Recipe 1.4 Using the Bitwise Complement Operator with Various Data Types

Problem

The bitwise complement operator (~) is overloaded to work directly with int, uint, long, ulong, and enumeration data types consisting of the underlying types int, uint, long, and ulong. However, you need to perform a different bitwise complement operation on a data type.

Solution

You must cast the resultant value of the bitwise operation to the type you wish to work. The following code demonstrates this technique with the byte data type:

byte y = 1;
byte result = (byte)~y;

The value assigned to result is 254.

Discussion

The following code shows incorrect use of the bitwise complement operator on the byte data type:

byte y = 1;
Console.WriteLine("~y = " + ~y);

This code outputs the following surprising value:

-2

Clearly, the result from performing the bitwise complement of the byte variable is incorrect; it should be 254. In fact, byte is an unsigned data type, so it cannot be equal to a negative number. If we rewrite the code as follows:

byte y = 1;
byte result = ~y;

we get a compile-time error: "Cannot implicitly convert type `int' to `byte.'" This error message gives some insight into why this operation does not work as expected. To fix this problem, we must explicitly cast this value to a byte before we assign it to the result variable, as shown here:

byte y = 1;
byte result = (byte)~y;

This cast is required because the bitwise operators are only overloaded to operate on six specific data types: int, uint, long, ulong, bool, and enumeration data types. When one of the bitwise operators is used on another data type, that data type is converted to the next closest data type of the six supported data types. Therefore, a byte data type is converted to an int before the bitwise complement operator is evaluated:

      0x01    // byte y = 1;
0xFFFFFFFE    // The value 01h is converted to an int and its 
              //   bitwise complement is taken
      0xFE    // The resultant int value is cast to its original byte data type

Notice that the int data type is a signed data type, unlike the byte data type. This is why we receive -2 for a result instead of the expected value 254. This conversion of the byte data type to its nearest equivalent is called numeric promotion. Numeric promotion also comes into play when you use differing data types with binary operators, including the bitwise binary operators.

Numeric promotion is discussed in detail in the C# Language Specification document in section 7.2.6 (this document is found in the directory \MicrosoftVisual Studio .NET 2003\Vc7\1033 below the .NET 2003 installation directory). Understanding how numeric promotion works is essential when using operators on differing data types and when using operators with a data type that it is not overloaded to handle. Knowing this can save you hours of debugging time.