On Fri, 2004-07-16 at 07:17, skaller wrote:

I should rephrase this:

> This is wrong. The representation being suggested is:
> 
> None -> NULL
> Some 'a -> pointer to 'a
> 
> Clearly this represntation is faithful and nothing
> more than an interpretation of the corresponding C 
> concept.
> 
> Some Some 1 is obviously distinct from Some 1 as reqiuired:
> pointer to pointer to 1 is distinct from pointer to 1.
> Some None is a pointer to NULL, which is distinct from 
> None which is just NULL.

What I'm saying is that the idea is not unboxing,
but untagging. Some 'a isn't a pointer to an 'a heap
value, but a pointer to a CAML_VALUE which is a pointer
to a heap value 'a. I don't see how the GC could
support this in Ocaml, since Some 'a isn't a CAML_VALUE
since it points to a box, not a heap object -- 
but in Felix it would work because the GC uses 
RTTI to locate pointers.

So the idea is fine, but won't work with Ocaml GC.
There's be no saving dereferencing a Some value here.
The saving would be that simply loading the value
into a register would set a condition code in the
processor, avoiding any need to check the value
of a tag word.

Ugg .. sorry for the confusion :(

-- 
John Skaller, mailto:skaller@users.sf.net
voice: 061-2-9660-0850, 
snail: PO BOX 401 Glebe NSW 2037 Australia
Checkout the Felix programming language http://felix.sf.net



-------------------
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