Question

I’ve noticed that the Javascript sort function is extremely slow in Internet Explorer versions below 9 (often by an order of magnitude when compared to Firefox and others. I’m implementing my own version to see if I can do better. Merge sort works decently, but it seems that most of the documentation for sorting algorithms assumes that arrays are implemented as contiguous blocks of memory. Javascript arrays are implemented as objects (at least in older browsers).

I was wondering, is there a sorting algorithm that takes into account the fact array access is more expensive than usual? That is to say, one that tries to optimize not just the number of comparisons, but also the number accesses and the cost of creating new arrays via things like splice and slice. Here’s my attempt at merge sort.

function mergeSort(array, compareFunc) {
    if (array.length <= 1) {
        return;
    }
    var mid = Math.floor(array.length/2);
    var left = array.splice(0, mid);
    var right = array.splice(0, array.length);
    mergeSort(left, compareFunc);
    mergeSort(right, compareFunc);

    while ((left.length > 0) && (right.length > 0))
    {
        if (compareFunc(left[0], right[0]) <= 0) {
            array.push(left.shift());
        }
        else {
            array.push(right.shift());
        }
    }
    while (left.length > 0) {
        array.push(left.shift());
    }
    while (right.length > 0) {
        array.push(right.shift());
    }
    return;
}

Answer 1

try Web SQL sort function. Think that is faster than native sort available in Javascript. If you are writing code that supports cross browser compatibility, merge sort is a good algo to use, but check http://www.cs.princeton.edu/~rs/strings/ .. This tells about using an amalgam of Quick sort and radix sort.

And if you can get the sorted data from the server, that would be the best method.