The plaintext format of floating or fixed point numbers and of date and time specifications can be determined specifically for a file. Thus independence is obtained from the current SET default definition.
A format specification consists of one of the keywords DEC (decimal representation), DATE (date representation), TIME (time representation), or TIMESTAMP (time representation) followed by a mask enclosed in single quotation marks. The same syntax rules apply to this mask that are valid for the corresponding SET parameter.
Example:
DATALOAD TABLE time_entry
day 1-8 DEFAULT NULL
from 10-14 DEFAULT NULL
to 16-20 DEFAULT NULL
break 22-24 DEFAULT NULL
INFILE * EBCDIC
DATE 'dd.mm.yy'
TIME 'hh:mm'
DEC '//./'
NULL ' - '
01.04.97 08:05 17:00 0.5
02.04.97 07:55 16:00 0.5
...
Numbers in default format (no thousand sign, decimal point) can always be loaded. If the used decimal representation is, e.g., / /,/, then it is also possible to load numbers such as 99 999.99 or 1,2345E+04.
Floating point numbers must not contain blanks between mantissa and exponent. Unsigned or one-digit exponents are allowed.
Date and time values are converted into the 8-digit SQL default representation. In case of DATE, a reasonableness check is made for day and month (Is there a February 29 in this year?). A TIME value has four digits for the hour; these need not coincide with a clock time (0.00 to 24.00 hours). Minutes and seconds must lie between 0 and 59. Timestamp values consist of a date and a time field. The time field has two digits for the hour, two digits for the minutes, two digits for the seconds, and six digits for the micro seconds. This means, it has a length of 20 digits in its default representation.