DIGITS Subcommand (EXPORT command)
DIGITS
specifies
the degree of precision for all noninteger numeric values written
to the portable file.
-
DIGITS
has the general formDIGITS
=n, where n is the number of digits of precision. -
DIGITS
applies to all numbers for which rounding is required. - Different degrees of precision cannot be specified for different variables. Thus,
DIGITS
should be set according to the requirements of the variable that needs the most precision. - Default precision methods used by
EXPORT
work perfectly for integers that are not too large and for fractions whose denominators are products of 2, 3, and 5 (all decimals, quarters, eighths, sixteenths, thirds, thirtieths, sixtieths, and so forth.) For other fractions and for integers too large to be represented exactly in the active dataset (usually more than 9 digits, often 15 or more), the representation used in the active dataset contains some error already, so no exact way of sending these numbers is possible. The program sends enough digits to get very close. The number of digits sent in these cases depends on the originating computer: on mainframe IBM versions of the program, it is the equivalent of 13 decimal digits (integer and fractional parts combined). If many numbers on a file require this level of precision, the file can grow quite large. If you do not need the full default precision, you can save some space in the portable file by using theDIGITS
subcommand.
Example
EXPORT OUTFILE=NEWSUM /DROP=DEPT TO DIVISION /MAP /DIGITS=4.
-
DIGITS
guarantees the accuracy of values to four significant digits. For example, 12.34567890876 will be rounded to 12.35.