This reverts commit 09ee4282e5.
The commit made the layouting time go from linear to quadratic, which is
problematic when rendering big documents impractical.
I wrote an alternative fix for QTBUG-20354 which will come as separate
change since it approaches the problem in a different way.
Fixes: QTBUG-89599
Task-number: QTBUG-20354
Pick-to: 5.15
Pick-to: 6.0
Change-Id: Ie450332f06ee40e60c8e4c6c7d10834bff1acf74
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Remove the qmake project files for most of Qt.
Leave the qmake project files for examples, because we still test those
in the CI to ensure qmake does not regress.
Also leave the qmake project files for utils and other minor parts that
lack CMake project files.
Task-number: QTBUG-88742
Change-Id: I6cdf059e6204816f617f9624f3ea9822703f73cc
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
d25589e052 mistakenly changed the code
to use toString() from the QTest namespace. Call toString()
unqualified and move the compareSequence() helper out of the
internal namespace.
Fixes: QTBUG-89737
Pick-to: 6.0
Change-Id: I8452a0aead7771ad13a5b95438b54e202ccce76b
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
This commit remove old legacy code that try to deal with clipboard
image conversion in the qwindowsmime class. It now uses
qbmphandler under the hood which is much more complete.
It also add a small fix for the conversion
of BI_RGB clipboard image with an explicit alpha layer (which Firefox
use on Windows).
Change-Id: Iae026378831799dc676e1aba7d5bd6a0d1c01e7f
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
QApplication tries to close all windows on quit using closeAllWindows,
but closeAllWindows skips windows that are already closing. This can
happen when calling quit() from a close event for example.
QApplication then tries to verify that all windows have been closed,
and that logic should skip the same kind of windows as closeAllWindows
does.
The fact that these two logics diverge was identified earlier in
5af73cd9db, but aligning them required further work. As that
commit notes, the right fix to align them is building on top of
tryCloseAllWidgetWindows(), which already returns true/false based on
whether it could close all windows or not. But, unlike the existing
logic in QApplication::event(), it doesn't skip Popups or Dialogs,
so that discrepancy needs further research.
Pick-to: 6.0
Fixes: QTBUG-89580
Change-Id: I87bff56f2eb8a539f1c859c957f5f239dc1eb93d
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
As lists of QStrings and QByteArrays are sequentially iterable the base
types should really also be.
The only problem is that they don't have methods to remove items from
the back or the front, but that is well within what we can support with
QSequentialIterable.
Change-Id: I2ab551e7b11a092aba363fb4012d131bbc4b11b4
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Make it clear that the fast lookup is by key, not by value.
See also discussion in
https://forum.qt.io/topic/121907/misleading-documentation-of-qhash-qmap/
Pick-to: 5.15 6.0
Change-Id: I396297e0e4674e0a1f889f4138ab52ff224c0ee2
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
When AA_ShareOpenGLContexts is not set, docking or undocking
will lead to changing the associated top-level window. This
leads to changing the OpenGL context, and tearing down and
then recreating all OpenGL resources (assuming a well written
application). The problem is, there are no paint events after
the Show, meaning the user code's paintGL is often not invoked,
which leads to showing an empty QOpenGLWidget until something
else triggers a paint event. To remedy this, send a paint event
upon Show, which should be harmless enough, while fixing the
case of docking/undocking.
Pick-to: 6.0 5.15
Fixes: QTBUG-89812
Change-Id: I3c4560f8f069d86645a6314bf7ad1b4ee8e2c716
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
...when a QVkRenderBuffer is used as depth-stencil (i.e. a "throwaway"
depth-stencil buffer that is a lazily allocated / transient VkImage
under the hood). Such a resource is not tracked by the
QRhiPassResourceTracker, so do an explicit barrier in beginPass
instead of having it generated by the tracker.
Fixes: QTBUG-89764
Change-Id: Ice794f44342175f712ea56f450270cbb8929f516
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
...encountered in the command list.
Move all, previously local, tracking variables into a struct. This
allows creating helper functions to reduce error-prone repetition in the
executeCommandBuffer() function body.
The only real change in the patch is in the handling of
Command::BufferSubData and Command::GetBufferSubData: here, instead of
calling glBindBuffer directly, use a helper function that also resets
the relevant state tracking variables. A subsequent
Command::BindVertexBuffer or BindIndexBuffer will therefore correctly
rebind the appropriate buffers.
This is particularly relevant with certain command stream patterns
exercised by some Qt Quick 3D scenes:
- A View3D renders a mesh,
- another View3D has some 2D Qt Quick content, as well as a model with
the same mesh.
When both View3Ds use the default Offscreen render mode, the resulting
command list consists of segments along the lines of:
1. prepare resources for first View3D
2. render content for first View3D - this binds the vertex and index
buffers for the mesh (state is tracked; all 1-4 steps are within
the same command list, processed by a single call to
executeCommandBuffer())
3. prepare the content for the "inline" 2D Qt Quick scene - this may
update vertex and index buffers, that may lead to adding
BufferSubData commands to the list (tracked state (last
vertex/index buffer) may need invalidation/updating - and that's
where our problem lies)
4. the second View3Ds 3D content is rendered: a model with the same
mesh as the last (Quick)3D draw call, so same vertex and index
buffers. If #3 did not invalidate and/or update the tracked state,
the glBindBuffer calls are (incorrectly) skipped.
Fixes: QTBUG-89780
Pick-to: 6.0
Change-Id: Icc933252f3993b8727d192e7ba4aa0f842bab51e
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
There was a check "if (cd && cd->signalVectorCount())" in the code
which was faulty because "signalVectorCount()" could return -1 to
indicate that no signalVector is available. This patch fixes
the problem.
Fixes: QTBUG-89846
Pick-to: 5.15
Change-Id: I12c512ff8d889ba2bd298789ab6081930f739089
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The code using it got removed in qttools commit 01da5d9a
Change-Id: I14d36ac939f2647139fcc42591a044a03e707553
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
We did set this variable, however after src/gui/configure.cmake
was loaded (it is early-loaded in src/CMakeLists.txt), meaning the
assignment had no effect.
Also, what this assignment tried to achieve is covered by our global
CMake policies now.
Pick-to: 6.0
Task-number: QTBUG-89754
Change-Id: Ie7aeee1505d112d799455b2915acbfa7e9a12a64
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
The qt6_android_generate_deployment_settings() command had been
creating deployment settings files with the wrong key name for qml
import paths. This resulted in the QT_QML_IMPORT_PATH target property
having no effect.
The QT_QML_IMPORT_PATH property can also potentially hold a list, not
just a single value. This change now handles the list case as well,
previously qt6_android_generate_deployment_settings() was assuming it
always held a single path if it was set.
Fixes: QTBUG-89628
Pick-to: 6.0
Change-Id: Ibdd74ec8d130f160433a60a14a0a9f496f496a1b
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
QMultiHash::equal_range crashes when called in a const member function.
The Data `d` is a NULL pointer when calling equal_range()
before inserting data into an empty QMultiHash.
Then calling`d->find` crashes.
Fixes: QTBUG-89687
Pick-to: 6.0
Change-Id: I10c3d196cbc72aed8c8c922ef16534bba51037b7
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
There is no need to disable startupSocketNotifier because the call to
QProcessPrivate::processStarted() will do that.
Change-Id: I20b816533d9a5c4b9bf57135d26166c961a07d07
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
If there is an editor in the table then this needs to be updated as
the resize happens, otherwise it will not adapt to the new size
correctly and can appear drawn over other cells.
Fixes: QTBUG-85484
Pick-to: 5.15 6.0
Change-Id: I918c37f1ee1b0614c03c6b38f95f0f8b35a571f2
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
The spec hasn't changed, but I made the same mistake in xdgmime
(the reference implementation) and in Qt: when multiple globs match,
and the result from magic sniffing is unrelated to any of those globs,
then I used the magic result, but that's wrong, globs have priority
and one of them should be picked up.
This is now fixed in xdgmime
(https://gitlab.freedesktop.org/xdg/xdgmime/-/merge_requests/3)
and in the expected results in shared-mime-info
(https://gitlab.freedesktop.org/xdg/shared-mime-info/-/merge_requests/99)
which this commit is also tested against.
This change also optimizes QMimeBinaryProvider::addFileNameMatches
to have the same logic as xdgmime for glob matching:
literals > extensions > other globs
As soon as one category matches, we can stop there.
This makes no difference in the overall results, in practice.
The user bug report (against the Qt implementation, actually)
is https://gitlab.freedesktop.org/xdg/shared-mime-info/-/issues/138
as well as https://bugs.kde.org/show_bug.cgi?id=411718
Pick-to: 6.0 5.15
Change-Id: Ia0a34080427daff43c732609443ee6df8f41447c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Previously it only worked when isNull() was false, which is true for
very short lines, even though length() may be non-zero.
[ChangeLog][QtCore][QLineF] QLineF::setLength() will now set the
length if the line's length() is non-zero. Previously, it was
documented to only set the length if isNull() was false; this is a
fuzzy check, so isNull() could be true for a line with non-zero
length().
Fixes: QTBUG-89569
Pick-to: 6.0 5.15
Change-Id: I803e622ad09c85815dde25df8dd3ba6dfcba0714
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
An invalid (as input data) certificate may have non-zero number of invalid (nullptr)
extensions (if OpenSSL failed to parse them, for example).
Fixes: QTBUG-89547
Pick-to: 6.0 5.15 5.12
Change-Id: I4b93ac9f482f850f02d01b0aea10560dc11b688d
Reviewed-by: Lars Schmertmann <lars.schmertmann@governikus.de>
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
The default value is 'DontShowIndicatorWhenChildless' and not
'ShowForChildren'.
Fixes: QTBUG-83648
Pick-to: 6.0
Pick-to: 5.15
Change-Id: I5d4dd9f3f58bcac09ed38d8ae451e3b771695ce7
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
QAbstractItemView::selectionCommand() returned the wrong SelectionFlags
when no event is given since c4366ff018.
Therefore re-add the call to QGuiApplication::keyboardModifiers() when
no event is given and add a unittest for them so it's not removed again.
Fixes: QTBUG-89711
Change-Id: I107357df08c4ff1b1a14d49523401c5e7b428f56
Pick-to: 6.0
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This attempts to reconcile a minor difference between Qt 5 and Qt 6:
Running Qt Quick applications with a platform plugin such as vnc,
led to an automatic fallback to the 'software' backend based on
the OpenGL capability reported from the platform plugin. In Qt 6.0
this logic is gone from Qt Quick, because we do not have, and wish
not to have, individual flags for each and every 3D API on this
level.
Therefore in Qt 6.0 a Qt Quick application running with the vnc
(or linuxfb, or minimal) platform needs an explicit selection of
the software backend via QT_QUICK_BACKEND or
QQuickWindow::setGraphicsApi().
To keep migration for users of such platform plugins easier, we
can still reintroduce a Qt 5-like logic: by having a high level
Is-QRhi-Supported type of flag, we can make Qt Quick query that,
and trigger the fallback to the software backend when it is
reported as false by the platform plugin. As this is the minority
case and a conscious choice by platform plugins, the flag can be
opt-out (i.e. true by the default hasCapability implementation).
When it comes to the existing OpenGL flag, that needs to stay for
compatibility reasons, but it is worth noting that the new flag
semantically falls in the exact same category: it does not indicate
things will really work at run time, but rather serves as an opt-out,
"do not even try" type of declaration the platform plugin can make,
which then allows modules like Qt Quick to make early, upfront
decisions about which rendering paths/backends to take.
Change-Id: I8d6fddeb82ca6eece7b7abc1a5b64ebe6d8af29d
Task-number: QTBUG-89561
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Take the logic we use in QRhi's Vulkan backend.
Pick-to: 6.0 5.15
Fixes: QTBUG-85791
Change-Id: Ifdc2b3d351af71fbc86e20abcede902303e41fc4
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
The only callers are QObject::findChild(ren), and they always
pass valid pointers, so there's no point of checking them.
Change-Id: I789abc3a53db523acf06c1a8a340094a71c79e41
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There are no queries for this (somewhat dubious) flag in practice.
Its usage in Qt Quick is gone since 2014 after
0129887195c7255f41515f72ceb213a38b98f72d.
Change-Id: I4fd9cef2705ce13b79ff1dc40a6eda7ce2035e30
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
Running with synchronization validation (SDK 1.2.154) we get:
Hazard WRITE_AFTER_WRITE vs. layout transition in subpass 0 for attachment 1
aspect depth during load with loadOp VK_ATTACHMENT_LOAD_OP_CLEAR
Attempt to fix this with an appropriate subpass dependency.
Fixes: QTBUG-89760
Pick-to: 6.0
Change-Id: I67e15a09aab564f63aab646761ccdb1f8b82111d
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Even though there is no D3D-specific logic in the windows platform
plugin, meaning a QWindow with either OpenGLSurface or VulkanSurface
(or anything really) is DXGI/D3D-compatible, it now looks like it is
beneficial, and more future proof, if there is a dedicated surface
type.
As the linked report shows, there are OpenGL-specific workarounds
accumulated in the platform plugin, while not being clear if these
are relevant to non-OpenGL content, or if they are relevant at all
still. (and some of these can be difficult/impossible to retest and
verify in practice)
When D3D-based windows use the same surface type, all these are
active for those windows as well, while Vulkan-based windows have
their own type and so some of these old workarounds are not active
for those. To reduce confusion, having a dedicated surface type for
D3D as well allows the logic to skip the old OpenGL workarounds,
giving us (and users) a more clear overall behavior when it comes
to OpenGL vs. Vulkan vs. D3D.
The change is compatible with any existing code in other modules
because any code that uses OpenGLSurface for D3D will continue to
work, using the new type can be introduced incrementally.
Task-number: QTBUG-89715
Change-Id: Ieba86a580bf5a3636730952184dc3a3ab7669b26
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
qt_plugins target is needed by qt_internal_add_plugin function, but when
standalone building sqldrivers, this target does not exist, which causes
configure error. So we need to add this target manually.
Fixes: QTBUG-89714
Pick-to: 6.0
Change-Id: Ia7d613faabb82399c2f3e5771aab24302b3c8f56
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
View embedding when QWidget is involved is a bit finicky. This change
breaks down the steps needed to embed it into an NSMenu item, and
simplifies the process by not relying on a container widget.
The main issue is that QCocoaWindow::recreateWindowIfNeeded() will
potentially create an NSWindow for the embedded view, resulting in
a stray view. To prevent this we set the Qt::SubWindow flag on the
window, but QWidget tends to reset this flag when the widget doesn't
have a parent, so we need to be careful about which order we do the
setup.
Pick-to: 6.0 5.15
Change-Id: I505f7c0a2d8e4350511fdb01a5e9b9c623a40a41
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
It's an implementation detail that the default constructed palette
happens to use the d-pointer of the application palette, and as a
result has the same colors as the application palette if requested.
The palette has an empty resolve mask, and hence all the colors in
the palette are undefined. Users should not rely on QPalette() to
represent the application palette when for example overriding a
parent palette.
Task-number: QTBUG-85226
Pick-to: 6.0 5.15
Change-Id: Ia6747887eef6e9a5f21b2002066608c433260de2
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The enumeration value was removed in Qt 6 by
1a342b2db1. Ignore it.
Add a check to helper function checkProperty() and move
the code from a5e2f2ddad there
as well.
Pick-to: 6.0
Task-number: QTBUG-89093
Fixes: QTBUG-89700
Change-Id: I0890be8fff155d7342e2386b7f6b0783a5f3a192
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Jarek Kobus <jaroslaw.kobus@qt.io>
It's defined in terms of operator<=> on QStringView, which does
not exist, causing a compile error.
Change-Id: I64fc60da4e52c7e53be7849d9b42952be139a816
Fixes: QTBUG-89729
Pick-to: 6.0
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
CMAKE_ANDROID_NDK is an official variable provided by CMake. The
ANDROID_NDK variable will be set by the NDK toolchain file, but we
don't need to rely on that (the user could theoretically not be using
the NDK's toolchain file). Using the CMake-provided variable means
we don't have to explain the source of the variable in documentation
for the qt6_android_generate_deployment_settings() command. We should
prefer to use things provided by CMake already where it makes sense
and this seems to be one such case.
Task-number: QTBUG-89651
Task-number: QTBUG-88839
Pick-to: 6.0
Change-Id: Ieda54de8f5c65c36da6bb55c87a8b8fdd1d5cd7b
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Instead, make them inheritable at creation time.
Change-Id: I1aabc203851b2873acc772ba20258238cbe01d59
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
And renamed PeekRemove* to PeekConsumed to document
better that nodes are not removed when peeking.
Change-Id: I7349f57fcfc3287d1a1309a31ee7f481f8d18755
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When a button has no parent, QAbstractButtonPrivate::queryButtonList()
unconditionally accessed parent which results in a nullptr access. Did
not crash because qt_qFindChildren_helper checks for nullptr and
therefore could only be found with a sanitizer.
Fixes: QTBUG-83865
Change-Id: I591e546e96acba50770935b9c3baaf08b09b833d
Pick-to: 6.0
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
[ChangeLog][QtCore][QProcess] Added support for
setStandardOutputProcess() with startDetached().
Change-Id: I61278cdb7084127f583c8c017688da392017b44c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
[ChangeLog][QtCore][QProcess] Added support for QProcess::MergedChannels
mode with startDetached().
Change-Id: I953ad2063322015332269522a297f8e2842e438c
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
We have the same channel forwarding, redirecting, and merging rules for
all platforms. This makes it possible to introduce the openChannels()
function, which consolidates the logic and performs high-level general
processing of the channels configuration properties.
Change-Id: Id3574fc42a56829328369b6a1a6ec9c95fce8eca
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
The idea was to create descriptors with HANDLE_FLAG_INHERIT. However,
this can be assumed to be already the case for std descriptors, and if
it isn't, the user messed up, and we shouldn't try to work around it.
This is consistent with what we already do in startDetached().
Change-Id: I8135c5e612c361e77a0442541f2d50cfbb5b4601
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
In theory, low-level functions like CreateNamedPipe() or
DuplicateHandle() could suddenly fail. Thus, in order to ensure the
correct configuration of the channels, we must check for these
errors.
Pick-to: 6.0 5.15
Change-Id: I3d6291043e5e394ec5c5a7d85dfb7b09bfdcd172
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The PostgreSQL driver can change it's connection status
after the first established connection.
In this case, the function "isOpen()" returns false
and the "close()" function would not free all resources.
With this fix, the "close()" function
always frees any allocated resource
independent of the connection status.
Fixes: QTBUG-88984
Pick-to: 6.0 5.15 5.12
Change-Id: I7c9add6a183bf46a8573952ab39f8cb1f728c00c
Reviewed-by: Andy Shaw <andy.shaw@qt.io>
This method will return a QByteArrayView of the data range corresponding
to the level. This avoids a leaky abstraction by moving the needed data
pointer arithmetic from the caller to the method. It will also make it
easier adding cubemap support in the future.
Change-Id: I2415bd88365d8d69ba14aefc77f5f1117f9e0033
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Nico Vertriest
Frederik Gladhorn
Joerg Bornemann
Friedemann Kleint
Allan Sandfeld Jensen
Camille Viot
Tor Arne Vestbø
Ulf Hermann
Kai Köhne
Laszlo Agocs
Andreas Buhr
Paul Wicking
Craig Scott
Zhang Yu
Alex Trotsenko
Andy Shaw
David Faure
Liang Qi
Edward Welbourne
Timur Pocheptsov
Christian Ehrlicher
Giuseppe D'Angelo
Li Xinwei
Gatis Paeglis
Matthias Möller
Jonas Karlsson