[MLton] MLton rules! (was: filedes = int)

[Stephen Weeks]

> > I agree that you could re-implement the library using the same representation
> > in the compiler, because the compiler doesn't do any thing that much with
> > big integers, but why bother. 
> 
> Two reasons come to mind: 
> 
> (a) portability: tools dependent on other tools and libraries
> are much harder to port -- especially stuff that uses
> as many C macros and autoconf lunacy as GNU code.
> 
> (b) licence

If these are really any issue, I don't think it would be too hard for
someone (not me :-) to replace our gmp-based IntInf implementation
with an SML one (e.g. by grabbing the SML/NJ code), which would give
you a MLton compiler that doesn't depend on gmp, as well as
executables that don't.  And speed certainly wouldn't matter for
compiles.