profile/ivi/qtbase.git
11 years agoreplace \key and \gui qdoc commands with \uicontrol
Jeremy Katz [Wed, 1 Aug 2012 12:36:13 +0000 (14:36 +0200)]
replace \key and \gui qdoc commands with \uicontrol

Change-Id: I0753305d4fe1ea20417f451766101da1247dfeeb
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
11 years agoqdoc: Removed useless qdoc warnings
Martin Smith [Wed, 1 Aug 2012 11:53:02 +0000 (13:53 +0200)]
qdoc: Removed useless qdoc warnings

qdoc was printing warning messages when a QML type
was not given a minor version number in its
\inqmlmodule command. The minor version number is
not used for anything, so the warning messages are
no longer printed.

Change-Id: I5586b4d860fcc8e392de3d388c3327ed0029e266
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
11 years agoOpenGL: Fix typo in extension name
Sean Harmer [Fri, 27 Jul 2012 17:42:13 +0000 (18:42 +0100)]
OpenGL: Fix typo in extension name

Change-Id: Ibb65b832290085bd565388982594eb889e950aaa
Reviewed-by: Laszlo Papp <lpapp@kde.org>
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
11 years agoSet the Qt API level to compatibility mode in all tests.
Thiago Macieira [Wed, 1 Aug 2012 11:19:40 +0000 (13:19 +0200)]
Set the Qt API level to compatibility mode in all tests.

Qt 5.0 beta requires changing the default to the 5.0 API, disabling
the deprecated code. However, tests should test (and often do) the
compatibility API too, so turn it back on.

Task-number: QTBUG-25053
Change-Id: I8129c3ef3cb58541c95a32d083850d9e7f768927
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
11 years agoFix crash in uic
Andreas Holzammer [Wed, 1 Aug 2012 11:53:10 +0000 (13:53 +0200)]
Fix crash in uic

Compiling Qt5 uic with Microsoft Visual Studio 2008
with SP1 will lead into a assertion in the copy
constructor of qstring.

This is apparently a compiler Bug.

Change-Id: Ia3353434d00b2e87800b937d891eabef86293751
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoAdded default argument for color profile to QImage constructors.
Alexandros Dermenakis [Mon, 23 Jul 2012 14:27:10 +0000 (16:27 +0200)]
Added default argument for color profile to QImage constructors.

Added default argument for color profile to QImage constructor
initializing QImages from pixel maps. This is to be used in future Qt
versions where color profile support will be added.

Change-Id: I815c3db4ef52f8383b48dad844b5f188aa3a3eee
Reviewed-by: Yoann Lopes <yoann.lopes@nokia.com>
11 years agoSkip doc subfolders when checking includes.
Lars Knoll [Tue, 31 Jul 2012 18:40:25 +0000 (20:40 +0200)]
Skip doc subfolders when checking includes.

This removes the bogus warnings when running
syncqt from configure about snippets in the doc/
folders.

Change-Id: I8e3d258238c1cd1e763ce63b895d58b68bbc06f3
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agoQNX: Do not send geometry change events to Qt too early
Sean Harmer [Tue, 31 Jul 2012 14:27:37 +0000 (15:27 +0100)]
QNX: Do not send geometry change events to Qt too early

We explicitly do not send geometry change events to Qt from QQnxWindow
from the constructor. This prevents us from ending up in resizeEvent()
reimplementations from the QWindow ctor.

Change-Id: I045b35aa7eb23890772fe131c3d19314252f6a5a
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Kevin Ottens <kevin.ottens.qnx@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
11 years agoQNX: Ensure that the QQnxWindow has a valid size for EGL surfaces
Sean Harmer [Tue, 31 Jul 2012 13:09:15 +0000 (14:09 +0100)]
QNX: Ensure that the QQnxWindow has a valid size for EGL surfaces

If QWindow::create() gets called before resize() or setGeometry() then
the rect argument passed into QQnxWindow::setGeometry() by the ctor
is null.

This could potentially result in 0 sized buffers and EGL surface
creation being attempted. We bail out in this case and tell
user how to fix their application.

Change-Id: Ia709dba617fa266bd66b61409bdb1c980acfb1a2
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Kevin Ottens <kevin.ottens.qnx@kdab.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
11 years agoQNX: Remove dependency on QtWidgets from QNX QPA plugin
Sean Harmer [Tue, 31 Jul 2012 17:12:18 +0000 (18:12 +0100)]
QNX: Remove dependency on QtWidgets from QNX QPA plugin

Change-Id: I52027752630beee81a97644983dbccf973c0c2c2
Reviewed-by: Kevin Krammer <kevin.krammer@kdab.com>
Reviewed-by: Kevin Ottens <kevin.ottens.qnx@kdab.com>
Reviewed-by: Thomas McGuire <thomas.mcguire@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
11 years agostatemachine: Make states exit order spec-compliant
Kent Hansen [Wed, 1 Aug 2012 08:28:10 +0000 (10:28 +0200)]
statemachine: Make states exit order spec-compliant

The SCXML spec states that entry order should be equivalent to
"document order" and exit order should be "reverse document order".
Since QStateMachine uses child order for the entry order, the exit
order should be reverse child order.

Change-Id: Ia7b05fdd5c9261ccf202f64f8d23f5c88b20a8c3
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@nokia.com>
11 years agoRemove usage of deprecated qWaitForWindowShown(QWidget *) method.
Friedemann Kleint [Mon, 30 Jul 2012 10:36:02 +0000 (12:36 +0200)]
Remove usage of deprecated qWaitForWindowShown(QWidget *) method.

Change-Id: I445d24a09dbb7abb62a37bd9914284f21a4f08f1
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
11 years agoMake sure that the parsing mode reaches QUrlPrivate::setHost
Thiago Macieira [Tue, 31 Jul 2012 13:19:30 +0000 (15:19 +0200)]
Make sure that the parsing mode reaches QUrlPrivate::setHost

Ensure that the parsing mode is cascaded down from setAuthority and
setUrl so that the hostname parsing does not attempt to decode
percent-encoded hostnames when it shouldn't.

Take the opportunity to also remove the "Boolean Trap" from
QUrlPrivate::setHost.

Change-Id: Ia64754c4a4900182700b7af1382aea8410abc7e9
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
11 years agoMake QUrl::setScheme only parse in strict mode (no decoding)
Thiago Macieira [Tue, 31 Jul 2012 12:28:48 +0000 (14:28 +0200)]
Make QUrl::setScheme only parse in strict mode (no decoding)

The URI RFC defines schemes as containing only a very restricted set
of characters, none of which require encoding, so don't even
try. Testing this behaviour in some web browsers indicate that they do
not accept percent-encoded schemes either.

Change-Id: I692dd20e1aac7e8a1bcb276cb5113b5802393d38
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
11 years agoFix QUrl support for empty usernames and passwords
Thiago Macieira [Tue, 31 Jul 2012 12:06:10 +0000 (14:06 +0200)]
Fix QUrl support for empty usernames and passwords

If the password is empty (but present), the userinfo component of the
URL should end in a colon (":"). QUrl already supported that and it
was tested (case "password-empty").

If the username is *also* empty but present, the userinfo component is
just the colon (":"). Fix support for that case by checking if we
stored the presence flag instead of checking the size of the
component.

Change-Id: Ie224493a997dbf76b2e44dd6d55fd9674ac83c1c
Reviewed-by: David Faure <faure@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
11 years agoFix handling of encoded NULs (%00) in QUrl::fromPercentEncoding
Thiago Macieira [Tue, 31 Jul 2012 11:10:54 +0000 (13:10 +0200)]
Fix handling of encoded NULs (%00) in QUrl::fromPercentEncoding

QString::fromUtf8, without an explicit size, (currently) defaults to
stopping at the first NUL. That means we need to pass an explicit
size.

Also take the opportunity to test that QUrl::toPercentEncoding also
works with the same data.

Change-Id: I79362d67afda624b01ca07b0315b611c4aa3fdda
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: David Faure <faure@kde.org>
11 years agoRemove references to Borland and bmake
Andy Shaw [Wed, 25 Apr 2012 15:12:53 +0000 (17:12 +0200)]
Remove references to Borland and bmake

Remove references to an old compiler that has not been
supported for a long time.  Also remove Borland specific
configuration flags which have no meaning elsewhere.

Change-Id: I3634a52b78f737ea972073e14c2b6669dcd0ae63
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agodon't change a widget's active state in QWindowsWindow::setWindowState
Joerg Bornemann [Mon, 2 Jul 2012 10:36:18 +0000 (12:36 +0200)]
don't change a widget's active state in QWindowsWindow::setWindowState

QWindow::setWindowState is not supposed to set the window active.
The method requestActivateWindow() should be used for that.

When switching from and to fullscreen mode we're always passing
SWP_NOACTIVATE to SetWindowPos to not change the activation state
of the window. This is inverse to the old behaviour, which did not
have an effect.

Change-Id: I339337935cdad76b3ef252202e92177f37543038
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoDoc: QVarLengthArray::length is new in 5.0
Christian Kamm [Tue, 31 Jul 2012 08:57:59 +0000 (10:57 +0200)]
Doc: QVarLengthArray::length is new in 5.0

Change-Id: I796e67d677309460d79a7a64a6890dfbcfbace69
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoAdding configure option: -no-gcc-sysroot
Thomas Senyk [Thu, 12 Jul 2012 11:03:16 +0000 (13:03 +0200)]
Adding configure option: -no-gcc-sysroot

This option is important if you want to use configure's
and qmake's -sysroot (e.g. PKG_CONFIG settings, device-files),
but the toolchain (in combination with the rootfs)
is not able to handle gcc's --sysroot.
One known case is freescale's ltib setup where the toolchain
itself comes with all the essential files (e.g. crt1.o),
while the rootfs has none of those files, so gcc's
--sysroot can't be used. The rootfs on the other hand contains
all kinds of "less important" files/packages (e.g. libdbus).
For those "less important" files/packages Qt needs pkg-config
to be able to include/link properly. Therefore one needs
configures -sysroot without gcc's --sysroot.

Change-Id: Iaec9b07012f2945f3ecb3ced0ed95176721b5ecd
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agoAdd a .prf file for SIMD qmake variables.
Michael Goddard [Tue, 24 Jul 2012 02:49:14 +0000 (12:49 +1000)]
Add a .prf file for SIMD qmake variables.

So you can get AVX/NEON etc source compiled by assigning to the
corresponding variable (e.g. AVX_SOURCES).

This was previously used in just the gui module, but other
external modules might like it too.

Change-Id: I51aa64760c469c7dc4c71e6f089c2ddef4f509c5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agoUse QtCore instead of QtGui for the standardpaths comments to be correct
Laszlo Papp [Mon, 30 Jul 2012 18:18:52 +0000 (19:18 +0100)]
Use QtCore instead of QtGui for the standardpaths comments to be correct

Change-Id: I98004fed565ac9653947ec70a3197b0372abcf2b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: David Faure <faure@kde.org>
11 years agoLazy initialize library paths
Harald Fernengel [Fri, 27 Jul 2012 09:53:46 +0000 (11:53 +0200)]
Lazy initialize library paths

For a lot of command line tools, library loading is not required, so
don't waste a lot of time computing them. According to callgrind, this
makes the QCoreApplication constructor factor 6 faster, and also removes
a lot of stat() calls and other file system access.

Change-Id: I0211f5303712fa0dcfc4168cce7025283c63c9d1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoAvoid qobject_cast
Kent Hansen [Tue, 31 Jul 2012 09:57:08 +0000 (11:57 +0200)]
Avoid qobject_cast

This is slightly faster, and also avoids accessing the object's meta
object that would cause any lazily created meta objects (e.g., for
QML) from being built.

This is just a port of commit a5972f8ec32631e70578c64e9d6c063b3c47f0f5
(which fixed QGuiApplication, but not QApplication).

Change-Id: I0d5080495aa4fd4486db297968c4d06bbfacf89b
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
11 years agoAdd command line options to windowgeometry test
Andreas Holzammer [Tue, 31 Jul 2012 12:00:11 +0000 (14:00 +0200)]
Add command line options to windowgeometry test

Add -widgetminimized, -widgetmaximized, -widgetfullscreen,
-windowminimized, -windowmaximized, -windowfullscreen
as command line option to test creation of windows in
these modes

Change-Id: If192c131c8996d3b67648427e5784da47ffee971
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoFix compilation: Move qHash(const T &t, uint seed) up.
Friedemann Kleint [Mon, 30 Jul 2012 13:25:25 +0000 (15:25 +0200)]
Fix compilation: Move qHash(const T &t, uint seed) up.

It needs to be visible from
qHash(const QPair<T1, T2> &key, uint seed).

Change-Id: Ibb63ce6da1e655bfb841c5e580e184ef66c5b766
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoFix reloading of plugins
Lars Knoll [Mon, 30 Jul 2012 10:30:58 +0000 (12:30 +0200)]
Fix reloading of plugins

Unloading and reloading a plugin didn't work correctly,
because we didn't reset instance to 0 on unload.

Task-number: QTBUG-26098
Change-Id: Ic3e4497f359b1ca455be949dce9cafa9d67d8039
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoFix some syncqt warnings.
Lars Knoll [Tue, 31 Jul 2012 18:50:18 +0000 (20:50 +0200)]
Fix some syncqt warnings.

Change-Id: I6f432ee991f4bde217fa27d4004ef318f1d480e0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoFix showFullScreen for QWindow
Andreas Holzammer [Tue, 31 Jul 2012 14:12:18 +0000 (16:12 +0200)]
Fix showFullScreen for QWindow

If the setWindowState is called in a early stage,
where the window is not yet shown, the fullscreen
flag is cached in the QWindow. Then in the call to
setVisible(true) the window gets generated, but the
fullscreen is not handled there. So set the state
while window is generated.

Change-Id: I0ea53ed8a2465da6a9973d84fb4579381543e89b
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agofix QWidget::activateWindow for child widgets
Joerg Bornemann [Tue, 31 Jul 2012 09:43:15 +0000 (11:43 +0200)]
fix QWidget::activateWindow for child widgets

For child widgets we must use the windowHandle of the toplevel widget.

Change-Id: If7bad8ed4bba694806a2c8b77bae3d25cc200b8b
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agofix window activation in tst_qmdiArea::subWindowActivated2
Joerg Bornemann [Tue, 31 Jul 2012 11:14:49 +0000 (13:14 +0200)]
fix window activation in tst_qmdiArea::subWindowActivated2

QApplication::setActiveWindow doesn't activate the native window but
marks the widget as active inside Qt.
We need to use QWidget::activateWindow instead. See docs.

Also moved the activation call further down because on Windows a
minimized window cannot be activated using the activation-by-focus
fake we're currently using.

Change-Id: I752f6ada1f463931fa9cfb3c35f42dbec0207bfa
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoImplement synchronous WS events by flushing.
Friedemann Kleint [Tue, 24 Jul 2012 07:40:20 +0000 (09:40 +0200)]
Implement synchronous WS events by flushing.

Previously synchronous window system events were
implemented by bypassing the queue and processing
the event immediately. This is not ideal since the
event order is not preserved - there might be "happened
before" events waiting in the queue.

Add QWindowSystemInterface::flushWindowSystemEvents
and change all handleSynchronous* to 1) queue the
event 2) call flushWindowSystemEvents.

flushWindowSystemEvents is almost identical to the
already existing sendWindowSystemEvents with the
exception that it does not call QApp::sendPostedEvents.
Move the common implementation to a new private function.

Task-number: QTBUG-20778

Change-Id: Ie98a83875bc0a14e335e36bed0dd9e0ed4a1dea0
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoAdding missing QT_BEGIN_HEADER/QT_END_HEADER
Sergio Ahumada [Mon, 30 Jul 2012 13:31:52 +0000 (15:31 +0200)]
Adding missing QT_BEGIN_HEADER/QT_END_HEADER

Change-Id: I1144497ec8376ec41c714b7a55c05637b6feee66
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
11 years agoMac: Don't crash if language preferences can't be retrieved.
Eike Ziller [Tue, 17 Jul 2012 09:19:02 +0000 (11:19 +0200)]
Mac: Don't crash if language preferences can't be retrieved.

That can happen when e.g. running an application with sudo.

Task-number: QTBUG-26547
Change-Id: Ib16ef7798ebcd1c9b8d661dd2e3ce3aadc393489
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Denis Dzyubenko <denis.dzyubenko@nokia.com>
11 years agoStabilize the QHeaderView test.
Friedemann Kleint [Tue, 31 Jul 2012 07:06:50 +0000 (09:06 +0200)]
Stabilize the QHeaderView test.

The test has a member QWidget *topLevel which it recreates
and shows in init() without waiting for it to be exposed
although it is not used in every test case. This apparently
interferes with some tests that create separate top levels.
Do not show in init(), delete the topLevel.
Add wait to the cases where the topLevel is shown.

Change-Id: Ib428020b36dc82991d41e68478fd583bdfb004c7
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
11 years agoAccessibility: fix updates for models with rootIndex set.
Frederik Gladhorn [Wed, 25 Jul 2012 15:22:26 +0000 (17:22 +0200)]
Accessibility: fix updates for models with rootIndex set.

The accessible table/list implementation would ignore
when a root index is set, resulting in trying to use row/column
for the top level index, leading to accessing invalid indexes.

(cherry picked from commit def2ee829435c705055733a0f99c0ca44a012c50)

Conflicts:

src/plugins/accessible/widgets/itemviews.cpp

Change-Id: Ic2745ab3e262ccee2d43e0d532e165d2958f519e
Reviewed-by: Jan-Arve Sæther <jan-arve.saether@nokia.com>
11 years agobuild host tools in release mode again
Joerg Bornemann [Mon, 30 Jul 2012 11:52:01 +0000 (13:52 +0200)]
build host tools in release mode again

CONFIG does not contain build_all anymore,
thus we need to check QT_CONFIG in bootstrap.pr{i,o}.

Change-Id: Ia505fa101adc49f185908ca575d3211caed612db
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agotst_qwidget task reference fixed
Joerg Bornemann [Mon, 30 Jul 2012 10:56:34 +0000 (12:56 +0200)]
tst_qwidget task reference fixed

Change-Id: I8d65f5912bfc2daa1e35be33a3c4d986508eebeb
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoadd configure options for debug/release OpenSSL
Peter Kümmel [Thu, 26 Jul 2012 15:16:29 +0000 (17:16 +0200)]
add configure options for debug/release OpenSSL

msvc cannot use the same library for debug and release builds
if openssl libraries are linked statically into the network library.

Change-Id: Ic27ede2d9531b94aff4c50c1699947ce72caf286
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
11 years agoremove second argument from $$eval() (access to global scope)
Oswald Buddenhagen [Tue, 24 Jul 2012 15:09:14 +0000 (17:09 +0200)]
remove second argument from $$eval() (access to global scope)

sam added this based on a vague notion of backwards compatibility when
he fixed the function to use the local scope, but it kind of makes no
sense - there is very little reason for accessing the global scope from
within a function. google doesn't find any relevant hits except our
source code, so let's just nuke it.

Change-Id: Ie957bb47b531f7e9b5dfcceb4e09f65dd826b422
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agodon't set -rpath-link in qmodule.pri
Oswald Buddenhagen [Mon, 23 Jul 2012 12:04:43 +0000 (14:04 +0200)]
don't set -rpath-link in qmodule.pri

qt.prf takes care of that already

Change-Id: Ibc426583a007edbbe2f53cc21ce676eb60b33485
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agorun processPrlFiles() over QMAKE_LIBS_PRIVATE as well
Oswald Buddenhagen [Fri, 20 Jul 2012 13:34:54 +0000 (15:34 +0200)]
run processPrlFiles() over QMAKE_LIBS_PRIVATE as well

there is no reason why private libs should not have prls resolved.

the two variables are resolved independently, so it's possible that
(even more) libraries will appear duplicated on the linker command line,
but that seems easiest for the time being.

Change-Id: I9070ba53808a0661fa72949db8111106b7aca487
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoinline MakefileGenerator::processPrlVariable()
Oswald Buddenhagen [Fri, 20 Jul 2012 13:31:14 +0000 (15:31 +0200)]
inline MakefileGenerator::processPrlVariable()

the visitor-pattern like approach is not needed any more

Change-Id: I990db681cbeee91d89ecba97745a8104595247e7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agomake the windows prl processing more like the unix variant
Oswald Buddenhagen [Fri, 20 Jul 2012 13:23:01 +0000 (15:23 +0200)]
make the windows prl processing more like the unix variant

it's more elegant, and more similar code is better.

Change-Id: I2b8b036cb70a932fd171e23cf7d3389188401924
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoremove apparently useless code
Oswald Buddenhagen [Fri, 20 Jul 2012 13:04:39 +0000 (15:04 +0200)]
remove apparently useless code

it wouldn't actually *do* anything, as l is not a reference. i cannot
figure out the original intention, so let's just drop it.

Change-Id: Ic0a3457a1872cde827259ee5530959120456e934
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agostub out MakefileGenerator::processPrlFiles()
Oswald Buddenhagen [Fri, 20 Jul 2012 10:29:21 +0000 (12:29 +0200)]
stub out MakefileGenerator::processPrlFiles()

the unix and windows variants don't rely on it.
not making it purely virtual only because the project generator inherits
MakefileGenerator as well but does not need an implementation.

Change-Id: I80099b3f5d07cd037b408cf1099c58ff3a2904cd
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoSearch the include paths for json files containing plugin info
Lars Knoll [Fri, 20 Jul 2012 10:39:53 +0000 (12:39 +0200)]
Search the include paths for json files containing plugin info

This fixes shadow builds with autogenerated .json files as
e.g. used by Qt Creator.

Change-Id: Ibb783b05d97d996100da4b0dca859fa3f310dc83
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
11 years agoStabilize QWidget-test alienwidgets on XCB.
Friedemann Kleint [Mon, 30 Jul 2012 20:54:08 +0000 (22:54 +0200)]
Stabilize QWidget-test alienwidgets on XCB.

Use QTRY_VERIFY for the mappped attribute check.

Change-Id: I3cbde9122405bf7067f3702193e80636edc8c5c6
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
11 years agoqclass_lib_map.h: Fix include of the QDeclarativeView widget.
Friedemann Kleint [Mon, 30 Jul 2012 10:10:58 +0000 (12:10 +0200)]
qclass_lib_map.h: Fix include of the QDeclarativeView widget.

Task-number: QTBUG-25196

Change-Id: If25c4df29297435b159706257d9f37a9ef2f3d7a
Reviewed-by: Kai Koehne <kai.koehne@nokia.com>
11 years agoAlign the CP949 codec name with ICU
Lars Knoll [Mon, 30 Jul 2012 07:27:19 +0000 (09:27 +0200)]
Align the CP949 codec name with ICU

It's name is windows-949 according to ICU. Keep
CP949 as an alias for compatibility with Qt 4.

Change-Id: I115ba2593da6f7b47e25136c3fadb19c7f798ff0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoFixes to the qdom autotest
Lars Knoll [Mon, 30 Jul 2012 07:26:03 +0000 (09:26 +0200)]
Fixes to the qdom autotest

Don't rely on a black list of codecs for the serialization
test. This breaks badly when new codecs get added to Qt
(e.g. through ICU). Instead use a white list of known
codecs that can encode/decode the test data.

Change-Id: I1dc55a25e852198bb935f070a4a21e8369f56268
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoAdd missing namespace macros
Lars Knoll [Fri, 27 Jul 2012 07:05:43 +0000 (09:05 +0200)]
Add missing namespace macros

Change-Id: I7cfeece0914b03b6113207de96b931dc54377f77
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoSmall doc fixes about the list of supported codecs.
Lars Knoll [Wed, 25 Jul 2012 11:04:19 +0000 (13:04 +0200)]
Small doc fixes about the list of supported codecs.

Change-Id: I98b8ec9d5de9e69f1bb6187b4d820d61f7ce03e5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoDo not preload any codecs anymore when using ICU
Lars Knoll [Wed, 25 Jul 2012 10:57:15 +0000 (12:57 +0200)]
Do not preload any codecs anymore when using ICU

When using ICU we can now avoid pre loading any codecs
at startup. Instead QTextCodecs will always be created
lazily when first asked for.

Change-Id: Ic668f2824700896d2eca7d0de54f978404826163
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoAdd a factory method to QIsciiCodec and fix the names
Lars Knoll [Wed, 25 Jul 2012 10:54:56 +0000 (12:54 +0200)]
Add a factory method to QIsciiCodec and fix the names

Make the names all lower case to be in line with what ICU
reports. Add a factory method to create the codecs according
to their name.

Change-Id: Ia3dddaa7701a0645d9d8fbcb52f1f8d58f79cf98
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoUse ICU instead of iconv/windows codecs for the locale
Lars Knoll [Wed, 25 Jul 2012 08:17:44 +0000 (10:17 +0200)]
Use ICU instead of iconv/windows codecs for the locale

With ICU there's no need to use the windows locale or iconv
codecs anymore as locale codecs.

Change-Id: I50c94a97ed6acbf4c6f05b2a88593a57ebfa8342
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoMake codecForLocale atomic
Lars Knoll [Wed, 25 Jul 2012 08:16:59 +0000 (10:16 +0200)]
Make codecForLocale atomic

Change-Id: I8017b1c2aa1df8d613e83919a945fd5f320713d6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoICU code page conversion support
Lars Knoll [Fri, 20 Jul 2012 22:31:37 +0000 (00:31 +0200)]
ICU code page conversion support

Use ICU to do code page conversion instead of the
builtin text codecs. With this QTextCodec simply
becomes a wrapper around ICU's ucnv_* methods.

We only keep our own codecs for UTF-*, ISO-8859-1,
ISO-8859-15 for performance reasons, and for TSCII
and iscii-* because they aren't supported by ICU.

Change-Id: I4fc49eba55cf772b9772c6dac606a47a44346a60
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoRename some codecs to be in line with ICU
Lars Knoll [Sat, 21 Jul 2012 21:34:07 +0000 (23:34 +0200)]
Rename some codecs to be in line with ICU

Change-Id: If523d2a12ef64c79e3860da1f430f128d24ff600
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoDisable iconv on utf-8 based platforms
Lars Knoll [Fri, 20 Jul 2012 19:22:47 +0000 (21:22 +0200)]
Disable iconv on utf-8 based platforms

There's no need to use iconv on Mac,
iOS, Android or QNX, as all these
platforms are fully utf-8 based.

Change-Id: I2a03c8dea72ad91f6ec83da0a838de20e46babef
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoLazily initialize iconv in the iconv codec
Lars Knoll [Fri, 20 Jul 2012 18:49:13 +0000 (20:49 +0200)]
Lazily initialize iconv in the iconv codec

Avoid dlopen'ing libiconv and initializing
it's members until the codec gets used for
the first time. This avoids some memory
and startup time overhead in case we can
use the utf8 codec instead of iconv.

It also removes a circular dependency
between codec initialization during app
startup.

Change-Id: I119c010c288dc59ab32279d8a213ae1f4347cace
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoUse utf-8 as the locale codec where it makes sense
Lars Knoll [Fri, 20 Jul 2012 09:02:40 +0000 (11:02 +0200)]
Use utf-8 as the locale codec where it makes sense

Use utf8 as the fallback codec in case we can't determine
something else instead of latin1.
Also use utf8 for Mac, iOS, Android and QNX.

Change-Id: I15dc85d2406b1ebdfacff25f45f8b6854c52b97e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoRemove encoding detection according to LANG
Lars Knoll [Fri, 20 Jul 2012 08:54:25 +0000 (10:54 +0200)]
Remove encoding detection according to LANG

This code is left over from the 90's, and should not
get used anymore. In all relevant cases we will use UTF-8
or iconv nowadays.

Change-Id: Ie3776f671de33f782fa77f6359bf6e105bd9c1b8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoMove the text codec list to qcoreglobaldata.
Lars Knoll [Fri, 20 Jul 2012 08:00:52 +0000 (10:00 +0200)]
Move the text codec list to qcoreglobaldata.

This removes some global statics in QTextCodec and
makes the code better to maintain in the longer term.

Remove QT_NO_THREAD defines around mutexes as this
isn't required in Qt 5 anymore.

Change-Id: I15ede75f53b16f134f4053f3188c4b47e86fcd8a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoEnsure that QAtomicPointer of forward-declared classes works
Thiago Macieira [Fri, 27 Jul 2012 14:28:36 +0000 (16:28 +0200)]
Ensure that QAtomicPointer of forward-declared classes works

This is working, but let's just be sure by adding a testcase.

Change-Id: I8c6b5ded0c7b6c90645dbf70a7ce6c1ba447a284
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Jason McDonald <jason.mcdonald@nokia.com>
11 years agoQtNetwork: Better detection of connection failures on Windows
Martin Petersson [Thu, 12 Jul 2012 10:52:17 +0000 (12:52 +0200)]
QtNetwork: Better detection of connection failures on Windows

If the error code from WSAConnect is WSAEWOULDBLOCK, then the
operation proceeds but the outcome is not known at that time. We then
check SO_ERROR's value to detect errors. But if that call returns 0
this could indicate that the value is still not know. In this case
we try one more time to increase the chance of getting the correct
value.

This fixed the tst_QNetworkReply::getFromUnreachableIp auto test
on Windows.

Change-Id: I25008aca062b2f823e3d93ebb0ae456d7e4a6ecc
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
11 years agoQUrl: document that setIdnWhitelist isn't thread-safe
Marc Mutz [Sat, 28 Jul 2012 23:04:08 +0000 (01:04 +0200)]
QUrl: document that setIdnWhitelist isn't thread-safe

This is a forward-port of 6b10fc91 from Qt 4.

Change-Id: I58878bf24e4f1b50ebfd0457c37eef58696f4a4c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoMake benchmarks compile without QtWidgets
Rohan McGovern [Mon, 30 Jul 2012 03:58:52 +0000 (13:58 +1000)]
Make benchmarks compile without QtWidgets

Ensure benchmarks which need QtWidgets are gracefully disabled when
that module is unavailable.

Fixed one unnecessary usage of "QT+=widgets".

Change-Id: I8031b5dca585749f0f4d22e0637adc3f57f4e418
Reviewed-by: Kalle Lehtonen <kalle.ju.lehtonen@nokia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
11 years agoqdoc: Added safety checks to findUnambiguousTarget()
Martin Smith [Thu, 26 Jul 2012 12:16:58 +0000 (14:16 +0200)]
qdoc: Added safety checks to findUnambiguousTarget()

This is a blind attempt to fix QTBUG-26615. I can't
reproduce the crash on OSX.

Task nr: QTBUG-26615

Change-Id: Ifc6b68b7f302b9cd7d02129be8ae56841a35e5de
Reviewed-by: Martin Smith <martin.smith@nokia.com>
11 years agofix QEventLoop qdoc errors
Jeremy Katz [Fri, 27 Jul 2012 14:20:52 +0000 (16:20 +0200)]
fix QEventLoop qdoc errors

Change-Id: I17bb720d539a7ccf5d9a3b20853897ff052fc8a6
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
11 years agotests: fixed check for QtOpenGL
Rohan McGovern [Mon, 30 Jul 2012 03:45:51 +0000 (13:45 +1000)]
tests: fixed check for QtOpenGL

contains(QT_CONFIG,opengl) tells whether Qt itself is able to use
OpenGL, which is not the same thing as whether the QtOpenGL API is
available. Make the check correct; fixes compilation when Qt is
configured with -no-widgets (which also disables QtOpenGL).

Change-Id: Iaa296c2b10650971ef4846f8bc6f44761fadcf7c
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
11 years agotests: disable qstandarditemmodel test when building without QtWidgets
Rohan McGovern [Mon, 30 Jul 2012 03:44:10 +0000 (13:44 +1000)]
tests: disable qstandarditemmodel test when building without QtWidgets

This test uses QTreeView.

Change-Id: I6ba56ad6ec3687a5b90d516be10d6f88e248d53d
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
11 years agotests: removed QtWidgets dependency from tst_qsignalmapper
Rohan McGovern [Mon, 30 Jul 2012 03:13:44 +0000 (13:13 +1000)]
tests: removed QtWidgets dependency from tst_qsignalmapper

This test was using some QSpinBox instances to generate signals for
testing. Use our own QtTestObject to generate the signals instead.

Change-Id: I3714955ae040d541c3b613a478945c38a18be18d
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
11 years agoQNX: Reduce dependency on QtWidgets
Kevin Krammer [Sun, 29 Jul 2012 08:46:12 +0000 (10:46 +0200)]
QNX: Reduce dependency on QtWidgets

Use QObject::inherits() to check if the current input focus object is
a numerical input instead of trying to cast.

Change-Id: Ie476fba3d2f9f20a1ce0328043430cb6a96a8a3d
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
11 years agoSend the native scancode in evdevkeyboard
Laszlo Agocs [Wed, 25 Jul 2012 11:36:16 +0000 (14:36 +0300)]
Send the native scancode in evdevkeyboard

Until now the native scancode was not specified which meant
when qwindow-compositor (or any other Qt-based compositor using
the standard Wayland protocol, i.e. sendKeyPress() and such) tried
to send the native scancode over to the clients, all it got was a
code of 0.

This fixes keyboard input when running qwindow-compositor on top
of kms and EvdevKeyboard.

Change-Id: I5d9a882f249634fb7584900b5f298f794befee3b
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
11 years agoDocument QEvent::EnterEditFocus and QEvent::LeaveEditFocus properly.
Mitch Curtis [Fri, 27 Jul 2012 08:48:48 +0000 (10:48 +0200)]
Document QEvent::EnterEditFocus and QEvent::LeaveEditFocus properly.

These enums are only accessible if QT_KEYPAD_NAVIGATION is defined.

Task-number: QTBUG-15631
Change-Id: Ifa826ac8f2bf28225572e0c664f18ffa5e78c302
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
11 years agoremoved duplicated docs in android qsharedmemory and systemsemaphore
Jeremy Katz [Fri, 27 Jul 2012 12:10:20 +0000 (14:10 +0200)]
removed duplicated docs in android qsharedmemory and systemsemaphore

Qdoc sees all source and header files, rather than the subset selected
by qmake. This leads to qdoc errors when the same functions are
documented in multiple implementations.

In the cases fixed here, all documentation was duplicated, so no
information was lost.

Change-Id: Id7f8e2e617862a3778d0b05b38d8811aad456ac5
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
11 years agoSpecialize QTypeInfo for QUrlTwoFlags to mark it PRIMITIVE.
Stephen Kelly [Wed, 25 Jul 2012 15:17:14 +0000 (17:17 +0200)]
Specialize QTypeInfo for QUrlTwoFlags to mark it PRIMITIVE.

QTypeInfoMerger combines the typeinfos for the (PRIMITIVE) flag arguments.

Change-Id: I5abf00489491d099f2bf7ba25c191a771a383d78
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
11 years agoMark the QFlag and QIncompatibleFlag as primitive.
Stephen Kelly [Wed, 25 Jul 2012 15:05:56 +0000 (17:05 +0200)]
Mark the QFlag and QIncompatibleFlag as primitive.

Change-Id: I7dab4d029e7840fe4778a750a8dd7367675d7a27
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agodo not look for frameworks in library path
Oswald Buddenhagen [Thu, 19 Jul 2012 15:52:29 +0000 (17:52 +0200)]
do not look for frameworks in library path

it's just wrong

Change-Id: Ia21e9fb7e67623d6c4f70eab1f76360624ba314b
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoinsert explicit library search paths before system search paths
Oswald Buddenhagen [Thu, 19 Jul 2012 15:50:59 +0000 (17:50 +0200)]
insert explicit library search paths before system search paths

... as the linker does.

Change-Id: I4a030d9e2f281b892b78ba25b11f6df6b0893712
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agolook into /Library/Frameworks as well in processPrlFiles()
Oswald Buddenhagen [Thu, 19 Jul 2012 16:50:05 +0000 (18:50 +0200)]
look into /Library/Frameworks as well in processPrlFiles()

Change-Id: I445b9d1d3cd349eb7981b1aec824e6878b0c16e7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agounify default search paths between findLibraries() & processPrlFiles()
Oswald Buddenhagen [Thu, 19 Jul 2012 16:49:40 +0000 (18:49 +0200)]
unify default search paths between findLibraries() & processPrlFiles()

i see no reason why they should deviate

Change-Id: Iaa0445b79a95d348f51df74a74c7c89216c468ec
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agofix lookup of libraries with a full path but without extension
Oswald Buddenhagen [Thu, 19 Jul 2012 16:32:16 +0000 (18:32 +0200)]
fix lookup of libraries with a full path but without extension

given that nobody noticed so far how broken this was, this doesn't
appear to be a particularly common path. but anyway ...

Change-Id: Ic17b239d724a4d69ff414a24be2e8588732bc8dd
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agodon't actually resolve frameworks in findLibraries()
Oswald Buddenhagen [Wed, 25 Jul 2012 11:51:26 +0000 (13:51 +0200)]
don't actually resolve frameworks in findLibraries()

the code wouldn't actually *do* anything with them, because it was
completely broken. didn't seem to hurt, so just remove it.
we still need to parse -framework, though, so we don't do funny things
with its possible argument.

Change-Id: Ia3266538612d3314a72f4965ee9c1ea2d3046ac9
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agofix host vs. makefile directory separator mess
Oswald Buddenhagen [Wed, 25 Jul 2012 08:51:44 +0000 (10:51 +0200)]
fix host vs. makefile directory separator mess

the system path separator and shell are bound to the host system
(system() will use cmd even on mingw with sh.exe in path).
the makefiles otoh may depend on what the qmakespec defines.

consequently, add $$system_path() and $$system_quote() (for use with
system() & $$system()). $$native_path() is renamed to $$shell_path() and
should be used with $$shell_quote() to produce command lines in
makefiles.
$$QMAKE_DIR_SEP needs to be applied to Option::dir_sep right after
parsing the spec, so it is available to $$shell_{path,quote}().

Change-Id: If3db4849e7f96068cf03a32348a24f3a72d6292c
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
11 years agoMake 'nmake qmake' work with the stub (fast) Makefiles
Thiago Macieira [Fri, 27 Jul 2012 13:39:49 +0000 (15:39 +0200)]
Make 'nmake qmake' work with the stub (fast) Makefiles

Without this, if you ran 'nmake qmake' or 'jom qmake' in qtbase's
top directory, it would simply tell you that 'qmake' was up to date.
After all, there is a directory called 'qmake' there.

Change-Id: I6cdb405e8106c137ecbab27923e524b924b19c84
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
11 years agoWork on the QWidget-autotest.
Friedemann Kleint [Thu, 26 Jul 2012 12:04:09 +0000 (14:04 +0200)]
Work on the QWidget-autotest.

- Introduce smart pointers to delete widgets and resources
  to ensure tests are not affected by left-over widgets
  also in case of failure.
- Replace deprecated QTest::qWaitForWindowShown() by
  QTest::qWaitForWindowExposed() and use QVERIFY,
  remove some hard-coded timeouts.
- Set some titles and object names.
- Add verbose debug output of event lists in tests
  childEvents.
- Set minimum sizes on widgets to avoid Windows warnings.
- Stabilize GDIWidget, trigger on first event only.

Change-Id: I64119a2e7113e4a9f0156d00c72ce0935d03bb81
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoQDebug: Add getter/setter for auto-insert-spaces.
David Faure [Fri, 27 Jul 2012 11:53:16 +0000 (13:53 +0200)]
QDebug: Add getter/setter for auto-insert-spaces.

This is useful for inserting a string without space-handling, given that
dbg.nospace() followed by dbg.space() inserts a space.

It's also useful for QDebug operators for custom types, so that they
can disable space handling and then restore to whatever it was before
(rather than forcing it to space() mode).

Change-Id: I9d72e9ffbcbc581ed093168752c29af924405b33
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoQDebug: adjust documentation of space(), nospace() and maybeSpace()
David Faure [Fri, 27 Jul 2012 12:20:42 +0000 (14:20 +0200)]
QDebug: adjust documentation of space(), nospace() and maybeSpace()

Reality is that they control a "current mode" in the stream, not a
"bool that remembers if the last character was a space".

Change-Id: Ic907c34bcb458039b73ddff48021e19f0c24c78c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
11 years agoCorrect method name referred to in QFont's detailed description.
Mitch Curtis [Fri, 27 Jul 2012 08:58:17 +0000 (10:58 +0200)]
Correct method name referred to in QFont's detailed description.

The detailed description of the QFont class reference contains
the wrong method name (and link) setPointSizeFloat().
This change replaces it with the correct name: setPointSizeF().

Task-number: QTBUG-15536
Change-Id: Ieb2e398b8e1d0b637a6bac16a3ec48522c5dddc1
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
11 years agoqdoc: Updated qdoc manual and qdoc guide
Martin Smith [Fri, 27 Jul 2012 10:22:05 +0000 (12:22 +0200)]
qdoc: Updated qdoc manual and qdoc guide

These documents now include the \qmltype and
\instantiates commands, and the \qmlclass command
is deprecated.

Task nr: QTBUG-26648

Change-Id: I62da273be51609e651e33b9088e1667e4c6c1e76
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>
11 years agomerge QMAKE_LIBDIR into QMAKE_LIBS early
Oswald Buddenhagen [Mon, 16 Jul 2012 12:29:48 +0000 (14:29 +0200)]
merge QMAKE_LIBDIR into QMAKE_LIBS early

dealing with the directories separately doesn't buy us anything. it's
easier to mix them into the libs, as that contains some paths, too, both
in projects and in prl files.

this brings the windows generators in line with the unix ones.

Change-Id: I1f58f7edd8e21d28bfabf04384bac2e315aaf446
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agounifiy initialization of QMAKE_LIBS{,_PRIVATE} among windows generators
Oswald Buddenhagen [Mon, 16 Jul 2012 12:04:50 +0000 (14:04 +0200)]
unifiy initialization of QMAKE_LIBS{,_PRIVATE} among windows generators

Change-Id: Ib24ec6e290bfb277232c2483768c1b863f961462
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoclean up fixTargetExt()
Oswald Buddenhagen [Mon, 16 Jul 2012 11:02:57 +0000 (13:02 +0200)]
clean up fixTargetExt()

there doesn't appear to be a point in having the function virtual; the
part in the mingw generator can be inlined somewhere else just fine.

Change-Id: I50d66d505095b43fce601928c6240a684389a4b7
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agode-virtualize some more functions
Oswald Buddenhagen [Mon, 16 Jul 2012 10:48:22 +0000 (12:48 +0200)]
de-virtualize some more functions

they are not re-implemented anywhere

Change-Id: I444a967bb39ec6b5994747c9fa3f605b4c53ce4f
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agodispose of QMAKE_LIBDIR_FLAGS, "demote" QMAKE_FRAMEWORKPATH_FLAGS
Oswald Buddenhagen [Fri, 13 Jul 2012 11:16:39 +0000 (13:16 +0200)]
dispose of QMAKE_LIBDIR_FLAGS, "demote" QMAKE_FRAMEWORKPATH_FLAGS

merge their content as early as possible into QMAKE_LIBS. that's where
they ultimately end up anyway, and this approach is way simpler.
QMAKE_FRAMEWORKPATH_FLAGS is also used for the compiler flags, so it
remains as such in this second function.

Change-Id: Idc3ba4a9b2569fce3252d5f5ddc3f6ebf93650cf
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agoremove duplicate -L flags in findLibraries()
Oswald Buddenhagen [Fri, 13 Jul 2012 11:09:30 +0000 (13:09 +0200)]
remove duplicate -L flags in findLibraries()

unlike duplicated -l options, duplicated -L options are really redundant.

Change-Id: I4fa42d8ee2f10c6e6263457d9bb97291e13685ef
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agounify handling of non-standard linker -L flags
Oswald Buddenhagen [Fri, 13 Jul 2012 11:08:26 +0000 (13:08 +0200)]
unify handling of non-standard linker -L flags

findLibraries() now consistently expands to the linker-specific syntax,
and processPrlFiles() expects that syntax.

Change-Id: Ifd7b51d01378c91d6f2b132aca33629f21ca72f9
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
11 years agomake sure that the stale header removal works the first time around
Oswald Buddenhagen [Tue, 17 Jul 2012 13:39:54 +0000 (15:39 +0200)]
make sure that the stale header removal works the first time around

sort backwards by ascii, so that the lowercase headers come first -
otherwise the camelcase headers would survive the first round, as they
include the lowercase forwarding headers which were not deleted yet.

Change-Id: Iece2788206d31a11428f38da8d8f9f4d0b6dc992
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>