PyF is architectured with different layers:
- The low-level layer is what we call the core layer, it’s pyf.dataflow.
- The mid-level layer is what we call the network layer, it’s in pyf.manager.
- The high-level layer is what we call the componentized layer, it’s in pyf.componentized.
- A webservice, pyf.services, adds scheduling, versionning, storage and design of tubes (data networks).
Each level is based on the level directly under it. At the core level of PyF we have reused, embedded and extended another library called zflow.