Java VM bug?

Question

Dear all,
I have observed this instance where my JVM implementation behaves in
a manner that appears inconsistent.
The following lines of code all compile:
long a = Long.parseLong("5L");
Long a1 = new Long("5L");
Long a2 = Long.valueOf("5L");
float b = Float.parseFloat("5f");
Float b1 = new Float("5f");
Float b2 = Float.valueOf("5f");
However, at run time a NumberFormatException is thrown at every
one of the first three lines because the JVM does not accept the suffix
"L" in a String number description.
Ok, fair enough!
But why is that rule not enforced for the Float class ( the next three
lines of code do not throw any exceptions, which shows that in this
case the suffix "f" is recogined in the String for what it is)?
Is this a bug or inconsistency?
Or is there a good reason for it?
Thanks
Jorge

Answer 1

In the code of java.lang.FloatingDecimal you can find:
          * We parsed everything we could.
          * If there are leftovers, then this is not good input!
         if ( i < l &&
                ((i != l - 1) ||
                (in.charAt(i) != 'f' &&
                 in.charAt(i) != 'F' &&
                 in.charAt(i) != 'd' &&
                 in.charAt(i) != 'D'))) {
                break parseNumber; // go throw exception
            }