A PPENDIX B

D IFFERENCES BETWEEN P YTHON VERSIONS

Python programming language is continually being updated and im-
T HE
proved by its creators. The most recent version is version 3, though ver-
sion 2 is still available and finds wide use. (The much earlier version 1, which
dates back to the 1980s, is now seen very rarely.)
The programs that appear in this book are written using Python version 3,
although, as noted in Appendix A, they can also be used with version 2 if you
prefer. If you use version 2 you should include at the beginning of all your
programs the statement

from __future__ import division,print_function

(Note the two underscore characters “__” on either side of the word “future”.)
This statement tells version 2 of Python to behave in the way version 3 does
with respect to the two most important differences between versions, the dif-
ferences in the division of integers and the structure of the print command.
The remainder of this appendix describes the main technical differences
between Python versions 2 and 3, for those readers who are interested.
Division returns a floating-point value: In version 2 of Python the division
of two integers, one by another, returns another integer, rounding down if
necessary. Thus 3/2 gives 1, not 1.5. In version 3 the same operation gives
the floating-point value 1.5. Furthermore, in version 3 even if the result of a
division is in fact an integer, the operation will still give a floating-point value.
Thus 4/2 gives a floating-point 2.0, not an integer 2.
If you are using version 2 of Python, you can duplicate the behavior of
version 3 with respect to division by including the statement

from __future__ import division

at the start of your program. If you are using version 3 and you want the

510
 A PPENDIX B | D IFFERENCES BETWEEN P YTHON VERSIONS

rounding-down behavior of version 2, you can get it by using the integer divi-
sion operation ”//”—see page 23.
Print is a function: In version 3 of Python the print command is a function,
where in version 2 it is a statement. The main practical difference between the
two is that in version 3 you must enclose the argument(s) of a print command
within parentheses, while in version 2 you should not. Thus in version 2 you
might say

print "The energy is",E

while in version 3 you would say

print("The energy is",E)

In most other respects the two commands behave in the same way.
If you are using version 2 of Python, you can duplicate the behavior of the
version 3 print function by including the statement

from __future__ import print_function

at the start of your program.

If you wish to duplicate the behavior of version 3 with respect to both divi-
sion and the print function in the same program, you can use the single state-
ment

from __future__ import division,print_function

at the start of your program (as mentioned previously).

Input returns a string: In version 3 of Python the input function always re-
turns a string, no matter what you type in, even if you type in a number. In
version 2, by contrast, the input statement takes what you type and evaluates
it as an algebraic expression, then returns the resulting value. Thus if you write
a program that includes the statement

x = input()

and you enter “2.5”, the result will be different in Python versions 2 and 3. In
version 2, x will be a floating-point variable with numerical value 2.5, while
in version 3 it will be a string with string value “2.5”. In version 2 if you
entered an actual string like “Hello” you would get an error message, while in
version 3 this works just fine.

511
A PPENDIX B | D IFFERENCES BETWEEN P YTHON VERSIONS

Version 2 of Python includes another function called raw_input, which be-

haves the same way that input does in version 3. Thus if you are using ver-
sion 2 you can still duplicate the behavior of version 3 by using raw_input
everywhere that version 3 programs would use input. (In version 3 the func-
tion raw_input no longer exists.)
There is only one integer type: In version 2 of Python there are two types
of integer variables called int and long. Variables of type int are restricted to
numbers in the range ±231 , but arithmetic using them is very fast; variables
of type long can store numbers of arbitrary size but arithmetic using them is
slower. In version 3 of Python there is only one type of integer variable, called
int, which subsumes both the earlier types. For smaller integer values version 3
will automatically use old-style ints with their fast arithmetic, while for larger
values it will automatically use old-style longs but slower arithmetic. You do
not need to worry about the distinction between the two—Python takes care
of it for you.
In fact, this change appeared earlier than version 3 of Python, starting in
version 2.4. If you are using version 2 of the language, it’s most likely that you
are using either version 2.6 or 2.7, in which case you don’t need to worry about
this point—you already have the improved behavior of Python 3 with respect
to integers.
Iterators: An iterator is an object in Python that behaves something like a list.
It is a collection of values, one after another, but it differs from a list in that the
values are not stored in the memory of the computer waiting for you to look
them up; instead they are calculated on the fly, which saves memory.
In version 2 of Python the function range generates an actual list of num-
bers, which occupies space in the computer memory. This can cause problems
if the list is very large. For instance, in version 2 on most computers the state-
ment

for n in range(10000000000):

will give an error message because there is not enough memory to store the
huge list generated by the range function. To get around this problem version 2
provides another function called xrange, which acts like range but produces
an iterator. Thus “xrange(100)” behaves in many respects like a list of 100
elements, but no actual list is created. Instead, each time you ask for the next
element in the list the computer just works out what that element ought to be
and hands the value to you. The value is never stored anywhere. Thus you

512
 A PPENDIX B | D IFFERENCES BETWEEN P YTHON VERSIONS

could say

for n in xrange(10000000000):

and the program would run just fine without crashing (although it would take
a long time to finish because the loop is so long).
In version 3 of Python range behaves the way xrange does in version 2,
producing an iterator, not a true list. Since the most common use of range by
far is in for loops, this is usually an improvement: it saves memory and often
makes the program run faster. Sometimes, however, you may want to generate
an actual list from a range. In that case you can use a statement of the form

x = list(range(100))

which will create an iterator then convert it into a list. In version 2 of Python
you do not need to do this (although it will work fine if you do).
(The function arange in the package numpy, which is similar to range but
works with arrays rather than lists, really does create an array, not an itera-
tor. It calculates all the values of the array and stores them in memory, rather
than calculating them on the fly. This means that using arange with large argu-
ments can slow your program or cause it to run out of memory, even in Python
version 3.)
Another situation in which iterators appear in version 3 is the map function,
which we studied in Section 2.4.1. Recall that map applies a given function to
each element of a list or array. Thus in version 2 of Python

from math import log

r = [ 1.0, 1.5, 2.2 ]
logr = map(log,r)

applies the natural logarithm function separately to each element of the list
[1.0,1.5,2.2] and produces a new list logr with the three logarithms in it.
In version 3, however, the map function produces an iterator. If you need a real
list, you would have to convert the iterator like this:

logr = list(map(log,r))

There are a number of other differences between versions 2 and 3 of Python,

but the ones above are the most important for our purposes. A full description
of the differences between versions can be found on-line at the main Python
web site www.python.org.

513