the monopoly of classes

Aaron J Reichow reic0024 at d.umn.edu
Fri May 23 00:50:19 UTC 2003


On Fri, 23 May 2003, Andreas Raab wrote:

> There is a different way of looking at "class Object" which I actually
> prefer. ...

Another way of looking at class Object and some of the other classes which
are part of the Smalltalk core- Behavior, Class, Metaclass, etc- as the
language support often written in another language.  Python doesn't have a
global root class like Smalltalk, but it implements the same functionality
in C using a less direct method.  That is, you can copy, deep copy,
instantiate, etc objects without having a root class define all this
functionality.  I for one prefer the way Smalltalk does it- I'm a sucker
for consistency, simplicity and of not having to stoop to C when I want to
play around with certain functionalities.

> BTW, I am very seriously considering doing exactly that - get rid of the
> "visibility" of class object in favour of a global context. That this could
> be implemented by setting the superclass pointer of a "root class" to the
> environment object in order to make the lookup of the "global functions"
> more efficient is a completely different issue.

At the end of the day, what real benefit would this provide?  It'd be an
interesting trick, yes.  But would it be modifiable by any developer?
Since it is Smalltalk, I assume it would be.  If this invisible root class
was modifiable- methods able to be added, removed or changed by
developers, what woudl be the point to hide it?

Regards,
Aaron

  Aaron Reichow  ::  UMD ACM Pres  ::  http://www.d.umn.edu/~reic0024/
  "if i don't stay true to live and hate, how do i differentiate
              between chasing cream and chasing dreams"  :: atmosphere



More information about the Squeak-dev mailing list