Comments on: Common Mistake about Python default parameter value

By: Joni Wilmoth

Joni Wilmoth — Sat, 06 Mar 2010 23:29:00 +0000

This is great! Thanks for this post. I am starting development and this got me straight.

By: Nicolas Trangez

Nicolas Trangez — Sat, 03 Oct 2009 10:46:29 +0000

It’s pretty much like this construct in C/C++/Java/…:
myList = !myList ? [] : myList, or the inverse: myList = myList ? myList : []

Or:
if myList is None: myList = [] # No need for an else case, this would be "myList = myList" anyway

I think the Python-form is slightly better to read if you put the ‘most common case’ as truth value of the predicate. This makes sure the ‘most common value’ will be immediately visible to the reader, on the right of the assignment operator.

By: Nicolas Trangez

Nicolas Trangez — Sat, 03 Oct 2009 10:42:50 +0000

The ‘or’ thing comes from the days when A if PRED else B was not part of the language (it was introduced in Python 2.5 is I’m not mistaken)

Similar:
>>> True and 1 or 2 1 >>> False and 1 or 2 2

This construct has a major catch though:
>>> False and 0 or 1 1 >>> True and 0 or 1 1

By: amal

amal — Fri, 02 Oct 2009 07:09:02 +0000

thanks
I faced the same problem before
and it took time to understand the problem

By: Ahmed S. Farghal

Ahmed S. Farghal — Fri, 02 Oct 2009 06:41:45 +0000

it’s like ternary operator in C of Java, myList will be assigned a new empty list ( [] ) if the passed myList was set to None, but if it has a value it’ll be pointing to that value.

By: Medhat Dawoud

Medhat Dawoud — Fri, 02 Oct 2009 01:49:24 +0000

it’s a great point that you show .
but i want to understand this piece of code
myList = [] if myList is None else myList
can you illustrate it , please ?

By: Ahmed S. Farghal

Ahmed S. Farghal — Thu, 01 Oct 2009 18:19:59 +0000

Yes, this makes:
myList = [] if myList is None else myList
better because it won’t fail in your second scenario, good point

By: Nicolas Trangez

Nicolas Trangez — Thu, 01 Oct 2009 18:16:14 +0000

A common pattern is
def f(a=None): a = a or list()
which is slightly less verbose. Do note this approach does not work in case the function behaviour would be a modification of the input list, like
def f(a=None): a = a or list() for i in xrange(len(a)): a[i] = a[i] + 1 return a


b1 = [1, 2, 3]

assert f(b1) is b1

b2 = [] # This fails assert f(b2) is b2

But obviously that’s in first place extremely bad/stupid API design.