Collections 2.0 includes a significant number of new collections, in addition to several useful Comparator classes.

These collections are new to Collections 2.0:

• Bag - A Collection that counts the number of times an object appears in the collection. Suppose you have a Bag that contains {a, a, b, c}. Calling getCount on a would return 2, while calling uniqueSet would return {a, b, c}. Note: this is an interface with several implementations.
• DoubleOrderedMap - Red-Black tree-based implementation of Map. This class guarantees that the map will be in both ascending key order and ascending value order, sorted according to the natural order for the key's and value's classes.
• FilterListIterator - A proxy ListIterator which takes a Predicate instance to filter out objects from an underlying ListIterator instance. Only objects for which the specified Predicate evaluates to true are returned by the iterator.
• HashBag - An implementation of Bag that is backed by a HashMap.
• MultiMap - This is simply a Map with slightly different semantics. Instead of returning an Object, it returns a Collection. So for example, you can put( key, new Integer(1) ); and then a Object get( key ); will return you a Collection instead of an Integer. This is an interface implemented by MultiHashMap.
• SequencedHashMap - A map of objects whose mapping entries are sequenced based on the order in which they were added.
• SortedBag - A type of Bag that maintains order among its unique representative members
• TreeBag - An implementation of Bag that is backed by a TreeMap. Order will be maintained among the unique representative members.

These are the new Comparator classes:

• ComparableComparator - A Comparator that compares Comparable objects. This Comparator is useful, for example, for enforcing the natural order in custom implementations of SortedSet and SortedMap.
• ComparatorChain - ComparatorChain is a Comparator that wraps one or more Comparators in sequence. The ComparatorChain calls each Comparator in sequence until either 1) any single Comparator returns a non-zero result (and that result is then returned), or 2) the ComparatorChain is exhausted (and zero is returned). This type of sorting is very similar to multi-column sorting in SQL, and this class allows Java classes to emulate that kind of behaviour when sorting a List.
• ReverseComparator - Reverses the order of another comparator.

ArrayIterator

Bugs fixed:

• ArrayIterator can now iterate over arrays of primatives

LRUMap

LRUMap has been reimplemented as a subclass of SynchronizedHashMap. The new implementation of LRUMap should be faster, and it also offers true LRU capabilities; now any get(Object) or put(Object,Object) from this collection promotes the key to the Most Recently Used position.

LRUMap 2.0 compatibility changes:

• LRUMap can no longer be cast to a HashMap.
• The removeLRU() method now has a different signature, and is no longer public. Instead, use remove(getFirstKey()).
• "Externalized" LRUMap 1.0 Objects cannot be read by LRUMap 2.0.

New features:

• True LRU algorithm.
• New methods from SequencedHashMap superclass.
• processRemovedLRU(Object key, Object value) method allows subclasses to perform custom actions on keys and values that are expunged from the Map.

Bugs fixed:

• calling setMaximumSize(int) will remove excess LRUs when the current size of the Map exceeds the new maximum size

BeanMap

BeanMap's entrySet() now properly returns a set containing Map.Entry objects. Previously, entrySet() was equivalent to keySet() (returns a set of the readable properties) and there was no mechanism to retrieve all of the readable properties along with their values. Additionally, the BeanMap clone method has been revamped to work better for subclasses.

BeanMap 2.0 compatibility changes:

• BeanMap's clone() method now declares it throws CloneNotSupportedException. This allows subclasses of BeanMap to not require being Cloneable and facilitates subclasses that wish to be cloneable (allows the subclass to call super.clone() and have it return an instance of the child rather than the parent).
• If a BeanMap is not cloneable because a new instance of the underlying bean cannot be constructed, a CloneNotSupportedException is thrown rather than an UnsupportedOperationException or other RuntimeException.
• BeanMap's entrySet() method now returns a set of Map.Entry objects rather than the set of readable properties. To retrieve a set of readable properties, use keySet() instead.

Bugs fixed:

• If no bean is set in the BeanMap, or setBean(Object) was called with a null argument, many BeanMap methods threw NullPointerExceptions. These have been fixed to properly handle the case where there is no bean established in the map.