Topes - a proposal

[Guy Steele <gls@think.com>]

   Date: Tue, 4 May 1993 17:02:18 -0400
   From: Matthias Blume <blume@cs.Princeton.EDU>


   I just skimmed through the article and found that your ``topes'' are
   (almost) exactly the value-cells as introduced by David Krantz' Orbit
   compiler and similar others (like VSCM II :).  ML facilitates ``ref''
   which seems to do the same job.

   Value cell creation and dereferencing a la ORBIT can be introduced
   automatically by the compiler for all local variables because it can
   easily check the entire scope of that variable for occurences of ``set!''.
   The difference is that value-cells are not directly accessible by the
   programmer.  You can still define ``cons'' in terms of ``lambda'' and
   ``set!'', but it (usually) will be much less efficient.

   ML's ``ref'', however, seem to be precisely the same as your ``topes''.

You can go back even further than that, to Hewitt's actor languages
of the late 1970's, and before that to Algol 68 ref types?

--Guy