First RAG 2016 report

Dear Koichi,

I have bunch of questions about Ruby's Garbage Collection. But before I proceed to ask, I would like to update you properly on the current progress that I have accomplished with Rubex.

Previous objectives

As of now I have finished most of the objectives that I had stated in the original proposal that was sent to the Ruby Association. So now, one can declare simple primitive data types, arrays, Ruby methods and even loops in Rubex and all of it is handled gracefully by the Rubex compiler.

As an example, consider this code:

def looper(i32 size)
  i32 a[size], i = 0, j = size, m

  for i <= m < j do
    a[m] = 5*m
  end

  m = 0

  for j > m >= i do
    print a[m]
  end

  i32 f = 0
  while !(a[f] != 45) do
    print a[f]
    f = f + 1
  end

  return a[2]
end

In the above code, I have written a method called looper that accepts an argument of type i32 (32 bit integer). The next line declares an array of i32 called a that if of size size. The subsequent code consists of loops that are interating over this array and performing operations like assignment (first loop), printing (second loop) and printing a part of the array (third while loop). Finally the method returns an Integer located at a[2].

Upon running the Rubex compiler, the above code translates into the following C code:

/* C extension for loops.
This file in generated by Rubex. Do not change!
*/
#include <ruby.h>
#include <stdint.h>

VALUE __rubex_f_looper (int argc, VALUE* argv, VALUE __rubex_arg_self);
VALUE __rubex_f_looper (int argc, VALUE* argv, VALUE __rubex_arg_self)
{
  int32_t __rubex_arg_size;
  int32_t __rubex_v_i;
  int32_t __rubex_v_j;
  int32_t __rubex_v_m;
  int32_t __rubex_v_f;
  if (argc != 1)
  {
    rb_raise(rb_eArgError, "Need 1 args, not %d", argc);
  }
  __rubex_arg_size=(int32_t)NUM2INT(argv[0])  ;
  __rubex_v_i = 0;
  __rubex_v_j = __rubex_arg_size;
  __rubex_v_f = 0;
  int32_t __rubex_arr_a[__rubex_arg_size];
  for (__rubex_v_m = __rubex_v_i; __rubex_v_m < __rubex_v_j; __rubex_v_m++)
  {
    __rubex_arr_a[__rubex_v_m] = ( 5 * __rubex_v_m );
  }
  __rubex_v_m = 0;
  for (__rubex_v_m = __rubex_v_j - 1; __rubex_v_m >= __rubex_v_i; __rubex_v_m--)
  {
    printf("%d", __rubex_arr_a[__rubex_v_m]);
  }
  while (! ( __rubex_arr_a[__rubex_v_f] != 45 ))
  {
    printf("%d", __rubex_arr_a[__rubex_v_f]);
    __rubex_v_f = ( __rubex_v_f + 1 );
  }
  return   INT2NUM(__rubex_arr_a[2]);
}

void Init_loops (void);
void Init_loops (void)
{
  rb_define_method(rb_cObject ,"looper", __rubex_f_looper, -1);
}

And it also generates the following the extconf.rb file:

require 'mkmf'
create_makefile('loops/loops')

If you copy the above generated C code into a C file and the extconf into a Ruby file, you can run ruby extconf.rb; make and it will compile the C file into a shared object file called loops.so. This code can then be used in the following manner in any Ruby script:

require_relative 'loops.so'

puts loopers(10)
# => 
#10
#454035302520151050%

v0dro/weekly_report1.md

Previous objectives