Some test results are suspect.

[richardspence]

I noticed the destructuring tests for TS indicated 150X slower times than es5.

This is unusual because the emitted code is actually (almost) identical to the es5. The only difference is one additional variable declaration / assignment.

Emitted ts code for test: http://www.typescriptlang.org/play/index.html#src=var%20data%20%3D%20%7B%0D%0A%20%20a%3A%20'foo'%2C%0D%0A%20%20b%3A%20%7Bc%3A%20'd'%7D%2C%0D%0A%20%20arr%3A%20%5B1%2C%202%2C%203%5D%0D%0A%7D%3B%0D%0A%0D%0Afunction%20fn()%20%7B%0D%0A%20%20var%20%7Ba%2C%20b%3A%7Bc%3Ab%7D%2C%20arr%3A%5B%2C%20c%5D%7D%20%3D%20data%3B%0D%0A%20%20return%20c%3B%0D%0A%7D%0D%0A%0D%0AassertEqual(fn()%2C%202)%3B%0D%0Atest(fn)%3B

es5 test:

//emitted js from TS compiler
var data = {
    a: 'foo',
    b: { c: 'd' },
    arr: [1, 2, 3]
};
function fn() {
    var a = data.a, b = data.b.c, _a = data.arr, c = _a[1];
    return c;
}
assertEqual(fn(), 2);
test(fn);

//baseline es5
var data = {
  a: 'foo',
  b: {c: 'd'},
  arr: [1, 2, 3]
};

function fn() {
  var a = data.a,
      b = data.b.c,
      c = data.arr[1];
  return c;
}

assertEqual(fn(), 2);
test(fn);

[richardspence]

It looks like the code emitted from benchmark includes the _read helper from tslib.

This looks to be because donwlevelIteration is turned on in tsc compiler options. That changes the emitted code to enable other scenarios (not tested) - making this slightly an unfair example.

[C-Lodder]

Thanks @richardspence. I'll try disabling donwlevelIteration and see what the outcome is with the next set of tests