views:

100

answers:

4

Which should I use _foo (an underscore) or __bar (double underscore) for private members and methods in Python?

+3  A: 

The double underscore. It mangles the name in such a way that it can't be accessed simply through __fieldName from outside the class, which is what you want to begin with if they're to be private. (Though it's still not very hard to access the field.)

class Foo:
    def __init__(self):
        self.__privateField = 4;
        print self.__privateField # yields 4 no problem

foo = Foo()
foo.__privateField
# AttributeError: Foo instance has no attribute '__privateField'

It will be accessible through _Foo__privateField instead. But it screams "I'M PRIVATE DON'T TOUCH ME", which is better than nothing.

zneak
+4  A: 

Please note that there is no such thing as "private method" in Python. Double underscore is just name mangling:

>>> class A(object):
...     def __foo(self):
...         pass
... 
>>> a = A()
>>> A.__dict__.keys()
['__dict__', '_A__foo', '__module__', '__weakref__', '__doc__']
>>> a._A__foo()

So therefore __ prefix is useful when you need the mangling to occur, for example to not clash with names up or below inheritance chain. For other uses, single underscore would be better, IMHO.

EDIT, regarding confusion on __, PEP-8 is quite clear on that:

If your class is intended to be subclassed, and you have attributes that you do not want subclasses to use, consider naming them with double leading underscores and no trailing underscores. This invokes Python's name mangling algorithm, where the name of the class is mangled into the attribute name. This helps avoid attribute name collisions should subclasses inadvertently contain attributes with the same name.

Note 3: Not everyone likes name mangling. Try to balance the need to avoid accidental name clashes with potential use by advanced callers.

So if you don't expect subclass to accidentally re-define own method with same name, don't use it.

Daniel Kluev
+1  A: 

Double underscore. That mangles the name. The variable can still be accessed, but it's generally a bad idea to do so.

Use single underscores for semi-private (tells python developers "only change this if you absolutely must") and doubles for fully private.

Not quite right. A single underscore means you are tampering with undocumented properties of a object that are subject to change and you may well break the invariants maintained by the object so you should understand exactly what you are doing if you decide to break the object encapsulation.
msw
A: 

Because thats coding convention. See here for more.

Zaki
The OP was asking which one to use. Not why.
Wallacoloo