Conflicts:
src/widgets/kernel/qwidget.cpp
This merge also extends the expected output of the pairdiagnostics
teamcity output (added in dev in commit
c608ffc56a) after the recent addition of
the flowId attribute to the teamcity output (commit
8f03656211 in 5.9).
Change-Id: I3868166e5efc45538544fffd14d8aba438f9173c
The Unix stat fields "st_ctime" and "st_ctim" mean "change time", the
last time that the file/inode status fields were changed. It does not
mean "creation time". So this commit splits all of the internal API to
"birth" and "metadata change" instead of "creation" to avoid the
conflict.
Change-Id: I149e0540c00745fe8119fffd1463fe78b619649e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Code was lagging a bit behind published standards.
Change-Id: I0fe07c0162a54d9eed8b539887496b9313b37aa4
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reviewed-by: Jesus Fernandez <Jesus.Fernandez@qt.io>
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
The call to QPlatformWindow::setRect (for storing into d->rect) was
in the wrong place. It has to be the potentially-overridden value.
Amends 3a31c70879
Task-number: QTBUG-57608
Change-Id: Id3c35e2dc178f7bd2f9643e8ae4754c8f2f39240
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
The conversion fallback did not catch all the more recently added
QImage formats, leading to corrupt pixmaps or crashes. Also alpha
channels were needlessly thrown away if present.
Change-Id: I38588035aa9bf37b77398489981df65201cf0340
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Avoid polluting the output by default. Unfortunately there are various
systems (modern Fedora with XWayland, VMWare) where parsing will fail
for some not yet known reason. Showing warnings on each and every Qt
application startup is not desirable, especially since not having the
EDID data available is far from being critical.
Change-Id: Ibaca7db4d897e705819f7c359f8de41b9be862a6
Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
On X11, QScreen::availableGeometry() is broken with multi-head systems,
and there doesn't seem to be a real fix for this due to limitation in
the protocol and therefore support in WMs (more details in the
patch). In Gnome this issue is more visible because on this DE the
_NET_WORKAREA rectangle represents the intersection of the available
geometries on all monitors. This results in a big area of "dead space"
on the secondary screen, when primary screen is positioned lower in the
virtual space. If menu is opened by clicking in this dead space, the menu
is awfully misplaced (qmenu uses availableGeometry() to calculate the
position of menu).
On Ubuntu with Unity (same is true for KDE Neon+Kwin and LUbuntu+Openbox),
_NET_WORKAREA returns a bounding rectangle containing all monitors.
Which does not cause the menu misplacement as "dead space" is outside
clickable area. But this does not mean that the QScreen::availableGeometry()
reported values are correct. With the same setup as described above,
QScreen::availableGeometry() thinks that we have a tool panel on the
right screen, when in reality it is on the left screen.
AwesomeWM for example does not set _NET_WORKAREA at all, which means
QScreen::availableGeometry() == QScreen::geometry(). I am not aware that it
would cause any issues for popup/menu window positioning in Qt (Qt positions
these windows manually by bypassing WM (via Qt::BypassWindowManagerHint) and
using availableGeometry for calculations. With this patch, we would take the
same code path as if _NET_WORKAREA was not set (where we know that_NET_WORKAREA
is cleary wrong). The solution here is to recognize _NET_WORKAREA as true
available geometry only in specific cases (cases where the meaning is cleary
defined by the specification) and adjust the documentation accordingly.
Not knowing the true available geometry on X11 is mitigated by WMs. Window
manager can position windows as it wants. WMs are smart enough not to place
windows on top of reserved areas at edges (even if user has explicitly requested
this via setGeometry based on inaccurate information from availableGeometry()).
[ChangeLog][Platform Specific Changes][Linux] The _NET_WORKAREA atom is
used for calculating QScreen::availableGeometry() only on systems with one
monitor. In all other cases QScreen::availableGeometry() is equal to
QScreen::geometry(). To restore the legacy behavior with untrustworthy
values in QScreen::availableGeometry() set QT_RELY_ON_NET_WORKAREA_ATOM=1
environment variable.
Task-number: QTBUG-60513
Task-number: QTBUG-29278
Task-number: QTBUG-43768
Task-number: QTBUG-18380
Change-Id: I7e0f62f81d1444991b8a6c007c2527d8f96088c2
Reviewed-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
When requesting windows with a size of 0x0 for which a default size
is determined by QPlatformWindow::initialGeometry(),
QWindowsWindow::initialize() did not call handleGeometry() since
it compared against the requested geometry obtained after
calling QPlatformWindow::initialGeometry().
Store the initial geometry in the context.
Amends deb7f9a7c3.
Task-number: QTBUG-62177
Task-number: QTBUG-61977
Change-Id: I9e96f2f0b984b9009bebb192f576c92b4409d5d1
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
We're adding a lot of unnecessary files that end up later as cargo-cult,
for at most a handful of lines. So instead move the testcases directly
into the .json file.
The following sources were not inlined, because multiple tests share
them, and the inlining infra does not support that (yet):
- avx512
- openssl
- gnu-libiconv/sun-libiconv (there is also a command line option to
select the exact variant, which makes it hard/impossible to properly
coalesce the library sources)
The following sources were not inlined because of "complications":
- verifyspec contains a lengthy function in the project file
- stl contains lots of code in the source file
- xlocalescanprint includes a private header from the source tree via a
relative path, which we can't do, as the test's physical location is
variable.
- corewlan uses objective c++, which the inline system doesn't support
reduce_relocs and reduce_exports now create libraries with main(), which
is weird enough, but doesn't hurt.
Done-with: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Change-Id: Ic3a088f9f08a4fd7ae91fffd14ce8a262021cca0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Commit 9e64fc9e1c caused a regression
which stored all QDateTime entries as if they were in localtime,
which causes them to be offset by the amount of local timezone
offset. This is fixed by adding "Z" if the time should be in UTC or
using "+/-hh:mm" if it should use fixed UTC offset or specific
timezone.
Task-number: QTBUG-57138
Change-Id: Ie60905dfb3a517db442b636ca41daf8348753d84
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
If ibus has been set, but either isn't installed or doesn't launch, all
composition fails on xcb. This ensures we fallback to "compose" which
is needed for working composition on XCB.
Change-Id: Ic2061c330bcb907759a13920c3eede14f3036563
Reviewed-by: René J.V. Bertin <rjvbertin@gmail.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
MinGW 7.1 claims the qUuidOf<> specializations are unused:
In file included from accessible\qwindowsaccessibility.cpp:59:0:
accessible\iaccessible2.h:274:5: error: 'IID qUuidOf() [with DesiredInterface = IAccessible2]' defined but not used [-Werror=unused-function]
IID qUuidOf<IAccessible2>() { return IID_IAccessible2; }
^~~~~~~~~~~~~~~~~~~~~
In file included from accessible\qwindowsaccessibility.cpp:59:0:
accessible\iaccessible2.h:58:5: error: 'IID qUuidOf() [with DesiredInterface = IAccessibleComponent]' defined but not used [-Werror=unused-function]
IID qUuidOf<IAccessibleComponent>() { return IID_IAccessibleComponent; }
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Amends 4f599c5ea1.
Change-Id: Iafcfe8b8d8ca9cdf00663f39ad491347b51d3ace
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
Instead of forwarding the flush to the view, using CoreGraphics to blit
the backing store to the window, we do everything in flush(), and use
higher level AppKit APIs to do the blit.
This simplifies the flow and code quite a bit, and also supports blitting
of individual regions in a flush instead of the whole bounding rect.
Change-Id: I2173c1a7763fe652a94125c7e3ae93a655412cd3
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When running on a desktop machine (using Windws SDK 10.0.14393) the
palette had fully transparent text colors which were assigned in the
"phone only" part of the theme's initialization. By checking the API
contract we can avoid that part of the initialization and thus return
proper values.
Change-Id: Id770a686c1c7e447a9594830fd7670352116eb21
Reviewed-by: Maurice Kalinowski <maurice.kalinowski@qt.io>
Change-Id: I914521e1dfecb0157a8b9e1c9d9a86ca45e0826e
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Explicitly declare which DRM event context version we want to use,
rather than just the latest one libdrm supports. New versions may change
semantics, or extend the structure, in ways we're unaware of.
Stick with version 2, which is the version that introduced
page_flip_handler.
Change-Id: I1d2066d5ab485ea571f016a8660829f435821c82
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
This changes the drawing model on macOS from the following:
1. Sending synchronous expose events directly from callbacks such as
windowDidOrderOnScreen and windowDidChangeOcclusionState
2. Waiting for a resulting flush of the backing store, and issuing
setNeedsDisplay as a response
3. Waiting for the asynchronous drawRect call in response to
setNeedsDisplay, where the backing store is finally drawn
to the window
To the following:
1. Issue setNeedsDisplay as a response to callbacks such as
windowDidOrderOnScreen and windowDidChangeOcclusionState,
when needed (in many cases this is automatic by AppKit)
2. Send synchronous expose events from the resulting drawRect
callback
3. Draw the backing store to the window when flushed
The new model matches how normal Cocoa application draw in response to
drawRect, and makes the backing store flush synchronous instead of having
to trigger a async setNeedsDisplay. This gives AppKit more information
about how much time we're spending in drawRect, as the actual drawing
and flushing all happens within the synchronous expose event.
Qt applications that draw outside of drawRect, e.g. in response to timers,
are still supported by manually locking focus of the view and flushing the
window at the end of the backingstore flush.
Task-number: QTBUG-50414
Change-Id: I2efb9ff8df51ab6e840ad20c497b71f53e21e1c2
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
CID 10994 (#1 of 1): Resource leak in object (CTOR_DTOR_LEAK)
The font database was not destroyed in the destructor.
Coverity-Id: 10994
Change-Id: Ibd4cb0b65d1e554593295a3654e8d8c946551cc2
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reviewed-by: Jesus Fernandez <Jesus.Fernandez@qt.io>
MinGW 5.3 is able to compile the IAccessible2 classes. All that is needed
is some way to provide the missing IIDs, which is done via specialization
of a function template (similar to the MinGW's __uuidof operator).
[ChangeLog][Windows][Accessibility] MinGW builds now support IAccessible2.
Change-Id: I218a4b89c81b54aa96f7c743544388631ca9a53e
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
... for better localization. This is not possible in C, but in
C++ a struct has the same features as a class. Thus, use default
member initialization feature (available since C++11).
Change-Id: I91522c3a4fe4270c37bb6e85fddeb5970a847352
Reviewed-by: Alexander Volkov <a.volkov@rusbitech.ru>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Some customers still need this for interoperability with legacy code.
Let's continue to keep it working in 5.9.x, and then move to
64-bit exclusive features (thus dropping 32-bit entirely) in 5.10.
Task-number: QTBUG-58401
Change-Id: Ibb7200c1885e9caba70439df5f7c86c81b1312b5
Reviewed-by: Gabriel de Dietrich <gabriel.dedietrich@qt.io>
Skip EGL surface re-creation when a sequence of unprocessed resizing
requests leaves the buffer size unchanged. In this situation, the
buffers won't be resized. Recreating the surface without resizing the
buffers leads to screen providing incorrect information about the
buffers.
Change-Id: I1f75ab99eb1dffe0bcf9660bf014f047407b0c1b
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
.. in an application that has many native windows.
We don't need to select XI_HierarchyChangedMask and XI_DeviceChangedMask
for every window. It is enough to register one window for these events to
update state of our X11 client (application).
Furthermore, XIAllDevices and XIAllMasterDevices always apply, even if the
device has been added after the client has selected for events. So there is
no need to call XISelectEvents on XIAllDevices/XIAllMasterDevices again, if
we are not updating event masks.
With this patch and the test application from QTBUG-57013, removing/attaching
a device takes few hundred milliseconds instead of 23-24 seconds.
Task-number: QTBUG-57013
Change-Id: Ieb0b5ee25feef2922f901165825cb4a1289fc852
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Use QPlatformWindow::mapFromGlobal() instead of QWindow:::mapFromGlobal()
as QPA operates in device pixels.
Task-number: QTBUG-62028
Change-Id: I64ec4f4c9a536e122676d738db58805b98a45c82
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
GetDefaultPrinter can return ERROR_FILE_NOT_FOUND. In that case just
return a empty string.
Task-number: QTBUG-53290
Change-Id: Ib0e28b3425d100bc185d1d8827dc6b7fd90cbbeb
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Do not call if the property indicating embedded windows is set.
Task-number: QTBUG-61972
Change-Id: I8f34dd8a59f1e5c9c8064646bcb15acea115cd68
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Do not scale the QPlatformScreen's geometry.
Fixes tst_QWidget::showFullScreen() when run with a scale factor.
Change-Id: I4a2e743303ff70b01fd3d2086281a790737d2c1d
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Change-Id: Ic3555445b045edda884983aa01834a4ae243d6fa
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Move the code that sends geometry change events from
QWindowsIntegration::createPlatformWindow() to
QWindowsWindow::initialize(), using the obtained geometry
from the creation context. Drop the check for window flags
since they are not changed.
Complements change 4c855a9f9f
Task-number: QTBUG-61977
Change-Id: I0c23abefc45110cc4bf11e10d65dc7ddbb9d20d5
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
... and group XI22 methods together under one ifdef clause.
The error message was:
qxcbconnection_xi2.cpp:1025:42: error: no ‘bool QXcbConnection::isTouchScreen(int)’ member function declared in class ‘QXcbConnection’
bool QXcbConnection::isTouchScreen(int id)
XInput 2.2 was released many years ago, this build failure could
in practice happen only on some really old platform.
Change-Id: I3c1741cbdffe15c0f5149c6d76592a743d1d8a91
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Revert SHA1 - b1708efeeb as it causes a
problem with QtMultimedia.
Change-Id: I0ba366fa6ddccff3715917f5f455b20c73c2e49e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Sometimes AppKit will pass in a proposed size that's smaller than the
geometry of the screen. We don't know why, but shouldn't assert.
Change-Id: I9970c5f587e1e0fb3f2fa932de5a32ac4e1eb76d
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The QWindow::create method calls createPlatformWindow, and assigns the
result to d->platformWindow. If the platform sends any sort of events
synchronously during the creation, the event will be delivered to a
QWindow that doesn't have a handle() yet, resulting in noop handling
of the event, or crashes.
To mitigate this situations, platforms should do as little a possible
in the QPlatformWindow constructor, and leave initialization to the
new method, where the QWindow will have a handle().
The macOS platform plugin still has a m_initialized guard, to prevent
sending geometry changes during initialization, as this will result
in a resize event before a show event. This forced behavior seems
dubious, but is left for a followup patch.
Task-number: QTBUG-61977
Change-Id: I04d32d93391e89d068752b719270438e7024ad46
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
.. thus making it purely into a _runtime_ check of XInput 2 version.
If Qt was build with -no-xinput2, it does not make sense to compile
in code that always returns "false". Simply exclude code that is not
relevant for -no-xinput2 builds with XCB_USE_XINPUT2 ifdefs. In
addition, this improves readability of the code.
Now, trying to use ::isAtLeastXI21() in a -no-xinput2 build will
result in the following build error:
error: ‘class QXcbConnection’ has no member named ‘isAtLeastXI21’
Change-Id: If242510d43d71829b327edc1f76322f3a0db0e08
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
There is no need to check for it explicitly.
Change-Id: I66958bf1ff4539ee75fec635c96f056524da8ddb
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
By printing the corresponding QPlatformWindow and QWindow for a given
QNSView we make it easier to track issues regardless of which of the
views/windows are being logged.
Change-Id: I4a42eff7f87cf3c8e722cd6ad8baccd4eeab8eb3
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Things that were broken:
- The return values of XIQueryVersion interpreted wrongly. BadRequest
means no XI2 support on the server. Nesting more XIQueryVersion
calls to check for lower minor version of XI2 in this case does not
make sense. On Success, server's supported X Input version is returned.
Server's supported version can be lower than the version we have announced
to support. In this case Qt client will be limited by X server's supported
version (which is ok, as we do check the available version at runtime via
QXcbConnection::isAtLeastXI2*).
- The code was _always_ announcing to X server that we support XI 2.2,
by ignoring what actually is supported in the specific build (see
XCB_USE_XINPUT ifdefs).
- qCDebug messages and logging categories were wrong too.
Change-Id: Ia84457f125474aa851b7a91ed19fc5b904ac359e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
This includes:
- Better sharing of common code.
- Less argument passing.
- Improved code readability:
It is somewhat confusing to read when QXcbWindow::handleEnterNotifyEvent
calls connection()->handleEnterEvent().
- Better organizing of ifdefs.
Change-Id: I4405390cada13b51db78f9fd30e26bb1793395ff
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
... as XI2 reuses device ids and we might end up with some bogus
objects in m_touchDevices (the same touch device after reconnecting
might have a new device id).
With this patch, m_touchDevices is handled similarly to tablets
and scolling devices. This patch also removes unnecessary calls
to X (XIQueryDevice). We already have device information available
for all devices when populating m_touchDevices list, so calling
XIQueryDevice again for every device id (which we haven't already
concluded to be a tablet device) is a redundant round trip to X.
For consistency with other device types, extend touch device data
structure to store the necessary device details, instead of holding
pointers to XIDeviceInfo. And allocate touch device objects on
stack instead of heap, also for the consistency reasons (this way
we won't need to special case m_touchDevices when running some
management task on _all_ devices, e.g. removing a device with id=n
when it was disconnected).
Change-Id: I9f6edfaf24da911dd21ec45cd48dfdd6d0dcdf8b
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Change-Id: I419f884f4145dbe2b60751bf6cde3968cf34fe4a
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Thiago Macieira
Giuseppe D'Angelo
Laszlo Agocs
Eirik Aavitsland
Gatis Paeglis
Frederik Gladhorn
Friedemann Kleint
Oswald Buddenhagen
Kwangsub Kim
Joni Poikelin
Allan Sandfeld Jensen
Tor Arne Vestbø
Oliver Wolff
Jonathan Liu
Jesus Fernandez
Jake Petroules
James McDonnell
Daniel Teske
Simon Hausmann
Andy Shaw
Stephan Binner
Gabriel de Dietrich