Difference between association, aggregation and composition

Question

What is the difference between association, aggregation and composition? Please explain in terms of implementation.

Answer 1

Any relationship between objects is called an association. Aggregation and composition are specialized forms of association and are compared in the answers to this question.

Answer 2

I think this link will do your homework: http://ootips.org/uml-hasa.html

To understand the terms I remember an example in my early programming days:

If you have a 'chess board' object that contains 'box' objects that is composition because if the 'chess board' is deleted there is no reason for the boxes to exist anymore.

If you have a 'square' object that have a 'color' object and the square gets deleted the 'color' object may still exist, that is aggregation

Both of them are associations, the main difference is conceptual

Answer 3

As others said, an association is a relationship between objects, aggregation and composition are types of association.

From an implementation point of view, an aggregation is obtained by having a class member by reference. For example, if class A aggregates an object of class B, you'll have something like this (in C++):

class A {
    B & element;
  // or B * element;
};

The semantics of aggregation is that when an object A is destroyed, the B object it is storing will still exists. When using composition, you have a stronger relationship, usually by storing the member by value:

class A {
    B element;
};

Here, when an A object is destroyed, the B object it contains will be destroyed too. The easiest way to achieve this is by storing the member by value, but you could also use some smart pointer, or delete the member in the destructor:

class A {
    std::auto_ptr<B> element;
};

class A {
    B * element;

    ~A() {
        delete B;
    }
};

The important point is that in a composition, the container object owns the contained one, whereas in aggregation, it references it.

Answer 4

From a post by Robert Martin in comp.object:

Association represents the ability of one instance to send a message to another instance. This is typically implemented with a pointer or reference instance variable, although it might also be implemented as a method argument, or the creation of a local variable.

[Example:]

|A|----------->|B|

class A
{
  private:
    B* itsB;
};

Aggregation [...] is the typical whole/part relationship. This is exactly the same as an association with the exception that instances cannot have cyclic aggregation relationships (i.e. a part cannot contain its whole).

[Example:]

|Node|<>-------->|Node|

class Node
{
  private:
    vector<Node*> itsNodes;
};

The fact that this is aggregation means that the instances of Node cannot form a cycle. Thus, this is a Tree of Nodes not a graph of Nodes.

Composition [...] is exactly like Aggregation except that the lifetime of the 'part' is controlled by the 'whole'. This control may be direct or transitive. That is, the 'whole' may take direct responsibility for creating or destroying the 'part', or it may accept an already created part, and later pass it on to some other whole that assumes responsibility for it.

[Example:]

|Car|<#>-------->|Carburetor|

class Car
{
  public:
    virtual ~Car() {delete itsCarb;}
  private:
    Carburetor* itsCarb
};