Dynalink uses the registry for storing configuration details. This includes runtime settings as well as paths to any of the configuration databases located on the system. While developing the new version of the platform several things bothered me about this approach but one thing specifically with handing different versions of the product on the same machine. Our goal is to allow for side by side versions of the service application. Each capable of service a different set of databases or versions of MAS. This would work well when testing new environments or migration work. I am not a fan of the Registry and believe that having the configuration stored in the file system local to the runtime is a much more desirable solution.
To that end the latest build now uses local configuration files to handle the runtime details of the server.