Performance rule js-flatmap-filter doesn't appear to provide a performance improvement

[nspector]

Hi! I was evaluating this rule:

Performance: .map().filter(Boolean) loops twice

https://react.doctor/docs/rules/react-doctor/js-flatmap-filter

Since this is categorized as a performance optimization, I wanted to verify the claim with a simple benchmark.

I tested three implementations over an array of 10,000 items:

// map + filter
DATA.map((_, i) => i % 2 ? i : null).filter(Boolean)

// flatMap
DATA.flatMap((_, i) => i % 2 ? i : [])

// reduce
DATA.reduce((acc, _, i) => {
  if (i % 2) {
    acc.push(i);
  }
  return acc;
}, [])

On my machine, the results were approximately:

Implementation	Ops/s
map().filter(Boolean)	~20k
flatMap()	~15k
reduce()	~75k

In this benchmark, flatMap() was slower than map().filter(Boolean) despite only making a single pass.

I realize this is only one benchmark and that performance varies between JavaScript engines. However, since this rule is categorized as a performance optimization, I expected flatMap() to consistently outperform the two-pass version, which wasn't the case here.

Given that, I wonder if this rule should either:

• be categorized as a readability or style recommendation rather than a performance optimization
• include evidence or caveats showing when flatMap() is actually expected to improve performance
• updated to recommend alternatives such as reduce()

For reference, here's the benchmark I used:
https://jsbm.dev/5L2w7oKTpySSb (including map by itself)

[react-doctor-evals]

❌ Triage errored — trace cba876b76e7f1dd98c60358f0e1ca671. Check server logs for that trace ID.