Python Finite State Machines for EPICS¶
Pysmlib is a Python library which helps creating event based finite state machines (fsm) for Epics Control System. Each fsm runs in a different thread and, by default, continues its execution until explicitly stopped. A convenient loader is provided to help loading multiple fsm together, thus creating an always-on daemon. Full integration with Epics Channel Access protocol is provided via PyEpics. The user can connect to Process Variables by defining an fsm input / output (I/O) and can therefore access its values and changes via convenient methods. The fsm current state is executed every time one of the connected inputs changes its value or its connection state, so that the user can evaluate the actions to be performed, including changing state. In some cases, the user may want to execute some actions after a certain amount of time (eg: when a timeout expires) and so the library includes a timer facility which execute the current state after the specified delay. Other useful features include a simple way to print logs in an unified way and the possibility to register a specific I/O as watchdog, meaning that the fsm will automatically write it periodically, so that external systems can be informed of the online or offline status of the fsm daemon.
The library is designed with network efficiency and system responsiveness in mind: it’s usually important to act as soon as possible upon the change of an input, without overflowing the network with useless traffic. This is achieved by choosing the daemon-like execution, which is the obvious choice in case of always-on algorithms (eg: a PID) but can be used also for one time procedures. In fact, an fsm can remain in a idle state, where no action is performed, until a certain condition is met (eg: a rising edge on the “enable” input) and then start executing the procedure, finally returning to the idle state. This means that when the enable arrives, all the I/Os are already connected and the fsm doesn’t have to wait for all the connection times. The downside here is the network overload due to many connections which remain active for a long time. For this reason the I/Os are shared between all the fsm loaded on the same daemon, so that the minimum number of connections is required. Then, when an event related to a certain PV arrives, the library executes all the fsm using that input and guarantees that the input doesn’t change during the state evaluation and that two or more fsm don’t interfere with each other. As a result it’s usually recommended to group all the fsm that use related I/Os in a single daemon, just remember that each fsm is a python thread!
For all these reasons, pysmlib is a great solution to develop high level automation and control systems in any facility using Epics. It enables the user to focus on the algorithms to implement without worrying about low-level problems.
Main features include:¶
Easy to use and fast development of complex event based fsm - just code the states!
Full EPICS Channel Access integration via PyEpics.
High expandability as provided by all the libraries of Python.
Integrated configurables logging systems.
Convenient methods to access all the information on I/O.
Timers can be used to execute actions after a time delay.
Integrated watchdog logic.
Multi-threading: each fsm is executed on a different thread, sharing I/O.
Convenient loader to launch a daemon with multiple fsm.
Possibility to apply a configurable naming convention on I/O.