Use Finite-state Machine to manage state

Introduction

Main Elements:

• States: must be limited
• Events
• Transitions: when event 1 from state A to state B
• Actions: when state|event|transition changed then act action

from https://en.wikipedia.org/wiki/Finite-state_machine, Fig. 3

Why We Need this

To make sure complex things never gone off the rails

• Add constraints on states, throw errors to avoid bad things
• Help to sort out reaction when state changed
• Code is documentation

Different with Flux-like State Managements

Flux-like State Managements(Redux/MobX/NgRx) is a way to manage states by utilizing unidirectional data flows, it is often to using this kind of state management with functional programming.

Redux lifecycle, from https://dev.to/radiumsharma06/abc-of-redux-5461, Fig. 1

It is easier to use FSM to manage state with object-oriented programming, but that did’t mean you can’t choose flux-like state management as your state management in object-oriented programming.

Actually, I kind of perfer using flux-like state management in morden frontend framework(React/Vue/Angular), because of these framework encourage you to use functional programming/unidirectional data flows.

Most of backend frameworks(Spring/Rails/Django/.NET) is develop by object-oriented programming languages, so using FSM to manage object state is totally reasonable.

Libaries

For Ruby

• https://github.com/aasm/aasm

For JavaScript or TypeScript

• https://github.com/davidkpiano/xstate

For C#

• https://github.com/dotnet-state-machine/stateless

References

• https://en.wikipedia.org/wiki/Finite-state_machine
• https://github.com/aasm/aasm
• https://github.com/davidkpiano/xstate
• https://github.com/dotnet-state-machine/stateless
• https://stackoverflow.com/a/54521035
• https://medium.com/pm的生產力工具箱/如何有邏輯的釐清事物的狀態-f9fb59b15054
• https://facebook.github.io/flux/docs/in-depth-overview/
• https://dev.to/radiumsharma06/abc-of-redux-5461