perf: use native Array methods instead of for loop

[edison1105]

see
https://www.measurethat.net/Benchmarks/Show/9251/0/array-every-vs-some-vs-for-loop-fork
https://www.measurethat.net/Benchmarks/Show/11704/0/for-loop-vs-arraysome

[CarterLi]

It was unfair.

At least for for loop vs Array.some test case, Array#some only tests element present in the array. In your benchmark:

var array = new Array(1000);

which generates an array with only holes will never call the callback function function(i) { return array[i] === 877; }.

array.some(x => !x) // => false

Besides, the two tests generates different results because they do different things.

---

In my benchmark, Array#some is about 94.6% slower than plain for loop. That is because Array#some is a native function. JIT compiler can't inline callback for native functions in most cases. Even for of loop which uses iterator is much faster than Array#some

---

We use Array methods not because they are fast, but because they are easier to write and result in cleaner code. For library uses, I do think plain for loop are better for performance reasons.

[edison1105]

thats weird.
https://www.measurethat.net/Benchmarks/Show/12451/0/for-loop-vs-arraysome-vs-for-of

We use Array methods not because they are fast, but because they are easier to write and result in cleaner code. For library uses, I do think plain for loop are better for performance reasons.

I dont think so.
see
#3206 (comment)
https://gomakethings.com/how-performant-are-modern-array-methods-vs-old-school-for-loops-in-vanilla-js/

[syuilo]

Basically, simpler code tends to benefit more from optimization by the engines.
So (even if the performance is temporary worse now), this PR should be worth it.
JFYI: #402 (comment)

[CarterLi]

Basically, simpler code tends to benefit more from optimization by the engines

Maybe yes for other cases. But for for loop no.

Because forEach, some and other Array methods are basically implemented ( and can only be ) as plain for loop, with other additional (and usually unnecessary) checks. A standard conformant implementation of forEach can be unexpectedly complex:

https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/forEach#polyfill

I don't think browsers can do lots of things for it. There is no silver bullet.

Note forEach are ES5 things, and being supported since IE 9, Chrome 1. It's been more than 10 years and still no optimization. So no.

[CarterLi]

thats weird.
measurethat.net/Benchmarks/Show/12451/0/for-loop-vs-arraysome-vs-for-of

We use Array methods not because they are fast, but because they are easier to write and result in cleaner code. For library uses, I do think plain for loop are better for performance reasons.

I dont think so.
see
#3206 (comment)
gomakethings.com/how-performant-are-modern-array-methods-vs-old-school-for-loops-in-vanilla-js

I got similar results with console.time

https://runkit.com/carterli/for-loop-vs-array-some

[edison1105]

@CarterLi Indeed, you are right that this change lowers performance.
Close this one.