What is interface in oops?

Answer 1

An object is the most primitive element of a program written in the OOP language.A method of an object defines the set of operations that the object will execute when a message corresponding to the method is received by the object.The binding of a message to an method is dynamic. The entire collection of methods of an object is called the "message interface" or " message protocol".

Answer 2

An interface provides the means with which to interact with a class in a controlled manner. All classes have at least one interface but can have as many as three. The private interface is only accessible to the class itself and to friends of the class. The protected interface is the same as private but is also accessible to derivatives of the class. The public interface is accessible to any code. Every class member must belong to one (and only one) of these three interfaces.

The main purpose of a class is to encapsulate one or more invariants -- variables that must remain in a valid state at all times. As an example, a vector object must acquire a memory resource to hold an array. In addition, it must keep track of the amount of memory currently allocated to the array as well as the number of elements that are allocated within the array. All three must be kept in sync with each other such that the amount of memory allocated is equal to or greater than the number of elements. All three must work together to form an invariant. If we declare any of these variables public or protected, then we risk external code invalidating the invariant, thus all invariants must be declared private and all member functions that operate upon those invariants must work together to maintain the validity of the object, regardless of whether those members are part of the private, protected or public interfaces. In addition, we must also ensure that all friends of the class maintain the object's validity.

Since only the class itself and friends of the class have access to the private interface, we can more easily maintain an object's validity. The class constructor's ensure the invariant is correctly initialised while its mutators (assignment operators) ensure the invariant is correctly updated. In this way we do not have to continually test the validity of the invariant -- it is fully encapsulated by the operators themselves.

Of course a vector object wouldn't be much use if we didn't provide a public interface. Most accesses are read-only, thus we can provide constant public accessors so consumers can determine the size of the vector (the number of elements currently allocated) and the amount of reserved memory (the array size minus the vector size). We can also provide a subscript operator so users can examine the individual values stored in the array. The values are not part of the invariant, thus we can provide both constant and non-constant accessors so user can change the values. We can also include range-checked operators to ensure consumers do not attempt to access values beyond the scope of the array.

With a well-defined interface, we can be sure that our objects remain valid at all times and minimise the number of runtime checks we need to perform, thus our code becomes more robust and more efficient. Unfortunately, we cannot eliminate stupidity completely, but a consumer would have to go out of his way to intentionally undermine a well-designed interface. The class designer ultimately decides what can and cannot be done with their class and if a user deliberately attempts to violate an invariant at runtime, the class designer can silently veto the operation (leaving the object unchanged). If that is not possible, they may choose to throw an exception to alert the consumer instead.