improper include guard in headers

Use an include guard in headers to ensure the header is not included
more than once. Make the header guard match its file name.

Change-Id: I29b41e9d33e4ea17165f44c49de0a963574dd809
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>

Attempt to stabilize QMenu auto tests

Change-Id: I7d8260570a280182990d78e07d696aa4b46e1747
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>

Make maemo scope syntax in project files more generic

It is necessary to use the n9 device file for now in scratchbox or/and on the
community open build service because the maemo platform mkspecs file assumes
that a cross-toolchain is used all the time. If no platform file is used, then
for instance certain plugins may not be built in general. There is currently an
ongoing issue with the meego plugin for context management in the Harmattan
components project. That is currently not built due to this issue, so no
orientation works in those applications.

The nice solution would be to make the maemo platform file work with cross and
native toolchains as well, but that requires a decent amount of investigation
and work. Thereby, the scope is extended this way for now.

Change-Id: I172c7d152bdbb2db279526d9fd1ca5648d0cd0a9
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>

Blackberry: fix tst_QFileInfo::fileTimes()

The Blackberry OS uses a filesystem with the noatime option,
which returns a "wrong" access time.

Change-Id: I04cdb899699e819a36e0917e30d750067b33388d
Reviewed-by: Tobias Koenig
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>

QErrorMessage: Use qInstallMessageHandler instead of qInstallMsgHandler

qInstallMsgHandler got deprecated in Qt 5. Use qInstallMessageHandler instead.

Change-Id: Ie1156ca11eb6be555c681007ddc230978211d029
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>

Autotests: Use qInstallMessageHandler

qInstallMsgHandler got deprecated in Qt 5.

Change-Id: Ib36983e66b3a8090b99f14e3fd4e210602a3f018
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

qdoc: Refactoring of qdoc data structures

This commit is the beginning of a significant
overhaul of qdoc. A new class, QDocDatabase, is
added, which will eventually encapsulate all the
data structures used by qdoc. In this commit, the
Tree class is made private and only accessible
from QDocDatabase. Several maps structures are
also moved into QDocDatabase from other classes.

Much dead code and unused parameters were removed.
Further simplification will follow.

Change-Id: I237411c50f3ced0d2fc8d3b0fbfdf4e55880f8e9
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>

Fix the comment for the maemo mkspecs since it does not work for fremantle

See the bug QTBUG-18361 for details.

Change-Id: Idbe2043c07cca795e8e2fca8204993dfb0a11988
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Rename the N9 device mkspecs folder to match the *-maemo* regex

Change-Id: I0c7b74d67f44d776a74671d38bbc0f2424974c2c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Romain Pokrzywka <romain.pokrzywka@kdab.com>
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>

Mark (non-public API's) ctor's as explicit

Make C++ class constructors that can be used with only one
required argument 'explicit' to minimize wrong use of the class.

Change-Id: I12ad5b6eb1794108c6b7464a2573e84068733b03
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

remove usage of QT_MAKEDLL in favor of the new QT_STATIC macro

Change-Id: I411cab0d92a8cad6b4f3676afe5cee050c4b789b
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Remove support for Intel C++ 7.1 and 8.0

Those compilers are old and untested. Qt is only tested with ICC 12.1
and 13.0.

Change-Id: I3a5fa9452de27a61ee087700cc054a76fe4160ee
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Fix repaint issue when resizing a window in Windows

QWindowsWindow::handleWmPaint() cached the device context it got from
BeginPaint call and used that for the window when the backing store
blitted itself to the window. The problem with this device context is
that the clipping region is set to only encompass the newly exposed
areas, which means any changes caused by the resize on the previously
exposed area were not repainted.

Fixed by removing the DC caching. The benefit was minimal anyway.

Change-Id: I8bd3c4031432ce6b52434c80bfe65d35d9feae49
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>

QDoc manual: ordered command list

Some commands were not listed in alphabetical order.

Change-Id: I299abe145ca0f9d63c9ceefb801f5c53548e7938
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Martin Smith <martin.smith@nokia.com>

Documented QtOpenGL classes that have better replacements as obsolete.

These have new replacements in QtGui since Qt 5.0.

Change-Id: I20e8fdd4e9c14d8c626eda3072217af1858bf492
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>

qpa: avoid useless resize in DirectFB platform

Check current size before resizing QDirectFbBackingStore through
its resize() method to avoid useless deallocation/reallocation.

Change-Id: Idc32eb20d3d3cac988fa9dd2feb80910303f5763
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>

Avoid deprecated QTest method in unit test

Change-Id: I1d6fc84962ba29a8963f99b87cccb4af465a2935
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>

Remove deprecated QInputMethod::visible()

Replaced with isVisible()

Change-Id: Id47683954e89c53e73fc5171977609de7f1b4245
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Fix compilation on Harmattan

The GLES2 libraries on Harmattan do not provide a typedef for GLChar.  Work
around it here by adding it. The Kkronos headers specify GLChar as a typedef to
char, so if an implementation already provides it, then this doesn't do any
harm.

Change-Id: I0848b72b81fcc602dc8d9eecefdacb5436163040
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

tests: Don't omit the body of a test function with QT_BUILD_INTERNAL

Changing it outside of the test function definition to avoid running
empty/inapplicable test functions.

Change-Id: I713560cde7f715696984ed082d682900f5f1bcdd
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Caroline Chao <caroline.chao@nokia.com>

Move QRecursiveMutexPrivate to qmutex.cpp and mark inline

Disassembly of the optimised code shows that the compiler was already
inlining the bodies of one or both functions (since they're in the same
.cpp, it's allowed to do that).

However, since there was no "inline" marker, the compiler was also
emitting an out-of-line copy, which wasn't used by anyone, as the class
is not exported.

So add the marker. To make sure that they don't get used by accident
elsewhere, the class is moved to the .cpp file too.

Change-Id: Iead578ec9c7d8dd6b4e6bb582ce5b829cdec5992
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Remove QBasicAtomic::tryLock with a timer

The non-futex implementation does not support timed locks in
QBasicMutex. That is only supported in QMutex (due to the
destructor).

Change-Id: I46d33a66a36e05c8a4344823537178e80a6ddd76
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

#include the qmutex_xxx.cpp at the end of qmutex.cpp

This produces slightly better inlining results.

Change-Id: Ie86471577f888cb2d9c9989306ec69ce0a296108
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Use storeRelease to unlock a mutex instead of fetchAndStoreRelease

We're not checking the result anyway, so use a simpler operation.

Change-Id: I8c2db35be86660b29d81dd97ce3e269de55a37df
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Avoid calling QElapsedTimer::nsecsElapsed before the first futex sleep

The first time we're going to sleep, the timeout should be exactly the
value that was passed by the user. We don't need to calculate the time
elapsed between start() and a few lines below.

Change-Id: I99c363b6f0ecfd07ad787b79b75e61771733c2b3
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Split the timed mutex lock from the non-timed lock functions

Non-timed mutex locks are by far more common, so let's try not to
penalise the locking of those with code that won't get used that
often.

Change-Id: I37f56d6429836467fdec2e588c0fb22d914b5d75
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Simplify the futex lock loop: no need for nested while

Once we enter the inner loop, we never exit it except to return from
the lockInternal() function, so the rest is never executed again.

As a consequence of this, we won't try to fastTryLock() twice per
mutex. Therefore, for a non-recursive mutex, if lockInternal() is
entered, we'll definitely need to use futexes.

Change-Id: Ice617ed27449c1fbdc112a159a86cd0660125e13
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

Make QBasicMutex be exclusively non-recursive

Dispatch to the recursive mutex functions from QMutex::lock, tryLock
and unlock.

This has the benefit that those using QBasicMutex will not go through
the testing for recursive mutexes. It simplifies a little the code for
those users.

For the users of QMutex, the code required to perform a lock does not
appear to change.

Change-Id: I0ca9965e012b283c30f1fab8e9f6d9b3288c2247
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Don't recheck about timeout == 0 during mutex locking

If the timeout wasn't zero, it can only become zero if we return from
futex() with a non-timeout reason but subsequently expires while we're
recalculating something.

A side effect is that we try-lock a non-recursive mutex exactly
once. Before this change, we'd fastTryLock() twice even with
timeout == 0.

Change-Id: I0af09fc2a84669a683a843fcf1513203b075dfb7
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>

Hoist the recursive mutex check out of the inner loop

A non-recursive mutex doesn't suddenly become recursive, so we don't
need to check it multiple times.

Change-Id: Id040254b6142d320a7bd3111491082ad09968404
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>

QtDBus: added missing QT_NO_DBUS guards

Change-Id: If3bbba7765a4949e5a7aefca063dc56c21c06687
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Fix QDBusServer with more than one connection

Create a new QDBusConnectionPrivate for every new connection in
qDBusNewConnection instead of creating a single QDBusConnectionPrivate
in the QDBusServer constructor which gets assigned the latest connected
DBusConnection in qDBusNewConnection (and loses track on all previous
DBusConnections).

Also extend tst_QDBusConnection::registerObjectPeer() test with multiple
connections to the server.

Task-Number: 24921
Change-Id: I4341e8d48d464f3fe0a314a6ab14f848545d65a0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Add qdbusconnectionmanager_p.h to the HEADER list.

Change-Id: Iacba510beb0ee72182931e97044eaa5b46df30af
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
Reviewed-by: Lorn Potter <lorn.potter@gmail.com>

throw() is C++, so don't use it in C mode

Q_DECL_NOTHROW is used by qVersion(), which can be extern "C".

Change-Id: If71f3bc3a0df3e8321237e5ac9f3bea82380c1e4
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

config.tests/arch: make distclean on the test, not qt

Additionally, qmake no longer needs -o since we are in the project
directory.

Change-Id: I89dbe829c064663b653e8f747f0849d63ceb367e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>

Use QElapsedTimer instead of QTime to measure elapsed time

QElaspedTimer is a lot more efficient.

Change-Id: I66d9514108f800a45181d8960b01a5e7d3fa9e80
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Increase the fuzziness of the QSemaphore test to 50 ms

The test tries to acquire a semaphore that isn't ready, so it will
timeout. The test is working properly.

The only problem is that 10 ms is too strict. For a wide variety of
reasons, especially since the Qt CI system runs multiple tests in
parallel, the program may not run again in that 10 ms window.

Change-Id: Ic0f684895f73646db5f1cc783fe9ef75fb1ab02b
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>

Fixed: QFileInfo::lastModified() returning wrong value.

Fixed so that empty QDateTime is returned for non existing file.
Fixed also created() and lastRead() to return empty QDateTime for
non existing file.

QFileSystemEngine::fillMetaData() returned true for non existing
files. This was also corrected.

Task-number: QTBUG-25811

Change-Id: I523eb99e4405b4b813b2950f85cc646239181d07
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Christian Stromme <christian.stromme@nokia.com>

Doc: Marked the QUrlInfo class documentation as \internal

Change-Id: Ib84c9ffc181edf17850d5334c95c035428d31b6c
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>

Test: Redo remove QSKIP from printsupport tests

Redoing 5704cbc462ad0b316a6b55a0f1dadc3c50762ef

This can't work, because the define comes from qconfig.h, not qmake.

Skipping initTestCase and cleanupTestCase when QT_NO_PRINTER is
defined.

Change-Id: I2527c018294b7518a4692f2c93da933848640b5d
Reviewed-by: J-P Nurmi <j-p.nurmi@nokia.com>

Fix QAbstractItemModel documentation errors.

Added a dependency on the qtwidgets module to qtcore.qdocconf so that
it has access to widgets examples. Also added examples/widgets to
exampledirs.

Change-Id: I7a69f51d51b65bd534950f0df5b348d6ef61a5d9
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>

Doc: Fix snippet bug in Getting Started Qt

Change-Id: I9c8caabee4c6f6131a69851586ef992138beb211
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>