Python: avoid setattr loop using dict, shouldn't it loop anyway?

Question

I know that you should be aware while redefine setattr method that you could end in a loop.

I know there's two solution to avoid loops:

1) using object superclass, calling :

object.__setattr__(instance,attr,value)

2) using the class dict:

self.__dict__[attr] = value

My question is, shouldn't the __dict__ solution end up in a loop too? and why it isn't?

Is because that way we're calling the __setattr__ of the __dict__ object (everything is an object) or what?

And if so, shouldn't it work the same for everything?

`a[b] = c` doesn't set an attribute. It sets an item. It's a different thing both conceptually and w.r.t. dunder methods involved. — , May 30 '13 at 16:01

score 5 · Accepted Answer · answered May 30 '13 at 16:02

5

Why would you expect it to end up in a loop?

instance.__dict__[attr] = value is basically what object.__setattr__(instance,attr,value) does (for normal attributes). Note that __dict__[attr] = whatever does not call __setattr__ at all. It calls __setitem__ which is a different method entirely.

answered May 30 '13 at 16:02

mgilson

300,191
65
633
696

1

It also calls it on an entirely different object. – May 30 '13 at 16:05
@delnan -- True. Of course, using `__getattribute__` you can get into the infinite loop using `self.__dict__`, but ... – mgilson May 30 '13 at 16:13
So it is because is calling __setitem__ or because is a different object? thanks guys – Mattia May 30 '13 at 16:54

Python: avoid __setattr__ loop using __dict__, shouldn't it loop anyway?

1 Answers1

Python: avoid setattr loop using dict, shouldn't it loop anyway?