We have a program that is spending a lot of time in set operations,
and we're thinking of trying an imperative implementation based on bit vectors.
I would hope that the basis of such an implementation would be an array
of native integers, but on scrutinizing the manual (espeically the chapter
on interfacing to C), I have concluded that such a thing is not possible.
Our choices appear to be

  * An array of native integers, which will be implemented as an array
    of pointers to native integers, because integers are boxed.

  * An array of tagged integers, which will be less efficient as
    dividing by 31 is more expensive than shifting.

How would the gurus recommend that we proceed?  Is there a better,
still efficient data structure for a set of small integers?

Could the compiler gods be persuaded to provide unboxed
representations for arrays of untagged integers, as is already done
for `float array'?


Norman

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