From: Dave Thomas <Dave@...>
Date: 2001-05-21T14:33:04+09:00
Subject: [ruby-talk:15457] Re: RCR: #const_name_is

Mathieu Bouchard <matju@sympatico.ca> writes:

> If you are looking for more concrete examples, well, there is RubyX11.
> 
> In RubyX11 I have this notion that a type is basically representable as a
> set membership function (this is a more general notion of type than what
> people usually think of). So I have extracted Module#=== and turned it
> into a separate interface called X11::Type (which I have extended with X11
> marshalling specificities). Many type-representing objects are not classes
> (nor modules), and so they don't fit with Ruby's assumption that all
> constants worthy of knowing their own name are modules.

I'm still struggling to understand how this would be useful. You have
objects whose semantics depend on the name of a constant symbol that
they're assigned to. Is it possible to show a couple of lines of code
that illustrate this. I'm intrigued.

Dave