cpressey / facts-about-state-machines Goto Github PK

I am missing a reference to SDL/Z.100 between OMT and UML. This was heavily used in the telecom industry in the 80's/90's.

Please could you add code snippets of the different types of state machine implementation. It is easier to visualize and perhaps even use the appropriate code snippet.

Thank you - this is a good, comprehensive description of state machines.

Thanks for writing these notes.

On the statement that state machine transitions are neither functional or imperative, they can also be both. We distinguish commands and events. Commands drive state machines and events are emitted by them and reflect what has happened. Commands can be side effecting and emit an event. An event is then applied to a state to yield another state and has no side effects ie it is a pure function. This is useful when event sourcing to reconstitute state as events can be replayed without side effects.

Here’s a blog I wrote on this topic a while back: http://christopherhunt-software.blogspot.com/2021/02/event-driven-finite-state-machines.html

Here’s an implementation of what the blog describes using Rust: http://christopherhunt-software.blogspot.com/2021/02/event-driven-finite-state-machines.html

I’d be interested in your thoughts regarding what we’ve done. Thanks again for your work.

Great article, first. Found it on Hacker News. But I did notice a section that could be revised. It is the section titled A complex state machine means your system has complex behaviour:

Firstly, the statement is false because it makes a statement about all machines of a type but doesn't include any exceptions. For example, a heavily complex state machine could always result in the output "false" (or "true"). This behaviour is actually quite simple but the state machine to produce it could be incredibly complex for some reason. I suppose you could make an argument that the "internal" behaviour is complex, but even this is a stretch of the possible truths.

Secondly, the converse is false as well; a very simple three-state machine could produce incredibly complex behaviour that could appear completely random to outside observers. I would recommend reading on this topic, especially with a nod to Wolfram's computational reduction arguments. He has produced very simple state machines that produce incredibly detailed and complex behaviours.

I would encourage you to redesign this section to qualify or give specific examples of complex machines with complex behaviours versus simple machines with simple behaviours --- and then describe how the opposite is also possible.