Why can't I call hash() on an apparently hashable method of an unhashable instance?

Question

Let's say I have a dictionary:

>>> d = {} 

It has a method clear():

>>> d.clear <built-in method clear of dict object at 0x7f209051c988> 

... which has a __hash__ attribute:

>>> d.clear.__hash__ <method-wrapper '__hash__' of builtin_function_or_method object at 0x7f2090456288> 

... which is callable:

>>> callable(d.clear.__hash__) True 

So why can't I hash it?

>>> hash(d.clear) Traceback (most recent call last):   File "<stdin>", line 1, in <module> TypeError: unhashable type: 'dict' 

Note: I know that dict objects are unhashable – I'm curious as to why this restriction extends to their methods, even though, as noted above, they appear to claim otherwise?

Answer 1

It is a bound method, and bound methods have a reference to self, e.g. the dictionary. This makes the method un-hashable.

You can hash the unbound dict.clear method:

>>> d = {} >>> d.clear.__self__ {} >>> d.clear.__self__ is d True >>> hash(d.clear) Traceback (most recent call last):   File "<stdin>", line 1, in <module> TypeError: unhashable type: 'dict' >>> hash(dict.clear) -9223372036586189204 

Methods on instances that are hashable will themselves be hashable, so the object type for built-in bound methods implements a __hash__ method but raises TypeError when the __self__ attribute is not hashable. This is consistent with the object.__hash__ method documentation; if you can set it to None or not implement it at all then that is preferable but for these cases where the hashability is only known at runtime raising a TypeError is the only option available.