I have a C extension module that is returning some doubles. When the doubles get out of range, the numbers print as 'nan'. Is there a better way to test for NaN than str(p1)=='nan' where p1 is a float? python2.3 Thanks, John Hunter