The object named 'type', in 2.2

[Jason Orendorff]

Philip Swartzleonard writes:
> 3. Type is it's own parent:
>
> >>> type.__class__ is type
> 1
>
> It is this last one particularly that I do not understand.

Well, the type of TypeType has always been itself.  For a long
time, it's been like this:

  type(3) is types.IntType
  type(types.IntType) is types.TypeType
  type(types.TypeType) is types.TypeType

It's the same today, except for two things:

  1.  type (like str, int, list, etc.) used to be just a function.
      Now it is a type.  But (like str, int, list, etc.) it still
      works if you call it as a function, just like the old way.

  2.  Now a bunch more objects have a nice x.__class__ attribute
      that is the same thing as type(x).

> I ran into it
> today trying to fix a recrusion-death problem in PyCrust under 2.2 -- it
> recursivly accesses thing's __class__es in it's attribute-finding
> routines. I've tried to find information on this, but it has eluded me so
> far...

I don't know where you can find more information, sorry.
It might be that recursively looking at __class__es is not what
you want after all; maybe you should walk x.__class__.__bases__
instead.

## Jason Orendorff    http://www.jorendorff.com/