From: "Aleksi Niemelä" <aleksi.niemela@...>
Date: 2000-12-16T05:51:38+09:00
Subject: [ruby-talk:7347] RE:  Different id's for the Object meta class?

> In a Ruby extension if I execute the following three lines I get
> three different values for the identifier of  the Object:

I know nothing about these things, so I'm just guessing. Beware.

>      rb_eval_string("print 'Object.id:  ', Object.id");
> Object.id:  5919960

Here you pass a string to Ruby C-api, to evaluate (compile & execute) it.
Thus the code in Ruby prints the id in Ruby world for the object which
constant Object refers to.

>      printf("\nrb_cObject:  %d", rb_cObject);
> rb_cObject:  58992127

Here you print right away on your own the address where rb_cObject pointer
(variable) points to, through a normal C-idiom. I'd expect it points to
somekind of struct representing Ruby object which in turn has a placeholder
for the id attribute. 

Now we're completely on the C world (assuming the code was in C and not in
Ruby :).

>      printf("\ncObject:  %d", rb_eval_string("Object"));
> Object:  11839920

Here you first ask Ruby to evaluate (compile&execute) string "Object", which
in turn returns you an Ruby object (C pointer pointing to struct
representing Ruby object). Then you just print the (C-) address of that
returned object (converting from pointer magically to integer in C way).

> What am I misunderstanding?

I guess nothing, except if you assume the all the above cases to print same
thing.

I'm sure people will correct me here, if I was guessing wrong.

	- Aleksi