A word about casts (type conversion)

All variables have a type in Java, and the type is checked so that you can't assign between two things that are incompatible. You cannot directly assign a floating-point variable to an integer variable.

It is reasonable, however, to convert between closely related types. That is what a cast does. You cast an expression into another type by writing the desired new type name in parentheses before the expression, as follows:

Some numeric conversions don't need a cast. You are allowed to directly assign from a smaller-range numeric type into a larger range—a byte to int, or int to long, or long to float—without a cast. Assigning to a more capacious type is called a widening primitive conversion, and it does not lose any information about the overall magnitude of the numeric value. When you want to do an assignment that can potentially lose data, such as a float-to-int assignment that would drop any fractional part, you must use a cast. The cast tells the compiler that you, the programmer, understand about the potential data loss, and it is ok.

A cast can be used to convert between two objects of related classes, as well as between two primitives. Casting is “type safe,” meaning that the validity of an object type conversion is checked at run-time. It will cause an error if you try to cast an object to a type that is just plain wrong for it.

You never used to be able to cast from an object to a primitive value or vice versa. There was no fundamental theoretical reason for the restriction, and it was loosened up in JDK 1.5. There is a predefined class type corresponding to each of the eight primitive types. You can now cast both ways between values of a primitive type and its corresponding object type. In fact, you don't even have to cast! You can make the assignment directly. There is more about this in Autoboxing and Unboxing on page 55.