The Java Tutorials have been written for JDK 8. Examples and practices described in this page don't take advantage of improvements introduced in later releases and might use technology no longer available.
See Java Language Changes for a summary of updated language features in Java SE 9 and subsequent releases.
See JDK Release Notes for information about new features, enhancements, and removed or deprecated options for all JDK releases.
Implementations are the data objects used to store collections, which implement the interfaces described in the Interfaces section. This lesson describes the following kinds of implementations:
java.util.concurrent
package.The general-purpose implementations are summarized in the following table.
Interfaces | Hash table Implementations | Resizable array Implementations | Tree Implementations | Linked list Implementations | Hash table + Linked list Implementations |
---|---|---|---|---|---|
Set |
HashSet |
TreeSet |
LinkedHashSet |
||
List |
ArrayList |
LinkedList |
|||
Queue |
|||||
Deque |
ArrayDeque |
LinkedList |
|||
Map |
HashMap |
TreeMap |
LinkedHashMap |
As you can see from the table, the Java Collections Framework provides several general-purpose implementations of the
Set
,
List
, and
Map
interfaces. In each case, one implementation
HashSet
,
ArrayList
, and
HashMap
is clearly the one to use for most applications, all other things being equal. Note that the
SortedSet
and the
SortedMap
interfaces do not have rows in the table. Each of those interfaces has one implementation
(TreeSet
and
TreeMap
) and is listed in the Set
and the Map
rows. There are two general-purpose Queue
implementations
LinkedList
, which is also a List
implementation, and
PriorityQueue
, which is omitted from the table. These two implementations provide very different semantics: LinkedList
provides FIFO semantics, while PriorityQueue
orders its elements according to their values.
Each of the general-purpose implementations provides all optional operations contained in its interface. All permit null
elements, keys, and values. None are synchronized (thread-safe). All have fail-fast iterators, which detect illegal concurrent modification during iteration and fail quickly and cleanly rather than risking arbitrary, non-deterministic behavior at an undetermined time in the future. All are Serializable
and all support a public clone
method.
The fact that these implementations are unsynchronized represents a break with the past: The legacy collections Vector
and Hashtable
are synchronized. The present approach was taken because collections are frequently used when the synchronization is of no benefit. Such uses include single-threaded use, read-only use, and use as part of a larger data object that does its own synchronization. In general, it is good API design practice not to make users pay for a feature they don't use. Furthermore, unnecessary synchronization can result in deadlock under certain circumstances.
If you need thread-safe collections, the synchronization wrappers, described in the
Wrapper Implementations section, allow any collection to be transformed into a synchronized collection. Thus, synchronization is optional for general-purpose implementations, whereas it is mandatory for legacy implementations. Moreover, the java.util.concurrent
package provides concurrent implementations of the BlockingQueue
interface, which extends Queue
, and of the ConcurrentMap
interface, which extends Map
. These implementations offer much higher concurrency than mere synchronized implementations.
As a rule, you should be thinking about the interfaces, not the implementations. That is why there are no programming examples in this section. For the most part, the choice of implementation affects only performance. The preferred style, as mentioned in the
Interfaces section, is to choose an implementation when a Collection
is created and to immediately assign the new collection to a variable of the corresponding interface type (or to pass the collection to a method expecting an argument of the interface type). In this way, the program does not become dependent on any added methods in a given implementation, leaving the programmer free to change implementations anytime that it is warranted by performance concerns or behavioral details.
The sections that follow briefly discuss the implementations. The performance of the implementations is described using words such as constant-time, log, linear, n log(n), and quadratic to refer to the asymptotic upper-bound on the time complexity of performing the operation. All this is quite a mouthful, and it doesn't matter much if you don't know what it means. If you're interested in knowing more, refer to any good algorithms textbook. One thing to keep in mind is that this sort of performance metric has its limitations. Sometimes, the nominally slower implementation may be faster. When in doubt, measure the performance!