Name: nkR10003 Date: 09/14/2001
According to the specification, the parse method returns "the most economical"
representation of the passed numeric string, in the form of either a long
or a double. While in the parse method, if a multiplier has been set, the
numeric representation of the string is divided by this multiplier. If the
numeric representation is in the form of a long, the quotient is rounded
towards zero. If the string is represented as a double, the quotient is
rounded according to IEEE 754 round-to-nearest mode. The specification
makes no mention of whether this division should take place with the string
being represented as a long or a double.
For example, one can pass a string representing Long.MIN_VALUE ("-9223372036854775808")
to DecimalFormat.parse() method. If the DesimalFormat object is given with a
multiplier of 100 the returned result will be undefined. If division takes place as
a long, the result is -92233720368547758. If division takes place as a
double, the result is -92233720368547760.
Plese specify the correct behavior.