Hello,
I encountered some problems when i tried to use the float type on my
embedded board based on arm926ejs core.
Indeed with the following code:
/
************************************************************************/
public class helloworld{
    public static void main(String argv[]){
        double dvar = 0.1;
        float fvar = 0.1f;
        System.out.println("Hello\n");
        System.out.println("\nboth of these should be 0.1:");
        System.out.println("double var = " + dvar);
        System.out.println("float var = " + fvar);

        System.out.println("\ndouble size = " + Double.SIZE);
        System.out.println("float size = " + Float.SIZE);

        System.out.println("\n");
    }
}
/
************************************************************************/
Here's what i get when i issue the following command:
# jamvm helloworld
Hello

both of these should be 0.1:
double var = 0.00000103455646584
float var = 0.0004:3256

double size = 64
float size = 32

After some researches i found a patch about the arm float parsing
http://www.mail-archive.com/classpath-patches@gnu.org/msg08956.html

It seems to correct the problem, but is there someone who can explain
exactly the problem ?
On the other hand, i'd like to understand how the floats are managed
into the java code.
Since my hardware does not embed a FP unit and the linux kernel is
disabled how the float are processed ?
regards
greg