Attention: Here be dragons

This is the latest (unstable) version of this documentation, which may document features not available in or compatible with released stable versions of Redot.

Checking the stable version of the documentation...

Custom modules in C++¶

Modules¶

Redot allows extending the engine in a modular way. New modules can be created and then enabled/disabled. This allows for adding new engine functionality at every level without modifying the core, which can be split for use and reuse in different modules.

Modules are located in the modules/ subdirectory of the build system. By default, dozens of modules are enabled, such as GDScript (which, yes, is not part of the base engine), the Mono runtime, a regular expressions module, and others. As many new modules as desired can be created and combined. The SCons build system will take care of it transparently.

What for?¶

While it's recommended that most of a game be written in scripting (as it is an enormous time saver), it's perfectly possible to use C++ instead. Adding C++ modules can be useful in the following scenarios:

Binding an external library to Redot (like PhysX, FMOD, etc).
Optimize critical parts of a game.
Adding new functionality to the engine and/or editor.
Porting an existing game to Redot.
Write a whole, new game in C++ because you can't live without C++.

Creating a new module¶

Before creating a module, make sure to download the source code of Redot and compile it.

To create a new module, the first step is creating a directory inside modules/. If you want to maintain the module separately, you can checkout a different VCS into modules and use it.

The example module will be called "summator" (godot/modules/summator). Inside we will create a summator class:

/* summator.h */

#ifndef SUMMATOR_H
#define SUMMATOR_H

#include "core/object/ref_counted.h"

class Summator : public RefCounted {
    GDCLASS(Summator, RefCounted);

    int count;

protected:
    static void _bind_methods();

public:
    void add(int p_value);
    void reset();
    int get_total() const;

    Summator();
};

#endif // SUMMATOR_H

And then the cpp file.

/* summator.cpp */

#include "summator.h"

void Summator::add(int p_value) {
    count += p_value;
}

void Summator::reset() {
    count = 0;
}

int Summator::get_total() const {
    return count;
}

void Summator::_bind_methods() {
    ClassDB::bind_method(D_METHOD("add", "value"), &Summator::add);
    ClassDB::bind_method(D_METHOD("reset"), &Summator::reset);
    ClassDB::bind_method(D_METHOD("get_total"), &Summator::get_total);
}

Summator::Summator() {
    count = 0;
}

Then, the new class needs to be registered somehow, so two more files need to be created:

register_types.h
register_types.cpp

Important

These files must be in the top-level folder of your module (next to your SCsub and config.py files) for the module to be registered properly.

These files should contain the following:

/* register_types.h */

#include "modules/register_module_types.h"

void initialize_summator_module(ModuleInitializationLevel p_level);
void uninitialize_summator_module(ModuleInitializationLevel p_level);
/* yes, the word in the middle must be the same as the module folder name */

/* register_types.cpp */

#include "register_types.h"

#include "core/object/class_db.h"
#include "summator.h"

void initialize_summator_module(ModuleInitializationLevel p_level) {
    if (p_level != MODULE_INITIALIZATION_LEVEL_SCENE) {
        return;
    }
    ClassDB::register_class<Summator>();
}

void uninitialize_summator_module(ModuleInitializationLevel p_level) {
    if (p_level != MODULE_INITIALIZATION_LEVEL_SCENE) {
        return;
    }
   // Nothing to do here in this example.
}

Next, we need to create an SCsub file so the build system compiles this module:

# SCsub

Import('env')

env.add_source_files(env.modules_sources, "*.cpp") # Add all cpp files to the build

With multiple sources, you can also add each file individually to a Python string list:

src_list = ["summator.cpp", "other.cpp", "etc.cpp"]
env.add_source_files(env.modules_sources, src_list)

This allows for powerful possibilities using Python to construct the file list using loops and logic statements. Look at some modules that ship with Redot by default for examples.

To add include directories for the compiler to look at you can append it to the environment's paths:

env.Append(CPPPATH=["mylib/include"]) # this is a relative path
env.Append(CPPPATH=["#myotherlib/include"]) # this is an 'absolute' path

If you want to add custom compiler flags when building your module, you need to clone env first, so it won't add those flags to whole Redot build (which can cause errors). Example SCsub with custom flags:

# SCsub

Import('env')

module_env = env.Clone()
module_env.add_source_files(env.modules_sources, "*.cpp")
# Append CCFLAGS flags for both C and C++ code.
module_env.Append(CCFLAGS=['-O2'])
# If you need to, you can:
# - Append CFLAGS for C code only.
# - Append CXXFLAGS for C++ code only.

And finally, the configuration file for the module, this is a Python script that must be named config.py:

# config.py

def can_build(env, platform):
    return True

def configure(env):
    pass

The module is asked if it's OK to build for the specific platform (in this case, True means it will build for every platform).

And that's it. Hope it was not too complex! Your module should look like this:

godot/modules/summator/config.py
godot/modules/summator/summator.h
godot/modules/summator/summator.cpp
godot/modules/summator/register_types.h
godot/modules/summator/register_types.cpp
godot/modules/summator/SCsub

You can then zip it and share the module with everyone else. When building for every platform (instructions in the previous sections), your module will be included.

Note

There is a parameter limit of 5 in C++ modules for things such as subclasses. This can be raised to 13 by including the header file core/method_bind_ext.gen.inc.

Using the module¶

You can now use your newly created module from any script:

var s = Summator.new()
s.add(10)
s.add(20)
s.add(30)
print(s.get_total())
s.reset()

The output will be 60.

Compiling a module externally¶

Compiling a module involves moving the module's sources directly under the engine's modules/ directory. While this is the most straightforward way to compile a module, there are a couple of reasons as to why this might not be a practical thing to do:

Having to manually copy modules sources every time you want to compile the engine with or without the module, or taking additional steps needed to manually disable a module during compilation with a build option similar to module_summator_enabled=no. Creating symbolic links may also be a solution, but you may additionally need to overcome OS restrictions like needing the symbolic link privilege if doing this via script.
Depending on whether you have to work with the engine's source code, the module files added directly to modules/ changes the working tree to the point where using a VCS (like git) proves to be cumbersome as you need to make sure that only the engine-related code is committed by filtering changes.

So if you feel like the independent structure of custom modules is needed, lets take our "summator" module and move it to the engine's parent directory:

mkdir ../modules
mv modules/summator ../modules

Compile the engine with our module by providing custom_modules build option which accepts a comma-separated list of directory paths containing custom C++ modules, similar to the following:

scons custom_modules=../modules

The build system shall detect all modules under the ../modules directory and compile them accordingly, including our "summator" module.

Warning

Any path passed to custom_modules will be converted to an absolute path internally as a way to distinguish between custom and built-in modules. It means that things like generating module documentation may rely on a specific path structure on your machine.

Customizing module types initialization¶

Modules can interact with other built-in engine classes during runtime and even affect the way core types are initialized. So far, we've been using register_summator_types as a way to bring in module classes to be available within the engine.

A crude order of the engine setup can be summarized as a list of the following type registration methods:

preregister_module_types();
preregister_server_types();
register_core_singletons();
register_server_types();
register_scene_types();
EditorNode::register_editor_types();
register_platform_apis();
register_module_types();
initialize_physics();
initialize_navigation_server();
register_server_singletons();
register_driver_types();
ScriptServer::init_languages();

Our Summator class is initialized during the register_module_types() call. Imagine that we need to satisfy some common module run-time dependency (like singletons), or allow us to override existing engine method callbacks before they can be assigned by the engine itself. In that case, we want to ensure that our module classes are registered before any other built-in type.

This is where we can define an optional preregister_summator_types() method which will be called before anything else during the preregister_module_types() engine setup stage.

We now need to add this method to register_types header and source files:

/* register_types.h */

#define MODULE_SUMMATOR_HAS_PREREGISTER
void preregister_summator_types();

void register_summator_types();
void unregister_summator_types();

Note

Unlike other register methods, we have to explicitly define MODULE_SUMMATOR_HAS_PREREGISTER to let the build system know what relevant method calls to include at compile time. The module's name has to be converted to uppercase as well.

/* register_types.cpp */

#include "register_types.h"

#include "core/object/class_db.h"
#include "summator.h"

void preregister_summator_types() {
    // Called before any other core types are registered.
    // Nothing to do here in this example.
}

void register_summator_types() {
    ClassDB::register_class<Summator>();
}

void unregister_summator_types() {
   // Nothing to do here in this example.
}

Improving the build system for development¶

Warning

This shared library support is not designed to support distributing a module to other users without recompiling the engine. For that purpose, use a GDExtension instead.

So far, we defined a clean SCsub that allows us to add the sources of our new module as part of the Redot binary.

This static approach is fine when we want to build a release version of our game, given we want all the modules in a single binary.

However, the trade-off is that every single change requires a full recompilation of the game. Even though SCons is able to detect and recompile only the file that was changed, finding such files and eventually linking the final binary takes a long time.

The solution to avoid such a cost is to build our own module as a shared library that will be dynamically loaded when starting our game's binary.

# SCsub

Import('env')

sources = [
    "register_types.cpp",
    "summator.cpp"
]

# First, create a custom env for the shared library.
module_env = env.Clone()

# Position-independent code is required for a shared library.
module_env.Append(CCFLAGS=['-fPIC'])

# Don't inject Redot's dependencies into our shared library.
module_env['LIBS'] = []

# Define the shared library. By default, it would be built in the module's
# folder, however it's better to output it into `bin` next to the
# Redot binary.
shared_lib = module_env.SharedLibrary(target='#bin/summator', source=sources)

# Finally, notify the main build environment it now has our shared library
# as a new dependency.

# LIBPATH and LIBS need to be set on the real "env" (not the clone)
# to link the specified libraries to the Redot executable.

env.Append(LIBPATH=['#bin'])

# SCons wants the name of the library with it custom suffixes
# (e.g. ".linuxbsd.tools.64") but without the final ".so".
shared_lib_shim = shared_lib[0].name.rsplit('.', 1)[0]
env.Append(LIBS=[shared_lib_shim])

Once compiled, we should end up with a bin directory containing both the godot* binary and our libsummator*.so. However given the .so is not in a standard directory (like /usr/lib), we have to help our binary find it during runtime with the LD_LIBRARY_PATH environment variable:

export LD_LIBRARY_PATH="$PWD/bin/"
./bin/redot*

Note

You have to export the environment variable. Otherwise, you won't be able to run your project from the editor.

On top of that, it would be nice to be able to select whether to compile our module as shared library (for development) or as a part of the Redot binary (for release). To do that we can define a custom flag to be passed to SCons using the ARGUMENT command:

# SCsub

Import('env')

sources = [
    "register_types.cpp",
    "summator.cpp"
]

module_env = env.Clone()
module_env.Append(CCFLAGS=['-O2'])

if ARGUMENTS.get('summator_shared', 'no') == 'yes':
    # Shared lib compilation
    module_env.Append(CCFLAGS=['-fPIC'])
    module_env['LIBS'] = []
    shared_lib = module_env.SharedLibrary(target='#bin/summator', source=sources)
    shared_lib_shim = shared_lib[0].name.rsplit('.', 1)[0]
    env.Append(LIBS=[shared_lib_shim])
    env.Append(LIBPATH=['#bin'])
else:
    # Static compilation
    module_env.add_source_files(env.modules_sources, sources)

Now by default scons command will build our module as part of Redot's binary and as a shared library when passing summator_shared=yes.

Finally, you can even speed up the build further by explicitly specifying your shared module as target in the SCons command:

scons summator_shared=yes platform=linuxbsd bin/libsummator.linuxbsd.tools.64.so

Writing custom documentation¶

Writing documentation may seem like a boring task, but it is highly recommended to document your newly created module to make it easier for users to benefit from it. Not to mention that the code you've written one year ago may become indistinguishable from the code that was written by someone else, so be kind to your future self!

There are several steps in order to setup custom docs for the module:

Make a new directory in the root of the module. The directory name can be anything, but we'll be using the doc_classes name throughout this section.

Now, we need to edit config.py, add the following snippet:

def get_doc_path():
    return "doc_classes"

def get_doc_classes():
    return [
        "Summator",
    ]

The get_doc_path() function is used by the build system to determine the location of the docs. In this case, they will be located in the modules/summator/doc_classes directory. If you don't define this, the doc path for your module will fall back to the main doc/classes directory.

The get_doc_classes() method is necessary for the build system to know which registered classes belong to the module. You need to list all of your classes here. The classes that you don't list will end up in the main doc/classes directory.

Tip

You can use Git to check if you have missed some of your classes by checking the untracked files with git status. For example:

git status

Example output:

Untracked files:
    (use "git add <file>..." to include in what will be committed)

    doc/classes/MyClass2D.xml
    doc/classes/MyClass4D.xml
    doc/classes/MyClass5D.xml
    doc/classes/MyClass6D.xml
    ...

Now we can generate the documentation:

We can do this via running Redot's doctool i.e. godot --doctool <path>, which will dump the engine API reference to the given <path> in XML format.

In our case we'll point it to the root of the cloned repository. You can point it to an another folder, and just copy over the files that you need.

Run command:

bin/<godot_binary> --doctool .

Now if you go to the godot/modules/summator/doc_classes folder, you will see that it contains a Summator.xml file, or any other classes, that you referenced in your get_doc_classes function.

Edit the file(s) following Class reference primer and recompile the engine.

Once the compilation process is finished, the docs will become accessible within the engine's built-in documentation system.

In order to keep documentation up-to-date, all you'll have to do is simply modify one of the XML files and recompile the engine from now on.

If you change your module's API, you can also re-extract the docs, they will contain the things that you previously added. Of course if you point it to your godot folder, make sure you don't lose work by extracting older docs from an older engine build on top of the newer ones.

Note that if you don't have write access rights to your supplied <path>, you might encounter an error similar to the following:

ERROR: Can't write doc file: docs/doc/classes/@GDScript.xml
   At: editor/doc/doc_data.cpp:956

Writing custom unit tests¶

It's possible to write self-contained unit tests as part of a C++ module. If you are not familiar with the unit testing process in Redot yet, please refer to Unit testing.

The procedure is the following:

Create a new directory named tests/ under your module's root:

cd modules/summator
mkdir tests
cd tests

Create a new test suite: test_summator.h. The header must be prefixed with test_ so that the build system can collect it and include it as part of the tests/test_main.cpp where the tests are run.
Write some test cases. Here's an example:

// test_summator.h
#ifndef TEST_SUMMATOR_H
#define TEST_SUMMATOR_H

#include "tests/test_macros.h"

#include "modules/summator/summator.h"

namespace TestSummator {

TEST_CASE("[Modules][Summator] Adding numbers") {
    Ref<Summator> s = memnew(Summator);
    CHECK(s->get_total() == 0);

    s->add(10);
    CHECK(s->get_total() == 10);

    s->add(20);
    CHECK(s->get_total() == 30);

    s->add(30);
    CHECK(s->get_total() == 60);

    s->reset();
    CHECK(s->get_total() == 0);
}

} // namespace TestSummator

#endif // TEST_SUMMATOR_H

Compile the engine with scons tests=yes, and run the tests with the following command:

./bin/<godot_binary> --test --source-file="*test_summator*" --success

You should see the passing assertions now.

Adding custom editor icons¶

Similarly to how you can write self-contained documentation within a module, you can also create your own custom icons for classes to appear in the editor.

For the actual process of creating editor icons to be integrated within the engine, please refer to Editor icons first.

Once you've created your icon(s), proceed with the following steps:

Make a new directory in the root of the module named icons. This is the default path for the engine to look for module's editor icons.
Move your newly created svg icons (optimized or not) into that folder.
Recompile the engine and run the editor. Now the icon(s) will appear in editor's interface where appropriate.

If you'd like to store your icons somewhere else within your module, add the following code snippet to config.py to override the default path:

def get_icons_path():
    return "path/to/icons"

Summing up¶

Remember to:

Use GDCLASS macro for inheritance, so Redot can wrap it.
Use _bind_methods to bind your functions to scripting, and to allow them to work as callbacks for signals.
Avoid multiple inheritance for classes exposed to Redot, as GDCLASS doesn't support this. You can still use multiple inheritance in your own classes as long as they're not exposed to Redot's scripting API.

But this is not all, depending what you do, you will be greeted with some (hopefully positive) surprises.

If you inherit from class_Node (or any derived node type, such as Sprite2D), your new class will appear in the editor, in the inheritance tree in the "Add Node" dialog.
If you inherit from class_Resource, it will appear in the resource list, and all the exposed properties can be serialized when saved/loaded.
By this same logic, you can extend the Editor and almost any area of the engine.