Question

In Java, having a HashMap fully filled in with data of such form:

HashMap<Integer, int[]> map = new HashMap<Integer, int[]>(1000000, 1);

what is faster when checking the existence of a random key, say 100:

if (map.get(100) == null))

or

if (!map.containsKey(100))

?

Question is interesting from the micro-optimization point of view.

Answer 1

The containsKey should be very slightly slower because it results in an extra function call (it just calls getEntry) (it could get optimised away, I’m not sure whether Java will do so). containsKey looks like:

public boolean containsKey(Object key) {
  return getEntry(key) != null;
}

But note that containsKey could on the other hand be very slightly faster on other Map implementations (but probably not those in the standard Java API).

Generally my implementations look like: (avoiding the need for containsKey)

int[] arr = map.get(100);
if (arr == null) // doesn't exist
  // do stuff
else // exists
  // do stuff with arr

The below would definitely be slower than the above: (if the items you look for exist a reasonable amount of the time)

if (!map.containsKey(100)) // doesn't exist
  // do stuff
else // exists
{
  int[] arr = map.get(100);
  // do stuff with arr
}

Edit: Thanks to zvzdhk for providing the source of containsKey. I should actually have checked.