We were starting to run out of space for builtin core types.
Remap the type id's to create lots of additional space.
We now reserve the first 64k id's for Qt, and have 16k id's for
Qt Core. That should hopfully be enough for a while ;-)
Fixes: QTBUG-85914
Change-Id: I0dab6bf23652e46a9557d9b38af7990b68c572b6
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
Refer to QMetaType, not QVariant::Type
Change-Id: I3e9284742b332095a26c03c68580ad0f6d4f9be7
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Vulkan cannot be a required public dependency of QtGui, because a
vulkan-enabled build of Qt will fail on a machine where the vulkan
headers are not installed. We cannot make the vulkan dependency private,
because we have public Qt headers that include vulkan headers.
The vulkan dependency is now marked as optional, meaning consuming
projects will try to find it, but there will be no error on failure.
Task-number: QTBUG-86421
Change-Id: I30edfd3eaa603048f4274e77036c54d90789d3c9
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Every public dependency of a Qt module results in a find_package call in
the consuming project. But not all public dependencies are mandatory.
For example, vulkan is only needed if the user project actually uses Qt
classes that pull in vulkan headers.
This patch adds the option MARK_OPTIONAL to qt_find_package.
Dependencies that are marked as optional will not produce an error on
find failure.
Task-number: QTBUG-86421
Change-Id: Ia767e7f36991e236582c7509cbd37ea3487bb695
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
...in code. The actual number of calls at runtime does not change.
The reason for this stunt is that we want to replace find_dependency
with find_package at runtime for optional dependencies, and this will
reduce the diff size of the next commit.
Change-Id: I304fdf09c69fea7b6d4adabf515712eb097f8c86
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The find_dependency() macro calls return() on failure, meaning any code
after it will be ignored, thus checking any XXX_FOUND variables after
find_dependency() is pointless.
Fix the places where we use find_dependency() and set the "failed" state
before the call and the "success" state afterwards.
Change-Id: Ia5239c704f02a9bec972210374ffed7808b14055
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
They were disabled in ad0e3e26fa,
but that was probably accidental.
Change-Id: Ia9cbccfbeecfe84768c3465f5699ed44b7f932a5
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Be a little bit more specific what needs to be done when compiling the
sql driver without compiling the whole Qt
Change-Id: I3e8010d4cc84a83108340e525e620abf6b6e89c1
Pick-to: 5.15
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
This change partially reverts change
a0e0b51001.
Replacing the QTest specific sleep function with QThread::msleep()
was not a good idea. The reason is that QThread::msleep() will force
the thread to sleep to x mseconds, even if a signal woke the thread
in the meantime.
This would cause qWaitFor() to not call processEvents(), in some cases,
leading to flakyness and test failures in tests that rely on timing,
such as the animation tests in Qt Qml.
Change-Id: I0ad132cdf32be5813b2e73552d772251fe1d7f89
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Provide functionality to add and remove values, so that you can use a
sequential iterable as stack or queue if the underlying container
supports this. To this end, provide a way to specify whether the
value should be added or removed at the beginning or the end of the
iterable.
Change-Id: If63d302f3ca085e56d601116ce4dfaa6b94a0c4f
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
For completeness' sake we should expose this. The iterators provided by
QIterable and friends will check the category at runtime, and should give
sensible feedback.
Change-Id: I778894f340c862f79a18c6c5607bcbba98dd7598
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
pageLayout().pageOrientation() does not exist, but pageLayout().orientation() does.
Pick-to: 5.15
Change-Id: I17a667b18aba284b91972422e604568e2fdbda05
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
This change should have no impact on Widgets, since style names
there are case-insensitive. But for QtQuick controls the style
names are case sensitive. So in order to use the style hint
from the platform theme for controls, we need to return
the name with an uppercase "F".
Change-Id: I360f43f174938202b0ef2cdfcde6daf39c9f39bb
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
A QAbstractSpinBox has a spinClickThresholdTimer that is detecting
if the user is doing a press'n'hold on one of the spin buttons, to
activate auto-repeat.
But if the valueChanged() handler in the application spends too much
time before it returns, the spinClickThresholdTimer will fire before
we get a chance to cancel it from the pending mouseRelease event.
The result is that the spinbox will think that the user is doing a
press'n'hold, and increment the value once more.
To avoid this, we start the timer _after_ the call to
valueChanged() instead.
Pick-to: 5.15
Fixes: QTBUG-86483
Change-Id: Iff5de4f8da562738e02848c98bc1fbc9fe227748
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
The special members do nothing that the compiler-generated default
doesn't do, so remove them.
Remove the comment introduced in d06b4ca9a1 about
QBitmap inherting QPixmap being a problem. It which wasn't followed
up on for Qt 6, and making QBitmap a QPaintDevice rather than a
QPixmap subclass (as implied in the commit message) would result in
significant code duplication.
Change-Id: I8d5032062f95ab19b38eb97d17191a9b13af870f
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
and obsolete itemDelegate(index), which the new function replaces.
This allows itemviews to override the item delegate per index. The
existing APIs to set delegates for rows and columns doesn't work for
tree views, or other views where the delegate should depend on the
parent index as well.
Adjust all itemview code that accessed the QAIVPrivate's relevant
data structures directly to go through the virtual function.
[ChangeLog][QtWidgets][QAbstractItemView] The
itemDelegate(QModelIndex) member function has been made obsolete,
and has been replaced by a new virtual
itemDelegateForIndex(QModelIndex) member function that can be
overridden to give subclasses control over which item delegate
should be used for an index.
Change-Id: Ib03529c38797386d3a1d4cf3c5646e911d95daf5
Reviewed-by: Samuel Gaist <samuel.gaist@idiap.ch>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
CentOS it seems not only backported some OpenSSL 3 functions,
but also raised the default security level to 2, making some of
our keys (and MDs?) 'too weak' and failing auto-tests here and
there as a result. For our auto-test we lower the level to 1,
as it is expected to be.
Fixes: QTBUG-86336
Pick-to: 5.15
Change-Id: I7062a1b292e8b60eb9c2b2e82bd002f09f9da603
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
There is no reason to use double precision for colors. We at most have
16 significant bits anyway.
Change-Id: I8b402cd978675b8ba7248176976d934363212ff1
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
This is needed for qtwayland, where QtWaylandCompositor package should
call find_package(QtWaylandScanner) in the 'Tools' section of the
ModuleDependencies.cmake file, rather than the regular 'Qt' section.
This takes care of handling host path prefixes, to ensure that a host
package is found even when tools have also been cross-compiled via the
QT_BUILD_TOOLS_WHEN_CROSSCOMPILING option.
Task-number: QTBUG-83968
Change-Id: I4725a630214d053105fb6d2a0f7c5ff6128d13f9
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
qt_apply_rpaths takes into account properties like MACOSX_BUNDLE. This
property might not yet be set when qt_internal_add_app is called, but
later.
To handle that, move the call of qt_apply_rpaths to
qt_internal_finalize_app.
As a result, the installed apps will have 2 rpaths, the $ORIGIN style
relocatable one, and an absolute path one pointing to the Qt
prefix/lib. The last one might be unnecessary.
Fixes: QTBUG-86514
Change-Id: I25e0d695c78c8b5703e94c99cc2457f772721456
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Add virtual hooks that can make it possible in Qt6 to implement paint
device specific pixmaps.
Change-Id: Ib1d1d7b5e32d9430203f718534344b25c0ead744
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Test data for QDateTime unit tests was constructed in system's local
time, which made unit test outcome dependent on the time zone of the
system running the tests.
This led to failing unit tests on systems with libc6 version
2.31 or newer. A QDateTime in a timezone, which has daylight saving
time, was created and then used in Hawaii's time zone,which has no
daylight saving time. Newer glibc checks this and returns errors.
This patch changes the behavior to create the test data in a
specified time zone setting.
Task-number: QTBUG-80441
Change-Id: I0330b647fa011be99141dde09001ff2d58bd3a5f
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
We can't unexport the class, as it's being used by the
OpenGL paint engine in QtOpenGL.
Change-Id: I74c1bef0e4720a82679a329462e80571f0583a84
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
win32-clang-g++/qmake.conf did define
QMAKE_CFLAGS_WARN_ON += -Wextra -Wno-ignored-attributes
probably for the same reason.
Change-Id: I14b66f2555949c03a2ea996923e31f5ec9f7105a
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Do categorize the license as LCS-Telegraphics License. Although SPDX
doesn't know about the license (yet), DejaCode does:
https://enterprise.dejacode.com/licenses/public/lcs-telegraphics/
So it's more helpful for the user to give the license a name, then
to just call it 'Custom'.
[ChangeLog][Third-Party Code] Changed classification of the wintab
license from "Custom" to "LCS-Telegraphics License"
Pick-to: 5.12
Pick-to: 5.15
Change-Id: Iba7538f67b43ceca2e599fc1ede520962d4a7e43
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Some repos/modules don't have configure.json files and thus no
qt_cmdline.cmake files. Make qt-configure-module check for the file's
existence. As drive-by fix, surround the configure.cmake's path by
double quotes.
Change-Id: If1a91a0bba0c2fd282cfa08fa6ff2bb20f0a15ba
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
We need to make sure to create a config.opt file, even if it's empty.
Change-Id: If1e41cea2f3b5f3ce67f30285b77c58f7376546a
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
For modules that have MODULE_INCLUDE_NAME set (for example ActiveQt) we
must use this name for the installation rules too.
Fixes: QTBUG-86484
Change-Id: I1b97cf534ea5e41655c7b3fdd0330b2f58356a3a
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This reverts commit f51b690e91.
The commit made all animation tests in qtdeclarative on macOS
flaky.
Change-Id: I4ccaa879df7e2ba7e253657de01cbabc9b2c655f
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
AFAICT with GSSAPI the normal workflow is to run kinit or similar and
authenticate before running programs relying on it. Therefore
we can try to get the credentials before we choose whether or not
to use Negotiate.
Pick-to: 5.15
Task-number: QTBUG-85123
Change-Id: If0478fdd45389b2939ad87c2f582776fe56959bb
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
-no-iconv configure parameter was removed and various changes to the
configure parameters.
Change-Id: I8716388b20846cdd1ba63a715308813a312dea65
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This is so that they are in line with the "value_type" usually found in
containers. Associative containers have "key_type" and "mapped_type" and
we will use those names for access to elements in QMetaAssociation.
Using "value" as name for sequential containers improves consistency.
Change-Id: I628b7e1446bb2d56843b843bca72d279a6b247e6
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Introduce a smart pointer that saves its constness, even if the
constness of the parent object changes. This allows us to implement the
const and mutable iterators in the same class.
Change-Id: I156e564c7fa2ecc8981c6d71861d796a5cbfec66
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
The GSSAPI thing is a bit noisy and not useful unless you're
debugging it specifically.
Pick-to: 5.15
Change-Id: I4a8c14159ec889776d06e0970ddf66083d788b63
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
In QAuthenticator "detach()" does not do what you expect.
First off it doesn't detach at all, and secondly it will reset the phase
that the authentication is in. This last part is intended, but it has
one issue: if setUser/setPassword is called with the same arguments
every time we ask for credentials then we never reach a fail-state since
it thinks we will have a new chance to authenticate.
Pick-to: 5.15
Change-Id: I02e2e42242220f3fced3572323e6492429cf173e
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Our authentication code is race-y by design:
1. When two requests are fired off and queued at the same time in the
same QHttpNetworkConnection then if one of them encounters
"authentication required" then it will copy whatever credentials it got
to all the other channels in the connection. This is likely what the
first part of the test is testing.
2. If a later request is fired off and it includes credentials in the
url then the newly included credentials should be used instead of the
cached ones.
The race here can occurr when one socket either takes too long to
connect or the connected signal is not received early enough. Then the
first socket is used for both requests and then we can hit case #2 when
the url contains credentials.
Pick-to: 5.15
Change-Id: I646a5378d8c1256b2de98b51912953df29f68cb2
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
It's problematic if we have already used these credentials and it
failed. Since then we emit authenticationRequired only for QNAM to
intercept it and "helpfully" suggest we use the same credentials again.
By moving on we can check the cache or ask the user about the necessary
credentials.
Pick-to: 5.15
Change-Id: Idaac5ae71167462369b66194ab6b1f77113d636a
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Although this is more of a QNAM thing.
In some cases what we load from the cache might be credentials we have
already tried (and failed with, thus leading us to emit the
authenticationRequired signal). With this patch we will fall through
more often and ask the user for credentials.
Pick-to: 5.15
Change-Id: If2a556883c3ea5b0b225f4df273d38353b552b54
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
This configure switch sets the CMake variable
CMAKE_INTERPROCEDURAL_OPTIMIZATION_<CONFIG> per release config to ON.
The feature 'ltcg' is enabled if any of the variables
CMAKE_INTERPROCEDURAL_OPTIMIZATION,
CMAKE_INTERPROCEDURAL_OPTIMIZATION_<CONFIG> are ON.
In order to implement the check, configurejson2cmake had to be extended
to be able to write extra CMake code before and after the feature
definition. This extra code can be added to a feature mapping below the
keys "cmakePrelude" and "cmakeEpilogue".
Task-number: QTBUG-85373
Change-Id: Ia2eb907edcf087f137977a9b090705397f83eb05
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>