HashSet vs LinkedHashSet

Question

What is the difference between them? I know that

A LinkedHashSet is an ordered version of HashSet that maintains a doubly-linked List across all elements. Use this class instead of HashSet when you care about the iteration order. When you iterate through a HashSet the order is unpredictable, while a LinkedHashSet lets you iterate through the elements in the order in which they were inserted.

But in sourcecode of LinkedHashSet there are only calling constructors of HashSet. So where is double-linked List and insertion order?

Solution 1

The difference between the two are, as you've stated:

A LinkedHashSet is an ordered version of HashSet that maintains a doubly-linked List across all elements. Use this class instead of HashSet when you care about the iteration order. When you iterate through a HashSet the order is unpredictable, while a LinkedHashSet lets you iterate through the elements in the order in which they were inserted.

As for your question:

But in sourcecode of LinkedHashSet there are only calling constructors of HashSet.

The answer lies in which constructors the LinkedHashSet uses to construct the base class:

public LinkedHashSet(int initialCapacity, float loadFactor) {
    super(initialCapacity, loadFactor, true);      // <-- boolean dummy argument
}

...

public LinkedHashSet(int initialCapacity) {
    super(initialCapacity, .75f, true);            // <-- boolean dummy argument
}

...

public LinkedHashSet() {
    super(16, .75f, true);                         // <-- boolean dummy argument
}

...

public LinkedHashSet(Collection<? extends E> c) {
    super(Math.max(2*c.size(), 11), .75f, true);   // <-- boolean dummy argument
    addAll(c);
}

And (one example of) a HashSet constructor that takes a boolean argument is described, and looks like this:

/**
 * Constructs a new, empty linked hash set.  (This package private
 * constructor is only used by LinkedHashSet.) The backing
 * HashMap instance is a LinkedHashMap with the specified initial
 * capacity and the specified load factor.
 *
 * @param      initialCapacity   the initial capacity of the hash map
 * @param      loadFactor        the load factor of the hash map
 * @param      dummy             ignored (distinguishes this
 *             constructor from other int, float constructor.)
 * @throws     IllegalArgumentException if the initial capacity is less
 *             than zero, or if the load factor is nonpositive
 */
HashSet(int initialCapacity, float loadFactor, boolean dummy) {
    map = new LinkedHashMap<E,Object>(initialCapacity, loadFactor);
}

Solution 2

HashSet is unordered and unsorted Set.
LinkedHashSet is the ordered version of HashSet.

The only difference between HashSet and LinkedHashSet is that:
LinkedHashSet maintains the insertion order.

When we iterate through a HashSet, the order is unpredictable while it is predictable in case of LinkedHashSet.

The reason for how LinkedHashSet maintains insertion order is that:
The underlying used data structure is Doubly-Linked-List.

Solution 3

LinkedHashSet's constructors invoke the following base class constructor:

HashSet(int initialCapacity, float loadFactor, boolean dummy) {
  map = new LinkedHashMap<E, Object>(initialCapacity, loadFactor);
}

As you can see, the internal map is a LinkedHashMap. If you look inside LinkedHashMap, you'll discover the following field:

private transient Entry<K, V> header;

This is the linked list in question.

Solution 4

I suggest you to use LinkedHashSet most of the time, because it has better performance overall):

1. Predictable iteration order LinkedHashSet (Oracle)
2. LinkedHashSet is more expensive for insertions than HashSet;
3. In general slightly better performance than HashMap, because the most of the time we use Set structures for iterating.

Performance tests:

------------- TreeSet -------------
 size       add  contains   iterate
   10       746       173        89
  100       501       264        68
 1000       714       410        69
10000      1975       552        69
------------- HashSet -------------
 size       add  contains   iterate
   10       308        91        94
  100       178        75        73
 1000       216       110        72
10000       711       215       100
---------- LinkedHashSet ----------
 size       add  contains   iterate
   10       350        65        83
  100       270        74        55
 1000       303       111        54
10000      1615       256        58

You can see source test page here: The Final Performance Testing Example

Solution 5

You should look at the source of the HashSet constructor it calls... it's a special constructor that makes the backing Map a LinkedHashMap instead of just a HashMap.

View more solutions

Comments

[Delta]

use Intellij(Ctrl + B) option to track down the answer. :)

[Delta]

of course you need source code attached. :)

[Shikarn-O]

Thanks, in HashSet there is constructor for creating LinkedHashMap, which is called in LinkedHashSet and all logic is in LinkedHashMap

[ASA]

A parent class having functionality explicitly for a child class, an ignored argument to distinguish

[Eric J.]

Not exactly a clean design using a dummy parameter for constructor disambiguation.

[lbalazscs]

It is reasonably clean design, because the API is clean (this HashSet constructor is package private). The implementation details do not matter for the users of the class. Maintaining this code could be harder, but in the case of java.util classes, even very small performance improvements can justify that.

[Felix S]

I don't see any warming up of the JVM before those "benchmarks", so I wouldn't take any of that data seriously. Read more