Here's a bit of a newbie Python question about instance variables.
Consider the following Python 2.7 class definition:
class Foo(object):
a = 1
def __init__(self):
self.b = 2
def __repr__(self):
return "%s" % self.__dict__
Now, when I create an instance of Foo, Foo.__dict__ contains b, but not a.
>>> x=Foo()
>>> x
{'b': 2}
>>> dir(x)
['__class__', '__delattr__', '__dict__', '__doc__', '__format__',
'__getattribute__', '__hash__', '__init__', '__module__', '__new__',
'__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__',
'__str__', '__subclasshook__', '__weakref__', 'a', 'b']
>>> x.__dict__
{'b': 2}
And here I thought I had a pretty good grasp on the Way of the Python.
What's the difference between x.a and x.b? As far as I can tell they're both instance variables.
Edit: OK, re-reading the Python docs I see that Foo.a is a class attribute rather than an instance variable. Hm... I guess the confusion comes from the fact that I can assign a new value to x.a and the new value only affects the x instance -- I suppose I'm now aliasing a member variable over the top of the Foo.a attribute:
>>> y=Foo()
>>> y.a = 2
>>> y
{'a': 2, 'b': 2}
>>> x
{'b': 2}
>>> x.a
1
>>> z=Foo()
>>> z
{'b': 2}
>>> z.a
1
>>> Foo.a
1
>>> x.a
1
>>> y.a
2
So, now I overwrite the previous value of Foo.a, and it affects all instances of Foo that haven't aliased Foo.a:
>>> Foo.a=999
>>> x.a
999
>>> y.a
2
__dict__too much: clases using__slots__and extension classes (from C or Cython) don't put everything in there either. Most uses can be replaced by some combination ofgetattr,setattr,hasattr, anddir.