Multi-stage scheduling is a concept which reuses the same data model to calculate the result in an iterative fashion (e.g. we schedule airplane routes in the first stage and based on the result we schedule pilot shifts). The scheme may require definition of multiple planning solution (@PlanningSolution) data objects.

By default, solver configuration uses <scanAnnotatedClasses> element, preventing usage of multiple planning solutions, as the engine fails fast with this configuration.

Initial requirements:

• When multiple planning solutions are present, the solver configuration should provide a way to specify a planning solution class to be used. <scanAnnotatedClasses> option should remain a default value though.
• Each stage will use its own solver configuration file and will be identified by a different knowledge base.
• Each stage can use its own set of guided rules/.drl files. Investigate a way to group the rules/.drl files by a knowledge base they belong to.