ROS2-001-What do initialisation and resource release actually do in a program?

In ROS2 nodes, whenever we write a C++ or Python program, we have to bring in two operations: "initialisation" and "resource release". For example:

In C++:

#include "rclcpp/rclcpp.hpp"

class MyNode: public rclcpp::Node{
public:
        MyNode():Node("hello_node_cpp"){
            RCLCPP_INFO(this->get_logger(), "helloworld");
        }

};

int main(int argc, char const *argv[])
{
    // Initialisation
    rclcpp::init(argc, argv);
    // Instantiate the custom class
    auto node = std::make_shared<MyNode>();
    // ......

    // Resource release
    rclcpp::shutdown();

    return 0;
}

In Python:

import rclpy
from rclpy.node import Node

class MyNode(Node):
    def __init__(self):
        super().__init__("hello_node_py")
        self.get_logger().info("helloworld!")

def main():
    # Initialisation
    rclpy.init()
    # Create the object
    node = MyNode()
    # ......
    # Resource release
    rclpy.shutdown()

if __name__ == '__main__':
    main()

But once you compile and run it, neither operation shows anything at all. So what are they actually doing?

The answer

1. A program built on ROS2 may consist of several steps or stages:
   • initialisation -> node object -> log output -> data publishing -> data subscription -> …… -> resource release;
2. Data gets passed between the different steps or stages;
3. So how do you pass that data?
   • Use a Context object. You can treat it as a data container that flows through the different steps, which is what gets the data across
4. So initialisation includes the step of creating the Context object, and resource release includes destroying the Context object.
   • It's a bit like how the VUE framework, during initialisation, creates all sorts of cache space across the lifecycle in order to achieve reactivity

Important

This article was translated by Hermes Agent; the original text can be found at here. The translation is for reference only. Translation time: 2026-11-05