[MLton] the next public release

Henry Cejtin henry@sourcelight.com
Mon, 2 Aug 2004 13:54:45 -0500

I  definitely  agree  with Matthew that a `MLton philosophy' section would be
very good.  Not only for things about the compiler itself (strict  SML),  but
also  for  the  coding  style (no currying unless there really is some useful
computation that can be done with only some of the args, arg order  in  fold,
the  fold/foldi  stuff, etc.).  All of these serve the same end: you can look
at some new code, or some new feature, and can guess how it must work  or  be

I  really  don't  like too many things in the shell environment because it is
such an unstructured mess.  If the environment must be  used,  then  I  would
vote  for  MLTON_ in front of all of it (including the SML/NJ locations) just
to  keep  from  confusing  other  programs  we   don't   even   know   about.
Alternatively,  instead  of  an  environment,  we could just have a file some
where which defined all the variables and their values.   That  way  any  one
could  add  more  by just editing the file.  (I think of the environment as a
hack for programs which don't know about the information to carry  it  around
to  programs  that do.  In this case it seems that all the parties know about