David Thomas a écrit :
> Isn't "!(x) -> a | _ -> b" the same as "x -> b | _ ->
> a"?  I don't really see much of a benefit, at the cost
> (I assume) of complexity.

In this case, yes. But you can express interesting (more easier) thinks 
as (examples of TOM)

!f(x,x) -> // matches either something different from f, or an f with x1 
!= x2

f(x,!g(x)) -> {
       // matches an f which has either x2!=g or x2=g(y) with y != x1

You can simulating this using classical pattern matching. So I thinks 
that is possible using camlp4. But, looking for someone you have do this 
work or plan to do this.

FG