Logical Design

• DECIMAL or NUMERIC is the type of numbers represented as strings of decimal digits. This will have some upper bound on the length of representations that the system is willing to tolerate. Although internally DECIMALs are stored just like strings, they are interpreted as numbers. Therefore any leading zeros are removed. On the other hand, trailing zeros are preserved because they could indicate a level of precision. There is also a type REAL which behaves like Java’s “float” but this should only be used when small rounding errors are acceptable in the manipulation of values. The point here is that floating point numbers are represented in binary format and rounding in that format can be at variance with the rounding that one would expect from numbers in their decimal format. Always use DECIMAL for amounts of money!