Is Chrome’s JavaScript console lazy about evaluating arrays?


I’ll start with the code:

var s = ["hi"];
s[0] = "bye";

Simple, right? In response to this, Firebug says:


Wonderful, but Chrome’s JavaScript console (7.0.517.41 beta) says:


Have I done something wrong, or is Chrome’s JavaScript console being exceptionally lazy about evaluating my array?


I observe the same behavior in Safari — so it’s probably a webkit thing. Pretty surprising. I’d call it a bug.
@mplungjan – that’s not true. the first line declares a “plain old” array with a single element at index 0. The third line simply assigns a new value to that element. both cases are working with a simple numerically indexed array.
if this is a bug, why this bug wasn’t found and fixed is beyond my comprehension.
To me it looks like a bug. On Linux Opera and Firefox display the expected result, Chrome and other Webkit-based browsers do not. You might want to report the issue to the Webkit devs:
DOH – you are of course correct. I was not fully awake



Thanks for the comment, tec. I was able to find an existing unconfirmed Webkit bug that explains this issue: (EDIT: now fixed!)

There appears to be some debate regarding just how much of a bug it is and whether it’s fixable. It does seem like bad behavior to me. It was especially troubling to me because, in Chrome at least, it occurs when the code resides in scripts that are executed immediately (before the page is loaded), even when the console is open, whenever the page is refreshed. Calling console.log when the console is not yet active only results in a reference to the object being queued, not the output the console will contain. Therefore, the array (or any object), will not be evaluated until the console is ready. It really is a case of lazy evaluation.

However, there is a simple way to avoid this in your code:

var s = ["hi"];
s[0] = "bye";

By calling toString, you create a representation in memory that will not be altered by following statements, which the console will read when it is ready. The console output is slightly different from passing the object directly, but it seems acceptable:



Actually, with associative arrays or other objects, this could be a real problem, since toString doesn’t produce anything of value. Is there an easy work-around for objects in general?
– Eric Mickelsen
Oct 30 ’10 at 19:00
@draeton: Cool.
– Eric Mickelsen
Jan 4 ’11 at 4:08
webkit landed a patch for this a few months ago


From Eric’s explanation, it is due to console.log() being queued up, and it prints a later value of the array (or object).

There can be 5 solutions:

1. arr.toString()   // not well for [1,[2,3]] as it shows 1,2,3
2. arr.join()       // same as above
3. arr.slice(0)     // a new array is created, but if arr is [1, 2, arr2, 3] 
                    //   and arr2 changes, then later value might be shown
4. arr.concat()     // a new array is created, but same issue as slice(0)
5. JSON.stringify(arr)  // works well as it takes a snapshot of the whole array 
                        //   or object, and the format shows the exact structure



You can clone an array with Array#slice:

console.log(s); // ["bye"], i.e. incorrect
console.log(s.slice()); // ["hi"], i.e. correct

A function that you can use instead of console.log that doesn’t have this problem is as follows:

console.logShallowCopy = function () {
    function slicedIfArray(arg) {
        return Array.isArray(arg) ? arg.slice() : arg;

    var argsSnapshot =, slicedIfArray);
    return console.log.apply(console, argsSnapshot);

For the case of objects, unfortunately, the best method appears to be to debug first with a non-WebKit browser, or to write a complicated function to clone. If you are only working with simple objects, where order of keys doesn’t matter and there are no functions, you could always do:

console.logSanitizedCopy = function () {
    var args =;
    var sanitizedArgs = JSON.parse(JSON.stringify(args));

    return console.log.apply(console, sanitizedArgs);

All of these methods are obviously very slow, so even more so than with normal console.logs, you have to strip them off after you’re done debugging.



This is already answered, but I’ll drop my answer anyway. I implemented a simple console wrapper which doesn’t suffer from this issue. Requires jQuery.

It implements only log, warn and error methods, you will have to add some more in order for it to be interchangeable with a regular console.

var fixedConsole;
(function($) {
    var _freezeOne = function(arg) {
        if (typeof arg === 'object') {
            return $.extend(true, {}, arg);
        } else {
            return arg;
    var _freezeAll = function(args) {
        var frozen = [];
        for (var i=0; i<args.length; i++) {
        return frozen;
    fixedConsole = {
        log: function() { console.log.apply(console, _freezeAll(arguments)); },
        warn: function() { console.warn.apply(console, _freezeAll(arguments)); },
        error: function() { console.error.apply(console, _freezeAll(arguments)); }



This has been patched in Webkit, however when using the React framework this happens for me in some circumstances, if you have such problems just use as others suggest:



Can confirm. This is literally the worst when trying to log out ReactSyntheticEvents. Even a JSON.parse(JSON.stringify(event)) doesn’t get the right depth/accuracy. Debugger statements are the only real solution I’ve found to get the correct insight.


Looks like Chrome is replacing in its “pre compile” phase any instance of “s” with pointer to the actual array.

One way around is by cloning the array, logging fresh copy instead:

var s = ["hi"];
s[0] = "bye";

function CloneArray(array)
    var clone = new Array();
    for (var i = 0; i < array.length; i++)
        clone[clone.length] = array[i];
    return clone;


That’s good, but because it’s a shallow copy, there is still the possibility of a more subtle problem. And what about objects that aren’t arrays? (Those are the real problem now.) I don’t think that what you’re saying about “pre compile” is accurate. Also, there is an error in the code: clone[clone.length] should be clone[i].
– Eric Mickelsen
Oct 31 ’10 at 16:54
No error, I’ve executed it and it was OK. clone[clone.length] is exactly like clone[i], as the array start with length of 0, and so does the loop iterator “i”. Anyway, not sure how it will behave with complex objects but IMO it’s worth a try. Like I said, that’s not a solution, it’s a way around the problem..
@Shadow Wizard: Good point: clone.length will always be equal to i. It won’t work for objects. Perhaps there is a solution with “for each”.
– Eric Mickelsen
Nov 1 ’10 at 16:57
Objects you mean this? var s = { param1: “hi”, param2: “how are you?” }; if so I just tested and when you have s[“param1”] = “bye”; it’s working fine as expected. Can you please post example of “it won’t work for objects”? I’ll see and try to climb that one as well.
@Shadow Wizard: Obviously, your function will fail to clone properties and will not work on any objects without a length property. The webkit bug affects all objects, not just arrays.
– Eric Mickelsen
Nov 3 ’10 at 14:19


Add a Comment