feat: parameter object is null object

[Uzlopak]

see fastify/fastify#4357

@Eomm
@mcollina
@jsumners

I did not bench

[Uzlopak]

tests fail because of tap issues...

[climba03003]

Cleaner and Simpler

new Function('paramsArray', `return { __proto__:null,${lines.join(',')}}`)

[Uzlopak]

Need to benchmark.

[mcollina]

I think this shoukd be a semver-major change.

[Uzlopak]

@climba03003

It seems that new NullObject is faster than using proto trick

node benchmark/null-object.js 
new NullObject() x 1,377,691,231 ops/sec ±0.11% (194 runs sampled)
{__proto__: null } x 33,097,458 ops/sec ±0.50% (185 runs sampled)

'use strict'

const { NullObject } = require('../lib/null-object')

const Benchmark = require('benchmark')
Benchmark.options.minSamples = 100

const suite = Benchmark.Suite()

suite.add(`new NullObject()`, function () {
  const result = new NullObject()
})
suite.add(`{__proto__: null }`, function () {
  const result = {__proto__: null }
})
suite
  .on('cycle', function (event) {
    console.log(String(event.target))
  })
  .on('complete', function () {
  })
  .run()

[mcollina]

Can you exclude windows v14 from CI?

[Uzlopak]

@mcollina done.

I would like to write another benchmark for the actual parameter generation. So please dont merge to fast.

I posted my benchmark to discuss if proto is the right way or not.

[mcollina]

lgtm looks nice!

[mcollina]

PTAL @ivan-tymoshenko

[ivan-tymoshenko]

lgtm

[ivan-tymoshenko]

@Uzlopak Can you post the bechmarks comparison pls?

[mcollina]

can you merge main?

[Uzlopak]

@mcollina done