Arithmetic errors (for example overflow, underflow, division by zero, conversion errors) can be controlled using the following values:

• 0 Errors not allowed and generate an error
• 1 Errors silently allowed (default)

If overflow is allowed, the result for numbers with decimal places is zero, and for integer types is the bit pattern that can be stored in the item corresponding to a correct result.

For example, subtracting 1 from an unsigned byte containing 0 produces 255, adding 257 to it produces 1, and adding 128 to a signed byte containing 0 produces -128.

Values after ignored conversion errors are undefined where the number of bits required to contain the result is more than are available.