[ruby-talk:20812] Re: Proc internals

In article <[e-mail]>, "Brian F.
Feldman" <[e-mail]> wrote:

> They're not exactly interchangeable...  There may be an answer depending
> on what exactly it is that you want to do.

The post from Henning von Rosen (`how update anonymous classes?'') gave
me an excuse for investigating the internals of Ruby.

I imagined a new instance method on Class:

	Class#add_method( methodName, &methodProc )

....which would add the supplied Proc/Block and add it as a method to the
particular instance of class. So far I have:

######################################

#include "ruby.h"

VALUE
rb_class_add_method(VALUE obj, VALUE methodName, VALUE methodProc)
{
    int argc;
    ID id;

    Check_Type(methodName, T_STRING);

    if (!rb_obj_is_kind_of(methodProc, rb_cProc))
	rb_raise(rb_eArgError, "Must pass a Proc");
    
    argc = FIX2INT(rb_funcall(methodProc, rb_intern("arity"), 0));
    
    id = rb_intern(STR2CSTR(methodName));
    
   /* WHAT TO PUT HERE? */

    return Qtrue;
}

void
Init_add_method()
{
    rb_define_method(rb_cClass, "add_method", rb_class_add_method, 2);
}

######################################

First I thought of rb_define_method( class, name, C-function, arity ),
but realizing that it expects a C funtion I gave up that idea.
Then I had a look on rb_add_method which expects a NODE instead. If there
is any way to get from a Proc to a NODE I figured I could use this.

Any comments are greatly appreciated.

Please excuse my lack of insight in the internals of Ruby.

/Christian