Overflow in

About Overflow

The exercise shows the behavior of various numeric types when they overflow, i.e. when their capacity is insufficient to contain the value resulting from a computation such as an arithmetic operation or cast.

• unsigned integers (byte, ushort, uint, ulong) will wrap around to zero (the type's maximum value + 1 acts as a modulus) unless broadly speaking they appear within a checked block in which case an instance of OverflowException is thrown. int and long will behave similarly except that they wrap around to int.MinValue and long.minValue respectively.

int one = 1;
checked
{
    int expr = int.MaxValue + one;  // overflow exception is thrown
}

// or

int expr2 = checked(int.MaxValue + one);  // overflow exception is thrown

• If a literal expression would cause the variable to which it is assigned to overflow then a compile-time error occurs.
• the checked state applies only to expressions directly in the block. Overflow states in called functions are not caught.
• float and double types will adopt a state of infinity that can be tested wtih float.IsInfinity() etc.

double d = double.MaxValue;
d *= 2d;
Double.IsFinite(d)
// => false

• Numbers of type decimal will cause an instance of OverflowException to be thrown.
• There is a corresponding unchecked keyword for circumstances where you want to reverse the effect of unchecked inside a checked block or when the compiler setting has been used.

Overflows that occur without an exception being thrown can be problematic because it's generally true that the earlier an error condition can be reported the better.

Problems with overflows for int and float can be mitigated by assigning the result to a variable of type long, decimal or double.

If large integers are essential to your code then using the BigInteger type is an option.

Naturally there are occasions on which it is legitimate to allow an integer to wrap around particularly in the case of unsigned values. A classic case is that of hash codes that use the width of the integer as a kind of modulo.

You will usually find in code bases that there is often no check where an uint or a ulong is used as an identifier because it is considered more trouble than it's worth. This also applies where it is evident from the domain that no very large values will be involved. But, look at this for a cautionary tale.