Question

Possible Duplicate:
“Least Astonishment” in Python: The Mutable Default Argument

I’m finding that dictionary arguments to the init() function of my class are defaulting to values I’ve previously set in previous instances. I really don’t understand this behavior, and it doesn’t seem to happen with lists or basic variables. Example code:

class TestClass:
  def __init__(
    self,
    adir={},
    alist=[],
    avar=None
  ):
    print("input adir: " + str(adir)) #for test2, shows test1.mydir
    self.mydir = adir
    self.mylist = alist
    self.myvar = avar

test1 = TestClass()   
test1.mydir['a'] = 'A'
test1.mylist = ['foo'] 
test1.myvar = 5

test2 = TestClass()
print(test2.mydir) #has same value of test1!
print(test2.mylist)
print(test2.myvar)

The output looks like this:
initializing test1
input adir: {}
initializing test2
input adir: {‘a’: ‘A’}
{‘a’: ‘A’}
[]
None
Why does the dictionary argument (adir) to test2 get set to test1.mydir? Especially, why is the behaviour different than other mutable types like list?

Thank you!

Answer 1

As DSM says, don’t modify mutable default arguments. Do this instead:

class TestClass:

  def __init__(self, adir=None, alist=None, avar=None):
    if alist is None:
      alist = []
    if adir is None:
      adir = {}