Building a QML application Building projects on the command line
The example below demonstrates how to create a library which exposes C++ to QML. The directory structure for the example looks like this:
├── CMakeLists.txt └── example └── mylib ├── CMakeLists.txt ├── mytype.cpp ├── mytype.h
The toplevel CMakeLists.txt
file does some basic setup, and then uses add_subdirectory
to include the one in mylib. The subdirectory structure corresponds to the QML module’s URI, but with the dots
replaced by slashes. That’s the same logic the engine uses when it searches for a module in the import paths. mytype.h
declares a class and uses the declarative
registration macros to expose it to the engine.
In the subdirectory’s CMakeLists.txt
we again call qt6_add_qml_module
. However, the invocation is slightly different:
qt6_add_qml_module(mylib URI example.mylib VERSION 1.0 SOURCES mytype.h mytype.cpp )
To add C++ types, the SOURCES parameter needs to be specified. The target for mylib is not created. Therefore, if the target passed to qt6_add_qml_module
does not exist, a library target is automatically
created, which is needed in this case.
When the project is built, in addition to the library, a QML plugin is also built. The plugin's auto-generated class extends from QQmlEngineExtensionPlugin
. The mylib library itself already contains the code to
register the types with the engine. However, that is only useful in cases where we can link against the library. To make the module usable in a QML file loaded by qml
, the QML Runtime Tool, a plugin is needed that can be loaded. The plugin is then responsible for actually linking against the library, and ensuring that the types get
registered.
Note that the automatic plugin generation is only possible if the module does not do anything besides registering the types. If it needs to do something more advanced like registering an image provider in
initializeEngine
, you still need to manually write the plugin. qt6_add_qml_module has support for this with NO_GENERATE_PLUGIN_SOURCE
.
Also, following the directory layout convention helps tooling. That layout is mirrored in the build directory. Which means that you can pass the path to your build directory to the QML tool (via the -I flag), and it will find the plugin.
Before concluding add a QML file to the module. In the lib subfolder, add a Mistake.qml
file
import example.mylib MyType{ answer: 43 }
and adjust the qt6_add_qml_module
call:
qt6_add_qml_module(mylib URI example.mylib VERSION 1.0 SOURCES mytype.h mytype.cpp QML_FILES Mistake.qml )
As mentioned, we made a mistake because answer
is actually a read-only property. This illustrates qmllint
integration: CMake creates a qmllint
target, and once we run it,
qmllint
warns about the issue:
$> cmake --build . --target mylib_qmllint ... Warning: Mistake.qml:4:13: Cannot assign to read-only property answer answer: 43 ^^
Building a QML application Building projects on the command line
As part of the free Business evaluation, we offer a free welcome call for companies, to talk about your requirements, and how the Felgo SDK & Services can help you. Just sign up and schedule your call.
Sign up now to start your free Business evaluation: