Hello Arjan,
I am not sure if editing the root pom would be viable for VI testing as that would really mean maintaining a separate fork of the TCK which would need to be rebased / changes reapplied every time there are changes within the TCK.
Regarding using a runner pom is there any example with the authentication TCK? Within WildFly we have been using runner poms for MicroProfile TCKs for a long time and other standalone TCKs as well but the pattern there tends to be more along the lines of:
- Add a dependency on the pre-built TCK module
- Provision / install the VI.
- Add VI specific arquillian configuration.
- Add a suite.xml to identify the tests to run.
- Define the surefire module to perform the tests as configured.
Looking at the authentication TCK it looks like it needs to be more than a single module as each test is now in it's own maven artifact. But it also doesn't look like it will be viable to reference them all from a runner pom as they look like they are written to require isolation?
On a related point is it necessary for the individual maven projects to create deployments during the build as looking at the code ShrinkWrap is still used to assemble the deployment under test? This looks to be the biggest benefit of the hierarchy?
I am going to try today and create a WildFly runner pom this might be a suitable basis for some VI instructions.
Regards,
Darran Lofthouse.