Fix build configuration for qdnslookup implementations for android as
UNIX is also set when building for Android. We should only have the more
specific conditions enabled.
Change-Id: Ib891df34e5569a20b03ce270e1eab080325549b9
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
In order for test lib to locate the file requested via QFINDTESTDATA, it
needs the build directory of the test (for example
$builddir/tests/auto/foo/bar) and __FILE__ expanding to a path to the
source relative to this build directory.
With ninja, __FILE__ is a path that is always relative to the top-level
build directory, not the per-test case one. Therefore the path
resolution in testlib fails.
To accommodate this, add_qt_test() now always sets QT_TESTCASE_BUILDDIR
as well as the newly introduced QT_TESTCASE_SOURCEDIR, which, as an
absolute path, removes the need to use __FILE__.
Change-Id: I16c2b0001e38162e6da9fdb1a61f4f8ce634fe46
Reviewed-by: Ville Voutilainen <ville.voutilainen@qt.io>
This commit introduces infrastructure work to allow static builds of Qt
to handle importing of plug-ins.
Change-Id: Ife0ca3ca7276ea8ec96fe0eb6adf934fad7620ec
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Need
qt_find_package(X11_XCB)
in src/gui/configure.cmake since we're using it in the file
qt_feature("xcb_xlib" PRIVATE
LABEL "XCB Xlib"
CONDITION QT_FEATURE_xlib AND X11_XCB_FOUND
)
Need
qt_find_package(XRender PROVIDED_TARGETS PkgConfig::xrender)
in src/plugins/platforms/xcb/CMakeLists.tx since we're using it in the
file
extend_target(XcbQpa CONDITION QT_FEATURE_xcb_native_painting AND QT_FEATURE_xrender
PUBLIC_LIBRARIES
PkgConfig::xrender
)
Use capital XRender in pkgconfig to be more consistent on how XRender is
called everywhere else
Change-Id: I403ead2cc123b08f741c5142f20db88987657ba8
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Now that ANDROID_EMBEDDED is not set by default anymore, we start
descending into Android specific sub-directories. This patch generates a
stub file but the Java bits still need to be ported.
Change-Id: Icbfc3ad2c1ef266980161bca081c7c4275873efd
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We used to copy a qconfig.cpp.in file verbatim.
Add some plumbing to actually compute the strings and their lengths
as it is done in qtbase/configure.pri.
Also make sure to replace the hardcoded linux mkspec with one that is
automatically determined.
Of course both the detection of the mkspec, and the hardcoded strings
for include, lib, etc. should be fixed in the future.
This is a stepping stone to allow building a Qt application using
the qmake built by CMake.
Change-Id: I2e6754f44b20b09b09d14fd85785d56288e6517b
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Generate module .pri files
* Generate qconfig.pri
* Propagate MODULE_CONFIG from the .pro files
This enables the basic use-case of simple application builds that for
example use the moc. Omitted from the patch is support for private
module configurations, prl files (should we do this?) and possibly more
hidden gems that need to be implemented to for example support building
Qt modules with qmake.
Task-number: QTBUG-75666
Change-Id: Icbf0d9ccea4cd683e4c38340b9a2320bf7951d0d
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Tobias Hunger <tobias.hunger@qt.io>
Replace manual memory management with unique_ptr and manual
re-implementations of QMarginsF with The Real Thing™.
The only noticeable difference should be that the checks for 0 now use
qFuzzyCompare() (by way of QMarginsF::isNull() and its operator==).
[ChangeLog][QtWidgets][QGraphicsWidget] Added QMarginsF overloads of
setContentsMargins() and setWindowFrameMargins().
Change-Id: I6b3bb87015fa52fdde245b7528cca4b8f9ce41e1
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@qt.io>
So we add the QTextDocument::ResourceType::MarkdownResource enum value to indicate
the type. QMimeDatabase is generally unable to detect markdown by "magic", so we
need to use the common file extensions to detect it (the same extensions as declared
in the mime database XML).
Change-Id: Ib71f03abd535c17e5a8c99bd92d0a6062e972837
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Since its initial implementation, QEventDispatcherWin32 manages a
delivery of the posted events from the window procedure through
WM_QT_SENDPOSTEDEVENTS message. That makes the implementation quite
difficult and unclear. As a result, posted events get stalled,
in case of using nested event loops or other recursion.
The proposed solution is to send posted events at the beginning of
processEvents() only once per iteration of the event loop. However,
in case of using a foreign event loop (e.g. by opening a native
modal dialog), we should leave the emission in the window procedure,
as we don't control its execution.
Task-number: QTBUG-74564
Change-Id: Ib7ce85b65405af6124823dda1451d1370aed9b1a
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
At the moment, Coin builds tests as a separate qmake invocation
against an installed Qt. We need to support the same with CMake.
Change the tests subdirectory to be a standalone CMake project when
CMake does not detect an existing QtTest target while processing the
subdirectory. If the target exists, it means we are building the whole
repo, if the target does not exist, we need to call find_package
to find the installed Qt.
Refactor and move around a few things to make standalone tests build
successfully:
- add a new macro to set up paths to find QtSetup
- add a new macro to find all macOS frameworks
- add a new macro to set up building tests
- add a new macro that actually builds the tests
- export the INSTALL_CMAKE_NAMESPACE value into the BuildInternals
Config file
- export the CMAKE_BUILD_TYPE value, because a test project doesn't
have a .git subdir and thus defaults to be built in Release
mode, even though qtbase might have been built in Debug, so to
avoid the mixing, the propagate the build type
- stop overriding INSTALL_CMAKE_NAMESPACE and
QT_CMAKE_EXPORT_NAMESPACE inside QtSetup if they are set, because
the tests project doesn't specify a major version, and if we
override the values, the moc / uic targets don't get the correct
major version prefix and configuration fails
Change-Id: Ibdb03687302567fe325a15f6d1cb922c76240675
Fixes: QTBUG-75090
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Some Linux distros might not have a double-conversion library. Because
it is such an essential library for QtCore, compile and use the
the copy of the library bundled with Qt.
Also change library name to be a proper target with double colons,
so that in case the library is not found for some reason, the CMake
configure step would fail, instead of failing at link time.
Task-number: QTBUG-74133
Change-Id: I9f3b4298ae6e952891a7a89541d46878176bf1ce
Fixes: QTBUG-75891
Reviewed-by: Liang Qi <liang.qi@qt.io>
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The state of the module is done, not deprecated, so we shouldn't
recommend users to move away from it in all cases. There's also
no direct replacement for the DOM API.
Fixes: QTBUG-70629
Change-Id: Ifaff9757234bd68a411a3da1403c57bbbcb94693
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Leena Miettinen <riitta-leena.miettinen@qt.io>
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
TSAN does not understand the futex system call.
Now, in QMutex and QSemaphore usage, futex is always wrapped by
atomic operations that always do an acquire (before waiting) and
a release (before waking waiters). That alone realizes a
synchronizes-with, and since Qt uses std::atomics, TSAN knows
what's going on and does not complain.
But what if one uses futex directly, or we change the
algorithms, or introduce some other new synchronization
primitive somewhere? Luckily TSAN offers annotations for this
that we can use.
This patch annotates the main entry point for the futex syscall
with a pair of acquire/release semantics. A futex call
guarantees total ordering on the operations on the futex
word(s), whether the call succeeds or fails.
Change-Id: Ib80ff898c09fbb6fc73989247eb757bf70971a8a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The noexcept part should be clear. The constexpr should make the
compiler not create initialization code for static instances. This is
how std::mutex works[1], and Clang and GCC understand it. MSVC
doesn't.
[1] https://en.cppreference.com/w/cpp/thread/mutex/mutex
Change-Id: If9e4c6c0279f0024a659d5602d55e126224ca950
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Fixing this in Qt 5 would be BIC. Although it could be okayish, there
are many more classes that shouldn't be copyable. I need to go through
clazy's output and fix them, and would rather do this noise for Qt 6 and
leave Qt 5 alone and purely BIC free.
Added a move-ctor and move-assign, as well as swap(), deprecated
copy-ctor and copy-assign.
The new copy special member functions warn at runtime if they are called.
In order to not pollute client code with the warning strings, lock them
away by defining the functions out-of-line.
[ChangeLog][QtCore][QBasicTimer] QBasicTimer is now a move-only class.
Copying is now deprecated and will be removed in Qt 6.
[ChangeLog][QtCore][QBasicTimer] Added swap() member and free function.
Change-Id: Ic3e6a26f3989d4c8d125c06e8b0b825411c6e106
Done-with: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
It seems like an optimization on the itemsDirty flag
caused a bug to be re-introduced. When a popup is shown
on a new screen, the itemsDirty must however be set to
ensure that new correct sizes are calculated.
Task-number: QTBUG-59794
Change-Id: Ifb5c233b1f9d4d38bd0cd7a9a71cc32ad3212f8c
Reviewed-by: Morten Kristensen <msk@nullpointer.dk>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Christian Ehrlicher <ch.ehrlicher@gmx.de>
this also optimizes network post method handling
Task-number: QTBUG-75660
Change-Id: Ibb0d01f2cc2b2bc7802598c4f6f04b04882c12ca
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
When runing the scenegraph example sgengine, gets rid if this error:
QSGContext::initialize: stencil buffer support missing, expect
rendering errors
Change-Id: I7f3a82409bc2cf81cf8217876e527f9c45be3bf4
Fixes: QTBUG-74694
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
It doesn't make much sense to look for X11 related packages on
macOS and Windows by default. Usually they would not be there, and
as a result the configuration step would show a long list of scary
not found packages, and also eat precious configure time.
Change the conversion script to allow putting conditions around
generated find_package calls.
These conditions can be manually set in the conversion script
library mapping, using the emit_if argument,
which we do for the X11 and Wayland related packages.
They are also computed by checking which features use a given library,
and if the feature is protected by a simple emitIf condition like
config.linux, the relevant library find_package call will be protected
by the same condition.
If a developer still wishes to look for all packages, they can define
the CACHE variable QT_FIND_ALL_PACKAGES_ALWAYS to ON.
The relevant configure.cmake files are regenerated in this patch.
Change-Id: I6f918a94f50257ec41d6216305dae9774933389a
Reviewed-by: Tobias Hunger <tobias.hunger@qt.io>
To implement this, create a new Qt5BuildInternals package.
All child Qt modules like qtsvg should use
find_package(Qt5BuildInternals) or
find_package(Qt5 COMPONENTS BuildInternals) in the their
top level CMakeLists.txt.
This will make the qt_build_repo() macros available.
For qtbase we slightly cheat, and specify a CMAKE_PREFIX_PATH
pointing to the source folder that contains the BuildInternals
package.
For the other modules we actually use a configured and installed
package Config file.
This change moves variables that used to be written into the
QtCore Config file into the BuildInternals package. This way
things that are relevant only for building additional Qt modules
does not pollute the QtCore package.
Task-number: QTBUG-75580
Change-Id: I5479adff2f7903c9c2862d28c05c7f485ce3e4eb
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Currently to build qtsvg we have some copy-pasted code to set up
the paths for QtSetup and QtPostProcess to be found.
To make it cleaner, introduce two new macros called
qt_build_repo_begin and qt_build_repo_end(). The first one
should be called in a child repo like qtsvg, right after
a find_package(Qt5) call, and the second one at the end of the
repo top-level CMakeLists.txt file.
In order for the macros to work, extract some of the variables
which were set in Qt5Config into Qt5CoreConfig instead. This
makes sure that it works also for find_package(Qt5Core) calls.
Task-number: QTBUG-75580
Change-Id: I85267c6bd86f9291ec2e170fddab1006ab684b5c
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
A non-prefix build is a build where you don't have to run
make install.
To do a non-prefix build, pass -DFEATURE_developer_build=ON when
invoking CMake on qtbase. Note that this of course also enables
developer build features (private tests, etc).
When doing a non-prefix build, the CMAKE_INSTALL_PREFIX cache variable
will point to the qtbase build directory.
Tests can be run without installing Qt (QPA plugins are picked up from
the build dir).
This patch stops installation of any files by forcing the
make "install" target be a no-op.
When invoking cmake on the qtsvg module (or any other module),
the CMAKE_INSTALL_PREFIX variable should be set to the qtbase build
directory.
The developer-build feature is propagated via the QtCore Config file,
so that when building other modules, you don't have to specify it
on the command line again.
As a result of the change, all libraries, plugins, tools, include dirs,
CMake Config files, CMake Targets files, Macro files, etc,
will be placed in the qtbase build directory, mimicking the file layout
of an installed Qt file layout.
Only examples and tests are kept in the separate module build
directories, which is equivalent to how qmake does it.
The following global variables contain paths for the
appropriate prefix or non prefix builds:
QT_BUILD_DIR, QT_INSTALL_DIR, QT_CONFIG_BUILD_DIR,
QT_CONFIG_INSTALL_DIR. These should be used by developers
when deciding where files should be placed.
All usages of install() are replaced by qt_install(), which has some
additional logic on how to handle associationg of CMake targets to
export names.
When installing files, some consideration should be taken if
qt_copy_or_install() needs to be used instead of qt_install(),
which takes care of copying files from the source dir to the build dir
when doing non-prefix builds.
Tested with qtbase and qtsvg, developer builds, non-developer builds
and static developer builds on Windows, Linux and macOS.
Task-number: QTBUG-75581
Change-Id: I0ed27fb6467662dd24fb23aee6b95dd2c9c4061f
Reviewed-by: Kevin Funk <kevin.funk@kdab.com>
Reviewed-by: Tobias Hunger <tobias.hunger@qt.io>
... instead of qPrintable(), %s, and explicit qt_error_string().
Saves 2KiB in text size on optimized Linux AMD64 GCC 9.1 builds.
Change-Id: I98b6717da1ed1b678f01167d704a96f10da47966
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
... instead of qPrintable() and %s. Avoids temporary QByteArrays.
Saves 1.2KiB in text size on optimized Linux AMD64 GCC 9.1
builds.
Change-Id: Ie626fc478667007ce9a6bc920b8d4ec0451f2cd0
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
The first two changes avoid creation of a temporary QString and QByteArray each,
by realisiing that QChar is more-or-less wchar_t on Windows and so we can just
use %ls to print the wchar_t array directly.
In msgFindNextFailed(), remove the inline keyword and mark the function as cold
(not sure this has any effect on Windows). When building the result, don't use
QTextStream. Everything that is streamed is text, so just use QString::op+=.
When using the result, use qUtf16Printable and %ls instead of qPrintable and %s,
to avoid the creation of a temporary QByteArray.
Change-Id: I09f576b894761fe342109b386c1de3532200e03c
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Collapse the two qWarnings() into one by just storing what is different between them.
Saves more than 400b in text size on optimized AMD64 Linux GCC 9.1 builds.
Change-Id: I16489d6165a550a9ad4ce6a77ca736a1d17a8c8a
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
There are no QLatin1String overloads, so using QLatin1String as the
format string produced a temporary QString. Use QStringViewLiteral
instead.
Change-Id: I682eb8cfaa98fdcfb491edd290460db636a98a9f
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Pack four qWarning() calls into a separate cold function.
Use qUtf16Printable().
Saves >600b in text size on optimized AMD64 Linux GCC 9.1 builds.
Change-Id: Ib25ea473d1d77faaecaf8750726c83675d87279e
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
When the families member was added to QFontDef, it was included
in op== and qHash(), however the seed was passed to two qHash()
calls for subobjects. With xor used as the combiner, it could
happen that the seed was xored out (e.g. on empty strings),
leaving a slight opening for prediciable hash values.
Fix by using QtPrivate::QHashCombine, which handles the seed in
such a way as to avoid the issue.
Change-Id: I8a3e4c2f368306446554249763695158df5ac634
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Leander Beernaert
Simon Hausmann
Jean-Michaël Celerier
Albert Astals Cid
Alexandru Croitor
Marc Mutz
Shawn Rutledge
Alex Trotsenko
Tasuku Suzuki
Friedemann Kleint
Kai Koehne
Giuseppe D'Angelo
Sérgio Martins
Thorbjørn Lund Martsum
Lorn Potter