Python: Why call '__init__' instead of className()?

Question

Since calling className() will execute the code in __init__(args), why in the code below, is someone explicitly calling __init__?

class Example(Frame):
 def __init__(self, parent):
 Frame.__init__(self, parent) 

Is there any difference in the actual code that is executed between the two method calls, or is chossing __init__() over className() simply arbitrary?

Running Python 3.4

Answer 1

className() does more than call __init__. It also calls __new__, which creates a new instance of the class. So calling Frame() would not do the superclass initialization on the same self object; it would create a new object and initialize that.

You call __init__ when you want to run just __init__ on a particular instance you've already created. Typically this is in a situation like the one you show, where you want to let a superclass do its initialization on top of a subclass's initialization.

• AllTradesJack: so calling Frame's __init__ is like calling super() in java?
• BrenBarn: @joshsvoss: Possibly. I don't really know Java. My experience with answering questions on this site is that people usually wind up with a better understanding if they try to understand what Python actually does instead of mapping it onto Java.
• daniel: You can also use super(Frame, self).__init__(parent), but I confess that I don't completely understand the difference in this context.
• BrenBarn: @daniel: Yeah, that's sort of a separate issue though. Basically the super way is more flexible in case you later rearrange the class hierarchy.