We may have to use a more sophisticated approach than our current linear approach to releasing options

[mulkieran]

The right way is a graph, a topological sort, etc. That is, we have objects representing tasks, we insert them into a graph based on the options chosen and their dependencies, we do a topological sort of the graph to find an order in which to execute them, then we run the tasks in the order the topological sort gives us. Maybe we do a dry run in a different way, i.e., not using mock.