DECIMAL(p,s)
The DECIMAL
data type is provided
to handle large numeric values with exact decimal storage.
Syntax
DECIMAL [ ( precision[,scale]) ]
- precision defines the number of significant digits (limit is 32, default is 16).
- scale defines the number of digits to the right of the decimal point.
- When no scale is specified, the data type defines a floating point number.
- When no (precision, scale) is specified,
it defaults to
DECIMAL(16)
.
Usage
Use the DECIMAL
data type when you need to store values that have a fixed
number of digits on the right and left of the decimal point (DECIMAL(p,s)
), or to
store a floating point decimal with an exact number of significant digits
(DECIMAL(p)
).
DEC
, DECIMAL
and NUMERIC
are
synonyms.
DECIMAL
variables are initialized to NULL
in
functions, modules and globals.
When using DECIMAL(p,s)
with a precision and scale, you define a decimal for
fixed point arithmetic, with p significant digits and s digits on the right of the decimal point.
For example, DECIMAL(8,2)
can hold the value 123456.78 (8 (p) = 6 digits on the
left + 2 (s) digits on the right of the decimal point).
DECIMAL(p)
with a precision but no scale, you define a
floating-point number with p significant digits. For example, DECIMAL(8)
can store
12345678, as well as 0.12345678.DECIMAL
types with precision and scale to
implement portable code, and avoid mistakes if default sizes apply when precisions and/or scale are
omitted in SQL statements. For example, with Oracle®, a NUMBER(p)
is equivalent to a DECIMAL(p,0)
in BDL,
not DECIMAL(p)
. When using DECIMAL
without a precision and scale, it defaults to
DECIMAL(16)
, a floating-point number with a precision of 16 digits.
MAIN
DEFINE d1 DECIMAL(10,4)
DEFINE d2 DECIMAL(10,3)
LET d1 = 1234.4567
LET d2 = d1 / 3 -- Rounds decimals to 3 digits
DISPLAY d1, d2
END MAIN
DECIMAL
values can be converted to strings based on the DBFORMAT (or DBMONEY) environment variable (defines
the decimal separator) setting.
Value ranges
The largest absolute value that a DECIMAL(p,s)
can store without errors is
10p-s - 10s. The stored value can have up to 30 significant decimal
digits in its fractional part, or up to 32 digits to the left of the decimal point.
When using DECIMAL(p,s)
the range of values is defined by the p, the
number of significant digits. For example, a variable defined as
DECIMAL(5,3)
can store values in the range -99.999 to 99.999. The
smallest positive non zero value is 0.001.
When using DECIMAL(p)
the magnitude can range from -N*10-124 to
N*10124, where N can have up to p significant digits and be 0<N<10. For
example, a variable defined as DECIMAL(5)
can store values in the range
-9.9999e-124 to 9.9999e+124. The smallest positive non zero value is 9.9999e-130.
Exceptions
When the default exception handler is used, if you try to assign a value larger than the
decimal definition (for example, 12345.45 into DECIMAL(4,2)
), no
out of range error occurs, and the variable is assigned with NULL
.
If WHENEVER ANY ERROR
is used, it raises error -1226. If you do not
use WHENEVER ANY ERROR
, the STATUS variable is not set to
-1226.
Data type conversion can be controlled by catching the runtime exceptions. For more details, see Handling type conversion errors.
Computation and rounding rules
When computing or converting decimal values, the "round half away from zero" rule will apply: If the fraction of the value v is exactly 0.5, then r = v + 0.5 if v is positive, and r = v - 0.5 if v is negative. For example, when the result must be rounded to a whole number, 23.5 gets rounded to 24, and -23.5 gets rounded to -24.
In the next example, the division result of 11 / 3 gives the infinite decimal value
3.666666... (with an infinite decimal part). However, this value cannot be stored in a fixed
point decimal type. When stored in a DECIMAL(10,2)
, the value will be
rounded to 3.67, and when multiplying 3.67 by 3, the result will be 11.01, instead of 11:
MAIN
DEFINE v DECIMAL(10,2)
LET v = 11 / 3
DISPLAY "1. v = ", v USING "---&.&&&&&&&&"
LET v = v * 3
DISPLAY "2. v = ", v USING "---&.&&&&&&&&"
END MAIN
1. v = 3.67000000
2. v = 11.01000000
High-precision math functions
A couple
of precision math functions are available, to be used with DECIMAL
values. These functions have a higher precision as the standard C
library functions based on C double data type, which is equivalent
to FLOAT
: