Re: staticmethod(cls)

On 9/4/25 13:28, Pierre Asselin wrote:
> For reasons I won't go into I ended up defining classes inside of
> a class body. That's not in the manual, but I see no reason it would
> fail.

Nested classes certainly work. I've seen them used quite a bit in 
testing code, where you may want an object which is only interesting in 
the scope of the test class it appears in; less so elsewhere. In the 
end, it's mostly about namespacing. As far as "in the manual", 
everything's an executable statement, including class definitions, so 
anywhere you can syntactically put a statement you can put a class 
definition - top level, inside classes, inside functions, nested 
multiple levels deep, etc. Nesting means the inner symbols appear in the 
namespace created for the outer. How those are rendered by various tools 
may be an implementation decision of those tools? Have you tried 
introspecting to see if the namespaces actually look any different 
between the staticmethod case and the non-staticmethod case?

> But then, I accidentally decorated the inner classes with
> @staticmethod! It didn't break anything, but it had an interesting
> side effect: help() and pydoc on the outer class shows the docs
> of the inner class members. Normally only the inner classes are
> listed, not their contents.
> 
> Here's a minimal example.
> ----------------------------------------------------------------------
> class Outer:
>      '''Outer class'''
>      def outer_method(self):
>          '''outer method'''
> 
>      class Inner0:
>          '''Inner class'''
>          def inner_0(instance):
>              '''inner method, not seen by pydoc.'''
> 
>      @staticmethod
>      class Inner1:
>          '''Inner class passed through staticmethod()'''
>          def inner_1(instance):
>              '''inner method, visible to pydoc !'''
> ----------------------------------------------------------------------
> 
> Save and run pydoc, or import and run help(_module_.Outer).
> 
> Now that's *seriously* not-in-the-manual. I won't rely on that
> behavior, plus the expanded documentation is a bit too much,
> but I'm curious as how that interaction between staticmethod()
> and help()/pydoc came about.
>