Proposal: Role override
Hello,
I would like to propose the idea of adding the ability to configure "role overrides" -- this idea came to me after !1310 (closed) MR was rejected.
To give some context, my firm is very keen on having this supported as high availability is crucial in our BCP/DR planning, and having a single point of failure is not ideal.
Problem
Right now, we are not on GET officially but we are in the process of adopting it. This has been wonderful in our testing but it gets problematic if we need to implement a workaround to accomplish something mandatory which does not align with GET's scope.
The current ability to add custom tasks is wonderful, but it is still limited in cases where we have to change certain logic in the roles embedded in the collection. Right now in our use case we create a wrapper collection which includes the GET collection as a dependency and it would be challenging for us to change the files directly during runtime.
Consideration: Maintaining own fork
While we have the choice of maintaining our own fork, keeping our fork updated with latest changes from the official repo will definitely be a challenge.
Proposal
What I would like to propose is a mechanism similar to custom tasks that have already been implemented, but instead the full role can be written and stored in the file. We have a check at the beginning to see if this file exists and if it does load and run all tasks within it -- skipping all other steps if it does (thus acting like a full role override)
I hope this makes sense and is something that could be supported.