Overview

This project was created to set up a development environment for ML4T that was compatible with remote-containers in vscode.

If you do not need to make a custom build, you can find the latest image at: https://hub.docker.com/repository/docker/girardkyle/ml4t-vscode.

Notes:

This project will not build on Apple Silicon as is. Many of the package versions referenced in the environment.yml do not support Apple Silicon. If you are not using this project for ML4T feel free to try updating the package versions in environment.yml—you should get a successful build on Apple Silicon.

Creating a new build

Docker command: docker build -t imagename:tagname dir

dir is the directory of your Dockerfile

Example

docker build -t girardkyle/ml4t-vscode:latest ./

Running a container

The following command will run a container with two key features enabled:

X11 display forwarding-–this enables the output of GUI items from Docker to the host; you can now use matplotlib plots your heart's content from within a Docker container.
Directory binding—-this will allow you to open your current directory as a directory from within the docker container; you can change files locally and run them in Docker instantly.

docker run -e DISPLAY=192.168.1.101:0 --mount type=bind,source="$(pwd)",target=/workspace/code_root -it --port 8888:8888 girardkyle/ml4t-vscode:latest

Jupyter

Start Jupyter manually from within container

jupyter notebook --ip=0.0.0.0 --port=8888 --no-browser > /dev/null 2>&1 &

Killing Jupyter

ps aux | grep jupyter
kill -9 pid

Reference links

General

https://docs.docker.com/storage/bind-mounts/

X11

http://wiki.ros.org/docker/Tutorials/GUI
https://www.youtube.com/watch?v=cNDR6Z24KLM (Note: running the command xhost + is unsafe. See first X11 link.)

Docker image inheritance