Team PLClub ICFP entry -- comparing the performance of OCAML and SML

[Suresh Jagannathan]

   > I guess you were trying to say that MLton does a better uncurrying
   > optimization and supports "reference cell" more efficiently than SML/NJ.

   I was not.  In fact, MLton does not at present have an explicit uncurrying
   optimization, although the combination of some of its other optimizations
   has the effect of some uncurrying.  There are some known cases where MLton has a 
   performance penalty for using curried functions.  The MLton developers are
   trying to eliminate them.

Have you had a chance to incorporate the uncurrier I sent you into the
latest release.  I'd be curious if it leads to any measurable improvement.

       -- Suresh