The expression:

   f.equals(new Float(f.toString()));

is true for all instances of the Float class except: Float.NaN,
Float.NEGATIVE_INFINITY and Float.POSITIVE_INFINITY each of
which fail with a parsing error. The same is true of the
Double class.

While it may well make sense to leave the definition of
the equals method with anomalies in these cases we think
it would be useful for the Float class to recognize the
Strings it generates in its toString() method to represent
these values. The mapping to and from a string representation
is used by the XMLEncoder class in the java.beans package and
it would seem to be more useful to put any special case code
required to handle these values in the most general place where
it applies - in the valueOf() methods of the Float and Double
classes which are used by their constructors.

See XMLEncoder bug: 4426652.