python - Why does setting a descriptor on a class overwrite the descriptor?

Question

Simple repro:

class VocalDescriptor(object):
    def __get__(self, obj, objtype):
        print('__get__, obj={}, objtype={}'.format(obj, objtype))
    def __set__(self, obj, val):
        print('__set__')

class B(object):
    v = VocalDescriptor()

B.v # prints "__get__, obj=None, objtype=<class '__main__.B'>"
B.v = 3 # does not print "__set__", evidently does not trigger descriptor
B.v # does not print anything, we overwrote the descriptor

This question has an effective duplicate, but the duplicate was not answered, and I dug a bit more into the CPython source as a learning exercise. Warning: i went into the weeds. I'm really hoping I can get help from a captain who knows those waters. I tried to be as explicit as possible in tracing the calls I was looking at, for my own future benefit and the benefit of future readers.

I've seen a lot of ink spilled over the behavior of __getattribute__ applied to descriptors, e.g. lookup precedence. The Python snippet in "Invoking Descriptors" just below For classes, the machinery is in type.__getattribute__()... roughly agrees in my mind with what I believe is the corresponding CPython source in type_getattro, which I tracked down by looking at "tp_slots" then where tp_getattro is populated. And the fact that B.v initially prints __get__, obj=None, objtype=<class '__main__.B'> makes sense to me.

What I don't understand is, why does the assignment B.v = 3 blindly overwrite the descriptor, rather than triggering v.__set__? I tried to trace the CPython call, starting once more from "tp_slots", then looking at where tp_setattro is populated, then looking at type_setattro. type_setattro appears to be a thin wrapper around _PyObject_GenericSetAttrWithDict. And there's the crux of my confusion: _PyObject_GenericSetAttrWithDict appears to have logic that gives precedence to a descriptor's __set__ method!! With this in mind, I can't figure out why B.v = 3 blindly overwrites v rather than triggering v.__set__.

Disclaimer 1: I did not rebuild Python from source with printfs, so I'm not completely sure type_setattro is what's being called during B.v = 3.

Disclaimer 2: VocalDescriptor is not intended to exemplify "typical" or "recommended" descriptor definition. It's a verbose no-op to tell me when the methods are being called.

Answer 1

You are correct that B.v = 3 simply overwrites the descriptor with an integer (as it should). In the descriptor protocol, __get__ is designed to be called as instance attribute or class attribute, but __set__ is designed to be called only as instance attribute.

For B.v = 3 to invoke a descriptor, the descriptor should have been defined on the metaclass, i.e. on type(B).

>>> class BMeta(type): 
...     v = VocalDescriptor() 
... 
>>> class B(metaclass=BMeta): 
...     pass 
... 
>>> B.v = 3 
__set__

To invoke the descriptor on B, you would use an instance: B().v = 3 will do it.

The reason for B.v also invoking the getter is to allow user's customization of what B.v does, independently of whatever B().v does. A common pattern is to allow direct access on the descriptor instance, by returning the descriptor itself when a class attribute access was used:

class VocalDescriptor(object):
    def __get__(self, obj, objtype):
        if obj is None:
            return self
        print('__get__, obj={}, objtype={}'.format(obj, objtype))
    def __set__(self, obj, val):
        print('__set__')

Now B.v would return some instance like <mymodule.VocalDescriptor object at 0xdeadbeef> which you can interact with. It is literally the descriptor object, defined as a class attribute, and its state B.v.__dict__ is shared between all instances of B.

Of course it is up to user's code to define exactly what they want B.v to do, returning self is just the common pattern. A classmethod is an example of a descriptor which does something different here, see the Descriptor HowTo Guide for a pure-python implementation of classmethod.

Unlike __get__, which can be used to customize B().v and B.v independently, __set__ is not invoked unless the attribute access is on an instance. I would suppose that the goal of customizing B().v = other and B.v = other using the same descriptor v is not common or useful enough to complicate the descriptor protocol further, especially since the latter is still possible with a metaclass descriptor anyway, as shown in BMeta.v above.