new SML/NJ

[Matthew Fluet]

> Well damn.  I knew that performance would suck.  It was just the overloading
> I wanted.  While I'm developing a program it is nice to use SML/NJ for its
> type checking, and with the bogus overloading that doesn't work.  I have gotten
> around this so far by just using
>         structure IntInf = Int
> at the start, but have sometimes found that that works but the program is
> still wrong because some where I called some function explicitly from Int.
> Oh well.

I suspect this is covered by the "What's next?" point:

3.Changes to support the single toplevel declaration: 
            structure LargeInt = IntInf

> Did you do any timing test on the new SML/NJ?  Did ML-RISC speed the Intel
> code up much?

No, I haven't really looked into that.