Consider a feature that has ENABLE and CONDITION expressions. For
example, system-libb2 on a Linux machine that does not have the
libb2-dev package installed:
qt_feature("system-libb2" PRIVATE
LABEL "Using system libb2"
CONDITION Libb2_FOUND
ENABLE INPUT_libb2 STREQUAL 'system'
DISABLE INPUT_libb2 STREQUAL 'no' OR INPUT_libb2 STREQUAL 'qt'
)
If the feature is enabled by the user by setting
FEATURE_system_libb2=ON, then there's an error:
Feature "system_libb2": Forcing to "ON" breaks its condition...
This is expected.
If the feature is enabled by the user by triggering the ENABLE condition
via the -system-libb2 configure flag (that sets INPUT_libb2 to
"system"), we did not get an error message.
This was unexpected and inconsistent.
Now, we consider features that are enabled via their ENABLE condition as
set by the user and yield an error if the CONDITION doesn't match.
Fixes: QTBUG-126144
Change-Id: I249baf9ab4f434f5a96894f4ec055f70859ffca9
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
(cherry picked from commit ad227368c3ab3a9dc68208bce2d075edefe7e6e0)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The test checks if paths in /etc and /home are properly completed.
/etc and /home do not exist on VxWorks.
Use /tmp instead.
Pick-to: 6.7
Task-number: QTBUG-115777
Change-Id: I6203fabb003a9a81eb4cdb666a972f47a53f06d6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 843f8fb043a0812ab01f0a378be1a821f2e1534d)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The rationale given for the addition of these functions was that they
are a more efficient way to check these things in public API.
If so, they shouldn't cause compilers to insert exception handling
around their calls, so mark them noexcept.
This is safe, as both conceptually and practically, these functions
should not and cannot fail, so we'll never need to remove noexcept
again.
Found in API-Review.
Amends 7a374c38d2 and
a3d50112e4.
These two commits also lacked each the changelog entry, so adding them
here:
[ChangeLog][QtCore][QThread] Added isMainThread() static member
function.
[ChangeLog][QtCore][QThread] Added isCurrentThread() member function.
Change-Id: Iaf39ce2cc0abd45049bff60b24693e84bf76d9e0
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit d73a2bf0fb3e0a00691c858645a5a5bb2b55b40a)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
There was a small mistake in:
d2ae5ef51e3339c11b1c0ed0b09f9d17b9708f53 commit, which caused a
regression in the input method.
While the function parameter name has been changed, the assignment name
has not. This resulted in the values never actually getting updated.
Fixes: QTBUG-125410
Pick-to: 6.7
Change-Id: I8e64e991d02546ddd078fd273f0b57491c63b55c
Reviewed-by: Tinja Paavoseppä <tinja.paavoseppa@qt.io>
(cherry picked from commit 79e00b0b25765b85b112b601fee70184b1efe3c0)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Extending headersclean_check to private headers has shown that
Q_DECLARE_PUBLIC needs to know the Public class in-size, so a
forward-declaration is not enough. Says Clang:
qfuturewatcher_p.h:31:5: error: static_cast from 'QObject *' to 'QFutureWatcherBase *', which are not related by inheritance, is not allowed
Q_DECLARE_PUBLIC(QFutureWatcherBase)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Amends the start of the public history.
Pick-to: 6.7 6.5 6.2 5.15
Task-number: QTBUG-126219
Change-Id: I1e132aaa844f5c87d57446e8818280fd141d2b1e
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit b58d926090620079f602f4f2ce9f17fe7025c962)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Replace code that manually operates on jarray with QJniArray.
The comment about using separate jobject and QJniObject didn't make
sense; we didn't construct the QJniObject via fromLocalRef(), so
allocated a new local reference anyway, in addition to the local
reference returned by GetObjectArrayElement.
The implicit code now creates a QJniObject (-like object;
QtJniTypes::String contains a QJniObject) as well, but it's also
just a temporary. It would be cheaper to iterate over a
QJniArray<jstring>, but only if we then duplicate the code from
QJniObject::toString to not create an intermediate QJniObject.
Change-Id: I9f10541f533dd2fbd2f7ba6fdacc7d79b3ac3ae9
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
(cherry picked from commit 64127a7a0db657cbc844938f02b623f90cdd78ee)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
This reverts commit f8359084b9.
Lookup-by-string functions should be all taking QAnyStringView, not be
overloaded on QString/const char*.
Found in API-review.
[ChangeLog][QtCore][QStringEncoder/QStringDecoder] The (name, flags)
constructor now takes QAnyStringView (was: const char*).
Fixes: QTBUG-126108
Change-Id: If68343ce5ec2ff60ce55156f9f2dc2bb6938e07f
Reviewed-by: Ahmad Samir <a.samirh78@gmail.com>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
(cherry picked from commit d4c4e6b876b948c1a34abdd46cf7f5be1c302137)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
These kinds of lookup functions are no-brainers for QAnyStringView.
This is required to replace the QStringEncoder/Decoder constructors
from QString, newly added for 6.8, and found in API review to be
better written using QAnyStringView.
Because this is API-Review-relevant, keep the implementation
pessimized. It's not worse than the code that it replaces (toLatin1()
on an unshared QString performs in-place conversion), and can be
improved later.
Task-number: QTBUG-126108
Change-Id: I4a0d1840d4754670b2e2f20d768ed98f7e1937d5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit f6c11ac4f20a16d0b2113014e2dac63b95d946ae)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
These kinds of lookup functions are no-brainers for QAnyStringView.
This is required to replace the QStringEncoder/Decoder constructors
from QString, newly added for 6.8, and found in API review to be
better written using QAnyStringView.
Because this is API-Review-relevant, keep the QStringView case
pessimized. It's not worse than the code that it replaces (toLatin1()
on an unshared QString performs in-place conversion), and can be
improved later.
[ChangeLog][QtCore][QStringConverter] The encodingForName() function
now takes QAnyStringView (was: const char*).
Task-number: QTBUG-126108
Change-Id: I02cc53bfbc8f1eebe539816487a10b39e96fac2d
Reviewed-by: Ahmad Samir <a.samirh78@gmail.com>
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit b3edce58e5aea76d1d84c868f76171c2b40520a4)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
We've had this workaround since Qt 5.4. It's due time we get
rid of it. This commit deprecates it, in preparation for removal.
[ChangeLog][QtCore][QExplicitlySharedDataPointer] Support for
QT_ENABLE_QEXPLICITLYSHAREDDATAPOINTER_STATICCAST has been
deprecated, and will get removed in a future version of Qt.
Change-Id: I3000ea606b37714542916e105ebd50871dc42935
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 0ed039fd134fb8025236847295681f13f1775847)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
On macOS, we see unexplained white flashes of the UI when
overriding the scheme in the widget's constructor. This only
happens when running the bundle, but not when running the
executable.
Work around that problem by setting the scheme immediately
after constructing QApplication.
Task-number: QTBUG-126248
Change-Id: I4a5fe467d628fca5e52e1e36f43af8143239c7fa
Reviewed-by: Doris Verria <doris.verria@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
(cherry picked from commit 88322f69a20fddedd9c45a1bb25f6c1d2dcf31f9)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
That way, an override through QStyleHints::setColorScheme is
immediately respected by the style.
As a drive-by, make the helper static and inline.
Change-Id: Ia6cde1f5099d9194b25fac812c3184f5a61bf1dc
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
(cherry picked from commit 4245987f6875d01853053c50d13ecf42b61fc0ad)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Simplify the relevant QJniObject calls to not use callObjectMethod,
but rely on the return type that we have to pass in anyway. Replace
manual loop unwrapping with QtJniArray, and declare the QtNetwork
Java class so that we can use it directly.
This removes the last usage of Q_DECLARE_JNI_TYPE in qtbase.
Change-Id: If872684c8b52bc7789e248cfc3046817c411903c
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 909493bae27eeb55c985aa9522a57bc4e60d0893)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
If the forward argument was meant to be 0, it was not forwarded
because it evaluated to false in if(condition).
To pass it along, instead check for empty strings.
Change-Id: Ia366df147de0c2d333017da43dc0643b56a89e9c
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
(cherry picked from commit 96abceb64e5dc0570ca7c3419f401cfafe946ba0)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
When QGuiApplicationPrivate::processTouchEvent() sees that the
touch event was not handled, and calls processMouseEvent(), the latter
uses the QEventPoint with pointId 0 regardless of the original
touchpoint ID. Now it updates the persistent QEventPoint from the
original touchpoint so that a double-click event will not be ruled out
because of the timestamp delta or position delta (movement since press)
being too large.
Fixes: QTBUG-125993
Pick-to: 6.7 6.5
Change-Id: I8e9b007818107ac2329454e0ccfb2ac9e506b617
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit 2a0b907f11b9c0ad46322ba06482861423246d93)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The docs claim that this is the default, but in practice it's not
the case.
Change-Id: I5f6184cbfded2efedbac88023644d3c1e80e9901
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
(cherry picked from commit a05f5457ca033214c1efcf528b85a09a9c758920)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
We always create UIView via initWithQIOSWindow, so there is no
need to keep part of the implementation in initWithFrame.
Change-Id: I5311ed78fff3294ed49af70d482449cbf3b8211b
Reviewed-by: Doris Verria <doris.verria@qt.io>
(cherry picked from commit 29b6674d3f8d1012def9dbcf89521e4773175dc1)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The test checks if two paths test equal if they are pointing to the same
place, but one of them is not a canonical path.
Example of such paths: /tmp/.. and /
The test assumes that /usr is an existing directory on platforms other
than win, haiku and android.
The reason /usr is used instead of /tmp is that /tmp is a symlink on
macos and that interferes with the test (as described in 0b67ad9d).
However /usr does not exist on VxWorks (unless created by the user).
/tmp can (and is in our CI env) enabled by turning on the
INCLUDE_TMP_DIR component.
Solve the problem by using /tmp instead of /usr on vxworks.
Pick-to: 6.7
Task-number: QTBUG-115777
Change-Id: I98b0d46d1b879f404c0b5bd655031e7c3f19ba33
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 70e2fe2bba71ea0a4d1d1e533c1130c97f77df55)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The #if-ery on the removed API's definition still needs to match its
declaration. Spotted by Axivion.
Pick-to: 6.7
Task-number: QTBUG-122619
Change-Id: I4e3dc03291a8fb7fbc01065bce8e5dc58a051166
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>
(cherry picked from commit 76f82b8f1334d8a377630876ef2d9681c05d7ee8)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
From the API review.
Change-Id: I74140a9268670575488ddc3c31b09c84bc83eae0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit df85c6fd9d65abf44ccdcd1d6e6594cab3fc7d65)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
TSAN has the same issue as ASAN when it comes to the SSE2 usage here.
WARNING: ThreadSanitizer: heap-use-after-free (pid=621)
Read of size 8 at 0x721000049f78 by main thread:
#0 _mm_load_si128(long long __vector(2) const*) /usr/lib64/gcc/x86_64-suse-linux/13/include/emmintrin.h:700 (libQt6Core.so.6+0x2ba056)
#1 qustrlen_sse2 qtbase/src/corelib/text/qstring.cpp:431 (libQt6Core.so.6+0x2ba056)
[...]
Previous write of size 8 at 0x721000049f78 by main thread:
#0 realloc <null> (libtsan.so.2+0x5bdbd)
#1 QArrayData::reallocateUnaligned(QArrayData*, void*, long long, long long, QArrayData::AllocationOption) qtbase/src/corelib/tools/qarraydata.cpp:244 (libQt6Core.so.6+0x309913)
[...]
(gcc 13.3.0)
Pick-to: 6.7
Change-Id: Id0127d5fb498dcf78bb977e6ff1b6b380ae423c7
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 600368279dafa60e788d8e3cd3af5c1915990fbb)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
In non-MSVC configurations this feature is meaningless.
Restrict it to MSVC.
Change-Id: Ib5111e8adf49a862d8ca0f191837d8c28aab6890
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
(cherry picked from commit 433411934c7ebb2a29e950c14871199166e5534a)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Prevents weak (= duplicated) vtables and the associated problems.
Amends the start of the public history.
Pick-to: 6.7 6.5
Task-number: QTBUG-45582
Change-Id: I26618483900437aca05c753a5d53a359e43bd702
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 645509935ad5a2e8a3fcc858c41d08919a2596bd)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
QIODevice is a base class of one of the classes defined in this
header. QtCreator's clangd goes on a rampage in this header as a
consequence.
Amends the start of the public history.
Pick-to: 6.7 6.5 6.2 5.15
Change-Id: I06c6a26dde7e7ae72b722136fcd666684733d40f
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 09fce4a587047ff03c7ca37dbb2bb8f71a18a75e)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
When building for 32bit Android:
qcolorclut_p.h:45:31: error: comparison of integers of different signs:
'qsizetype' (aka 'int') and 'unsigned int' [-Werror,-Wsign-compare]
Q_ASSERT(table.size() == gridPointsX * gridPointsY * gridPointsZ * gridPointsW);
~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fix by casting the result of the multiplication explicitly to qsizetype.
Change-Id: Ica9a2f9738959adfa841270ffdd4893bd7d1d4e8
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
(cherry picked from commit 37ba5e94d681c8d65b0e44a14b0e1595d072800e)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Found in API review.
Change-Id: I873c8c0b7e9521d07718b7fd38065d33b98e3ba9
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit f0a8fe6cdf3ec5e8c9030fa581df1aad3b78d210)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
CMake somehow doesn't treat the access to the cache variable in
condition as the usage. Replace the check of the
CACHE{QT_USE_TARGET_ANDROID_BUILD_DIR} value with the check of the
CACHE variable definition and the check of the CACHE and non-CACHE
value equality.
Pick-to: 6.7
Change-Id: I07d940d457ebbc4ce3cefef2f73394009d52f367
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
(cherry picked from commit 69a661e88e6a7799799033030497086a9194caab)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The parameter buildPath comes from QDir:absolutePath(), which gives a
path that starts with '/' or a drive speficiation. When calling
getImportPaths on a Windows machine the path becomes something like
"/C://dev/qt5/...". As a result, androiddeployqt exists with a failure.
On Linux/macOS systems this was not a problem because // = /
Change-Id: I85376253055549344f06c9da9ebd67364e429112
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
(cherry picked from commit 776982a14e53e8661c3508c3d05e0efe538bc67e)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
The check was re-introduced in 7a9bc220c7
and moved to QtRootLayout in 6ff88f97a6.
It introduces an issue where the resizing of the application window does
not work if the its aspect ratio is different from the screen's aspect
ratio.
Fixes: QTBUG-126122
Pick-to: 6.7 6.7.2
Change-Id: Id6d26b247b9ca7a8d0e89e1fdea2d77db12a6cfe
Reviewed-by: Tinja Paavoseppä <tinja.paavoseppa@qt.io>
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
(cherry picked from commit a6b78e6f4dbbf688fc0b1d1487b86890b149d323)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
QChronoTimer is in Qt 6.8.
Found in API review.
Change-Id: I7f20958948cb635ccd1f4cef383aed810a188f12
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
(cherry picked from commit 25f2b7c32226fdabcb22fbf928bb0c058cd53022)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
So that it's not forgotten come Qt7.
Amends 19258608e9.
Change-Id: I323f2d2738a295faa1e118c3ea7c8a5b6c6d0ac7
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
(cherry picked from commit ef8e1e5861292514a676c5fed4d670fb6ecb887d)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
When a popup widget is shown, it's first raised(), and then made visible.
In this scenario, the logic in QIOSWindow::setVisible() to compute the
window level was too late, and we ended up showing the popup below other
popups.
Instead of moving the window level computation earlier, we remove the
member variable and just compute the level when needed, as we do in
the Cocoa platform plugin.
Fixes: QTBUG-126052
Fixes: QTBUG-126044
Pick-to: 6.7
Change-Id: If5dbcdcf0500aed57b725bb8eb636fe076459130
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
(cherry picked from commit 0e656164e640ff2aca00e82b51b0f1019fd7802f)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Clang's -Wimplicit-fallthrough is a little stricter than gcc's.
Change-Id: Idf933e19ab2d664885743e5db6c1fbb66c8b615e
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
(cherry picked from commit 05a139d5fb5eb3d652f24ef5ebe28c232b69c6a9)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Rename the predictate for checking whether a container type has the
required member functions and types to IfContiguousContainer (we
need size(), data(), and value_type members).
And since std::initializer_list always meets that criteria, remove the
constraint from the respective constructor.
Add an IfConvertible predicate to test whether the element type of
an existing array can be converted to the element type of the new
array.
Change-Id: I7e5ba31de9664088b027c277c068c948f2189238
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
(cherry picked from commit 468126d34a8c3a2fdd486768f935d13ebe565dd0)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Make sure that QByteArray does not detach when not needed and avoid some
copies by using QByteArray::fromRawData().
Change-Id: I4454a3113c6bd1fe30b404af091f5cc0f904f78a
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
(cherry picked from commit 54853352bbeff7ad3ee3b3aa3a90e45de5082e17)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Joerg Bornemann
Łukasz Matysiak
Marc Mutz
Bartlomiej Moskal
Volker Hilsheimer
Giuseppe D'Angelo
Jan Arve Sæther
Alexandru Croitor
Shawn Rutledge
Tor Arne Vestbø
Edward Welbourne
Mårten Nordheim
David Faure
Friedemann Kleint
Mate Barany
Alexey Edelev
Topi Reinio
Olli Vuolteenaho
Jan Moeller
Ahmad Samir
Tim Blechmann
Christian Ehrlicher