Proper multipoint touch support was only introduced in XInput2.1, but Harmattan uses
a tweaked version of XInput2.0 that transfers touch data through mouse events.
This patch applies on the xcb plugin a subset of the changes that were applied on the
Qt 4.7 that was shipped to Harmattan to get similar multipoint touch support.
Change-Id: Ifda7ad40de29d7ded1443d4f78b3ec3807303a9f
Reviewed-by: Simon Hausmann <simon.hausmann@nokia.com>
lastPos is copied from the last point's position, and the position has
to be resolved according the the receiving item's transformation.
However, if a QTouchEvent contains only stationary points, it won't be
delivered, it's pos()/rect() will still be uninitialized, and the
touch point of the next event will be delivered with the uninitialized
pos() as it's lastPos().
This patch makes sure that the lastPos() won't be filled with a
previous incomplete stationary point state even if the platform is
sending us stationary-point-only touch events.
Change-Id: Ia7d10423c8fbe78348edbb0a89fbfa66d1b8b5d1
Reviewed-by: Denis Dzyubenko <denis.dzyubenko@nokia.com>
Before commit 070d9c00c4, the raster paint engine used
to call loadGlyphs() on the FT engine, which ensures that the glyphs are loaded
and (more importantly) that the correct transformation matrix is applied on the
freetype face. After commit 070d9c0 lockedAlphaMapForGlyph is called for each glyph,
which unfortunately doesn't respect the provided transformation matrix. Therefore
when drawing scaled text, it was never actually scaled.
This patch applies the combination of the fontengine transform and the requested
transform on the freetype face actually used, similarly to loadGlyphs.
Change-Id: I0956a9e71784582db6bb90475a001a63800773f4
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>
qdoc does not understand the void (*) syntax, there is a typedef to
QFunctionPointer that does work.
Change-Id: Idbe9d43d00f8676304d088d72795b6ddb7e4ee72
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
There is no way for a .pro file to know if exceptions are supported in
the Qt build, so these tests cannot be omitted from the build if
exceptions are not supported. Because of this limitation, these tests
were using QTEST_NOOP_MAIN when exceptions were not available, making
the tests appear to pass. This commit changes QTEST_NOOP_MAIN to QSKIP,
so that the user receives a clear indication that the test isn't testing
anything when exceptions are not available.
Change-Id: I2facf08ba69e6f985e2da50dd36ad2658246b22e
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
There was no value returned previously when none of GLX, EGL, DRI was
enabled. Now we show a warning message and return null (resulting in a
crash). Previously not returning a value resulted in obscure crashes
with strange-looking call stacks, now it will be clear what the
problem is.
This is particularly important because not having libx11-xcb-dev
installed results in silently disabled xlib support which in turn
disallows the usage of glx too. The result is an obscure crash in any
GL (e.g. QML2) app. Now there will be a proper warning message at
least.
Change-Id: I15cf3a6c16ca373d4ada7abb6d3c8f7e30c1e03d
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
QXcbWindow::updateNetWmUserTime() creates a new window using the
actual window as its parent. This operation was failing when Xlib
usage and GLX (or EGL) was enabled because in these cases the real
window was created with a different visual using XCreateWindow.
For the user time window however we were passing the xcb_screen's
root_visual always, and this cannot work when Xlib+GLX/EGL is in use,
resulting in a BAD_MATCH (followed by some annoying flicker on the
screen on Ubuntu at least, or a full log off (X crash?) on Fedora).
The patch saves the visual id used to create the actual window and
passes that when creating the user time window. This way the creation
of the user time window succeeds with Xlib+GLX too.
Change-Id: If3925d9ac1670d385dcc7c2b475c196908569f06
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
The QCopChannel class was removed during Qt5 refactoring.
Change-Id: I62914eca1db65c680d90233ce9dad5d891890c4a
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
The qgraphicswidget and qgraphicsproxywidget tests are empty if Qt is
built without the cleanlooks style, so don't build the tests in that
case.
Change-Id: I2308e723a9b0abcc3e95b32a562fcb02afe0444d
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
The test appears to be applicable only if building with g++, and even
then the test is broken.
Change-Id: I9eefb13325a10ec295a59c2bd1111e3c324ff94f
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
This test was empty, the logic having been recently removed because it
only applied to the Carbon UI, which will not be supported by Qt5.
Change-Id: I2b5d6aba6be4a51eadbac3ca9d19d1a68c4fc1ab
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
This test is only valid with obsolete Qt Solutions packages. Any Qt
Solutions that are still of value need to be rewritten as Qt5 addons and
will be delivered with their own unit tests.
Change-Id: I099ef816655cc7de13715f5a4e5f4a06099df2d5
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Make the parent .pro files follow the pattern of listing all tests in
one place and then removing inapplicable tests as needed.
Change-Id: Ide1ae98d50dd63cd8c1b4d6fd53135cdff801bc3
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
This test is only useful on Windows, so avoid building it elsewhere
rather than building an empty test.
Change-Id: I01d91266bcb2dd242e02e70741ad5b81a13c33f0
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
The tools tests are not built for cross-compiled Qt builds, so the logic
to make those tests empty when cross-compiled is redundant.
Change-Id: Ia3a3a680d7ade2cee78aa633b41c9ef5be6f2f15
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
Rather than building the test with all the logic #ifdef'd out, just
avoid building the test at all.
Change-Id: I394f48b2dacb40325d03de719096da36fcb7312a
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
All but one of these tests was already excluded from the build on
non-Mac platforms, but still had #ifdef's to make the test empty if
compiled on a non-Mac platform.
Change-Id: I1a3677c1cdb7bd467cf4317ed4cfbac89e934e76
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
If the glyph run is generated based on characters which split a
ligature glyph, it needs to include the ligature glyph in the run,
and when painting, we need to clip the painter to the reported
bounding rect of the run in order to avoid painting too much of
the ligature. To make it possible to reduce clipping to a
minimum, we need a flag to inform of whether the glyph run contains
a split ligature or not.
Change-Id: Id787e0bec6d6e8e06cc818700ae0fca22cf81e98
Reviewed-by: Jiang Jiang <jiang.jiang@nokia.com>
It's not a good idea to look at the host's /bin/ls when cross-compiling
Change-Id: I1fe67b17abd92f01c1d89146227c5419eaa3242f
Reviewed-by: Arvid Picciani <arvid.picciani@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
When the platform plugin does not reimplement QPlatformWindow::winId()
to return a non-zero value, the default implementation provides a WId
of 0. This breaks various scenarios with QWidgets which, for example
in nativeParentWidget(), rely on internalWinId() being non-zero for
TLWs.
As a result, apps like collidingmice will crash with the 'minimal'
platform plugin when they try to do markDirtyOnScreen for the graphics
view's viewport widget because viewport()->nativeParentWidget()
unexpectedly returns null, even though it should return the
QGraphicsView.
The patch makes the default value WId(1) in order to to avoid this.
Change-Id: Iafef2e510e32b5610b8cef742313e9119ed64963
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Without a call to QWindowSystemInterface::handleWindowActivated() or
QPlatformWindow::requestWindowActivated() QGuiApplication's
activeWindow() and focusWindow() will return 0 all the time, and this
breaks key event delivery when key events are received not from a
windowing system but directly from e.g. evdev. In such cases the only
choice is to send the event to the active window but if there isn't
one the event will be lost.
These platforms will have one fullscreen window at a time so marking
it the active one should be a safe assumption.
Change-Id: Ia6f0e9f021fbd61bbd368ddc46f8a13fccda1c2f
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
This test verifies that all gui types with built-in QMetaType
support can be created, either using the default constructor
or the copy constructor.
Change-Id: Ibb1c5aab8571b598638c74112471d6869516a202
Reviewed-on: http://codereview.qt-project.org/6344
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
It only tested creating a copy of QSize.
This change adds testing of all core types, both
copy and non-copy creation.
Change-Id: If2b8fb45718c4a976dd8b883b332878f7fb0da6f
Reviewed-on: http://codereview.qt-project.org/6343
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
This is the first towards bringing shortcut support
back into QtGui.
Change-Id: Ic2cf376d08c326402fb90a0cf170a987b07ac304
Reviewed-on: http://codereview.qt-project.org/6577
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Enrich playWithVaraint template by a macro that returns correct line
number in case of failure.
Change-Id: I7684169852481a6905b71e4420135e7e3f298b12
Reviewed-on: http://codereview.qt-project.org/6570
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Instead of creating new instances of the class whenever
accessibility() is called, create one on startup.
Change-Id: Idd5b06ee935d7acf4934b98882d254a1ef0ca04e
Reviewed-on: http://codereview.qt-project.org/6435
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Reviewed-by: Jan-Arve Sæther <jan-arve.saether@nokia.com>
swapBuffers would wait until DRI2_InvalidateBuffers was received from the
X server, but would never receive it since we are not running the xlib event loop.
This patch calls the XESetWireToEvent xlib callback of an unhandled event from the
xcb event loop directly, if one is available for its response type.
Change-Id: I3b4ef3f6e3efbae25f49f161e229e9b15e951778
Reviewed-on: http://codereview.qt-project.org/6557
Reviewed-by: Samuel Rødal <samuel.rodal@nokia.com>
Instead of ignoring the touch major/minor and returning a rectangle
with width and height of 1, the size of the area is now calculated
based on the major. This is incomplete and far from perfect but at
least it makes visualizers, like the fingerpaint example app, work out
of the box.
Change-Id: I3bb70b4d7247f289e5cae4e7de2460ecf2c9d009
Reviewed-on: http://codereview.qt-project.org/6566
Reviewed-by: Paul Olav Tvete <paul.tvete@nokia.com>
When configure detects the presence of font-config/freetype via pkg-config
_and_ we are cross-compiling for Harmattan, make sure that we pick up the
include path for freetype. The .pri file is included in various places in
qtbase when the usage of freetype is required.
Change-Id: I4a5390dfc063504def80a891bc9153d6a376ccd7
Reviewed-on: http://codereview.qt-project.org/6509
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
This plugin was still using QApplication and QDesktopWidget. The code
is now migrated to QGuiApplication, QWindow, and QScreen.
Change-Id: I22a7e15982efea16cbf560d09c8831cd82ed29d0
Reviewed-on: http://codereview.qt-project.org/6528
Reviewed-by: Laszlo Agocs <laszlo.p.agocs@nokia.com>
startHostInfoLookup will try to detect if IPv4 or IPv6 will be used
when connecting to the host. If a proxy is set we should lookup
the proxy hostname instead, in case host name can't be resolved
via DNS and should be resolved by the proxy.
Task-number: QTBUG-21889
Change-Id: I2012798784fc40f153469a1298e261c52981297e
Reviewed-on: http://codereview.qt-project.org/6447
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The reference file has suffered whitespace changes during the cherry
pick from 4.8. Refetched from the test server.
Change-Id: I03a0b7a66cf6743de3391905422b2f4cdcd86dcf
Reviewed-on: http://codereview.qt-project.org/6454
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Using the first pkg-config in PATH is prone to errors when
cross compiling. This is a known issue, and `configure' already
avoids using pkg-config for cross compiling (unless -force-pkg-config is
used). However, there was previously no way for a .pro file to test
whether or not the user had passed -force-pkg-config.
Let !contains(QT_CONFIG, no-pkg-config) be used to test whether or not it
is safe to run pkg-config.
(cherry picked from commit 53794d94994cb3149581b1a91762dfa2d6defac0)
Change-Id: I53794d94994cb3149581b1a91762dfa2d6defac0
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Reviewed-on: http://codereview.qt-project.org/6488
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
It is better to avoid building a test when we know that it is not
applicable, rather than building an empty test that appears to pass.
Note that some of the tests were already excluded by
tests/auto/other.pro, but still had the preprocessor directives to make
an empty test.
Change-Id: I71b75c37a05f6e20dbef43d82f72b29b35ef496a
Reviewed-on: http://codereview.qt-project.org/6410
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
If Qt is built without QtConcurrent, the relevant autotests should be
excluded from the build, rather than building empty tests that appear
to pass and thus produce confusing test results.
Change-Id: I43bafcb0ffa42e8981bec4e27bf6a23a28d86df9
Reviewed-on: http://codereview.qt-project.org/6395
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
The platformsocketengine autotest uses the native socket engine directly
rather than through QAbstractSocket. The bind tests were failing because
the autotest was creating a socket with IPv4 (AF_INET) and then binding
with QHostAddress::Any (AF_INET6).
A linux kernel update caused this to start failing on the test machines.
Change-Id: Iea62f3d56dbfb35fcb952dcf00313578eb2bd764
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Previously we were generating include lines based on the relative path of
the source file, which resulted in the .d files being placed in the source
dir as well. We now expect the .d files to live in the output dir, but keep
the dependency from the .d file to the original source file.
Before:
.deps/%.d: %.cpp
-include .deps/../../src/foo.d
After:
.deps/%.d: ../../src/%.cpp
-include .deps/foo.d
Change-Id: I749adeb671cf8424f0849521c5bb1489eb3e76d5
Reviewed-on: http://codereview.qt-project.org/6455
Sanity-Review: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
When two threads construct a QNetworkAccessManager at exactly the
same time on an SMP system, there are construction races for some
Q_GLOBAL_STATIC data. This is normal and expected - the losing
thread deletes its instance as part of the Q_GLOBAL_STATIC macro.
For QNetworkAccessBackendFactoryData, a guard mechanism intended
to prevent the data being reconstructed by destructors of other
global static classes was being set by the loser.
To fix this, the bool is changed to a QAtomicInt. In the normal
case, it will have value 0->1 on startup and 1->0 on shutdown.
In the race case, it will have values 0->1->2->1 on startup and
1->0 on shutdown.
Task-Number: QTBUG-20343
Change-Id: Ie3fe38944d10809d1ccdbe772df82d67faffe19c
Reviewed-on: http://codereview.qt-project.org/6181
Sanity-Review: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Now QInputPanel::visible() can be set true even when platform doesn't provide
a virtual keyboard. Like keyboard geometry, visibility should be dictated by
the platform plugin and not QInputPanel, whose role is more like that of a mediator.
QInputPanel::show() and ::hide() calls should be treated as requests that may fail.
Changed the QInputPanel's visible property to read-only as a setter that may
fail is not really a setter, show() and hide() should be used instead.
Enabling the new functionality cannot be activated immediatelly without breaking
existing keyboards, added a temporary function handlesInputPanelVisibility that
handovers the responsiblity of updating input panel visibility to QInputContextPlatform
only once QInputContextPlatform says that it is able to handle it.
Change-Id: Ideecaf7225cc3971f33a0ac976bd92cf7767475b
Reviewed-on: http://codereview.qt-project.org/6429
Sanity-Review: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Pekka Vuorela <pekka.ta.vuorela@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
