Re: [Caml-list] Differentiating lists from arrays with Obj

[Remi Vanicat <remi.vanicat@laposte.net>]

John Goerzen <jgoerzen@complete.org> writes:

> Hello,
>
> I am looking at using the Obj to access internal representations of some
> types.

It is generally a bad idea. Why do you need this ?

>
> I have figured out how to access those types for ints, doubles (floats),
> strings, and arrays of doubles (why those are special, I have no
> idea!).

for optimization reason, and for interface with C.

> I can also figure out how to access lists and arrays, for which Obj.tag
> returns zero.  (Although that value does not occur in obj.ml, it is in
> some of the header files.)
>
> Now, here's the question: how can I accurately differentiate lists from
> arrays?

You can't. Type information are lost in the midst of the
compilation. You cannot recover them.

>
> Lists appear to be internally represented by structures that look
> somewhat like this:
>
> [1; [2; [3; 0]]]

Yes.

>
> That is, Obj.size will return 2 for the list [1; 2; 3] because it is a
> collection of an int and a list.  That's fine, but its internal
> representation of the list [1] is:
>
> [1; 0]

Exactly

>
> Which happens to be identical to the representation of the array
> [| 1; 0 |].  Eep. 

And of several other thins (like the couple (1,0), or the couple
(true, false), or the couple (1, None), 

>  Now, it's not a big deal to me whether I've got a list or an array,
> but I need to be able to know whether that list or array has an
> extra 0 at the end.

You can't.

>
> Worse, the internal representation of [] appears to be identical to the
> internal representation of the integer 0.

And it the same for false, None, and a lot of others....

> I am a little puzzled here, and I do not know the type of the data I'm
> dealing with in advance.

You should read the "Interfacing ocaml with C" chapter of the manual
language, it explain the internal representation of most ocaml
value. But there is no miracle : what you try to do is not doable.
-- 
Rémi Vanicat

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners