Persistent Data Structures

Lesley Lai, BoulderJS, March 20, 2019

Definitions

• A persistent data structure
  • Example: git is fully preserved on the disk, and when you make changes and updates, old commits are still there and accessible
  • Old values are preserved
  • First developed as 'imperative data structures'
• Pure functions From Eric Elliot:

  • Given the same input, will always return the same output.
  • Produces no side effects.

  • You MUST return something
• Why would you need immutable state?
  • (from Dan) there's less guesswork involved tracing how data came to be in any given state...
  • so errors due to invalid object state should be impossible or easier to isolate for
  • some immutable libs/tooling let you view your state changes over time, almost like time-travel debugging
  • Instead of Immutable, consider Immer
• Advantages of Persistent Data Structures
  • minimum copying
  • compact history
  • fast comparison

Immutable.js

• The library is called 'immutable' because it never changes data, it creates a new new set of data with the changes you've applied
• Lists are called 'stacks' in Immutable.js