In node.js, why is there a util.isArray and an Array.isArray?

To actually answer why util.isArray exists, we need a bit of a history lesson.

When it was first added to node, it did a bit more than call Array.isArray.

function isArray (ar) {
  return ar instanceof Array
      || Array.isArray(ar)
      || (ar && ar !== Object.prototype && isArray(ar.__proto__));
}

This was a local function in utils and actually wasn’t exported until v0.6.0.

In this form, util.isArray handled a case that Array.isArray doesn’t:

> x = [1,2,3]
[ 1, 2, 3 ]
> y = Object.create(x)
[ , ,  ]
> Array.isArray(y)
false
> Array.isArray(Object.getPrototypeOf(y))
true

There’s some discussion here about this behavior of util.isArray, and consensus was that this behavior is actually bad because y is not really an Array.

Thus, the prototype-checking functionality was soon removed and replaced with a check that used both Array.isArray and a check of the argument’s [[Class]].

function isArray(ar) {
  return Array.isArray(ar) ||
         (typeof ar === 'object' && objectToString(ar) === '[object Array]');
}

However, checking the [[Class]] is actually duplicate effort because Array.isArray also checks the [[Class]], so it too was eventually removed – leaving only a call to Array.isArray.

Today, util.isArray is just an alias of Array.isArray.

So in other words, the existence of util.isArray is mostly a legacy thing and can be safely ignored.