Geert Fannes
6/8/2005 12:34:00 PM
Hello,
for performance reasons, I have to add C-implemented methods to my
classes. This works fine, but I wonder if there exist more elegant
techniques for extending inherited classes. The code hereunder indicates
what I mean. The ruby file defines the class A and method A#f (which
writes out "A#f"), and the class B, which inherits from A, but B#f is
not defined. My goal is to define B#f in a C-file. My question has to do
with how I can get the class variable cB in an elegant way.
******the Ruby file******
#this is the base class
class A
def f() puts('A#f') end
end
#B inherits from A and I want to write a C-method for B#f
class B < A
require('f.so')
end
B.new.f
******the C file******
#include "ruby.h"
VALUE f0(VALUE self){printf("B#f");return self;}
VALUE cA,cB;
void Init_f(){
//correct: results into writing out "B#f"
cA=rb_define_class("A",rb_cObject);
cB=rb_define_class("B",cA);
//wrong: results into writing out "A#f"
cB=rb_gv_get("B");
rb_define_method(cB,"f",f0,0);}
The two lines after the "correct" comment gives the correct result and
writes out "B#f", but when I use the single line after the "wrong"
comment, the method A#f is called when ruby executes B.new.f(), which
writes out "A#f". In this problem, it is still feasible to follow the
whole inheritance path to get the correct cB value since B only inherits
from A, but this becomes more of a problem when using more complex
inheritance structures. Even more, I guess it should not matter if I
modify something to the inheritance structure (unless my C-function uses
some of the inherited classes), so this is some sort of
inheritance-structure duplication that leads to maintenance problems. Is
there a more elegant method to get the class variable?
Greetings,
Geert Fannes.