matlab-unitest

The purpose of this project is to enable ci on matlab projects in order to launch unit tests. A full description of the process is given here, this project is just an adaptation of the workflow to work on gitlab.inria.fr with inria matlab licence servers. We use only one simple function written in the src folder associated with one unit test put in the +test folder

General considerations

You first have to enable CI/CD feature for your project as described in the intro project.

Do not enable shared runners in this context because the shared runner will not be able to connect any matlab licence server. We have to define our own runner on a ci.inria.fr virtual machine.

Register the matlab server url as a CI variable

To be able to register your matlab version you have to register the Licence server url as a CI variable. Browse to Settings - CI/CD - Variables and add a new variable named MATLAB_SERVER with the licence server url as value (you can find it in the licence.txt files at the very beginning of the file, it has to be something like: 27000@myserver.inria.fr). Check the "Protect variable" and "Mask variable" properties for security.

Create your ci.inria.fr VM with gitlab-runner

Follow the instruction of the CI documentation to create a slave on ci.inria.fr. You can also connect to any of your domains in cloudstack and use a VM you already defined for other projects which avoid creating VM for a single use.

When your VM is created, log into it via ssh as super user (sudo su as ci user in ubuntu VM for exemple). Then install/update gitlab-runner with:

curl -L "https://packages.gitlab.com/install/repositories/runner/gitlab-runner/script.deb.sh" | sudo bash apt-get update

and register your project by using the instructions of the Settings - CI/CD - Runners - Specific runners url and token:

gitlab-runner register

In our example we set the executor as docker, the image as docker:20.10.16 and tags as matlab-ci.

Create your .gitlab-ci.yml

We defined too jobs in our project. The first one is just the execution of the rand matlab function to check that the matlab image as been correctly loaded. The second one is executing the unit test of the +test folder and report the results and the coverage in the artifacts as junit format (to have more details on the unit tests and coverage reports, see the pytest-gitlab-reports project).