Building Component Networks
This chapter will give you a complete view on the different aspects of how components are composed within a Rock/Syskit system. This touches a lot of different subjects, but we're going to start with the basics, that is how component networks are modelled and integrated in Syskit. The first few sections will thus go into more details about what we've seen in the basics chapter. The one major addition being the aspects of the modelling that allow to reuse networks and models across systems (most importantly, between the real and simulated systems).
After that, we will broach more complex subject related to the component network: - how the system designer can influence how data flows between the components - time-ordered data processing - modelling and integration of transformations
Component Networks in a Syskit Application
A Syskit application can be understood, or "read" in two different ways.
The functional structure. These are the separate functions - sometimes also called behaviors that are offered by the application, as well as how these functions are related to each other. One way to build these functions is to use Rock components. This will be the subject of this chapter. Other ways to build and compose functions will be presented in the coordination chapter.
The temporal structure. This is the main subject of the coordination chapter. It deals with how one can make a Syskit application represents what happens in the real world, and how the system can evolve in response to these events.
On the subject of component networks, the main concepts are compositions and profiles. The former allows to combine components together, the aim being the building of a library of generic functions. The latter allows to fine-tune these functions to adapt them to a specific robot configuration and/or to a specific situation. The networks are exposed on a profile are what is actually exposed to the coordination layer.
Structure of this Chapter
In this chapter, we will start by detailing and expanding what we have already seen in the basics chapter. We are assuming that you have read the Runtime section of the Integrating Functionality chapter, so we're going to jump straight at compositions and profiles. We will then go into the mechanisms that allow to build reusable models, a.k.a. how to make most of the models robot-independent, to then fine-tune the models for each system.
The rest of the chapter will details more advanced aspects of the design of the component network(s). For now, you can find more about the system handling of geometric transformations, a critical tool to build components that are (mostly) system-independent, promoting reuse.