For parallelism we use the matrix and set the environment variables to different tests:
pipeline: test: image: blah commands: - ./testRunner $TEST nightly_build: image: blah commands: - ./testRunner nightlyTest1 - ./testRunner nightlyTest2 - ./deploy when: event: deployment environment: nightly matrix: include: - TEST: test1 - TEST: test2
But when we run our deployment - we don’t want the matrix to be active. I.e. we’d like to fall back to sequential commands.
It doesn’t make sense for us to publish our JVM project for each entry in the matrix (nor can we since maven is immutable). We can work around this by using a script and passing an argument from the matrix and check it in the script - if it isn’t X then just exit 0. This feels hacky and I’m sure it’s solved in 0.6 by using groups for the regular CI and omitting the group in the deployment.