... in preparation for replacing the QString keys in the public API
with QAnyStringView ones.
This removes the "important optimization" that avoids a detach in the
common case where the input is the same as the output of
normalization. But that optimization is beside the point, because it
trades a memory allocation avoided in the library for O(N) allocations
inserted into user code for each call to QSettings::value(), the vast
majority of which are calls with string literals.
With the public interface ported to QAnyStringView in the follow-up
patch, we can then internally optimize memory allocations _in a central
place_ (e.g. by returning std::u16string or QVarLengthArray<QChar> from
normalizeKey() instead of QString). But first we need to get rid of all
the unwarranted allocations in user code.
Change-Id: I45fc83d972c552a220c9c29508001d3f172e1162
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Sona Kurazyan <sona.kurazyan@qt.io>
Introduced in commit 705ea62f9e, but
apparently these constructors aren't used anywhere yet.
Change-Id: Ic30914a4448d4bc28974fffd16dc521ab49d3be4
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
This way, the visitor implementations are directly where they are used,
without the need for a virtual function call. The allocation of the
visitors on the heap was anyway overkill (they could just have been
allocated on the stack where they were used).
Change-Id: Ic41fd285e3a45daaf2e17aa5aeee07e754127a00
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Selection change handlers of the items might call a method that
implicitly recreates the selectedItems QSet, which then invalidates
the iterators of the ranged for loop, resulting in crashes.
Iterate over a copy of the set instead.
Add a test case that crashen without the fix.
Fixes: QTBUG-101651
Pick-to: 6.3 6.2
Change-Id: I6da6f4043fe1906b0186931a37283f635cb5a404
Reviewed-by: Axel Spoerl <axel.spoerl@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
Add move constructor and move assignment operator.
To allow for the the change to be implemented QTemporaryDir, the
internal d_ptr was modified from a QScopedPointer, which is not movable,
to a raw pointer.
Add member + free swap implementations.
Add tests.
[ChangeLog][QtCore][QTemporaryDir] Enabled move semantics.
Change-Id: I9f196a77c70b4ca0b7f0c06505d00fdd87a9785c
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
It was added in 375dd7ad34
but never used.
Pick-to: 6.2 6.3 5.15
Task-number: QTBUG-101883
Change-Id: Ibad7e0b1c371bdee63bcdbea0c5280091107408a
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
ClipboardManager didn't have any APIs to clear clipboard data before
API level 28.
As a workaround an empty Intent with MIMETYPE_UNKNOWN is created and
inserted into the clipboard for lower API levels.
This makes the QApplication::clipboard()->clear() method work more or
less as expected.
This allows to unblacklist tst_QPlainTextEdit::copyAvailable().
Task-number: QTBUG-87423
Task-number: QTBUG-89402
Pick-to: 6.3 6.2
Change-Id: I454376199cf3b8eed0fa2ecf2f85b87f40892280
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
Because setObjectName() now takes a QAnyStringView, passing C string
literals is much more efficient than using QString::fromUtf8().
This patch only deals with setObjectName() as a known property,
probably from a name attribute. If the objectName property is set by a
<property> element (or so I assume), we still emit a
QString::fromUtf8(). Detecting objectName there would be too much
magic.
Besides, I haven't been able to find why sometimes there's a second
setObjectName call right after the first, when in the XML it seems all
kosher (name attribute and objectName <property> agreeing on the
content).
Change-Id: Icf07bad841ac735a9c744bbd955428ba15758089
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
In a toplevel build, we include qtbase's .cmake.conf twice: Once for
qtbase itself, but already earlier to provide some necessary information
for the toplevel build. We now detect the latter, and skip setting
QT_EXTRA_INTERNAL_TARGET_DEFINES in that case, to ensure that it is only
applied to the qtbase itself
Task-number: QTBUG-101640
Pick-to: 6.2 6.3
Change-Id: I14dbf6110c79331fac38cceb76f669ca329ad33a
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Improve the docs for the recently added move operations.
Change-Id: Iae17f5289d30faab4aaaec14a27036f7a03129ab
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Parent window in sizeLessWindow test was using software pipeline at
least on webOS, causing a QCRITICAL() exit when the test was executed.
Fixes: QTBUG-101869
Pick-to: 6.3
Change-Id: Ie8cc0ea0b6a09729475389be44197e01a45b217f
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
When CMake generates compilation rules, it extracts the values from
INTERFACE_INCLUDE_DIRECTORIES and checks if any of the values
are framework paths.
If they are, instead of adding an -I ./lib/My.framework to the
compilation rules, it adds -iframework ./lib or -F ./lib.
The same transformation does not happen when AUTOMOC passes include
paths to moc, nor during a headersclean check. The values there
are passed verbatim, with an -I prepended.
This causes issues when the include file name is the same as the
framework name.
E.g. #include <QtQml> + -I ./lib/QtQml.framework because moc
then ends up silently including the shared library
./lib/QtQml.framework/QtQml
instead of the header
./lib/QtQml.framework/Headers/QtQml
This can lead to a variety of silent issues during moc generation,
because all the definitions of QtQml will be missing.
Unfortunately, there does not seem to be a clean way to fix this in
the build system due to CMake semantics.
See https://gitlab.kitware.com/cmake/cmake/-/issues/23337 for details.
We can mitigate the issue by ensuring that
-I ./lib/QtQml.framework/Headers
comes before
-I ./lib/QtQml.framework
by manipulating the order of values in INTERFACE_INCLUDE_DIRECTORIES.
We might want to consider implementing an additional mitigation in
AUTOMOC, so that it filters out include paths like
-I ./lib/QtQml.framework, thus ensuring that a newer CMake version
will not exhibit the same issue when used with an older Qt.
We could consider doing the same in moc. The advantage of doing it
in moc is that that moc will consider fewer invalid include paths
when searching for headers.
Amends 4b2de41b13
Amends d7efb2a419
Pick-to: 6.2 6.3
Fixes: QTBUG-89545
Fixes: QTBUG-101718
Fixes: QTBUG-101775
Change-Id: Ib2c25b5744bd2b5c9c83813bb04ad88c0179f6ec
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
When building Qt we should have warnings of any internal use
of deprecated API, so set it to Qt7.
Also added comments to clarify what the macros do.
Change-Id: Ib47278fed9ab1ec4411ed9f69a9a9c0f811db02d
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Unlike the Gui and Widgets pointers, the Core one wasn't needed,
because the only user can inline the variables contents.
Saves one relocation.
Pick-to: 6.3
Change-Id: Ib7535e1a29c5125eea55c66dad6d2d1de72f6bc1
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
There is no reason to allow relock() on a locked locker, or unlock()
or an unlocked one, just like we don't allow that on a plain mutex
to begin with. The docs already said that e.g. relock() locks an
_unlocked_ locker.
[ChangeLog][QtCore][QMutexLocker] QMutexLocker allowed relock() and
unlock() on an already closed (resp. open) locker object. These
semantics have always been undocumented and are now unsupported
(in both cases they yield undefined behavior.)
Change-Id: Id5f67beb5dc30d6435dae88a3085fba93ec7d96e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Percent-decoding was previously only present as a static method taking
a QBA parameter; it might as well be an instance method of that
parameter. Change most QBA tests to use it rather the static method.
[ChangeLog][QtCore][QByteArray] percentDecoded() is now available as
an instance method of the byte array to be decoded, equivalent to the
static QByteArray::fromPercentEncoding().
Change-Id: I982101c44bdac5cc4041e85598d52ac101d38fa1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It is private and was not called by anything inline, even in 5.15, so
can safely be removed without causing binary incompatibilities, now
that it is no longer used even out-of-line.
Task-number: QTBUG-74286
Change-Id: Iaec3877079746bd9ba7d1a39efbfb2711e7c00a5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
After inlining the local static void to which it used to delegate most
of the work, the main loop can access *this and we have no need for
yet another copy of it or the variables to iterate it. We can also
access the result array directly, rather than via a (poorly-named)
pointer.
It turns out that wrapping QBA in QBAV each time we call a lambda on
it costs enough to cause a 50% slowdown, so change the lambda to now
take the const refs we have. (Constructing the views only once is as
good, for reference.)
Change-Id: I81c5996d91415d8a933de714177a89462555ff03
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This fixes issue with non working QHash when loading QML plugins in QNX.
Change-Id: I55c9edc58aa27cff694603e6281c980e1fedbeba
Taks-number: QTBUG-101341
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When trying to match a font request to fonts in the database we might
end up with a mismatch due to the style not being available, but the
font family itself was.
If that's the case there's no point in trying to populate font aliases,
so we explicitly propagate this fact back to QFontDatabasePrivate::findFont
so that it can choose to not populate family aliases.
Fixes: QTBUG-98369
Fixes: QTBUG-99216
Pick-to: 6.3 6.2 5.15
Change-Id: I14470554c73eace836f57cb65e63ada594ccf62e
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The class is similar to unique_lock in that it allows for unlocking
and relocking. Since the locked state is tracked by QMutexLocker itself,
it's trivial to make it movable.
[ChangeLog][QtCore][QMutexLocker] The class is now movable.
Change-Id: I534044f8024575e996c12efb2236761d493798a3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Forward declare more types in qvariant.h, and do the same for qdebug.h
(which will be removed from qvariant in a separate patch).
As we now only forward declare containers (e.g. QMap, QList, ...), code
which includes <QVariantMap>, <QVariantList>... breaks without further
adjustment: So far, those headers simply included qvariant.h. However,
by introducing new actual headers for those types and adjusting
sync.profile, we can avoid this issue.
To avoid breaking leaf modules and user code, we make the change opt-in:
Unless QT_LEAN_HEADERS is defined to a value of >= 1, we still
include the superfluous headers. We also set this macro in qtbase via
QT_EXTRA_INTERNAL_TARGET_DEFINES.
[ChangeLog][Potentially Source-Incompatible Changes] qvariant.h no
longer includes <QList>, <QMap>, <QHash>, <QVarLengthArray>, <QSet> and
<QObject>. Code that relied on transitive includes might need to
explicitly include those headers now. Notably, including <QVariant> (or
qvariant.h) is no longer sufficient when using QVariant(List|Map|Hash).
Using them now requires including respectively <QVariantList>,
<QVariantMap>, or <QVariantHash>. Alternatively, including <QVariant>
and the <QList>, <QMap> or <QHash> header also works.
[ChangeLog][Potentially Source-Incompatible Changes] qdebug.h no longer
includes <QSet>, <QHash>, <QVarLengthArray> and <QMap>. Code that relied
on transitive includes might need to explicitly include those headers
now.
Task-number: QTBUG-97601
Change-Id: I142e5de709ed0b305716369a3266389ab7fbbb71
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
As a drive-by, remove superfluous includes from qnetworkmanagerservice.h
and obey the coding conventions for includes in a few more places.
Change-Id: I65b68c0cef7598d06a125e97637040392d4be9ff
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
As a drive-by, follow the coding convention when including qobject.h
Change-Id: I73066646d7f03ee0138e70567738e296b2183f63
Reviewed-by: Liang Qi <liang.qi@qt.io>
gcc 10.2 struggles with returning just {} when compiling in release mode
with debug symbols. std::nullopt is also more readable.
Change-Id: Ib050d47e8741ee87bef865dd55ecba622faf25b4
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
Add a new entry to the Call enum that can be used by dynamic meta
objects to encode their own metaCall variations.
Change-Id: I7890d11924ab99311b6f016ce97d2ab483c26d27
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Connecting to nullptr, or connecting to a non-signal PMF, would result
in a QFuture which would never finish. Catch these cases and handle
them.
Windows+MSVC for some reason fails the test. I can't entirely understand
why, so I've marked it as XFAIL, with QTBUG-101761 to track it.
Change-Id: I314980e7e9b7156d8cddd3b33d5cbf1d0bcd6116
Pick-to: 6.2 6.3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
We want to use configure and qt-configure-module when building in
Coin, rather than pure cmake and qt-cmake.
There are a few benefits:
- CI tests the scripts, making sure we don't introduce regressions
- CI uses the same scripts that we mention in our documentation for
developers to use
- The platform configurations become a bit less verbose and easier to
read due to less shouty-case CMake options
To ensure a more gradual porting, Coin will only use the new
instructions if the 'UseConfigure' feature is set on a platform
configuration in qt5.git. This allows going back to the old
instructions in case if something isn't working properly.
Due the opt-in, we need to support both old and new instructions in
the implementation.
The change strives to remove as much duplicate code as possible,
by moving it into common includes.
The README.md is updated to mention the overview of how the different
environment variables are used.
There are a few important things to point out.
1) Because during the porting we have to allow mixing of the old style
and new style, platform configs have to separate CMake-style options
from configure-style options in different environment variables.
Otherwise the instructions wouldn't be able to create a valid
configure call, where all CMake-style options have to go at the end
after a double dash --.
After all platform configs are ported to the new style, it should be
possible to combine all the options in a single environment variable
if that is desired, but it will require another round of porting to
remove all the '-D' prefixes in CMake-style options, and just use
regular variable assignment which configure supports.
e.g. -DQT_BUILD_EXAMPLES=ON becomes QT_BUILD_EXAMPLES=ON, which can be
mixed in-between configure-style args.
2) Configure is more strict in that it doesn't allow passing
unknown options. Due to that, we can't pass non-qtbase configure
options via NON_QTBASE_CONFIGURE_ARGS. qt-configure-module would
error out in the repos where the configure option is unknown.
Because we don't have a Coin configure variable for each repo,
we circumvent the issue by continuing to pass CMake-style options via
NON_QTBASE_CMAKE_ARGS instead, which does not do validation checks.
In the future, we could introduce a configure flag that disables
the validation checks.
Pick-to: 6.2 6.3
Task-number: QTQAINFRA-4357
Task-number: QTQAINFRA-4815
Change-Id: I72d8ba0b3a543b42982e22ae8d6566c0e885c446
Reviewed-by: Jörg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Toni Saario <toni.saario@qt.io>
This part originates from when the CMake was being introduced and
now things are stable enough that the bug rarely is everywhere.
Pick-to: 6.2 6.3
Task-number: COIN-828
Change-Id: Ifb7b5ce82740f5ae49f712f9666870993ce00b15
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Conditions with the LINK_LANGUAGE genex are not parsed correctly when
generating .pc file. So link options will be added unconditionally.
Amends d1e02c3855
Pick-to: 6.2 6.3
Fixes: QTBUG-101723
Change-Id: Ib837b3f3429d195a469450ef25af9630ad7d15e2
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Use VERBATIM option to prepare the correct command line for the
add_custom_command. This especially sensitive when using build
directories with names containing special symbols, that cannot be
handled by shell correctly.
Change-Id: I51d7041cb806411135fd59bf6273c04a3c695443
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
QString already has a localeAwareCompare(QStringView, QStringView)
static method. Use it in QStringView::localeAwareCompare(QStringView).
[ChangeLog][QtCore][QStringView] Added
QStringView::localeAwareCompare().
Task-number: QTBUG-98431
Change-Id: Iec3865a5439d9fb653cc1150da21042186bdee98
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Marc Mutz <marc.mutz@qt.io>
On Wayland, there is no protocol to do it yet.
Task-number: QTBUG-100792
Task-number: QTBUG-101145
Change-Id: Id72625a8161359111a2d0f43f3215e787778ba6b
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
