ROS2 Service & Action

Create a node class inheriting from rclcpp::Node and call create_service with the .srv type, topic name, and a callback binding request and response. The callback should delegate to a domain use case and map its result onto the response fields.

Use rclcpp_action::Server with handle_goal, handle_cancel, and handle_accepted callbacks, then run execute in a separate thread that publishes feedback via the goal handle. Define goal, result, and feedback sections in the .action file first.

Services are synchronous request-response interactions for short tasks, while actions are asynchronous goal-based interfaces for long-running tasks that provide periodic feedback and support cancellation. Choose services for quick queries and actions for navigation or motion goals.

Yes, by placing service and action nodes in an infrastructure layer that converts ROS messages to domain objects and calls injected use case classes. This keeps business logic free of rclcpp or rclpy dependencies and independently testable.

Yes, the patterns cover rclpy for Python service servers and rclcpp for C++ service servers, clients, and action servers. Both languages share the same layering principle of isolating use cases from middleware code.