A benchmark that treats a relative path to a file in the same directory, it shows that node`s `path` and `url` APIs in are not always the fastest way.

gistfile1.js

var Benchmark = require('benchmark');
var path = require('path');
var url = require('url');
var suite = new Benchmark.Suite();
var p1 = 'path/to/file';
var p2 = './another_file';
function a() {
  return path.join(p1, '..', p2);
}

function a1() {
  return path.join(p1, '..', p2).replace(/\\/, '/');
}

function a2() {
  var dir;
  dir = p1.split('/');
  dir.pop();
  dir.push(p2.replace('./', ''));
  return dir.join('/');
}

function a3() {
  var base = path.basename(p1);
  return p1.replace(base, '') + p2.replace('./', '');
}

function a4() {
  return url.resolve('path/to/file', './another_file');
}
//check output
console.log(a(), '-> path.join');
console.log(a1(), '-> path.join with replace with to unix separator:');
console.log(a2(), '-> split');
console.log(a3(), '-> cut');
console.log(a4(), '-> url.resolve');
console.log('');


// add tests
suite
  .add('path.join', a)
  .add('path.join with replace with to unix separator', a1)
  .add('split', a2)
  .add('cut', a3)
  .add('url.resolve', a4)
// add listeners
  .on('cycle', function(event) {
    console.log(String(event.target));
  })
  .on('complete', function() {
    console.log('Fastest is ' + this.filter('fastest').pluck('name'));
  })
// run async
  .run({ 'async': true });

  //results:
  //
  // path/to/another_file -> path.join
  // path/to/another_file -> path.join with replace with to unix separator:
  // path/to/another_file -> split
  // path/to/another_file -> cut
  // path/to/another_file -> url.resolve
  // 
  // path.join x 125,439 ops/sec ±20.17% (57 runs sampled)
  // path.join with replace with to unix separator x 109,778 ops/sec ±15.86% (67 runs sampled)
  // split x 1,577,144 ops/sec ±17.13% (59 runs sampled)
  // cut x 559,393 ops/sec ±18.45% (51 runs sampled)
  // url.resolve x 18,055 ops/sec ±16.97% (59 runs sampled)
  // Fastest is split