Sunswift Racing ROS2 Jazzy Docker Test

This is a practice environment to get comfortable with the Docker, Github, and ROS 2 development workflow using VS Code. Below are instructions on how to set up and start developing/playing around in a Ubuntu 24.04 environment with ROS 2 Jazzy already installed. It's a good idea to go through the Jazzy tutorials once you have your environment set up.

Made by Ryan Wong for the Sunswift embedded high-level software team.

Full Development Workflow

Introduction

ROS 2 runs best (and with least effort) on a linux environment. However, since getting everyone on native linux is not feasible, and VMs are slow and clunky, especially for ROS (from experience!), a quick and easy fix is to use Docker.

Docker is a tool for packaging applications and their dependencies into containers, ensuring they run reliably across different environments. A Docker image is kind of a snapshot of an app and its environment, and a container is a running instance of that image. Use Dockerfile to define how an image is built, and docker-compose.yml to manage multi-container setups.

Something new for SR8's development process I want to try to use the devcontainers vscode extension. It allows you to open vscode inside a container, almost like a remote environment. This is great bcause tools like IntelliSense can understand the location of libraries like rclcpp to make coding easier. This is configured in .devcontainer/devcontainer.json, which outlines how vscode should open the container, and what it should use.

---

Prerequisites

You need these installed:

1. Git/Github
2. Visual Studio Code
3. Docker Desktop

Also, install this VS Code extension:

• Dev Containers (ms-vscode-remote.remote-containers)

---

Workflow

1. Clone this git repository

2. Open this directory in VS Code

3. Open the command palette (Ctrl+Shift+P or Cmd+Shift+P) and select Dev Containers: Rebuild and Reopen in Container
   → This builds the Docker image using the Dockerfile and docker-compose.yml.
   → Once you've built the image once, you DO NOT need to rebuild as it takes ages for nothing. Choose Dev Containers: Open Folder in Container instead. The only time you need to rebuild is if you change Dockerfile, docker-compose.yml or devcontainer.json

4. You should now see that your VS Code terminal is open inside the container

   • It’ll look something like: root@168495d3f21d:~/ros2_docker_test#

5. I've already created a ROS workspace called ros2_ws, but feel free to delete it:

   • rm -rf ros2_ws
   • Then follow the ROS 2 Jazzy tutorials to learn how to set up your own workspace and packages
   • If you want, have a look at the simple publisher I wrote and soak in the glorious C++ syntax... To run this publisher node, run colcon build from /root/ros2_docker_test/ros2_ws, then run source install/setup.bash (you must do this after everytime you colcon build). Then run ros2 run test_package minimal_cpp_publisher. Observe the std_msgs::msg::String being published at 2Hz.

6. In the real high-level repo, you would now edit your code once you're in the environment, then after finishing, close the remote environment. Finally, commit and push this code to github.
   → But since we're not, don't bother commiting and pushing as I want to use this repo for my own testing LOL. (don’t worry, you can’t push anyway!)

7. To get intellisense working, you must open VS Code in the ros2_ws directory. By default, the ROS2-docker-test directory is opened in VS Code when you open the container, so cd ros2_ws then code . and intellisense will work with c++, python, all your ROS packages.

There is a warning message when you colcon build which says something about a clock skew. I think this is due to the bind mount and a slight clock skew between the host and the container. I've tried a bunch of stuff but can't fix this warning. However, I don't think it's a major issue as the build still completes and nodes run OK.

---

How it works

1. How does my code from github, which I cloned to my host, appear in the container?

2. Why do code changes made IN the container still persist even after it shuts down?

In docker-compose.yml, there is a bind mount under volumes, which basically mounts this repository into the docker filesystem at /root/ros2_docker_test. Simply, a bind mount is a directory which is shared between the host and the container, reflecting changes in real time, both ways. If you look in the Dockerfile, it simply uses ros:jazzy-ros-base as the base image, which is a minimal Ubuntu 24.04 + important ROS 2 packages image. It then mounts all the code from this repo (on the host) to the container filesystem at /root/ros2_docker_test.

3. Why do I not need to rebuild the image everytime I edit code?

All the image does is start with a fresh ROS 2 Jazzy + Ubuntu 24.04 image. The bind mount is the thing the adds our source code into the container. The only time we want the source code to be built into the docker image is during production (I think).

If you need any help or have any questions, feel free to ask me at [email protected]. Or message me on Teams.