CI/CD Pipelines

Should you wish to adapt the repository, it is recommended that the user sets up a continuous integration (CI)/continuous development (CD) pipeline. A suite of tests has been written in the tests/ folder to make sure the code is behaving as it should. In this section we describe a GitLab-based method for setting up your own CI/CD pipeline.

GitLab

Once the repository is downloaded, link it to an empty repository on a GitLab instance. Presuming the necessary permissions are set up, this can be done using the following lines, replacing the necessary usernames and addresses.

git remote add gitlab git@gitlab.com:username/ROMpEIT.git
git push gitlab --all

Now, add a .gitlab-ci.yml file to the root of the repository. An example is provided in the setup/CI/ folder. This file triggers a CI/CD pipeline on GitLab that pulls a MATLAB-based Docker image and runs the container. In the container, a suite of tests from the tests/ folder are executed.

Before pushing this change to GitLab however, the MATLAB license file variable MLM_LICENSE_FILE needs to be changed to the address of the your instituations license server. Otherwise the tests will fail.

The Docker image pulled is a custom one made with Statistics and Machine Learning Toolbox loaded with MATLAB. This image is 09nwalkerm:matlab:v2. If you require more toolboxes in MATLAB then you must create your own publically available Docker container (on Docker Hub) with the relevant toolboxes. There is a Dockerfile in the setup/CI/ folder to get you started. More info on setting up custom MATLAB containers can be found here.

GitHub Mirroring

If you have forked the GitHub repository of this toolbox and wish to keep pushing changes to GitHub, then GitLab can still be used to run tests automatically through setting up a pull mirror on you Gitlab instance. We recommend following the instructions laid out in this guide to set up this pipeline.