perf: use fast dashify function in parser.js

[hyperz111]

Benchmark file:

import Benchmark from "benchmark";
import postcssJs from "postcss-js";
import postcssJs2 from "../postcss-js/index.mjs";

const suite = new Benchmark.Suite();
const style = {
  top: 10,
  '&:hover': {
    top: 5
  }
};

suite
        .add("postcss-js", () => postcssJs.parse(style))
        .add("postcss-js (fast dashify)", () => postcssJs2.parse(style))
        .on("cycle", function (event) {
                console.log(String(event.target));
        })
        .on("complete", function () {
                console.log("Fastest is " + this.filter("fastest").map("name"));
        })
        .run({ async: true });

Benchmark result:

postcss-js x 113,757 ops/sec ±0.58% (58 runs sampled)
postcss-js (fast dashify) x 178,389 ops/sec ±1.86% (60 runs sampled)

[ai]

Wow, great result.

[ai]

I will merge and release it during this week

[hyperz111]

@ai, can you try benchmark it yourself?

[ai]

My results:

postcss-js x 3,467,532 ops/sec ±3.65% (84 runs sampled)
postcss-js (fast dashify) x 3,670,041 ops/sec ±0.87% (88 runs sampled)
Fastest is postcss-js (fast dashify)

But hostly, you are using very simple input CSS. Proper benchmark should use more realistic example (try Bootstrap CSS as we do in PostCSS benchmark).

[ai]

Thanks. Your fix was released in 5.0.2.