large_float_exponent(1.0e+100).