好吧，var x = new Array()与var x =[]在某些特性上是不同的，我只解释其中最有用的两个(在我看来)。

在我解释这些区别之前，我将首先设定一个基础;当我们使用x =[]时，定义了一个数据类型为Array的新变量，它继承了属于数组原型的所有方法，这与扩展一个类非常相似(但不完全相同)。但是，当我们使用x = new Array()时，它会初始化分配给变量x的数组原型的克隆。

现在让我们看看有什么不同

The First Difference is that using new Array(x) where x is an integer, initilizes an array of x undefined values, for example new Array(16) will initialize an array with 16 items all of them are undefined. This is very useful when you asynchronously fill an array of a predefined length. For example (again :) ) let's say you are getting the results of 100 competitiors, and you're receiving them asynchronously from a remote system or db, then you'll need to allocate them in the array according to the rank once you receive each result. In this very rare case you will do something like myArray[result.rank - 1] = result.name, so the rank 1 will be set to the index 0 and so on.

The second difference is that using new Array() as you already know, instanciates a whole new clone of the array prototype and assigns it to your variable, that allows you to do some magic (not recommended btw). This magic is that you can overwrite a specific method of the legacy array methods. So, for example you can set the Array.push method to push the new value to the beginning of the array instead of the end, and you can also add new methods (this is better) to this specific clone of the Array Prototype. That will allow you to define more complex types of arrays throughout your project with your own added methods and use it as a class.

最后一件事，如果你来自极少数关心应用程序处理开销和内存消耗的人(我真的很喜欢)，你永远不会不顾一切地使用新Array():)。

我希望这已经足够解释野兽new Array():)

事情远比看上去的复杂。大多数其他答案是正确的，但也…

新数组(n)

允许引擎为n个元素重新分配空间 为创建数组而优化 创建的数组被标记为稀疏数组具有最少的性能数组操作，这是因为每个索引访问都必须检查边界，查看是否存在值并遍历原型链 如果数组被标记为稀疏，就没有办法返回(至少在V8中)，它在生命周期内总是会变慢，即使你在1毫秒或2小时后用内容填充它(打包的数组)也没关系

[1, 2, 3] | | []

创建的数组被标记为已打包(除非使用delete或[1，，3]语法) 针对数组操作进行了优化(用于..， forEach, map等) 随着数组的增长，引擎需要重新分配空间

对于旧版本的浏览器可能不是这样。

当你初始化没有任何长度的数组时，没有区别。因此var a = [] & var b = new Array()是相同的。

但是如果你初始化数组的长度像var b = new array(1);，它将数组对象的长度设置为1。所以它等价于var b = [];b.length = 1;。

当你使用array_object时，这就会有问题。按下，它在最后一个元素后添加项目并增加长度。

var b = new Array(1);
b.push("hello world");
console.log(b.length); // print 2

vs

var v = [];
a.push("hello world");
console.log(b.length); // print 1

好吧，var x = new Array()与var x =[]在某些特性上是不同的，我只解释其中最有用的两个(在我看来)。

在我解释这些区别之前，我将首先设定一个基础;当我们使用x =[]时，定义了一个数据类型为Array的新变量，它继承了属于数组原型的所有方法，这与扩展一个类非常相似(但不完全相同)。但是，当我们使用x = new Array()时，它会初始化分配给变量x的数组原型的克隆。

现在让我们看看有什么不同

The First Difference is that using new Array(x) where x is an integer, initilizes an array of x undefined values, for example new Array(16) will initialize an array with 16 items all of them are undefined. This is very useful when you asynchronously fill an array of a predefined length. For example (again :) ) let's say you are getting the results of 100 competitiors, and you're receiving them asynchronously from a remote system or db, then you'll need to allocate them in the array according to the rank once you receive each result. In this very rare case you will do something like myArray[result.rank - 1] = result.name, so the rank 1 will be set to the index 0 and so on.

The second difference is that using new Array() as you already know, instanciates a whole new clone of the array prototype and assigns it to your variable, that allows you to do some magic (not recommended btw). This magic is that you can overwrite a specific method of the legacy array methods. So, for example you can set the Array.push method to push the new value to the beginning of the array instead of the end, and you can also add new methods (this is better) to this specific clone of the Array Prototype. That will allow you to define more complex types of arrays throughout your project with your own added methods and use it as a class.

最后一件事，如果你来自极少数关心应用程序处理开销和内存消耗的人(我真的很喜欢)，你永远不会不顾一切地使用新Array():)。

我希望这已经足够解释野兽new Array():)

我在使用[]时出现了奇怪的行为。

我们有模型“类”，字段初始化为一些值。例如:

require([
  "dojo/_base/declare",
  "dijit/_WidgetBase",
], function(declare, parser, ready, _WidgetBase){

   declare("MyWidget", [_WidgetBase], {
     field1: [],
     field2: "",
     function1: function(),
     function2: function()
   });    
});

我发现当字段用[]初始化时，它将被所有Model对象共享。对其中一项进行更改会影响所有其他项。

用new Array()初始化它们时不会发生这种情况。对象的初始化也是一样({}vs new Object())

TBH我不确定这是否是我们使用的框架(Dojo)的问题

在使用承诺时，我发现了一个不同。在使用承诺数组(例如arr，初始化为arr=[])时，在Promise.all(arr)中得到一个错误。而当声明为arr = Array()时，没有得到编译问题。希望这能有所帮助。