Skip to content

This is a SpatialOS project which can serve as a template for building SpatialOS workers using the C++ SDK.

Notifications You must be signed in to change notification settings

spatialos/CppBlankProject

Repository files navigation

SpatialOS C++ Blank project

This is a SpatialOS project which can serve as a template for building SpatialOS workers using the C++ SDK. It uses CMake as the build system. If you're new to SpatialOS, have a look at Introduction to the C++ worker SDK.

About managed and external workers

Have a look at the Glossary entry for Workers for a complete discussion and examples.

Dependencies

This project requires the following software to be installed:

  • CMake 3.7+
  • (Windows) Visual Studio 2017 or above.
  • (macOS) Xcode
  • (Linux) gcc / clang and make (for the "Unix Makefiles" CMake project generator)

Quick start

The build scripts (build.json) for each worker:

  1. Generate C++ code from schema
  2. Download the C++ worker SDK
  3. Create a cmake_build directory in the worker directory
  4. Invoke cmake with arguments depending on the target platform

To build and launch a local deployment execute the following commands:

spatial worker build
spatial local launch --runtime_version=15.0.0

Windows users

If you have WSL installed, it is strongly recommended that you invoke spatial worker build directly from a non-WSL bash shell, e.g. git bash. The command executes shell scripts internally, which are likely to fail when invoked from other types of shells (e.g. cmder, powershell). Try this if you see the $'\r': command not found error, or the cmake command not getting found despite being installed.

To connect a new instance of the "External" worker type to a running local deployment (after spatial local launch):

spatial local worker launch External local

What the project does

When you launch a deployment, a single instance of the Managed worker will be started as configured in default_launch.json.

The Managed worker connects to SpatialOS and then assigns the entity with ID 2 (defined in the snapshot) as a partition entity to itself. This then gives it authority over the LoginListenerSet and PositionSet on entity 1, as per the authority delegation already set up on it in the snapshot.

The Managed worker updates the position of entity 1 in a loop, making it move around the origin in a circle - this will be visible from the inspector.

The Interest component on entity 1 is configured such that the Managed worker gains interest in all other entity with the improbable::restricted::Worker component. The Managed worker uses this detect when new workers connect to the deployment and log a message to the runtime for each connected worker.

The External worker, once manually started, simply connects to SpatialOS and then loops to continually process the Ops list it receives.

Project structure

The CMake project hierarchy doesn't exactly match the directory structure of the project. For example the projects for workers add as subdirectories the schema and dependencies projects.

This is how projects are structured in the directory:

+-- schema/CMakeLists.txt
+-- dependencies/CMakeLists.txt
+-- workers
    |-- External/
    |   |-- External/CMakeLists.txt
    |   |-- CMakeLists.txt
    |   |-- build.json
    |-- Managed/
        |-- Managed/CMakeLists.txt
        |-- CMakeLists.txt
        |-- build.json

This enables you to keep the worker directories free of CMake files for schema and dependencies while not needing a CMake file at the root of the project.

The schema directory contains a sample empty component called blank. It is not used by the workers directly so feel free to delete it but it's there to show how sources generated from the schema could be linked in the worker binary. See schema/CMakeLists.txt which creates a library with all generated sources.

The snapshot exists in both JSON and binary format in the snapshots folder. There is no script to generate the snapshot as the snapshot was written by hand in JSON format, but it's possible to make simple changes to the JSON snapshot and regenerate the binary snapshot from it. To update the binary snapshot after making a change, run the following command:

spatial project history snapshot convert --input-format=text --input=snapshots/default.json --output-format=binary --output=snapshots/default.snapshot

The worker project

The following applies to both the Managed and External worker projects but examples will only be about Managed.

After running spatial build the generated project by CMake will be in workers/Managed/cmake_build. Exactly what it contains will depend on the generator you use. A worker project includes 3 subdirectories in its CMakeLists.txt - dependencies, schema and Managed. The first two are not true subdirectories of workers/Managed in the file structure but their binary directories are set as if they were.

On Windows, both the release and debug builds of the Worker SDK are downloaded and set up correctly in the CMakeLists.txt. This means that both the Release and Debug configurations in the generated Visual Studio solution (.sln) should build and link correctly without any further changes.

Attaching a debugger

If you use a Visual Studio generator with CMake, the generated solution contains several projects to match the build targets. You can start a worker from Visual Studio by setting the project matching the worker name as the startup project for the solution. It will try to connect to a local deployment by default. You can customize the connection parameters by navigating to Properties > Configuration properties > Debugging to set the command arguments. Using receptionist localhost 7777 DebugWorker as the command arguments for example will connect a new instance of the worker named DebugWorker via the receptionist to a local running deployment. You can do this for both worker types that come with this project. Make sure you are starting the project using a local debugger (e.g. Local Windows Debugger).

Cloud deployment

Our cloud deployment environment is based on Linux, so therefore if you're not using Linux, you'll have to set up a cross-compile toolchain and build out a Linux binary (due to the nature of C++). More information can be found here.

If using Windows, options include (but are not limited to):

Once this is done and you have successfully built a Linux assembly, set the project_name field in spatialos.json to match your SpatialOS project name. Then upload and launch:

spatial cloud upload <assembly-name>
spatial cloud launch <assembly-name> default_launch.json <deployment-name> --snapshot=<snapshot-path> --runtime_version=15.0.0

See spatial cloud connect external if you want to connect to a cloud deployment. In addition, the External worker has a second external launch configuration called cloud which could be used to connect provided that you know the deployment name and have a login token:

spatial local worker launch External cloud <deployment-name> <login-token>

Integrating with an existing project

If you have an existing project, to add a new C++ worker to it:

  1. Decide whether the worker you're adding will be used as a managed or external worker.
  2. Copy the corresponding directory (e.g. workers/Managed) into the workers directory of your existing project.
  3. In the worker project CMakeLists.txt set SCHEMA_SOURCE_DIR and WORKER_SDK_DIR to point to the CMake projects in your project that generate the corresponding targets and if the targets have different names from Schema and WorkerSdk also rename those.
  4. Add it to the workers definition in your SpatialOS launch configuration (e.g. default_launch.json)