I think that the issue here is that your partitioning step does not necessarily shrink the input array. For example, let’s trace what happens if you try sorting [1, 2]. In this case, your pivot element will be the element 2. Since 1 > 2 is false, 1 is added to the list l
. Since 2 > 2 is false, 2 is added to the list l
. As a result, your recursive call on the list l
will have exactly the same arguments as your original call, causing infinite recursion.
To fix this, try splitting the input into three lists – one of smaller values, one of equal values, and one of greater values. This code is shown here:
function sort(data){
if (data.length < 2){
return data;
} else {
var l = [];
var r = [];
var e = [];
var i = 0;
var pivot = (data.length / 2) | 0;
for(i = 0; i < data.length; i++) {
if (data[i] > data[pivot]) {
r.push(data[i]);
} else if (data[i] < data[pivot]) {
l.push(data[i]);
} else {
e.push(data[i]);
}
}
return sort(l).concat(e, sort(r));
}
}
This new version explicitly groups the equal elements into their own list, so they aren’t recursively sorted by either of the recursive calls. It also gracefully handles duplicate elements.