There is no need to run the catch-all DPR update if
the window is not visible.
Pick-to: 6.6
Fixes: QTBUG-117762
Change-Id: Ib4e235fed4b21e5aa2ecafa960ab0900e2b68295
Reviewed-by: David Edmundson <davidedmundson@kde.org>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Add missing include of global module header.
Drop unneeded global namespace qualifier for constant in
our namespace.
Amends 6017695bfa
Change-Id: I04633030af13df3c3359fc09b08b4a5f031013a9
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When a QWindow is parented into another window, either via its constructor,
or via setParent(), we also update the QObject parent of the window. This
in turn sends ChildAdded and ChildRemoved events to the new and old parent
of the window, as well as ParentAboutToChange and ParentChange to the window
itself.
But at the point when these events are sent, the QWindow parent relationship
has not been fully established. We have not updated d->parentWindow, nor
have we propagated the parent relationship to the platform window.
This is problematic because clients can not use these events to react to
window parent changes synchronously. For example. trying to raise a child
window when added to a parent is not going to work, because at that point
the child window isn't a native child of the parent.
Instead of re-using the QObject events for QWindow, like QWidget does,
by delaying the events or sending them manually at a point when the
window parent relationship has been fully formed, we opt to add new
dedicated events.
Change-Id: I019c14eba444861f537e4f68ab3a82297f843cf7
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
QGuiApplicationPrivate::processMouseEvent() requires the following
conditions to be met in order to trigger a double click event:
1. The same button should be pressed (kept in mousePressButton member)
2. The time between two press events should not exceed doubleClickInterval
3. Movement between presses shouldn't be bigger than doubleClickDistance
Unfortunately, when we are processing *synthesized* mouse events (from
touch), in order to check criterion 2) we cannot rely on
QPointingDevicePrivate::pointById(0)->eventPoint.pressTimestamp();
because this was already updated in QGuiApplication::processTouchEvent(),
so it does not any longer reflect the *previous* press timestamp, but
the current one.
Therefore we could not reliably calculate the time between the previous
press and the current press (in fact, it never exceeded the
doubleClickInterval condition)
I suspect this occurs with all touchscreens, but at least it was
reproduced with e.g. a NVIDIA Terra / Thinkvision touch screen on
Windows 11 and Dell Express SVC touch screen on Windows 11.
=> Since QGuiApplication is a singleton, add a static local variable to
remember the time stamp of the last processed press event.
Autotesting has been modelled in PS7. However, QTest::touchEvent is
unable to synthesize hardware specific detection latency and stray
touch events, which actually reproduce the issue.
=> Not adding an autotest.
Fixes: QTBUG-112479
Fixes: QTBUG-119032
Fixes: QTBUG-100688
Pick-to: 6.6 6.5
Change-Id: Idc97fb9d49ed334f56d702d420451fc062e9b5a9
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
More logical than throwing it in the main rhi backend object.
In addition, holding on to the generated data from the last
setShaderResources() per command buffer has great value if/when
more granular resource updating is introduced at some point.
Change-Id: I26887423cf66b27995ec92ae9965d99a40267066
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Kristoffer Skau <kristoffer.skau@qt.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
CBV and SRV descriptor ranges default to
DATA_STATIC_WHILE_SET_AT_EXECUTE with RS 1.1.
Due to how constant buffers (that back
Dynamic+UniformBuffer QRhiBuffers) are implemented,
DATA_STATIC is more appropriate, leaving more optimization
opportunities. The buffer contents will not change once
setShaderResources() is called (which is what issues
the SetGraphicsRootConstantBufferView()). In fact
the buffer contents should not be changing anymore during
the whole render pass recording (begin..endPass).
Change-Id: I5901a6d9608fb7c61f04f9d6a3a9acebce37a673
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Nothing is relying on the binding list being sorted based on the
binding number. In many ways the D3D12 backend is very similar
to the OpenGL one: the srb merely holds some flags derived from
the binding list and that's it really, no native resources or
other content are owned directly by the binding set objects.
Change-Id: I35d652c9738308bff691095e43aea9706031dd90
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Going through the UIImage.CGImage flattens the image to monochrome.
Change-Id: If74c33badb4e94253b3bf21b6d69bb9f521d6bff
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
If the attribute does not follow the required
"name:value" syntax, ignore it, rather than crashing
if it doesn't contain any colon.
Same issue as spotted by Jan Arve Sæther during the review
of a QTBUG-118106 related change that would have introduced
the same issue in UIA code.
Pick-to: 6.6 6.5
Change-Id: Id391502ed7aec7f09ef2826a456f2e4737af045e
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
Follow-up change from enabling DPI awareness, which
caused some style elements (for instance check boxes)
to be rendered incorrectly on non-primary displays,
when there is a difference in DPI between displays.
Use two approaches to get system metrics and themes:
* Use forDpi() API variants and query at 96 DPI for style
metrics, that are in device independent pixels. These are
metrics which are used for layout calculations.
* Get theme metrics at the target display DPI, and scale
to device independent pixels when needed. This is used
for OpenThemeData(), since this theme is used for drawing
as well and needs to be in device pixels.
One approach is not used any more:
* Get metrics for the main display, and scale by the ratio
between the main and target display.
Change the theme cache to cache themes per window handle (HWND).
This is required since OpenThemeData() returns theme data for
a specific DPI, which means we can no longer use a shared
cache.
Clear the cache on theme change, DPI change, and when
the window is destroyed. This handles cache invalidation
when the window is moved to a different screen, and also
when the DPI for a screen is changed.
Move the cache implementation to QWindowsStyleSupport
in QtGui, where it can be accessed by both the style and
windows platform plugins.
Task-number: QTBUG-110681
Change-Id: I5a4ff3a3753762bad8a51d08e51e8013bc7816a1
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Jøger Hansegård <joger.hansegard@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Implement support for getting the strikethrough
style (UIA attribute UIA_StrikethroughStyleAttributeId [1])
by checking for the corresponding IAccessible2
text attribute "text-line-through-type" [2] and
mapping the value accordingly if set.
Only report those attributes from the
QAccessibleTextInterface::attributes return value,
if they apply for the whole range of the
QWindowsUiaTextRangeProvider.
With this in place, the NVDA screen reader on
Windows announces "strikethrough" as expected
when asked to report the formatting information
(e.g. via NVDA+f shortcut) for the QTBUG-118106 example.
[1] https://learn.microsoft.com/en-us/windows/win32/winauto/uiauto-textattribute-ids
[2] https://wiki.linuxfoundation.org/accessibility/iaccessible2/textattributes
Task-number: QTBUG-118106
Change-Id: I2fa060bd3bf493227bba766385f34d224497784c
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
When the text attributes an accessible interface reports
include one of the various "text-line-through-*"
IAccessible2 text attributes with a value that indicates
that strikethrough is applied [1], bridge that to the
AT-SPI layer via the "strikethrough" attribute with a
value of "true".
See also the ATK documentation [2] for the
`ATK_TEXT_ATTR_STRIKETHROUGH` text attribute.
This will not only be used by an upcoming change
that implements reporting that for Qt's own text
widgets, but is also needed for third-party applications.
Change for LibreOffice that makes use of this: [3]
[1] https://wiki.linuxfoundation.org/accessibility/iaccessible2/textattributes
[2] https://gnome.pages.gitlab.gnome.org/atk/AtkText.html#AtkTextAttribute
[3] https://gerrit.libreoffice.org/c/core/+/157939
Task-number: QTBUG-118106
Change-Id: Ieb98584a3c6270d8db508d59994f9ba244e2bc64
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
Under some circumstances (e.g. setDesktopSettingsAware(false) on
windows) it may happen that the call to platformTheme::palette() returns
a nullptr which is not checked before dereferencing the pointer.
Therefore add a check for this to avoid a crash.
Fixes: QTBUG-111527
Change-Id: I6443d5d1a9b813f499d8f65b4fee55b0b8299b16
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The IAccessible2 text attribute specification [1], which
is the applicable specification for text attributes in
Qt as well, says:
> These characters need to be escaped with a backslash:
> backslash, colon, comma, equals, and semicolon.
To adhere to this specification the commas in the
"rgb(0,0,0)" value strings need to be escaped.
However, AT-SPI does not expect those to be
escaped, s. for example the ATK text attribute
specification for ATK_TEXT_ATTR_FG_COLOR:
> The foreground color. The value is an RGB
> value of the format "u,u,u"
Therefore, replace any backslash-escaped comma
with just a comma in the AT-SPI adaptor.
The context where I ran into this is LibreOffice
change [3] (where LibreOffice's qt6 VCL plugin
is based on Qt, but the Windows variant
has its own IAccessible2-based a11y implementation).
[1] https://wiki.linuxfoundation.org/accessibility/iaccessible2/textattributes#formatting
[2] https://gnome.pages.gitlab.gnome.org/atk/AtkText.html#AtkTextAttribute
[3] https://gerrit.libreoffice.org/c/core/+/157845
Change-Id: I0a9003ff891f1bfb180a6d16a1dff2afe4002b3e
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
The condition was wrongly converted while qmake-to-cmake porting.
Pick-to: 6.2 6.5 6.6
Change-Id: I8b8de7c9b2ca079609ed466cf91f1a3dd0d22aa9
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We need to free the memory allocated when getting the
info about the variable font.
Change-Id: I25c5f070a6d02468c904428ea2ae71efce1e847c
Reviewed-by: Lars Knoll <lars@knoll.priv.no>
When an invalid index is given to QStandardItemModel::mimeData() the
function bails out with a warning and returns a nullptr. But the
allocated data from QAIM::modelData() was not freed.
Pick-to: 6.6 6.5 6.2 5.15
Task-number: QTBUG-116532
Change-Id: Ibce9d51ea09d45f1b0eb3ca980aaff141af5bf68
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
After the change for QTBUG-89145 the parent was not set to 0 when the
item was not attached to a model.
Pick-to: 6.6 6.5 6.2
Fixes: QTBUG-117900
Task-number: QTBUG-89145
Change-Id: I421e775130b03ce3eb2dd1dd05370e7391af087b
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Otherwise, for example fillRect() will fail silently.
Most operations already had the warning, but some were missing it.
Pick-to: 6.6
Change-Id: I1ef6bf880d5b0722baadcf0ced68a968f0b1a070
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: David Faure <david.faure@kdab.com>
The two destcolors and the buffer are all already premultiplied. This
would double premultiply them.
Pick-to: 6.6 6.5 6.2
Change-Id: I235bee616d8e0033b87c6f96464f0926af7bd29a
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
We can not thread mono image formats as the bit access isn't atomic.
Pick-to: 6.6 6.5
Fixes: QTBUG-117494
Change-Id: Ica8e08fc3da73983f6b4da2a133f5e715fd0d787
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Applicable to D3D11 only, although the flag is general enough that
other backends could use it if it made sense for them.
This allows Qt Quick to state that warnings about QRhi::create()
failures that lead to retrying with a different set of flags
(PreferSoftwareRenderer to get to use the WARP software rasterizer)
should be suppressed and turned to regular categorized debug prints.
Other users, e.g. an application directly working with QRhi may not
want this. A create() failure must be complemented by an unconditional
qWarning since normally that is pretty serious error. Hence the opt-in
flag.
Task-number: QTBUG-117926
Change-Id: I808bd1670b631e2068b566ab08589e1783f62ca5
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Reimplement the virtual function QIconEngine::isNull() for
QPixmapIconEngine. This lets QIcon::isNull() return true when no pixmaps
are loaded during QIcon::addFile() e.g. due to a wrong filename.
Pick-to: 6.6 6.5
Fixes: QTBUG-118667
Change-Id: I29f2c492e55b60638507fa398ef7af76f4e9ff48
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
So now instead of filling out a VkPhysicalDeviceFeatures and
sending it in via VkDeviceCreateInfo::pEnabledFeatures,
we rather create a VkPhysicalDeviceFeatures2, which contains
a VkPhysicalDeviceFeatures (for 1.0 stuff) and a pNext
chain, and chain it via pNext to the VkDeviceCreateInfo.
QVulkanWindow will not do anything with post-1.0 features
by default. (in this regard it deviates from the QRhi
Vulkan backend which already tackles 1.1, 1.2, and 1.3
features as well, but this is not going to change)
However, applications may want to enable some of those still.
To enable this, add a new callback. When set, the callback is
invoked and the alternative behavior described above is
activated.
Fixes: QTBUG-117966
Change-Id: I8560a6b23421f545ee760e714cbe2ec2f6078406
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
- editing same document in different views is possible? yes
- resizing one QTextEdit calls QTextDocument::setTextWidth(), which
affects the width in the other view too. You can resize either one,
but you can't have two views with different layouts, of course.
Added a categorized log message in QTextDocument::setTextWidth()
for a sanity check.
Task-nunber: QTBUG-35688
Change-Id: I59c4d10143fda3a66b946237832274d67f9d9d45
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Mirror what other platforms such as Windows or xcb do.
Those all unconditionally upgrade a non-alpha format such as RGB32 to
something like ARGB32_Premultiplied (or whatever
qt_maybeAlphaVersionWithSameDepth returns).
Without this, there is a discrepancy between macOS and the other
platforms when using QOpenGLWidget or QQuickWidget (which trigger
the usage of QRhiBackingStore).
For example, attempting to have semi-transparent raster widget content
over the RHI-based widgets will render incorrectly. This is visible in
6.7 (dev) with the cuberhiwidget example as well: the red
semi-transparent overlay is rendered incorrectly on macOS since
QRhiBackingStore sticking with a QImage of RGB32 has far-reaching
consequences.
Upgrading to an alpha format is also required by the "hole punching"
mechanism the texture-based widget system relies on, although there is
a suspicion that that just happens to look correct with RGB32 as well
since there is an alpha channel technically with that format, and alpha
writes may not be masked out, depending on painting internals, ending
up with RGBA content in practice.
But in other cases, like in the report and the one mentioned above
this clearly breaks down.
Fixes: QTBUG-118553
Pick-to: 6.6 6.5
Change-Id: I4b5b7a4f720377d64903e948b866f8d1c443a2d2
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
...instead of hardcoding a format such as ARGB32_Pre. A subclass may
override format() after all.
A few lines below there is the same check in fact, but this time
with hasAlphaChannel(). To be consistent, switch over to
hasAlphaChannel() in the first case as well.
Pick-to: 6.6 6.5
Change-Id: I7a5653863a72f8a0df98da66290f6d17e548b0a8
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
The addButton(StandardButton) and button(StandardButton) APIs return
a QAbstractButton, that the user can in theory modify to their heart's
content. This causes problems when the native dialog backend is not
aware of these modifications.
We documented this limitation in e9a1c5321, but it turns out to be
more common than we first though. In particular, a typical case is
modifying the text of the button.
We now try to detect if the button has a non-standard text, and if
so turn it into a custom button when passing it on to the native
backend. From the point of view of QMessageBox it's still a standard
button, and will be reported as such in the result of exec().
To make this work the QMessageDialogOptions::CustomButton needs to
learn about custom button identifiers, so we can pass the original
standard button though.
This moves us closer to a world where the QMessageDialogOptions
represent both standard buttons and custom buttons using the same
structure, which is what we want anyways, so that we e.g. respect
the added order of standard buttons with the same role.
Pick-to: 6.5 6.6
Change-Id: Ifb7f7bd537fe71293f14ef6a006999e350bd0b52
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Disable QGraphicsFrameCapture for the no-thread (and
no-process) configuration.
Change-Id: I7767098efff78338a763ff192512d484cd300c1d
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
QDoc automatically generates include statements for types, by a set of
rules used to determine Qt's convenience header. For the
`QWaylandApplication` struct, no such convenience header is generated.
Use QDoc's `\inheaderfile` command to specify the correct header for
access to `QWaylandApplication`.
Fixes: QTBUG-118315
Pick-to: 6.6 6.5
Change-Id: Ib3b120e3ce6b5d4df74078a21bb137adaaffa2ce
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
While it would also be acceptable to leave this conversion to the user,
the returned image's format (that it's QImage::Format_RGBA8888) is not
documented. That needs to be corrected in any case. Then, we may just as
well add and document that for the most commonly used swapchain color
buffer format the r-b swap is performed automatically.
Pick-to: 6.6 6.5 6.2
Fixes: QTBUG-118244
Change-Id: I7ec8324ec3a0f5c1f2a7a592501d5ff59e3e8cc5
Reviewed-by: Kristoffer Skau <kristoffer.skau@qt.io>
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
There's logic in QMessageBox to resolve the default and escape button
if not set by the user. And the user may of course override these.
We now propagate the resulting buttons via the dialog helper, and
pick them up in the macOS native dialog backend. The only common
information we have between the standard buttons and custom buttons
is the button identifier, which for standard buttons is the enum
value, and for custom buttons is an auto generated identifier.
The same identifier is used when reporting the clicked button
from the native dialog helper.
Fixes: QTBUG-118308
Pick-to: 6.6 6.5
Change-Id: I5ca45604b51f0bbf74e56134d7b55bb8911f3563
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
So far, only accessible interfaces of roles QAccessible::Dialog
and QAccessible::Window were taken into account when trying to
find the top-level window in order to calculate the window-relative
coordinates relative to that one.
However, an app doesn't necessarily need to have any such widget/
window as it's top-level object, but can have any widget there.
Therefore, consider any a11y object that is a direct child of the
application's a11y object a top-level window as well.
For example, in the spinboxes example
(qtbase/examples/widgets/widgets/spinboxes/spinboxes), the top-level
widget is a QGroupBox, which has an accessible role of QAccessible::Client,
which maps to ATSPI_ROLE_FILLER for AT-SPI on Linux.
Since that's the top-level widget, window-relative coordinates
should be calculated relative to it.
(Without this change in place, screen coordinates would be
used for the window-relative coordinates as well, as found by
the script attached to QTBUG-106527.)
Also deduplicate a bit and just have a single loop instead
of the extra check on the interface itself at the beginning.
Fixes: QTBUG-106527
Pick-to: 6.6 6.5
Change-Id: I8d2a00bcc22c71d696e4f48233afddc80e93bc1b
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
- remove unused variable
- port index-based loop to range-based for
- use const method more to avoid implicit detach
Change-Id: I223f6c221d0c6277e94efd2e7b7be0f8d8456c60
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Change-Id: Ief76068a37d32759666e3aa2a24f126f18eca88b
Reviewed-by: Hatem ElKharashy <hatem.elkharashy@qt.io>
Reviewed-by: Janne Koskinen <janne.p.koskinen@qt.io>
When a child window is made visible without its parent having been
created yet, we defer creating and making the child visible until
the parent is created.
But if a child window is explicitly created, we create the parent
first. And creating the parent will in turn apply visibility to
all children that had their visibility deferred. Which includes
creating the child.
This results in child -> parent -> child creation recursion,
which means that when we return from creating the parent window
we already have a platform window for our child, and should
bail out.
Pick-to: 6.6 6.5
Change-Id: I11dc4864b57f031de2cca70b79cdfc057d4fbd0d
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
Add an overload for fontTable (we can't deprecate the const char *
overload as it will be used for string literals, even if the tag-
overload would work), and use the tag type in the test instead of
QByteArray.
Task-number: QTBUG-117046
Change-Id: I104f11a25e9c453a5d1081b6cf320bdc186b7e80
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Morten Sørvig
Mårten Nordheim
Tor Arne Vestbø
Jan Arve Sæther
Laszlo Agocs
Volker Hilsheimer
Michael Weghorn
Christian Ehrlicher
Ahmad Samir
Alexey Edelev
Eskil Abrahamsen Blomfeldt
Sergio Martins
Axel Spoerl
Allan Sandfeld Jensen
Lu YaNing
Tasuku Suzuki
Shawn Rutledge
Anton Kudryavtsev
Paul Wicking
Joni Poikelin
Lucie Gérard