Conrad Schneiker <schneik / austin.ibm.com> writes:

> But I still have not seen convincing arguments that some form of
> constant/frozen class (aka static typing) couldn't be a worthwhile
> extension of Ruby.

The most compelling argument to me is "is it ain't broke, don't fix
it."

In this context, it means that someone needs to produce real evidence
that:

  1. there is a problem
  2. static typing fixes it
  3. static typing is meaningful within the context of Ruby
  4. static typing can be implemented in Ruby

I feel that (3) is probably one of the more difficult points to
prove. Assuming a type is represented by a class name (a dubious
assumption but I haven't seen better here), then for a static type to 
be meaningful, that class and all that class' ancestors must be
frozen (otherwise you could change the semantics of an object at run
time).

If a class is frozen, then singleton classes based off that class
become a problem. For example,

    a = "Hello"
    def a.+(b)
       "walter"
    end

What is the type of 'a' above? Clearly it isn't 'String' anymore.

But then you get a logical problem. Global methods in Ruby extend a
particular singleton instance of Object. Under static typing, Object
must be frozen (because it is a ancestor of all classes). Thus the
object under which you define these global functions can no longer be
an instance of Object (because adding functions to it changes its
semantic, just as adding + to 'a' stops 'a' being a String).

This leads you to a problem. The global object can no longer be an
instance of the frozen class Object. But all objects in Ruby must be
descendents of Object. Dizzy.....

I don't think that adding static types to Ruby is as obvious as some
people are making out.

Again, what exactly _are_ people trying to fix anyway?


Dave