Oracle PL/SQL Programming, 5th Edition

Numeric Datatypes

Like the Oracle database, PL/SQL offers a variety of numeric datatypes to suit different purposes:

NUMBER

A true decimal datatype that is ideal for working with monetary amounts. NUMBER is also the only one of PL/SQL’s numeric types to be implemented in a completely platform-independent fashion. Anything you do with NUMBERs should work the same regardless of the underlying hardware.

PLS_INTEGER and BINARY_INTEGER

Integer datatypes conforming to your hardware’s underlying, integer representation. Arithmetic is performed using your hardware’s native, machine instructions. You cannot store values of these types in the database.

SIMPLE_INTEGER

Introduced with Oracle Database 11g. Has the same range as BINARY_INTEGER, but does not allow for NULLs and does not raise an exception if an overflow occurs. The SIMPLE_INTEGER datatype results in significantly faster execution times for natively compiled code.

BINARY_FLOAT and BINARY_DOUBLE

Single- and double-precision, IEEE-754, binary floating-point types. I don’t recommend these types for monetary amounts. They are useful, however, when you need fast, floating-point arithmetic.

SIMPLE_FLOAT and SIMPLE_DOUBLE

Introduced with Oracle Database 11g. Have the same range as BINARY_FLOAT and BINARY_DOUBLE, but do not allow for NULLs, do not raise an exception if an overflow occurs, and do not support special literals or predicates such as BINARY_FLOAT_MIN_NORMAL, IS NAN, or IS NOT INFINITE. These SIMPLE datatypes result in significantly faster ...