Which is more correct for compaing to None?

[Greg Ball]

> Both of these seem to work. Which is more correct?
> 
> if foo is None: 
>     print 'is'
> 
> if foo == None:
>     print '=='


I'm going to use 'foo is None' in future... 

The context I use this in is default arguments:

def plot(seqy, seqy=None):
	if seqx is None:
		seqx = range(len(seqy))
	# plot seqy vs. seqx
	...

In the newest version of Numerical Python, 'a == None' raises an exception
if a is a Numpy array.

'is' should be completely safe since it just uses object identity (i.e.
it doesn't ask the object how to do the comparison).  I believe python
guarantees there is a unique None object.

In principle the object identity check is faster, particularly if foo
happens to be an instance, since a lot less work gets done under the
covers. It might even be faster than a truth test, but more importantly a
truth test might not be the right behaviour in general, where foo == ''
is meaningful for example.

The standard library says

$ grep 'is None' /usr/lib/python2.2/*.py | wc -l
    280
$ grep 'is not None' /usr/lib/python2.2/*.py | wc -l
    107
$ grep '== None' /usr/lib/python2.2/*.py | wc -l
      0
grep '!= None' /usr/lib/python2.2/*.py | wc -l
      0

with the first example that comes up appearing in BaseHTTPServer.py.

--Greg Ball