Informix-ESQL/C Programmer's Manual

Informix-ESQL/C Programmer's Manual
Chapter 5: Working with Numeric Data Types

Home Contents Index Master Index New Book

The Floating-Point Data Types

The database server supports the following data types for floating-point values.


SQL Floating-Point Data Type	ESQL/C or C Language Type	Range of Values
`SMALLFLOAT`, `REAL`	float	Single-precision values with up to 8 significant digits
`FLOAT`, `DOUBLE PRECISION`	double	Double-precision values with up to 16 significant digits
`DECIMAL(`p`)`	decimal	Absolute value range between 10-130 and 10124

Declaring float Host Variables

When you use the C float data type (for SMALLFLOAT values), be aware that most C compilers pass float to a function as the double data type. If you declare the function argument as a float, you might receive an incorrect result. For example, in the following excerpt, :hostvar might produce an incorrect value in tab1, depending on how your C compiler handles the float data type when your program passes it as an argument.

main()

{

	double dbl_val;



	EXEC SQL connect to 'mydb';

	ins_tab(dbl_val);



...



}



ins_tab(hostvar)

EXEC SQL BEGIN DECLARE SECTION;

	PARAMETER double hostvar;

EXEC SQL END DECLARE SECTION;

{

	EXEC SQL insert into tab1 values (:hostvar, ...);

}

For more information on the DECIMAL data type, see Chapter 4, "Working with the DECIMAL Data Type."

Implicit Data Conversion

When an ESQL/C program fetches a floating-point column value (data type of DECIMAL(n), FLOAT, or SMALLFLOAT) into a character host variable (char, fixchar, varchar, or string), it includes only the number of decimal digits that can fit into the character buffer. If the host variable is too small for the full precision of the floating-point number, ESQL/C rounds the number to the precision that the host variable can hold.

In the following code fragment, an ESQL/C program retrieves the value 1234.8763512 from a FLOAT column that is called principal into the prncpl_strng character host variable:

EXEC SQL BEGIN DECLARE SECTION;

	char prncpl_strng[15]; /* character host variable */

EXEC SQL END DECLARE SECTION;
.

.

.


EXEC SQL select principal into :prncpl_strng from loan

	where customer_id = 1098;

printf("Value of principal=%s\n", prncpl_strng);

Because the prncpl_strng host variable is a buffer of 15 characters, ESQL/C is able to put all decimal digits into the host variable and this code fragment produces the following output:

Value of principal=1234.876351200

However, if the preceding code fragment declares the prncpl_strng host variable as a buffer of 10 characters, ESQL/C rounds the FLOAT value to fit into prncpl_strng and the code fragment produces the following output:

Value of principal=1234.8764

ESQL/C assumes a precision of 16 decimal digits for DECIMAL(n), FLOAT, or SMALLFLOAT values. (For DECIMAL(n,m), ESQL/C assumes m decimal digits.)

Informix-ESQL/C Programmer's ManualChapter 5: Working with Numeric Data Types Home Contents Index Master Index New Book

The Floating-Point Data Types

Declaring float Host Variables

Implicit Data Conversion

Informix-ESQL/C Programmer's Manual
Chapter 5: Working with Numeric Data Types

Home Contents Index Master Index New Book