除了可维护性，第一种方法消除了意外创建全局变量的可能性:

(function () {
var variable1 = "Hello, World!" // Semicolon is missed out accidentally
var variable2 = "Testing..."; // Still a local variable
var variable3 = 42;
}());

而第二种方式就不那么宽容了:

(function () {
var variable1 = "Hello, World!" // Comma is missed out accidentally
    variable2 = "Testing...", // Becomes a global variable
    variable3 = 42; // A global variable as well
}());

除了可维护性，第一种方法消除了意外创建全局变量的可能性:

(function () {
var variable1 = "Hello, World!" // Semicolon is missed out accidentally
var variable2 = "Testing..."; // Still a local variable
var variable3 = 42;
}());

而第二种方式就不那么宽容了:

(function () {
var variable1 = "Hello, World!" // Comma is missed out accidentally
    variable2 = "Testing...", // Becomes a global variable
    variable3 = 42; // A global variable as well
}());

这样做的时候可读性更强:

var hey = 23;
var hi = 3;
var howdy 4;

但是这种方法占用的空间和代码行数更少:

var hey=23,hi=3,howdy=4;

它是节省空间的理想选择，但是让JavaScript压缩器为您处理它。

我认为第一种方法(多个变量)是最好的，否则你就会得到这样的结果(来自使用KnockoutJS的应用程序)，这在我看来很难理解:

    var categories = ko.observableArray(),
        keywordFilter = ko.observableArray(),
        omniFilter = ko.observable('').extend({ throttle: 300 }),
        filteredCategories = ko.computed(function () {
            var underlyingArray = categories();
            return ko.utils.arrayFilter(underlyingArray, function (n) {
                return n.FilteredSportCount() > 0;
            });
        }),
        favoriteSports = ko.computed(function () {
            var sports = ko.observableArray();
            ko.utils.arrayForEach(categories(), function (c) {
                ko.utils.arrayForEach(c.Sports(), function (a) {
                    if (a.IsFavorite()) {
                        sports.push(a);
                    }
                });
            });
            return sports;
        }),
        toggleFavorite = function (sport, userId) {
            var isFavorite = sport.IsFavorite();

            var url = setfavouritesurl;

            var data = {
                userId: userId,
                sportId: sport.Id(),
                isFavourite: !isFavorite
            };

            var callback = function () {
                sport.IsFavorite(!isFavorite);
            };

            jQuery.support.cors = true;
            jQuery.ajax({
                url: url,
                type: "GET",
                data: data,
                success: callback
            });
        },
        hasfavoriteSports = ko.computed(function () {
            var result = false;
            ko.utils.arrayForEach(categories(), function (c) {
                ko.utils.arrayForEach(c.Sports(), function (a) {
                    if (a.IsFavorite()) {
                        result = true;
                    }
                });
            });
            return result;
        });

是一个很好的特性。没有理由逃避。随着技术的发展，我们也必须自我进化。这个特性在Perl等一些语言中已经存在很长时间了。例如构建WebGL网格，新的javascript样式

   //initialize vertices with some calculated points
   [verts[ix],     verts[iy],     verts[iz]]     = ps[0];
   [verts[ix + 3], verts[iy + 3], verts[iz + 3]] = ps[1];
   [verts[ix + 6], verts[iy + 6], verts[iz + 6]] = ps[2];
   //initializing normals with cross products
   [norms[ix],     norms[iy],     norms[iz]]     = cr;
   [norms[ix + 3], norms[iy + 3], norms[iz + 3]] = cr;
   [norms[ix + 6], norms[iy + 6], norms[iz + 6]] = cr;

事实上，老式的代码更难调试，到目前为止也更难理解和发现任何bug。这个样本过于简化了。大量重复的例程阻碍了真正的逻辑，使代码看起来像某种魔法。和上面一样，不过是古风:

      //initialize vertices with some calculated points
      verts[ix]     =  ps[0][0];
      verts[iy]     =  ps[0][1];
      verts[iz]     =  ps[0][2];
      verts[ix + 3] =  ps[1][0];
      verts[iy + 3] =  ps[1][1];
      verts[iz + 3] =  ps[1][2];
      verts[ix + 6] =  ps[2][0];
      verts[iy + 6] =  ps[2][1];
      verts[iz + 6] =  ps[2][2];
      //initializing normals with cross products
      norms[ix]     =  cr[0];
      norms[iy]     =  cr[1];
      norms[iz]     =  cr[2];
      norms[ix + 3] =  cr[0];
      norms[iy + 3] =  cr[1];
      norms[iz + 3] =  cr[2];
      norms[ix + 6] =  cr[0];
      norms[iy + 6] =  cr[1];
      norms[iz + 6] =  cr[2];

注意，在迁移新样式的代码时，我不仅大量删除了大块的例程代码。我很容易就发现了那些逃避了大量代码审查的不一致之处，这只是因为我让代码更容易可视化，更简洁，更面向逻辑，更不面向例程。

我相信在我们开始使用ES6之前，使用单一的var声明的方法既不好也不好(如果你有lints和'use strict'的话)。这真的是一种口味偏好。但现在我的情况变了。以下是我对多行声明的一些看法:

Now we have two new kinds of variables, and var became obsolete. It is good practice to use const everywhere until you really need let. So quite often your code will contain variable declarations with assignment in the middle of the code, and because of block scoping you quite often will move variables between blocks in case of small changes. I think that it is more convenient to do that with multiline declarations. ES6 syntax became more diverse, we got destructors, template strings, arrow functions and optional assignments. When you heavily use all those features with single variable declarations, it hurts readability.