Class variable inheritance

[Terry Reedy]

Lie Ryan wrote:

> 
> Note that when the python interpreter meets this statement:
> 
> class B(P):
>     def foo(self):
>         print('ab')
>     X = 'f'
> 
> the compiler sees a class statement -> create a new blank class
>                                     -> assign P as the new class' parent

No, it saves the name 'B' and bases tuple P, and create a new *dict*, 
call it d here though it is anonymous as far as the class body is concerned.

>                                     -> *execute* the class' body

with the new dict d as the local namespace. In other words, the 
equivalent of

exec('''body''', globals(), {}) # new Py3 exec() form

> # new context
> the compiler sees a def statement -> *compile* the def's body

to a code object which is then attached to a function object

>                                   -> assign the compiled body to B.foo

No, d['foo'] = <function object>

> the compiler sees X = 'f' statement -> assign 'f' to B.X

No, d['X'] = 'f'

> # exit the new context

Up to this point, there is no new class object.

>                                     -> assign the new class to B

It calls meta(name, bases, d), where meta is the metaclass 'type' by 
default but can be any callable which does anything, though the 
intention is that it be a subclass of type or at least something that 
creates a class object, and that d become the backstage attribute dict 
for the result.

> Your problem is related to how the interpreter *execute* a class 
> definition rather than the name resolution.

Terry Jan Reedy