events: improve for-loop

[gengjiawen]

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• commit message follows commit guidelines

[lpinca]

Is for...of as fast as classic for now? (It probably doesn't matter in this case)

[gengjiawen]

When test count is small (TEST_COUNT = Math.pow(10, 2)), forOf is faster

forLoop: 0.205ms
forOf: 0.107ms

When test count is bigger (TEST_COUNT = Math.pow(10, 8)), forLoop is faster

forLoop: 884.075ms
forOf: 1629.238ms

Env:

  System:
    OS: macOS 10.14.3
    CPU: (8) x64 Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
    Memory: 1.27 GB / 16.00 GB
    Shell: 3.0.0 - /usr/local/bin/fish
  Binaries:
    Node: 11.10.1 - /usr/local/bin/node
    Yarn: 1.13.0 - /usr/local/bin/yarn
    npm: 6.7.0 - /usr/local/bin/npm

Test snippet

const TEST_COUNT = Math.pow(10, 2);

const pojo = {
    name: 'n',
    age: 10
}

function Benckmark(desc, func) {
    console.time(desc)
    let sum = 0;
    for (let index = 0; index < TEST_COUNT; index++) {
        func()
    }
    console.timeEnd(desc)
}

const noop = () => {}

function forLoop(o) {
    const keys = Object.keys(o)
    for (let i = 0; i < keys.length; ++i) {	
        const key = keys[i]
        noop(key)
    }
}

function forOf(o) {
    for (const key of Object.keys(o)) {	
        noop(key)
    }
}

Benckmark('forLoop', () => forLoop(pojo))
Benckmark('forOf', () => forOf(pojo))

[BridgeAR]

@gengjiawen turning around the execution order is also going to turn around the results for what is faster for big n. Therefore this benchmark does not seem reliable.

The difference between the two loops is not very big but it is ~15%. See https://bugs.chromium.org/p/v8/issues/detail?id=8070 for the tracking bug that I opened last year for it.

[BridgeAR]

CI https://ci.nodejs.org/job/node-test-pull-request/21171/

[BridgeAR]

Landed in d79176a 🎉