Remove process.exit commands

[georg-schwarz]

In the interpreter-lib, there were still some process.exit commands. IMO, only surrounding applications should trigger a process exit.

The details of the FAILURE are logged / can be extracted by injecting a custom logger.

[georg-schwarz]

I wonder if we should distinguish between different kinds of FAILURES, like CONFIGURATION_FAILURE, RUNTIME_FAILURE, RUNTIME_ERROR, and so on.

[rhazn]

Would be good, yeah. Note though, this is the exit code of the process so we can not freely define new ones (afaik). We definitely should have some form of more elaborate error output outside of the exit code though, like writing a logfile.

[georg-schwarz]

I want to draft a release after integrating this. Should we do version v0.2.0 to keep v0.1.x reserved for people using the current version and only supply bug fixes there?
In theory, there should be no breaking changes - but I wouldn't guarantee for not having introduced bugs with the refactorings like #460.

[rhazn]

I want to draft a release after integrating this. Should we do version v0.2.0 to keep v0.1.x reserved for people using the current version and only supply bug fixes there? In theory, there should be no breaking changes - but I wouldn't guarantee for not having introduced bugs with the refactorings like #460.

I am not sure this doesn't change behavior. Was the cli-interpreter always also just returning exit code 1? Because otherwise, this is also a change in functionality that would mean 0.2.0 imho. But if there is no behavior change we should go with 0.1.1 and trust the tests I think?

Maybe in the future we should do some MADE only releases (e.g. release v.WS2324.MADE) so we can be more flexible with keeping a stable version for MADE and still release new versions for general users.