TO BE DESIGNED AND IMPLEMENTED
By default an application instantiates all processes and nodes before starting any STM schedule. To optimize the startup time for a subset of the data pipeline, the startup should be splittable in stages.
In the first stage only the nodes needed for a minimal schedule are being initialized in order to start that schedule as early as possible. While that schedule is running the initialization of more nodes continues in the background. Whenever the necessary nodes are ready, it becomes possible to switch to the schedules including these nodes (see schedule switching).
The schema for .app.json files will need to be updated to contain information about the startup stages.