The code for deducing the type of output sequence was assuming that the
input sequence is always a template class and was trying to use the
corresponding container type for the output sequence. Fixed the
deduction code, to assume that the output sequence has the same type as
the input sequence, when it's not a template class. Also added tests to
verify that all QtConcurrent functions support non-template input
sequences.
Fixes: QTBUG-30617
Pick-to: 6.2 6.1
Change-Id: I486fe99f3207cfff5dcceb3712cc7de863067edb
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Two separate changes addressed the OUTPUT_TARGETS variable not being
set in the calling scope of _qt_internal_process_resource() for a
particular code path, but they did so at different places. Remove one
of them, since we don't need both.
Pick-to: 6.2
Change-Id: Ibc1052e886ec73a99231ada3b7a1bb9e7a873cc4
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
In certain situations, a file generated by add_custom_command(OUTPUT)
may be added as a source to a target defined in a different directory
scope. That makes its GENERATED source property not visible to the
target for CMake versions before 3.20. For 3.20 and later, policy
CMP0118 can make the generated state of a source file global, but that
policy is under the control of the user project, since its value at
the end of the target's directory scope is what is used. Therefore we
can force that global visibility using CMP0118. With CMake 3.18 or
later, we can set source properties in arbitrary scopes, which allows
us to force visibility in the scopes we need it. Use that for all
CMake versions 3.18 and later to make the source file properties
visible to the target.
Making the generated state of a source file visible to the target
isn't enough on its own. In the scope that the file is generated,
there must be a target that depends on the file or else CMake won't
set up the necessary dependencies to ensure that generation happens
before the target tries to use it. Add targets in the generated
file's own scope where we currently might not do that.
Task-number: QTBUG-95200
Pick-to: 6.2
Change-Id: Ic1dfca1a193041de01d7f903b79473b9d47f1d1d
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The variable might be defined to HOST_PERL-NOTFOUND which will return
early instead of issuing an error.
Change-Id: Id49dfd0fec8f477dad9e816e6128e36059e68702
Pick-to: 6.2
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
...just like create() would do when requesting a full rebuild.
Not relevant for the OpenGL backend, while D3D and Metal may get away
without doing this, but the Vulkan backend gives visible rendering
errors in Qt Quick once updateResources() is taken into use and a
scene manages to do the "right" amount and types of changes.
The most common source is a changing uniform buffer offset. Consider
how the Qt Quick scenegraph merges uniform data into a single buffer
in unmerged batches (i.e. when a set of geometry nodes cannot be
batched together in a single draw all). While each node gets its own
draw call, each associated srb refers to the same uniform buffer at
binding point 0, just with a different offset.
Without the layout-compatible-update optimization (that is
QRhiShaderResourceBindings::updateResources()) this is not something
that needs extra care: once an srb is built or rebuilt (by calling
create()), the offset, just like the resource itself is fixed and
cannot change. And once create() is called, it conveniently
invalidates all related data structures, leading to the subsequent
setShaderResources() rewrite descriptors (incl. the resource, the
offset, etc.) with Vulkan or do whatever is relevant with other
backends.
updateResources() only does the minimum amount of changes because we
know that the binding list layout has not changed. It turns out this
was a bit too minimal, because certain state tracking data structures
still need resetting, just as if we called create().
The problem is becoming apparent with non-layout data such as the
uniform buffer offset, storage buffer offset, or the storage image mip
level. It however works as expected when changing the resource itself.
E.g. if a binding point now references a QRhiBuffer different than
before, then there is no visible problems, regardless of the buffer
offset being different or the same. Hence being difficult to discover,
until the aforementioned Qt Quick renderer case triggers it.
Why is this?
Changing the resource (buffer, texture, sampler) itself works due to
the guarantees given by the QRhi resource model. Consider the
following:
ubuf is a uniform buffer
ubuf->create();
srb->setBindings({ references ubuf });
srb->create();
// use the srb in some draw calls
// ...
// later, when preparing the next frame we decide we need new data with
// a different size even:
ubuf->setSize(new_size)
ubuf->create();
// fill in new data to ubuf
// use the srb in some draw calls
at this point "magic" happens: note how there is no rebuilding of the
srb itself (no create(), no nothing), yet it will correctly pick up
the now-different native buffer objects underneath ubuf.
This works because there is a certain degree of state tracking
happening that allows recognizing that a resource referenced from an
srb got rebuilt and now has different native objects (e.g. a VkBuffer)
underneath, which in turn needs (with Vulkan) rewriting the associated
descriptor with the new native resource.
Incidentally, this also makes updateResources() work just fine as long
as it replaces the QRhiBuffer/Texture/Sampler reference for the
binding point. However, with the example snippet above there is no way
to change the other associated data such as the buffer offset. (that
would need rebuilding the srb with create(), and that resets all
related state tracking structures)
So once we encounter an updateResources() where the same QRhiBuffer is
now used with an offset different from before, that is not recognized
by setShaderResources() and (with Vulkan) it will not rewrite the
descriptor with the new offset. (unless the changes for another
resource in the binding list trigger it; the granularity here is quite
coarse, i.e. we either rewrite (with Vulkan) all descriptors or none
at all; this makes the problem even less apparent because now
rendering errors occur only when Qt Quick ends with an update where
only the uniform buffer offset, but nothing else changed)
Pick-to: 6.2
Change-Id: I82ee43aa358947288135ff72ec213e091342e9cb
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Be idiomatic and return the output iterator one past the last element.
Otherwise passing in a plain pointer (as exercised by the autotest now)
fails to function because we write over the same 4 elements again and
again for each binding.
Pick-to: 6.2
Change-Id: If74463fa5140ffa2b1d5be97b71868848ad46614
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Instead of making it return a non-useful nullptr.
Change-Id: Ie72b0dd0fbe84d2caae0fffd16a245cce5ea65f1
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
[ChangeLog][Potantially Source-Incompatible Changes] The macro
Q_DECLARE_INTERFACE used to cast away the constness of the QObject
parameter. That is now fixed in this release, but may cause failure to
build source code that depended on this incorrect behavior. If fixing
the const correctness in your code is not an option, insert an explicit
const_cast<IFace *> of the object prior to the qobject_cast call.
Change-Id: Ie72b0dd0fbe84d2caae0fffd16a246a36f0c088f
Reviewed-by: Rui Oliveira
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
We only want the first entry, so avoid all the allocations of
split()ing and just look at the first.
Change-Id: I81beee1856608c932254213f2971fc37bc457c41
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
If we are the one initiating the close (from Qt Widget land), we want
to mark the widget as closing as early as possible.
Clarified the role of close_helper by renaming it to handleClose.
Change-Id: Iae250a0ae1583d743c59e99fcb99fdf18d2a1882
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This reverts commit 63c1e7c4a1.
It caused the following compilation error with g++-11.
error: ‘void* memcpy(void*, const void*, size_t)’ copying an
object of non-trivial type ‘class QChar’ from an array of
‘const char16_t’
Fixes: QTBUG-96268
Change-Id: I2680b15aba8d0d867092391fcee3815e7fa4c0bc
Reviewed-by: Ievgenii Meshcheriakov <ievgenii.meshcheriakov@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
On macOS, to display a modal with Qt::WindowModality set, or of type
Qt::Sheet, we call beginSheet:completinHandler:. However, this method
won't display the specified sheet unless the current active one is
dismissed, which is a different behavior than what we expect for
this case. Use beginCriticalSheet:completionHandler: whenever we
already have an active sheet attached to the NSWindow, which allows
us to display multiple sheets for the same window.
Fixes: QTBUG-91059
Pick-to: 5.15 6.1 6.2
Change-Id: I86bdbcbd63758edbbc48a8aade0178917dcb0e5b
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
When doing a cross build for Android we need only the host tools which
are available under cmake's host_tools target, but androiddeployqt and
androidtestrunner are not part of that. This fixes that.
Pick-to: 6.2
Change-Id: Icdbc4a78ca050b66ec8df656d9ec766ef6c9f4b9
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reading through the code contemplating what might have caused QTBUG-89141
brought up some opportunities for improvement.
* updated coding style and variable names
* use ranged for where possible and meaningful
* replacing a QList of pointers to heap-allocated structs with a list of
values
Since the QList population code makes sure that we never have gaps (we
only insert within the existing range), the test for null-entries is not
needed, and was perhaps just precausion to avoid nullptr dereference.
Task-number: QTBUG-89141
Change-Id: I4694d820427a221f1334d2428f50069751919aef
Reviewed-by: Oliver Eftevaag <oliver.eftevaag@qt.io>
Reviewed-by: Jan Arve Sæther <jan-arve.saether@qt.io>
The operator checks cause compilation errors when trying to check for
their existence for recursive containers. This happens because of trying
to check for the operators on the template parameter type(s), that
inherit from the container itself, which leads to compilation errors.
Introduced alternative versions of the operator checks (with _container
suffix), that first check if the container is recursive, i.e. any of its
template parameter types inherits from the given container, and skips
the operator check, if that's the case.
The fix is done for all Qt container types that had the problem, except
for QVarLengthArray and QContiguousCache, which don't compile with
recursive parameter types for unrelated reasons.
Fixes: QTBUG-91707
Pick-to: 6.2 6.1
Change-Id: Ia1e7240b4ce240c1c44f00ca680717d182df7550
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
...by the Qt Quick renderer, for example.
A typical Qt Quick material binding set serializes to 8 uints. This
would not demand a container like QVector. However, being implicitly
shared is essential here due to the intended usage (query the
serialized blob, put it into a cache key, hash it, compare it, all
without any copying and new allocs; we can afford an extra alloc
upon each srb construction, but don't want more afterwards in the
rendering engines)
Also make it clear in the pipeline docs that the optimization Qt Quick
is (soon going to be) doing is legal. (the srb ref in the pipeline can
be dead and dangling as long as every call to setShaderResources()
specifies a layout-compatible alternative)
Pick-to: 6.2
Change-Id: I97efbea1fa3516b10c9832adbab0a21b7bc0845d
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Use the same variable name in function
Amends 6869d2463a2e0d71bd04dbc82f5d6ef4933dc510
Pick-to: 6.1 6.2
Change-Id: If710a53993e84d048f9052f4fcf92eb57635f585
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
The file was not handled during the qmake -> CMake transition.
qmake had a more generic mechanism to allow embedding dSYM scripts per
Qt module, but it wasn't used as far as I know, so the CMake
implementation only limits it to QtCore.
Pick-to: 6.2
Change-Id: Ibcd818df1897f603ac194092b28d4bd4d1d6c6ae
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
or from Xcode.
A command line lldb that uses python3 fails to load the 'utils'
module dependency when importing lldbbridge.py via the dSYM script
debug_script.py.
Add the directory where lldbbridge.py is as an additional import path
to sys.path.
This fixes the bridge to load in both CLI lldb and from within Xcode.
Traceback (most recent call last):
File "src/corelib/debug_script.py", line 92, in __lldb_init_module
bridge = import_bridge(bridge_path, debugger, session_dict)
File "src/corelib/debug_script.py", line 42, in import_bridge
bridge = imp.load_source(MODULE_NAME, path)
File "Versions/3.8/lib/python3.8/imp.py", line 171, in load_source
module = _load(spec)
File "<frozen importlib._bootstrap>", line 702, in _load
File "<frozen importlib._bootstrap>", line 671, in _load_unlocked
File "<frozen importlib._bootstrap_external>", line 783,
in exec_module
File "<frozen importlib._bootstrap>", line 219,
in _call_with_frames_removed
File "Qt Creator.app/Contents/Resources/debugger/lldbbridge.py",
line 34, in <module>
import utils
ModuleNotFoundError: No module named 'utils'
Amends 1b73c202ce
Pick-to: 6.2
Change-Id: I521d7530e35ee9c51ae0418d2c532e58ec1952d0
Reviewed-by: hjk <hjk@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
We already had code that filtered out QT_DEPRECATED_X("text"). But that
isn't enough, because, by now, we have a true cornucopia of
QT_DEPRECATED_* macros. And only some are called with an argument list.
Move the filtering code into the subroutine filterDeprecationMacros,
because our filtering is slightly more complex now:
- Try to match a QT_DEPRECATED_* macro call.
- Try to match balanced parentheses with a recursive regular expression.
- Check whether the found balanced parentheses are directly behind
QT_DEPRECATED_*, because only then it is the argument list of that
macro.
- Filter out what we've found.
With this patch, syncqt doesn't discard deprecated classes anymore.
Task-number: QTBUG-80347
Change-Id: I7872159639be330d5a039c98eac0c5007d9acb93
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The parser was lenient in accepting backslashes followed by invalid
characters, but accidentally sign-extended everything above 0x7f causing
broken outputs that weren't valid UTF-16 either.
For example, the sequence "\\\xff" (backslash followed by 0xff) produced
sequence "\ud7bf\udfff" (U+D7BF is not a surogate pair).
Change-Id: Ie72b0dd0fbe84d2caae0fffd16a113c703a7696f
Reviewed-by: Ievgenii Meshcheriakov <ievgenii.meshcheriakov@qt.io>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
This reapplies the fix from 1ecf2212fa,
using QApplication::translateRawTouchEvent to dispatch the touch event
received by the QGraphicsProxyWidget to the relevant child widgets
under each touch point.
In addition, limit the implicit grabbing of each touch point before
sending the event to those cases where we have to: touch pads, and
if the target widget comes from a closest-widget matching. And don't
call the QTouchEvent override of QEvent::setAccepted() on QTouchEvent
instances in QGraphicsView classes, as this will override each event
point's acceptance state.
This way, we can identify which touch points have been accepted after
event delivery, any only implicitly grab those points that were.
Otherwise, touch points not accepted by a proxied widget hierarchy
will still be part of an accepted event, and be grabbed by the
viewport of the QGraphicsView. This would then lead to infinite
recursion when the QGraphicsProxyWidget passes the TouchUpdate event
on to each touch point's grabber.
Re-activate the test case, and extend it with more combinations.
Refactor touch-event recording to make it easier to test multi-touch
scenarios.
Task-number: QTBUG-45737
Fixes: QTBUG-67819
Pick-to: 6.2
Change-Id: Id5611f4feecb43b9367d9c2c71ad863b117efbcb
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Some backends were missing support for standalone days and months,
also the standaloneDayName() implementation was always using the
same codepath as dayName().
This patch fixes the issues.
Support for narrow format will be added in the following patch.
Task-number: QTBUG-84877
Pick-to: 6.2
Change-Id: I38ee06342cafab544e3c69097bd0e6ae68e85645
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Counting repaints is unreliable on macOS.
Pick-to: 6.2
Change-Id: Ifa1cf516a2cab94540bea005354d8311a2651d64
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Also when closed by destruction, which as of today doesn't call
QWidget::close and therefore also not QWindow::close.
Pick-to: 6.2
Change-Id: I426255e2274eae9262243c769df2264fbaa915b0
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
std::chrono values are passed by const reference.
Still warnings from undocumented parameters, but rephrasing
the documentation doesn't make it better in this case, so perhaps
qdoc needs a way to suppress the warning. Adding an \omit block
where the parameters or return values are mentioned doesn't help.
Change-Id: I7d495d73d8367d9d90dd33a4880ac7c978382d19
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
We default to configuring examples as separate ExternalProjects when
using a developer / non-prefix Qt build.
This ensures we test that the examples configure successfully without
the pollution of the main Qt build (e.g. already found packages).
One down-side of this is that a developer's IDE doesn't see these
example targets, unless each project is loaded into the IDE
separately. This is cumbersome to do when refactoring or renaming
code across multiple example projects.
Allow configuring the example projects as part of the main Qt build
by setting QT_BUILD_EXAMPLES_AS_EXTERNAL to FALSE when configuring Qt.
Save the value of the variable in QtBuildInternalsExtra.cmake.in
so it's propagated to leaf repositories as well.
Amends dab8f64b6d
Amends d97fd7af2b
Task-number: QTBUG-90820
Task-number: QTBUG-94608
Task-number: QTBUG-96232
Task-number: QTCREATORBUG-26168
Pick-to: 6.2
Change-Id: Ie1f724f74365b3a788b04c3fffe9eb2d0611dd50
Reviewed-by: Craig Scott <craig.scott@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
At least we try to do it with all events triggered by user.
Pick-to: 5.15 6.1 6.2
Change-Id: I28b399a2517600f7da2c91a50fecdf58b9d81fb6
Reviewed-by: JiDe Zhang <zhangjide@uniontech.com>
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Those casts are not needed when passing pointers to simple types
to memcpy.
Change-Id: I686265b0e152aa22e0195ff252c442ab1a122ba7
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Use standard char16_t and char32_t types instead of ushort and uint.
Remove members of QUtf8BaseTraits that use those integer types.
Change-Id: I77b1a9106244835c813336a50417f6bbdfada288
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
On OpenSUSE we seem to get an Enter event after or while the QWindow is
closed, and that appears to reset the currentMouseWindow after we
cleared it in QWindowPrivate::destroy. Apply a workaround similar to the
focus_window to make sure that currentMouseWindow and
currentMousePressWindow don't point to destroyed objects.
Pick-to: 6.2
Change-Id: I61afdaa7d8d1d437058624012319460be2f2567f
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Some versionless wrappers were not passing back output variables to
their calling scope. Ensure they always are.
Fix qt6_extract_metatypes() to set its output variable in the parent
scope (it was previously setting it erroneously in the local scope).
Some functions had code paths that would not set output variables.
This would allow situations where if the variables had an initial
value set by a higher up parent scope, the output variable would
still have that value in the caller's scope upon return. That could be
misleading, so fix these code paths to explicitly set the output
variable to an empty string instead.
Task-number: QTBUG-96121
Task-number: QTBUG-96219
Pick-to: 6.2
Change-Id: I291775813f025cabdccd4372ac077cdfd3ec090e
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
This reverts commit 08180e76e6.
Macros add another level of escaping that functions do not. The
conversion of the versionless wrappers to macros may alter the
behavior, so revert that change.
Task-number: QTBUG-96219
Pick-to: 6.2
Change-Id: Ic5dcff3081123d957888584ba1d76ae0580d9083
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Unfortunately QInputMethodEvent::Cursor defines length == 0 as the
cursor being hidden, and length > 0 as being visible. This is the
opposite of what native macOS application do.
A future improvement here would be to base the QInputMethodEvent
logic on QStyle::SH_BlinkCursorWhenTextSelected, which we already
respect for normal selections. That would also allow us to use
QInputMethodEvent::Cursor to set the preedit selection, which
we currently have to fake via QInputMethodEvent::TextFormat.
Pick-to: 6.2
Change-Id: I75b5d8c5403283a0988355e440a98b4df35ec995
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
QDialog today only hides itself during reject and accept, it doesn't
close itself properly. This is problematic, as it doesn't close the
QWindow.
However, fixing this behavior must not result in duplicate calls to
virtual function, or additional calls to virtual functions (such as
closeEvent) without explicitly flagging the change in the changelog.
Add more tests to document existing behavior so that we can identify
such changes and verify the desired side effects.
Pick-to: 6.2
Change-Id: I1f30701cd766eb3c7957751b51e8579d4542dd16
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Only QByteArray has a toHex() member, QByteArrayView doesn't.
Since toHex() is linked to from result() already, remove it here
to avoid the wrong impression that there was a toHex() that doesn't
require any memory allocation.
Change-Id: I76f876aca90403baebf9328b794aeaf9be698c46
Reviewed-by: Luca Di Sera <luca.disera@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
