allan.jacobs@Eng 1998-09-17

Floating point NaN's should be converted to int or long 0 in narrowing primitive
and casting conversions. Conversions to int are not done correctly.


algol% ${JDK}/bin/java -fullversion
java full version "JDK-1.2fcs-J"
algol% uname -a
SunOS algol 5.6 Generic sun4u sparc SUNW,Ultra-2
algol% ${JDK}/bin/javac -d . X.java
algol% ${JDK}/bin/java -verify X
NaN NaN
2147483647 2147483647
0 0
algol% cat X.java

public class X {
public static void main( String[] argv ) {
        double d1=Double.NaN;
        double f1=Float.NaN;
        System.out.println(f1+" "+d1);
        System.out.println(((int)d1)+" "+((int)f1));
        System.out.println(((long)d1)+" "+((long)f1));
}
}