Get started with the Java Collections Framework

Find out how Sun's new offering can help you to make your collections more useful and accessible

JDK 1.2 introduces a new framework for collections of objects, called the Java Collections Framework. "Oh no," you groan, "not another API, not another framework to learn!" But wait, before you turn away, hear me out: the Collections framework is worth your effort and will benefit your programming in many ways. Three big benefits come immediately to mind:

  • It dramatically increases the readability of your collections by providing a standard set of interfaces to be used by many programmers in many applications.
  • It makes your code more flexible by allowing you to pass and return interfaces instead of concrete classes, generalizing your code rather than locking it down.
  • It offers many specific implementations of the interfaces, allowing you to choose the collection that is most fitting and offers the highest performance for your needs.

And that's just for starters.

Our tour of the framework will begin with an overview of the advantages it provides for storing sets of objects. As you'll soon discover, because your old workhorse friends Hashtable and Vector support the new API, your programs will be uniform and concise -- something you and the developers accessing your code will certainly cheer about.

After our preliminary discussion, we'll dig deeper into the details.

The Java Collections advantage: An overview

Before Collections made its most welcome debut, the standard methods for grouping Java objects were via the array, the Vector, and the Hashtable. All three of these collections have different methods and syntax for accessing members: arrays use the square bracket ([]) symbols, Vector uses the elementAt method, and Hashtable uses get and put methods. These differences have long led programmers down the path to inconsistency in implementing their own collections -- some emulate the Vector access methods and some emulate the Enumeration interface.

To further complicate matters, most of the Vector methods are marked as final; that is, you cannot extend the Vector class to implement a similar sort of collection. We could create a collection class that looked like a Vector and acted like a Vector, but it couldn't be passed to a method that takes a Vector as a parameter.

Finally, none of the collections (array, Vector or Hashtable) implements a standard member access interface. As programmers developed algorithms (like sorts) to manipulate collections, a heated discourse erupted on what object to pass to the algorithm. Should you pass an array or a Vector? Should you implement both interfaces? Talk about duplication and confusion.

Thankfully, the Java Collections Framework remedies these problems and offers a number of advantages over using no framework or using the Vector and Hashtable:

  • A usable set of collection interfaces

    By implementing one of the basic interfaces -- Collection, Set, List, or Map -- you ensure your class conforms to a common API and becomes more regular and easily understood. So, whether you are implementing an SQL database, a color swatch matcher, or a remote chat application, if you implement the Collection interface, the operations on your collection of objects are well-known to your users. The standard interfaces also simplify the passing and returning of collections to and from class methods and allow the methods to work on a wider variety of collections.

  • A basic set of collection implementations

    In addition to the trusty Hashtable and Vector, which have been updated to implement the Collection interfaces, new collection implementations have been added, including HashSet and TreeSet, ArrayList and LinkedList, and HashMap and Map. Using an existing, common implementation makes your code shorter and quicker to download. Also, using existing Core Java code core ensures that any improvements to the base code will also improve the performance of your code.

  • Other useful enhancements

    Each collection now returns an Iterator, an improved type of Enumeration that allows element operations such as insertion and deletion. The Iterator is "fail-fast," which means you get an exception if the list you're iterating is changed by another user. Also, list-based collections such as Vector return a ListIterator that allow bi-directional iteration and updating.

    Several collections (TreeSet and TreeMap) implicitly support ordering. Use these classes to maintain a sorted list with no effort. You can find the smallest and largest elements or perform a binary search to improve the performance of large lists. You can sort other collections by providing a collection-compare method (a Comparator object) or an object-compare method (the Comparable interface).

    Finally, a static class Collections provides unmodifiable (read-only) and synchronized versions of existing collections. The unmodifiable classes are helpful to prevent unwanted changes to a collection. The synchronized version of a collection is a necessity for multithreaded programs.

The Java Collections Framework is part of Core Java and is contained in the java.util.collections package of JDK 1.2. The framework is also available as a package for JDK 1.1 (see Resources).

Note: The JDK 1.1 version of collections is named Keep in mind that code developed with the 1.1 version must be updated and recompiled for the 1.2 verson, and any objects serialized in 1.1 cannot be deserialized into 1.2.

Let us now look more closely at these advantages by exercising the Java Collections Framework with some code of our own.

A good API

The first advantage of the Java Collections Framework is a consistent and regular API. The API is codified in a basic set of interfaces, Collection, Set, List, or Map. The Collection interface contains basic collection operations such as adding, removing, and tests for membership (containment). Any implementation of a collection, whether it is one provided by the Java Collections Framework or one of your own creations, will support one of these interfaces. Because the Collections framework is regular and consistent, you will learn a large portion of the frameworks simply by learning these interfaces.

Both Set and List implement the Collection interface. The Set interface is identical to the Collection interface except for an additional method, toArray, which converts a Set to an Object array. The List interface also implements the Collection interface, but provides many accessors that use an integer index into the list. For instance, get, remove, and set all take an integer that affects the indexed element in the list. The Map interface is not derived from collection, but provides an interface similar to the methods in java.util.Hashtable. Keys are used to put and get values. Each of these interfaces are described in following code examples.

The following code segment demonstrates how to perform many Collection operations on HashSet, a basic collection that implements the Set interface. A HashSet is simply a set that doesn't allow duplicate elements and doesn't order or position its elements. The code shows how you create a basic collection and add, remove, and test for elements. Because Vector now supports the Collection interface, you can also execute this code on a vector, which you can test by changing the HashSet declaration and constructor to a Vector.

import java.util.collections.*;
public class CollectionTest {
   // Statics
   public static void main( String [] args ) {
      System.out.println( "Collection Test" );
      // Create a collection
      HashSet collection = new HashSet();
      // Adding
      String dog1 = "Max", dog2 = "Bailey", dog3 = "Harriet";
      collection.add( dog1 );
      collection.add( dog2 );
      collection.add( dog3 );
      // Sizing
      System.out.println( "Collection created" + 
        ", size=" + collection.size() + 
        ", isEmpty=" + collection.isEmpty() );
      // Containment
      System.out.println( "Collection contains " + dog3 + 
         ": " + collection.contains( dog3 ) );
      // Iteration. Iterator supports hasNext, next, remove
      System.out.println( "Collection iteration (unsorted):" );
      Iterator iterator = collection.iterator();
      while ( iterator.hasNext() ) 
         System.out.println( "   " + );
      // Removing
      collection.remove( dog1 );

Let's now build on our basic knowledge of collections and look at other interfaces and implementations in the Java Collections Framework.

Good concrete implementations

We have exercised the Collection interface on a concrete collection, the HashSet. Let's now look at the complete set of concrete collection implementations provided in the Java Collections framework. (See the Resources section for a link to Sun's annotated outline of the Java Collections framework.)

Hash TableResizable ArrayBalanced Tree (Sorted)Linked ListLegacy
Interfaces SetHashSet* TreeSet* *
List* ArrayList* LinkedListVector
MapHashMap* TreeMap* Hashtable

Implementations marked with an asterix (*) make no sense or provide no compelling reason to implement. For instance, providing a List interface to a Hash Table makes no sense because there is no notion of order in a Hash Table. Similarly, there is no Map interface for a Linked List because a list has no notion of table lookup.

Let's now exercise the List interface by operating on concrete implementations that implement the List interface, the ArrayList, and the LinkedList. The code below is similar to the previous example, but it performs many List operations.

import java.util.collections.*;
public class ListTest {
   // Statics
   public static void main( String [] args ) {
      System.out.println( "List Test" );
      // Create a collection
      ArrayList list = new ArrayList();
      // Adding
      String [] toys = { "Shoe", "Ball", "Frisbee" };
      list.addAll( Arrays.toList( toys ) );
      // Sizing
      System.out.println( "List created" + 
        ", size=" + list.size() + 
        ", isEmpty=" + list.isEmpty() );
      // Iteration using indexes.
      System.out.println( "List iteration (unsorted):" );
      for ( int i = 0; i < list.size(); i++ ) 
         System.out.println( "   " + list.get( i ) );
      // Reverse Iteration using ListIterator
      System.out.println( "List iteration (reverse):" );
      ListIterator iterator = list.listIterator( list.size() );
      while ( iterator.hasPrevious() ) 
         System.out.println( "   " + iterator.previous() );
      // Removing
      list.remove( 0 );

As with the first example, it's simple to swap out one implementation for another. You can use a LinkedList instead of an ArrayList simply by changing the line with the ArrayList constructor. Similarly, you can use a Vector, which now supports the List interface.

When deciding between these two implementations, you should consider whether the list is volatile (grows and shrinks often) and whether access is random or ordered. My own tests have shown that the ArrayList generally outperforms the LinkedList and the new Vector.

Notice how we add elements to the list: we use the addAll method and the static method Arrays.toList. This static method is one of the most useful utility methods in the Collections framework because it allows any array to be viewed as a List. Now an array may be used anywhere a Collection is needed.

Notice that I iterate through the list via an indexed accessor, get, and the ListIterator class. In addition to reverse iteration, the ListIterator class allows you to add, remove, and set any element in the list at the point addressed by the ListIterator. This approach is quite useful for filtering or updating a list on an element-by-element basis.

The last basic interface in the Java Collections Framework is the Map. This interface is implemented with two new concrete implementations, the TreeMap and the HashMap. The TreeMap is a balanced tree implementation that sorts elements by the key.

Let's illustrate the use of the Map interface with a simple example that shows how to add, query, and clear a collection. This example, which uses the HashMap class, is not much different from how we used the Hashtable prior to the debut of the Collections framework. Now, with the update of Hashtable to support the Map interface, you can swap out the line that instantiates the HashMap and replace it with an instantiation of the Hashtable.

public class HashMapTest {
   // Statics
   public static void main( String [] args ) {
      System.out.println( "Collection HashMap Test" );
      HashMap collection1 = new HashMap();
      // Test the Collection interface
      System.out.println( "Collection 1 created, size=" + collection1.size() + 
         ", isEmpty=" + collection1.isEmpty() );
      // Adding
      collection1.put( new String( "Harriet" ), new String( "Bone" ) );
      collection1.put( new String( "Bailey" ),  new String( "Big Chair" ) );
      collection1.put( new String( "Max" ),     new String( "Tennis Ball" ) );
      System.out.println( "Collection 1 populated, size=" + collection1.size() + 
         ", isEmpty=" + collection1.isEmpty() );
      // Test Containment/Access
      String key = new String( "Harriet" );
      if ( collection1.containsKey( key ) )
         System.out.println( "Collection 1 access, key=" + key + ", value=" + 
            (String) collection1.get( key ) );
      // Test iteration of keys and values
      Set keys = collection1.keySet();
      System.out.println( "Collection 1 iteration (unsorted), collection contains keys:" );
      Iterator iterator = keys.iterator();
      while ( iterator.hasNext() )
         System.out.println( "   " + );
      System.out.println( "Collection 1 cleared, size=" + collection1.size() + 
         ", isEmpty=" + collection1.isEmpty() );

We've covered most of the interfaces and implementations in the Java Collections framework, and we're ready to check out some of the additional capabilities Collections offers us.

Other capabilities

Many of the additional features such as sorting and synchronization are encapsulated in the Collections and Arrays classes. These classes, which will appear throughout the following discussion, provide static methods for acting on collections.

Sorting a collection

We'll begin by exploring sorting. Two of the concrete implementations in the Java Collections Framework provide easy means to maintain a sorted collection: TreeSet and TreeMap. In fact, these two classes implement the SortedSet and SortedMap interfaces, which are similar to their unsorted counterparts except that they provide methods to access first and last elements and portions of the sorted collections.

There are two basic techniques for maintaining a sorted collection. The first uses one of the sorted collection classes and provides the collection with an object that implements a comparison via the Comparator interface. For example, going back to our first code example, we can sort our collection by creating a StringComparator and adding it to the end of the code, as shown here:

// This class sorts two String objects.
class StringComparator implements Comparator {
   public int compare( Object object1, Object object2 ) {
      return ((String) object1).compareTo( (String) object2 );

Next, we need to change the collection from a HashSet (unsorted) to a HashMap (sorted with our StringComparator by using the following constructor:

   TreeSet collection = new TreeSet( new StringComparator() );

Rerun the example and you should see that the iteration is performed in sorted order. Because the collection is ordered, you should now be able to find the min and the max elements using the static class Collections.

The second technique is to implement natural ordering of a class by making the class implement the Comparable interface. This technique adds a single compareTo method to a class, which then returns 0 for equal objects, less than 0 if the first parameter is less than the second, or greater than 0 of the first parameter is greater than the second. In Java 1.2, the String class (but not StringBuffer) implements the Comparable interface. Any comparable object can be placed in a sorted collection, and the collection order is maintained automatically by the collection.

You can also sort Lists by handing them to the Collections class. One static sort method takes a single List parameter that specifies a naturally ordered class (one that implements the Comparable interface). A second static sort method takes a Comparator object for other classes that do not implement the Comparable interface.

Unmodifiable collections

The Collections class provides many static factory methods (like Collection.unmodifiableCollection and Collection.unmodifiableSet) for providing unmodifiable or immutable collections. In fact, there is one method for each of the basic collection interfaces. These methods are extremely useful to ensure that no one modifies your collection. For instance, if you want to allow others to see your list but not change it, you may implement a method that returns an unmodifiable view of your collection. Here's an example:

   List getUnmodifieableView() {
      return Collections.unmodifableList( this );

This code will throw an UnsupportedOperationException, one of the RuntimeExceptions, if someone trys to add or remove an element from the list.

Unfortunately, the unmodifiable views of a collection are of the same type as the original collection, which hinders compile-time type checking. Although you may pass an unmodifiable list to a method, by virtue of its type, the compiler has no way of ensuring the collection is unchanged by the method. The unmodifiable collection is checked at runtime for changes, but this is not quite as strong as compile-time checking and does not aid the compiler in code optimization. Perhaps it's time for Java to emulate C++'s const and add another modifier signifying immutability of any method, class, or object.

Synchronized collections

Finally, note that none of the concrete methods mentioned thus far support multithreaded access in the manner that the Vectors and Hashtable did. In other words, none of the methods on the concrete implementations are synchronized and none of the implementations are thread-safe. You must support thread safety yourself.

This may seem like a major omission, but in actuality it's not really a big problem. The Collections class provides a synchronized version of each of the collection implementations. You ensure thread safety by using the synchronized version of a collection and synchronizing on the returned object. For example, we can ensure thread safety on a List by using the following construct:

   List dogs = synchronized List( new ArrayList() );
   synchronized( list ) {
      Iterator iterator = list.iterator(); // Must be in synchronized block
      while ( iterator.hasNext() )
         nonAtomicOperation( );

Many programmers are already using these types of synchronized blocks around Vectors and Hashtables, so the new considerations aren't too significant.


That concludes our survey of the new Java Collections Framework. We covered a lot of territory, so let's briefly review. We began with a look at the interfaces, which are used through the API and are useful for any new collections we may create. These interfaces are intuitive and provide a common way for all Java programmers to access collections. By implementing a common interface in the Collections framework, you make it easy for other programmers to access your collection, you reduce the time it takes for others to learn your class, and you make your class more useful.

We also examined the basic concrete implementations provided with the framework. You can use these basic collections to implement any generic collection of Java objects. Like the existing Vector and Hashtable, these new collection implementations will cover a majority of your needs as a developer.

Finally, we looked at several general-purpose methods for sorting and manipulating collections. The sort interfaces are easy to add to any class, and the Collections sort methods will knock a few deliverables out of your code package.

Dan Becker works in the Network Computing Software Division of IBM Corporation in Austin, TX. He is currently working on Java 1.2 and other Java extensions for IBM Operating System/2. Before that, Dan worked on porting previous Java virtual machines, the multimedia plugins for Netscape Navigator for OS/2, OpenDoc, and the Multimedia parts for OS/2 Warp Version 4.0.

Learn more about this topic

  • Download the Java Collections Framework as package for Java 1.1 JDK 1.1 at the Sun JavaBeans InfoBus page
  • Sun's overview and description of the Java Collections Framework for Java 1.2
  • Annotated reference of the Java Collections Framework
  • A frequently asked questions file for the Java Collections Framework.
  • Read about ObjectSpace's Java Generic Library -- the main competitor to the Java Collections Framework
Join the discussion
Be the first to comment on this article. Our Commenting Policies