Short is the basic built-in integer data type.  Values may range between
-2**(15) and 2**(15)-1.  All operations are performed mod 2**16.
(On the Sum implementation operations are performed modulo 2**32.)  No
overflow is reported (unless the appropriate hardware trap is enabled).
(The Long type provides unlimited precision arithmetic).

    Short: type S {
      (* constants *)
	'0' :   func [] val S;
	'1' :   func [] val S;
	'2' :   func [] val S;
	'3' :   func [] val S;
	'4' :   func [] val S;
	'5' :   func [] val S;
	'6' :   func [] val S;
	'7' :   func [] val S;
	'8' :   func [] val S;
	'9' :   func [] val S;
      (* operations *)
        % :     func [x,y:val S] val S;  (* remainder *)
        * :     func [x,y:val S] val S;
        ** :    func [x,y:val S] val S;  (* exponentiation *)
        + :     func [x,y:val S] val S;
        - :     func [val S] val S;
        - :     func [x,y:val S] val S;
        / :     func [x,y:val S] val S;
        (* assignment operations *)
          := :    func [var S; val S] val S;
          += :    func [var S; val S] val S;
          -= :    func [var S; val S] val S;
        < :     func [x,y:val S] val Boolean;
        <= :    func [x,y:val S] val Boolean;
        <> :    func [x,y:val S] val Boolean;
        = :     func [x,y:val S] val Boolean;
        > :     func [x,y:val S] val Boolean;
        >= :    func [x,y:val S] val Boolean;
        New:    func [] var S;
	New:    func [val S] var S;   (* initializes to arg value *)
        V:      func [var S] val S;
        ^+ :    func [x,y: val S] val S;
        get :   func [var Void] val S;
        put :   func [val S] val S;
        puts:   func [val S] val ChStr;
        shift : func [what, how_much: val S] val S;
    };

The expression x += 1 is equivalent to x := x + 1.  Inequality is represented
as <>.  The ^+ operation multiplies its left argument by 10 and adds
its right argument to the result.  Division truncates toward 0.
The sign of the remainder is that of the dividend.  (The authors know that
this is the wrong thing to do.  We're just waiting for the hardware designers
to catch on.)

The put functions writes a short integer onto the standard output file.
The puts function converts a short integer to a printable representation.
The "get" function reads one from the standard input.  The application
shift[x,n] returns the largest integer which is no greater than x*(2**n),
modulo 2**16 (2**32 on a Sun).  The second argument to shift should be
between -16 and 16 (-32 and 32 on a Sun).

Implementation note:  Values are efficiently implemented.  Variables leave
a little to be desired, but aren't too bad.  Representation for values is
compatible with C.

See also: long, strings.