Real Generic

Python' generic is kind a fake generic that there is no specialization but generic alias.

class A[T]:
    @classmethod
    def cls_info(cls):
        print(T)

    def info(self):
        print(T)

A[int].static_info() # T
A[int].cls_info() # T
A[int]().info() # T
print(issubclass(A[int],A)) # False
print(isinstance(A[int](),A[int])) # False
print(isinstance(A[int](),A)) # True

In this example, you can't get the resolved value of the type var T in any kind of method of class. Type alias A[int] is not subclass of its origin A, and call type alias A[int] only got a instance of the origin type A.

from warp_hint import Generic

class A[T](Generic):
    @staticmethod
    def cls_info(cls):
        print(cls.type_attr(T))

    def info(self):
        print(self.type_attr(T))

A[int].cls_info() # int
A[int]().info() # int
print(issubclass(A[int],A)) # True
print(isinstance(A[int](),A[int])) # True
print(isinstance(A[int](),A)) # True

warp_hint enable all of this, resolved type variable value can be retrieved via a class method type_attr, type alias A[int] become a real specification that subclass A.

class A: ...
class B[t]:...
class C[t](A[t]):...
class D(A[int]):...
class E[k,v](B[k],C[v]): ...

You can build very arbitary generic classes structure.

Generic Cache

Python don't cache generic alias so:

print(list[int] is list[int]) # False
print(list[int] == list[int]) # True 

warp_hint make it True by cache it.

print(list[int] is list[int]) # True