Thanks for the well-presented argument.  One of the points you made is
that functions with inferred general recursive types are often
incorrect.  I agree, and I am willing to make sacrifices to tie down
type inference (although it seems a little extreme to remove the
offending constructions).

By that way, thanks for 1.06 release.  OCaml 1.05 has been solid and
reliable, and you have added many useful features to 1.06.  The language
and its type system have been clearly thought out.  Nice!

Xavier Leroy wrote:
> 
> Here is the straight dope...

-- 
Jason Hickey                    Email: jyh@cs.cornell.edu
Department of Computer Science  Tel: +1 (607) 255-4394
Cornell University