These were omitted in the original change, which cause cmake builds to
fail configuring
Change-Id: I37996099aaa0d912ce06b1f06c175cc5cf159e33
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
This matches how it is in the qmake mkspec, silencing a lot of
warnings - but contrary to when building with qmake, it's only
applied when building Qt itself. For external projects built with
cmake, the Qt headers are included with -isystem, which silences any
warnings from those headers.
Change-Id: I1a498d3c2715cb73e647668cb7226ceeffb7ff0b
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Now all shared libraries and executables will get .debug files on
the platforms that support FEATURE_separate_debug_info
With the directory property _qt_skip_separate_debug_info certain
targets can retain the debug symbols in the binary e.g. lupdate with
MinGW 8.1.0 will cause objcopy / strip to fail.
Fixes: QTBUG-87015
Change-Id: I03b106e68ef0a42011d1ba641e6f686b2e7b7fb4
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
"configure" script translates feature-related parameters to INPUT_
variables instead of FEATURE_.
Both INPUT_ and FEATURE_ variables passed to cmake script are
equivalent. FEATURE_ has higher priority in case if both are defined.
Fixes: QTBUG-87755
Change-Id: If697a0d62ab839877a3196ea74e631582a570dda
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We must write config.opt in the same directory we're reading it from.
We must not write the -top-level argument to config.opt.
This amends commit 2a29426e39.
Change-Id: I96da9094579fec29c290411677d6b538878399f4
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Add a new function that returns the minimum CMake version required to
build Qt. Pass that value to cmake_minimum_required() when building
qtbase and its standalone tests.
The minimum supported CMake version is read from qtbase/.cmake.conf
and its value should be updated when the need arises. It's the main
source of truth for all repos.
Provide a way to lower the minimum CMake version at configure time by
passing a value via QT_FORCE_MIN_CMAKE_VERSION.
This is not an officially supported way of building Qt. If the
specified version is lower than Qt's supported minimum, show a
warning.
Nevertheless the option is useful for testing how Qt builds with a
different minimum CMake version due to different policies being
enabled by default.
Issue warnings for CMake versions that are higher than the minimum
version but are known to cause issues when building Qt.
A counterpart change is needed in qt5 to ensure the minimum CMake
version is set at the proper time for top-level builds.
Ideally we would use the same 'check the CMake minimum version` code
in all our repositories, but that will cause lots of duplication because
we can't really find_package() the code and doing something like
include(../qtbase/foo.cmake) hardcodes assumptions about repo
locations.
So for now we don't bump the minimum version in child repo
cmake_minimum_required calls (qtsvg, qtdeclarative, etc).
Instead we record both the minimum supported version and the computed
minimum version (in case a different version was forced) in
QtBuildInternalsExtra.cmake.
Then we require qtbase's computed min version in
qt_build_repo_begin().
This won't set policies as cmake_minimum_required would, but at least
it propagates what minimum CMake version should be used for child
repos.
We might still have to bump the versions in child repos at some point.
Task-number: QTBUG-88086
Change-Id: Ida1c0d5d3e0fbb15d2aee9b68abab7a1648774b9
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
The flags go before the library in the final linker line, as opposed
to the dependencies declared in LIBS.
This allows us to declare the flags for the entrypoint
in the project file of the entrypoint, instead of in
a standalone prf.
Change-Id: I35c054fe9fdaa6add7cd0e8ba3f7304f975ff80f
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
On the qmake-side we had exports, but they were quoted.
Change-Id: I95af4b927079691cab6403fec850f345ba181a00
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The arguments -platform, -xplatform and -device determine the mkspec
that's used for the qmake companion files.
If the user specifies a mkspec that indicates usage of clang or icc,
set the respective CMake variables to use one of those compilers.
Fixes: QTBUG-87836
Change-Id: I2b10d819b0eb92a97d7f79672547b1e2d821cf33
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
At the moment our examples require a minimum of 3.14 due to changes
in upstream CMake's Autogen functionatlity to support Qt 6. Anything
lower would simply not work with Qt 6.
It's not clear yet if we actually want to require 3.14, or something
higher. At the very least there were many significant changes to
support iOS in CMake 3.15.
But for now just bump the version checked by Qt6Config.cmake to be
consistent with what's in our examples.
Task-number: QTBUG-88086
Change-Id: I119c2ad05a18c357fe7c659b30685af87475fc84
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
1. clang-cl doesn't support "-fno-exceptions", it uses msvc's parameter.
2. some parameters supported by msvc are not supported by clang-cl
and they are causing huge warning message flood, don't add them.
3. use correct optimize parameter for clang-cl.
Change-Id: Idbadf139127143c5fa6c49068588cb26f47da7a2
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
When linking static libraries, MSVC's link.exe complains about
the unknown parameter "/INCREMENTAL:NO" and output a lot of warning
messages about this. This doesn't happen when it's linking exes
or dlls.
The situation is a lot more worse when we are using clang-cl.
clang-cl will print some error message like it can't find a
file named "/INCREMENTAL:NO" and just stop compiling. It seems
clang-cl treat unknown parameters as input files.
Fixes: QTBUG-87875
Change-Id: I37ed29de082b0258e81494db54f275417ab42708
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The QT.<module>.DEFINES assignment in pri files needs to take into
account the module name when computing the define name. This is the
MODULE value that qmake specifies.
In CMake that would be the value of CONFIG_MODULE_NAME.
Previously the value of the define was computed in
qt_internal_module_info() without taking into account the module name.
While qt_internal_module_info() ended being used also for plugins and
other target types, the defines computed by it were meant to be used
only for Qt modules.
Thus remove the <result>_define assignment from
qt_internal_module_info and move its computation directly into
qt_internal_add_module, taking into account the value of
CONFIG_MODULE_NAME.
The only other use of module_define was in qt_internal_add_plugin but
that was merely a long overdue copy-paste error, qmake doesn't
propagate QT_FOO_LIB defines for plugins.
As result, a define special case in testlib is not needed anymore,
because the define is now computed properly.
Finally, QT_FOO_LIB should not be used while building the Qt module
itself, so instead of using PUBLIC_DEFINES option of
qt_internal_extend_target, use target_compile_definitions(INTERFACE)
directly.
Change-Id: I4d44f7461bac2f0c09aec3e995d02dfe36e00883
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Makes the diff between Qt 5.15 and 6.0 easier to read, to see what's
missing.
Change-Id: Idf8aa17b3ab8494f6855c172665423a53ca8a024
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
We do not have host-specific INSTALL_*DIR variables in the CMake build.
It is equivalent to a qmake build with host prefix = prefix.
Change-Id: I65731e6038508b6c28e3f2819b624ba19abfc82a
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The value of this variable must be exactly "mkspecs" or end with
"/mkspecs".
Change-Id: I39f83e9660794dfe23f5fd39fb1084b87ba1f140
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The value of this variable - whithout the mkspecs part - is what's
called the host data dir in Qt5.
Fixes: QTBUG-87681
Change-Id: I3dfeed17e8a614476aef4d9c651a141ce62e6551
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We must escape backslashes in CMake code that's to be evaluated and in
the arguments we read from config.opt.
Change-Id: I65d033c77f71888974983aa3d834acb2fe89f3fb
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Passing arguments with equal signs was broken for configure.bat and
qt-configure-module.bat. An argument FOO=BAR was split at = and written
as
FOO
BAR
to config.opt, breaking every attempt of assigning CMake variables.
We must not iterate over %* in batch files to avoid splitting arguments
at equal signs. Instead, pass %* unmodified to a CMake script that
writes config.opt.
Fixes: QTBUG-88019
Change-Id: I7c743a206961d1ed168f2313f864905f6b345b49
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
In a subsequent change we will call another CMake script from
qt-configure-module.bat. Write the location of qtbase/cmake into the
generated scripts instead of the path to QtProcessConfigureArgs.cmake.
Change-Id: Ie333b16d310b215c6e49efa27740c7525453d28f
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
CMake considers ON/OFF as booly string values regardless of the case.
Make the value comparison in QtFeature.cmake case-independent.
It's now possible to build Qt with '-DFEATURE_gui=off'.
Fixes: QTBUG-87948
Change-Id: I3d948e8219ad9728414803c8c4cd756034073b46
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Rolf Eike Beer <eb@emlix.com>
Previously we determined if a library represented by an absolute path
is a Qt module by checking if it's located in the build dir of the
current repo.
That is not sufficient for non-qtbase prefix builds, where
a Qt module might link against both a module in the current
build dir and in the prefix dir.
Detect such cases, and rewrite the absolute paths to relocatable paths
(either framework flags or paths starting with $$[QT_INSTALL_LIBS].
This should fix building examples with qmake that use QtQuick.
Fixes: QTBUG-87840
Change-Id: Icaf8f1a7c66292c80662fd0d5771a5a1628a9899
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Add an EXTRA_CMAKE_FILES argument to qt_internal_add_tool()
that allows tools to install an additional Macro.cmake file.
This is modelled after similar functionality in qt_internal_add_module.
Task-number: QTBUG-87870
Change-Id: I80838b8966f1018fdd379b1da877b6bc418de075
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Fixes a regression introduced in commit 748b3b9c89 for builds
that are both not a QT_SUPERBUILD and not QT_WILL_INSTALL.
Change-Id: I52b920176a2696fa6206b89b2b69ae1a7f3fbc25
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
The top-level prepare_docs target is supposed to run before the
generate_docs target.
The repository-level targets like prepare_docs_qtbase now also run
before their respective generate targets.
To achieve that, several intermediate targets had to be introduced.
The dependencies for top-level generate_docs look like this:
/--> generate_top_level_docs_Core --\
generate_docs ---> generate_top_level_docs_Gui -----> prepare_docs
\--> ... --/
The dependencies for repo-level generate_docs_<repo> look like this:
/--> generate_repo_docs_Core
generate_docs_qtbase ---> generate_repo_docs_Gui
\--> ...
generate_repo_docs_Core --\
generate_repo_docs_Gui -----> prepare_docs_qtbase
... --/
Analoguous for qch_docs.
And last but not least, the module-level generate_docs_<module> depends
on prepare_docs_<module>, just like before.
The strangely named top-level install targets have been renamed to
install_html_docs, install_qch_docs and install_docs.
The html_docs target is now merely an alias for generate_docs.
Apart from prepare_docs, the top-level targets do not depend on the
repository-level targets anymore. Targets that are just an alias (docs,
install_docs, ...) have been re-defined to depend just on the respective
top-level targets.
Fixes: QTBUG-87758
Change-Id: I24466d3a252d518a30a2064bf876be881c46231f
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The value is useful for QPA plugins built in repos other than qtbase,
to decide if it should be a default plugin or not.
Currently useful for qtwayland.
Also export a qmake value assignment when doing static builds, just
like src/gui/configure.pri does.
Change-Id: I1253f1a7e178b24b16e2615ba20d1e92b0b87b1a
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Otherwise we get error messages from CMake.
Change-Id: I374b87e38550062b841b7eacf5fbb6bb5ccbdbde
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Allow CMake for Android to use ANDROID_NDK_ROOT to deduce the path for
QT_CHAINLOAD_TOOLCHAIN_FILE instead of the user providing it manually.
Change-Id: Ida728011d5ca8d5a723d341ea77b173e8f105f8c
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This is meant to be called by our CI instructions to build standalone
tests of a Qt repository.
Currently it just calls qt-cmake with
-DQT_BUILD_STANDALONE_TESTS=ON, but it might contain more things in
the future.
The script also simplifies configuring standalone tests locally, due
to not having to remember the name of the magical variable.
Change our CI instructions to use the new script.
Change-Id: I6bc02b4e94adc9d0d05fecb0fe70a561043271f1
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
With the introduction of the new 'default_qpa_plugins' custom target,
a target dependency cycle occurred in qtwayland:
qtwaylandscanner -> default_qpa_plugins -> a wayland qpa plugin ->
WaylandClient -> qtwaylandscanner
The issue is twofold:
- default_qpa_plugins accidentally depended on non-qpa plugins.
- All qpa plugins were enabled by default, including the wayland ones.
Fix the default_qpa_plugins target not to depend on regular non-qpa
plugins.
Also fix qpa plugins not to be enabled by default, but instead only
choose one qpa plugin to be the default (via evaluating the
DEFAULT_IF) condition.
Amends df9c7456d1
Change-Id: I22cd2c72f6b75be54263fd21097258bd179e3616
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
The way we detected whether SysrootifyPrefix needs to be set was
incorrect.
We checked if extprefix == prefix => SysrootifyPrefix should be
true. But that was previously always the case for a non-prefix build,
which means it was not possible to build apps (due to qmake expecting
to find include dirs in the sysroot, despite the non-prefix Qt not
being installed anywhere into the sysroot).
Instead we should do what qmake does. Only set SysrootifyPrefix to
true if extprefix was not provided and a sysroot is available.
To make it work for iOS and Android, the sysroot detection code had to
be moved earlier than the SysrootifyPrefix detection.
Task-number: QTBUG-86053
Change-Id: Idcc1260a675fb2ef9ae27fc9dc42a68b30f763b8
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
The html_docs target does not need to call QDoc a third time as
both its dependencies already do that.
With qmake, html_docs did invoke QDoc if CONFIG did not include
'prepare_docs', but none of the Qt modules used that configuration.
We can manage without it for now.
Fixes: QTBUG-87751
Change-Id: I45d2d06006c566c431ae41d7f3a06d6b809987f2
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
For top-level builds it's desirable to have "ninja generate_docs" build
all tools needed to generate the documentation.
This is problematic since the doc-generating targets are created before
the doc tool targets. Thus, we must defer the dependency connection if
the doc tool target is not yet available.
This patch adds the functions qt_internal_defer_dependency and
qt_internal_add_deferred_dependencies.
Change-Id: Ica940b80882e67cb0e0943e95541f7f4d1885948
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Instead of propagating the deprecation wranings, we compiled all code
with -Ddeprecations.
Change-Id: I0233ddc85bdbdcb93d366073b2cea5d47bdbe52a
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
qmake treats internal modules as just the regular module file,
but with a _private suffix, as opposed to the current cmake
logic, that treats it as the private module file, resulting
in missing e.g. the Qt.foo.module entry.
Change-Id: Id55ca4c23921656d5abfd1d0fdf6430d4fe120bf
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This patch allows in top-level prefix builds to build Qt and to generate
the documentation without running cmake --install first.
For top-level builds we now always use the qdoc binary from the build
directory, not the installation directory.
We also have to copy the global doc files to the build directory to make
them accessible to the generate_docs target.
Change-Id: I2251603418fc3df9a21c7f2892789e9ff6c8cc21
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
It's not sufficient to pass -I{prefix}/include when doing the header
clean check.
We need to propagate all target include directories and compile
definitions, and also the compile flags for good measure.
For macOS frameworks we also need to explicitly pass an -iframework
flag (qmake passses -F instead), to ensure that <QtGui/qfoo.h> style
includes are found when building other repos than qtbase.
Task-number: QTBUG-82615
Change-Id: I76d12340bc01c5c948ff04df9a3df384dcb7e076
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
This ensures ccache or sccache is used to cache the compilation of the
headerclean checks.
Task-number: QTBUG-82615
Change-Id: Ie944eb1d643e7271551c9f8337609741e419e9d8
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
It appears there's a difference between the chosen macOS compiler path
in the CI versus the compiler path on my local machine.
In the CI the chosen compiler path ends up 'being /usr/bin/clang++'
whereas for me locally it's
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++
For some reason the headersclean commands succeed in the CI, but
locally they fail for me saying that standard library includes can not
be found, unless an explicit sysroot flag is specified.
I assume that in the CI the '/usr/bin/clang++' compiler shim chooses some
implcit sysroot, whereas the longer Xcode compiler expects an explicit
sysroot.
It's probably also affected by the fact that in the CI we pass an
explicit CMAKE_OSX_SYSROOT to a non-standard Xcode location e.g.
/Applications/Xcode11.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.15.sdk
Note the '11' in the Xcode app name.
Locally I don't pass a custom CMAKE_OSX_SYSROOT, and my Xcode is
installed in a regular location e.g /Applications/Xcode.app.
The sysroot flag and path is added to regular CXX compilation rules
inside CMake's core (in cmLocalGenerator.cxx).
Reuse the same variables that CMake uses and add them to our headersclean
command rules.
Task-number: QTBUG-82615
Change-Id: Ic03ea27e39471f5fa168eb5970bf3d3f1d1be251
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
