# build Docker image with GCC installed
docker build -t testing_container testing_container/
# install Python client library to control Docker
sudo pip install docker-py
You also need to make sure that Docker's environment variables DOCKER_HOST (and DOCKER_TLS_VERIFY and DOCKER_CERT_PATH if using TLS) are set and accessible when you run framework commands.
This is a very simple framework that isolates untrusted code and your system. Student code is compiled and run inside a clean Docker container without network access, and code on your machine sets up files and scripts for the container to use.
The input directory should contain student code files, test datasets, and test scripts. This directory is made available in the container to your supplied test script as /mnt/input/.
The test script will be executed as Docker container's root user. It is responsible for running student code as framework-provided testing user, as well as setting up file permissions for the testing user. Test scripts should redirect/write whatever output needed for performing evaluation later and providing student feedback to /mnt/output/ in the container. The framework ensures that /mnt/output/ directory (used for grading) is not accessible by the testing user. Finally and optionally, your test script is also responsible for limiting the time allowed for student code to run.
Whatever is in the container's /mnt/output/ appears in your supplied output directory.
The test manifest file provides the framework with the following information:
- Which test script (in container input volume) to run, once the container is started
- How much system resource should be allocated to the container (CPU utilization, number of CPUs, etc.) [planned]
/path/to/irun test_manifest input_directory output_directory
For easiest access in your shell, add this project's bin/ to your PATH.