⚡️ Extract inlined-classes outside of Arbitrary base-class

[dubzzz]

In a nutshell

❌ New feature
❌ Fix an issue
❌ Documentation improvement
✔️ Other: performance

(✔️: yes, ❌: no)

Potential impacts

Arbitraries in general.

[codesandbox-ci]

This pull request is automatically built and testable in CodeSandbox.

To see build info of the built libraries, click here or the icon next to each commit SHA.

Latest deployment of this branch, based on commit 65dbca1:

Sandbox	Source
Vanilla	Configuration
fast-check-examples	Configuration

[dubzzz]

Here are the stats I got from my local benchmark:

Reference..gen.integer x 4,372 ops/sec ±9.11% (179 runs sampled)
Test.......gen.integer x 4,462 ops/sec ±4.60% (176 runs sampled)
Reference..gen.integer:map x 838 ops/sec ±4.46% (152 runs sampled)
Test.......gen.integer:map x 3,778 ops/sec ±5.22% (178 runs sampled)
Reference..gen.integer:map:map x 506 ops/sec ±3.93% (152 runs sampled)
Test.......gen.integer:map:map x 3,012 ops/sec ±8.07% (162 runs sampled)
Reference..gen.integer:chain x 758 ops/sec ±4.58% (143 runs sampled)
Test.......gen.integer:chain x 3,205 ops/sec ±5.69% (165 runs sampled)
Reference..gen.integer:filter x 797 ops/sec ±4.30% (147 runs sampled)
Test.......gen.integer:filter x 4,002 ops/sec ±4.54% (175 runs sampled)
Reference..gen.integer:noShrink x 746 ops/sec ±3.47% (146 runs sampled)
Test.......gen.integer:noShrink x 3,859 ops/sec ±5.00% (173 runs sampled)
Reference..gen.integer:noBias x 2,810 ops/sec ±4.95% (148 runs sampled)
Test.......gen.integer:noBias x 4,709 ops/sec ±4.62% (179 runs sampled)
Reference..gen.float x 832 ops/sec ±4.29% (145 runs sampled)
Test.......gen.float x 3,738 ops/sec ±4.96% (172 runs sampled)
Reference..gen.floatNext x 838 ops/sec ±4.45% (155 runs sampled)
Test.......gen.floatNext x 3,465 ops/sec ±5.15% (169 runs sampled)
Reference..gen.double x 705 ops/sec ±3.98% (150 runs sampled)
Test.......gen.double x 2,528 ops/sec ±5.96% (169 runs sampled)
Reference..gen.doubleNext x 703 ops/sec ±4.93% (149 runs sampled)
Test.......gen.doubleNext x 2,743 ops/sec ±4.57% (170 runs sampled)

With node 12.18.2

Code of the benchmark

const Benchmark = require('benchmark');
const fcOld = require('../lib-old/fast-check');
const fcNew = require('../lib-new/fast-check');

const { runComplexFailure, runArraySuccess } = require('./tasks');

const MIN_SAMPLES = 100;
const benchConf = { minSamples: MIN_SAMPLES };
const sampler = (fc, arb) => fc.sample(arb);

const algos = [
  { name: 'gen.integer', run: (fc) => sampler(fc, fc.integer()) },
  {
    name: 'gen.integer:map',
    run: (fc) =>
      sampler(
        fc,
        fc.integer().map((i) => i)
      ),
  },
  {
    name: 'gen.integer:map:map',
    run: (fc) =>
      sampler(
        fc,
        fc
          .integer()
          .map((i) => i)
          .map((i) => i)
      ),
  },
  {
    name: 'gen.integer:chain',
    run: (fc) =>
      sampler(
        fc,
        fc.integer().chain((i) => fc.constant(i))
      ),
  },
  {
    name: 'gen.integer:filter',
    run: (fc) =>
      sampler(
        fc,
        fc.integer().filter(() => true)
      ),
  },
  {
    name: 'gen.integer:noShrink',
    run: (fc) => sampler(fc, fc.integer().noShrink()),
  },
  {
    name: 'gen.integer:noBias',
    run: (fc) => sampler(fc, fc.integer().noBias()),
  },
  { name: 'gen.float', run: (fc) => sampler(fc, fc.float()) },
  { name: 'gen.floatNext', run: (fc) => sampler(fc, fc.float({ next: true })) },
  { name: 'gen.double', run: (fc) => sampler(fc, fc.double()) },
  { name: 'gen.doubleNext', run: (fc) => sampler(fc, fc.double({ next: true })) },
];

for (const a of algos) {
  for (let idx = 0; idx !== 1000; ++idx) {}
  a.run(fcOld);
  a.run(fcNew);
}
for (const a of algos) {
  Benchmark.invoke(
    [
      new Benchmark(`Reference..${a.name}`, () => a.run(fcOld), benchConf),
      new Benchmark(`Test.......${a.name}`, () => a.run(fcNew), benchConf),
    ],
    {
      name: 'run',
      queued: true,
      onCycle: (event) => console.log(String(event.target)),
    }
  );
}

[coveralls]

Coverage increased (+0.004%) to 96.872% when pulling 65dbca1 on perf-optim-map into 179bd89 on master.