Question

In the code below, I needed to fetch an element, any element, from toSearch. I was unable to find a useful method on the Set interface definition to return just a single (random, but not required to be random) member of the set. So, I used the toArray()[0] technique (present in the code below).

private Set<Coordinate> floodFill(Value value, Coordinate coordinateStart)
{
    Set<Coordinate> result = new LinkedHashSet<Coordinate>();

    Set<Coordinate> toSearch = new LinkedHashSet<Coordinate>();
    toSearch.add(coordinateStart);
    while (toSearch.size() > 0)
    {
        Coordinate coordinate = (Coordinate)toSearch.toArray()[0];
        result.add(coordinate);
        toSearch.remove(coordinate);
        for (Coordinate coordinateAdjacent: getAdjacentCoordinates(coordinate))
        {
            if (this.query.getCoordinateValue(coordinateAdjacent) == value)
            {
                if (!result.contains(coordinateAdjacent))
                {
                    toSearch.add(coordinateAdjacent);
                }
            }
        }
    }

    return result;
}

The other technique I have seen discussed is to replace “(Coordinate)toSearch.toArray()[0]” with “toSearch.iterator().next()“. Which technique, toArray() or iterator(), is the most likely to execute the most quickly with the least GC (Garbage Collection) impact?

My intuition (after composing this question) is that the second technique using the Iterator will be both faster in execution and lower overhead for the GC. Given I don’t know the implementation of the Set being passed (assuming HashSet or LinkedHashSet as most likely), how much overhead is incurred in each of the toArray() or iterator() methods? Any insights on this would be greatly appreciated.

Questions (repeated from above):

1. Which technique, toArray() or iterator(), is the most likely to execute the most quickly with the least GC (Garbage Collection) impact?
2. Given I don’t know the implementation of the Set being passed (assuming HashSet or LinkedHashSet as most likely), how much overhead is incurred in each of the toArray() and iterator() methods?

Answer 1

toSearch.iterator().next() will be faster and less memory-intensive because it does not need to copy any data, whereas toArray will allocate and copy the contents of the set into the array. This is irrespective of the actual implementation: toArray will always have to copy data.