Performance is mostly effected by choosing a declarative or imperative programming approach:

• ▪ Declarative: a paradigm where you describe WHAT the program does, without specifying how. For example, Array methods .forEach, .reduce and .map are declarative.
• ▪ Imperative: a paradigm where you describe HOW the program should do something. For example, for..of and for loops.

Below is the performance of different looping approaches for 3 common tasks; for each, map, and reduce.

For Each

Map

Reduce

Results

• ▪ Imperative loops are more performant than declarative. Invoking callback functions is not free and adds up.
• ▪ for..of and for loops have approx. the same performance for arrays
• ▪ It's faster to pre-allocate an array with a known length than to rely on .push automatic growth (see Map example).
• ▪ Object destructing does not effect performance.

This makes sense when you break down what each approach is doing:

• ▪ for every iteration in a for loop:
  1. 1. Check i < array.length
  2. 2. i++
• ▪ for every iteration in a for..of loop:
  1. 1. Call an interator method
  2. 2. Declare and initialize a new variable const item

Selecting an Approach

Don't prematurally optimize. When iterating data, selecting an approach shouldn't be based on performance only. Performance is just 1 variable that needs to be considered with others, such as:

• ▪ readability (imperative code can be more verbose)
• ▪ the amount of data your processing. Is it small enough that performance doesn't matter?

References

• Iteration protocols
• Performance benchmarks for different array iteration methods

Feedback

Have any feedback about this note or just want to comment on the state of the economy?