Article ID: 28164 - View products that this article applies to.
This article was previously published under Q28164
This article has been archived. It is offered "as is" and will no longer be updated.
This article describes a case where the PRINT USING statement appears to round numeric constants to unexpected values.
For the following statement
the above Basic versions give the following output
while you might have expected the following rounding
0.25 0.25 0.26 0.28 0.28
or with IEEE rounding (to the nearest even integral value), you would expect the following:
0.25 0.26 0.27 0.28 0.29
This behavior occurs because the internal representation of the numbers used by Basic (IEEE floating-point format) differs slightly from the decimal numbers typed into the source code.
0.24 0.26 0.26 0.28 0.28
IEEE floating-point format cannot accurately represent numbers that are not of the form 1.x to the power of y (where x and y are base 2 numbers). The internal representation will be slightly more or slightly less than the decimal numbers typed into the source code.
The internal representations are correctly rounded and displayed in the above program. This is not a software problem.
If you want floating-point constants in the source code to be exactly stored, you can append the constant with the CURRENCY type-cast operator, an "at sign" (@) character. The CURRENCY data type is only found in Microsoft Basic PDS 7.00 and 7.10. The following example rounds to 0.25, 0.26, 0.27, 0.28, 0.29 as you may have wanted:
Note that the CURRENCY data type can store 19 digits, but only 4 digits can be used after the decimal point.
Article ID: 28164 - Last Review: January 6, 2015 - Revision: 2.1