[plt-scheme] Native code generation and immutable pairs

On 2/12/06, Noel Welsh <noelwelsh at yahoo.com> wrote:
> The primary use in my experience is to simulate a mutable
> binding.  Module bindings can only be mutated by the
> defining module.  Sometimes it is useful to allow others to
> mutate a binding, in which case storing the value in a box
> is a solution.  However, it is almost always the wrong
> solution -- parameters are better for most use cases, and
> monads for pretty much all other cases I can think of,
> leaving boxes for the cases where you want inexplicable
> behaviour in the presence of concurrency and continuations.

How do you implement unification efficiently?  That little
inverse-Ackermann thing for unifying logic variables with each other
is awful nice.