Floating point precision variation in Python 2.7.5

Question

If I run the following code in Python 2.7.5 console:

>>> import math
>>> math.radians(0.000001)

I get

1.7453292519943295e-08

However, if I put the same code in a file:

$ cat floatingtest.py
import math
print(math.radians(0.000001))

And run it, I get:

$ python.exe floatingtest.py
1.74532925199e-08

Why the difference in floating point precision when running code in a script vs. running code in the console?

(Python 3.3 doesn't seem to have this 'issue'. Both ways return the same high-precision value.)

Answer 1

This is the difference between repr and str :

>>> repr(math.radians(0.000001))
'1.7453292519943295e-08'
>>> str(math.radians(0.000001))
'1.74532925199e-08'

By default, print calls str on its arguments, but the REPL displays objects using repr when there is no assignment (and the return value is not None ).

Answer 2

It has no relation to precision only to the representation:

In [1]: import math    
In [2]: math.radians(0.000001)
Out[2]: 1.7453292519943295e-08    
In [3]: print math.radians(0.000001)
1.74532925199e-08    
In [4]: str(math.radians(0.000001))
Out[4]: '1.74532925199e-08'