|Navigate Aggregate topic: )|
The most basic collection interface is called
Collection. This interface gives the user a generic usage of a collection. All collections need to have the same basic operations. Those are:
- Adding element(s) to the collection
- Removing element(s) from the collection
- Obtaining the number of elements in the collection
- Listing the contents of the collection, (Iterating through the collection)
When you put an object in a collection, this object is not actually in the collection. Only its object reference is added to the collection. This means that if an object is changed after it was put in an collection, the object in the collection also changes. The code listing 5.2 computes the seven next days from tomorrow and store each date in a list to read it after. See what happens:
Each collection items were said to be updated to a different date but they all have been updated to the last one. It means that each update has updated all the collection items. And this is the case. The
currentDate has been used to fill all the collection items. The collection didn't keep trace of the added values (one of the seven dates) but the added object references (
currentDate). So the collection contains the same object seven times! To avoid this issue, we should have coded it this way:
Now each time we add an item to the collection, it is a different instance. All the items evolve separately. To add an object in a collection and avoid this item to be changed each time the source object is changed, you have to copy or clone the object before you add it to the collection.
Objects put into a collection are upcasted to
Object class. It means that you need to cast the object reference back when you get an element out from the collection. It also means that you need to know the type of the object when you take it out. If a collection contains different types of objects, we will have difficulty finding out the type of the objects obtained from a collection at run time. Let's use a collection with any objects in it:
|Code section 5.1: Collection feeding.
This error could have been fixed earlier, at compile time, using generic types.
The Generics has been added since JDK version 1.5. It is an enhancement to the type system of the Java language. All collection implementations since 1.5 now have one parameterized type <E> added. The E refers to an Element type. When a collection is created, the actual Element type will replace the E. In the collection, the objects are now upcasted to E class.
|Code section 5.3: Collection with generics.
ageList is a collection that can contain only Integer objects as elements. No casting is required when we take out an element.
|Code section 5.4: Item reading.
Generics is not mandatory but it is often used with the collection classes.
There is no direct implementation for the
java.util.Collection interface. The Collection interface has five sub interfaces.
A set collection contains unique elements, so duplicates are not allowed. It is similar to a mathematical Set. When adding a new item to a set, the set calls the method
int hashCode() of the item and compare it to the hash code of all the already inserted items. If the hash code has not been found, the item is added. If it is, the set now call the
boolean equals(Object obj); method with all the set items. If all calls returns false, the item is inserted. If not, the item is not inserted.
- This is the basic implementation of the
Setinterface. Not synchronized. Allows the
- Elements are sorted, not synchronized.
- Thread safe, a fresh copy is created during modification operation. Add, update, delete are expensive.
- java.util.EnumSet<E extends Enum<E>>
- All of the elements in an enum set must come from a single enum type that is specified, explicitly or implicitly, when the set is created. Enum sets are represented internally as bit vectors.
- Same as HashSet, plus defines the iteration ordering, which is the order in which elements were inserted into the set.
Detecting duplicate objects in SetsEdit
Set cannot have duplicates in it. You may wonder how duplicates are detected when we are adding an object to the
Set. We have to see if that object exists in the Set or not. It is not enough to check the object references, the objects' values have to be checked as well.
To do that, fortunately, each java object has the
boolean equals(Object obj), method available inherited from
Object. You need to override it. That method will be called by the Set implementation to compare the two objects to see if they are equal or not.
There is a problem, though. What if I put two different type of objects to the Set. I put an Apple and an Orange. They can not be compared. Calling the
equals() method would cause a
ClassCastException. There are two solutions to this:
- Solution one : Override the
int hashCode()method and return the same values for the same type of objects and return different values for different type of objects. The
equals()method is used to compare objects only with the same value of hashCode. So before an object is added, the Set implementation needs to:
- find all the objects in the Set that have the same hashCode as the candidate object hashCode
- and for those, call the
equals()methods passing in the candidate object
- if any of them returns true, the object is not added to the Set.
- Solution two : Create a super class for the Apple and Orange, let's call it Fruit class. Put Fruits in the Set. You need to do the following:
- Do not override the
hashCode()methods in the Apple and Orange classes
appleEquals()method in the Apple class, and create
orangeEquals()method in the Orange class
- Override the
hashCode()method in the Fruit class and return the same value, so the
equals()is called by the Set implementation
- Override the
equals()method in the Fruit class for something like this.
- Do not override the
|Code section 5.5:
- Only the objects that have the same hashCode will be compared.
- You are responsible to override the
hashCode()methods. The default implementations in Object won't work.
- Only override the
hashCode()method if you want to eliminate value duplicates.
- Do not override the
hashCode()method if you know that the values of your objects are different, or if you only want to prevent adding the exactly same object.
- Beware that the
hashCode()may be used in other collection implementations, like in a Hashtable to find an object fast. Overriding the default
hashCode()method may affect performance there.
- The default hashCodes are unique for each object created, so if you decide not to override the
hashCode()method, there is no point overriding the
equals()method, as it won't be called.
SortedSet interface is the same as the Set interface plus the elements in the SortedSet are sorted. It extends the Set Interface. All elements in the SortedSet must implement the Comparable Interface, furthermore all elements must be mutually comparable.
Note that the ordering maintained by a sorted set must be consistent with equals if the sorted set is to correctly implement the Set interface. This is so because the Set interface is defined in terms of the equals operation, but a sorted set performs all element comparisons using its compare method, so two elements that are deemed equal by this method are, from the standpoint of the sorted set, equal.
The SortedSet interface has additional methods due to the sorted nature of the 'Set'. Those are:
||returns the first element|
||returns the last element|
||returns from the first, to the exclusive toElement|
||returns from the inclusive fromElement to the end|
||returns elements range from fromElement, inclusive, to toElement, exclusive. (If fromElement and toElement are equal, the returned sorted set is empty.)|
In a list collection, the elements are put in a certain order, and can be accessed by an index. Duplicates are allowed, the same element can be added twice to a list. It has the following implementations:
- Synchronized, use in multiple thread access, otherwise use ArrayList.
- It extends class Vector with five operations that allow a vector to be treated as a stack. It represents a last-in-first-out (LIFO) stack of objects.
- The basic implementation of the
Listinterface is the
ArrayList. The ArrayList is not synchronized, not thread safe.
Vectoris synchronized, and thread safe.
Vectoris slower, because of the extra overhead to make it thread safe. When only one thread is accessing the list, use the ArrayList. Whenever you insert or remove an element from the list, there are extra overhead to reindex the list. When you have a large list, and you have lots of insert and remove, consider using the
- Non-synchronized, update operation is faster than other lists, easy to use for stacks, queues, double-ended queues. The name
LinkedListimplies a special data structure where the elements/nodes are connected by pointers.
Head Node 1 Node 2 Node n ______ | Size | _________________ _______________ _____________ |______| | | point | | | point | | | | | First|-------->| Data | to next |------>| Data | to next|-- ... -->| Data | null | | elem | |______|_________| |______|________| |______|______| |______| ^ | Last | | | elem |----------------------------------------------------------------- |______|
Each node is related to an item of the linked list. To remove an element from the linked list the pointers need to be rearranged. After removing Node 2:
Head Node 1 Node 2 Node n ______ _____________________ | Size | _________________ | _______________ | ______________ |_- 1__| | | point | | | | point | | | | | | First|-------->| Data | to next |---- | Data | to next| -...-->| Data | null | | elem | |______|_________| |______|________| |______|______| |______| ^ | Last | | | elem |----------------------------------------------------------------- |______|
- Represents a list of values for attributes of an MBean. The methods used for the insertion of Attribute objects in the AttributeList overrides the corresponding methods in the superclass ArrayList. This is needed in order to insure that the objects contained in the AttributeList are only Attribute objects.
- A RoleList represents a list of roles (Role objects). It is used as parameter when creating a relation, and when trying to set several roles in a relation (via 'setRoles()' method). It is returned as part of a RoleResult, to provide roles successfully retrieved.
- A RoleUnresolvedList represents a list of RoleUnresolved objects, representing roles not retrieved from a relation due to a problem encountered when trying to access (read or write to roles).
The Queue interface provides additional insertion, extraction, and inspection operations. There are FIFO (first in, first out) and LIFO (last in, first out) queues. This interface adds the following operations to the Collection interface:
||Retrieves, but does not remove, the head of this queue. This method differs from the peek method only in that it throws an exception if this queue is empty|
||Inserts the specified element into this queue, if possible.|
||Retrieves, but does not remove, the head of this queue, returning null if this queue is empty|
||Retrieves and removes the head of this queue, or null if this queue is empty|
||Retrieves and removes the head of this queue. This method differs from the poll method in that it throws an exception if this queue is empty.|
- waits for the queue to become non-empty when retrieving an element, and waits for space to become available in the queue when storing an element. Best used for producer-consumer queues.
- orders elements according to an order/priority specified at construction time, null element is not allowed.
- orders elements FIFO; synchronized, thread safe.
- each put must wait for a take, and vice versa, does not have any internal capacity, not even a capacity of one, an element is only present when you try to take it; you cannot add an element (using any method) unless another thread is trying to remove it.
Complete UML class diagramEdit
Synchronization is important when you are running several threads. Beware, synchronization does not mean that your collection is thread-safe. A thread-safe collection is also called a concurrent collection. Most of the popular collection classes have implementations for both single thread and multiple thread environments. The non-synchronized implementations are always faster. You can use the non-synchronized implementations in multiple thread environments, when you make sure that only one thread updates the collection at any given time.
A new Java JDK package was introduced at Java 1.5, that is
java.util.concurrent. This package supplies a few Collection implementations designed for use in multi-threaded environments.
The following table lists all the synchronized collection classes:
The Java JDK collection implementations are quite powerful and good, so it is unlikely that you will need to write your own. The usage of the different collections are the same but the implementations are different. If the existing collection implementations do not meet your needs, you can write your version of the implementation. Your version of the implementation just needs to implement the same
java.util.Collection interface, then you can switch to using your implementation and the code that is using the collection does not need to be changed.
Use the Collection interface if you need to keep related (usually the same type of) objects together in a collection where you can:
- Search for a particular element
- List the elements
- Maintain and/or change the order of the elements by using the collection basic operations (Add, Remove, Update,..)
- Access the elements by an index number
The advantages of using the
Collection interface are:
- Gives a generic usage, as we talked about above, it is easy to switch implementation
- It makes it easy to convert one type of collection to another.
Collection interface defines the following basic operations:
||Using Element type E|
Note that in
addAll() we can add any type of collection. This is the beauty of using the Collection interface. You can have a
LinkedList and just call the
addAll(list) method, passing in a list. You can pass in a
YourImpOfCollection, ... All those different types of collection will be magically converted to a
Let's have a closer look at this magic. The conversion is easy because the
Collection interface defines a standard way of looping through the elements. The following code is a possible implementation of
addAll() method of the
|Code section 5.6: Collection transfer.
The above code just iterates through the passed in collection and adds the elements to the linked list. You do not have to do that, since that is already defined. What you might need to code for is to loop through a
|Code section 5.7: Iteration on a collection.
Notice two things:
- The above code will work for all type of collections.
- We have to know the type of objects inside the collection, because we call a method on it.