[Oberon] RISC emulator and floating-point

[Paul Onyschuk]

On Tue, 25 Mar 2014 20:49:15 +0100
Peter De Wachter <pdewacht at gmail.com> wrote:

> The compiler in Project Oberon (both the original and the current
> edition) permits integers with the sign bit set when using hexadecimal
> notation. From the compiler source code it's clear that this is
> intentional. This is used in a couple of places, for example in the
> FileDir module:
> 
> CONST DirMark*    = 9B1EA38DH;
>       HeaderMark* = 9BA71D86H;

I think I found issue.  Right now in Oberon running under emulator:

  -1 DIV 10 = -1
  -1 MOD 10 = 9

Is that indented behavior?  Now version of procedure WriteInt() with
comments:

  PROCEDURE WriteInt* (VAR W: Writer; x, n: LONGINT);
    VAR i: INTEGER; x0: LONGINT;
      a: ARRAY 10 OF CHAR;
  BEGIN i := 0;
(* Taking this branch, since x = 80000000H = -2147483648 *)
    IF x < 0 THEN
      IF x = 0 THEN WriteString(W, " -2147483648")
(* Negating x results in integer overflow and x0 wraps
                                   around to -2147483648 *)
      ELSE DEC(n); x0 := -x
      END
    ELSE x0 := x
    END;
    REPEAT
      a[i] := CHR(x0 MOD 10 + 30H); x0 := x0 DIV 10; INC(i)
(* Loop never terminates, since -1 DIV 10 = -1 and x0 is never 0 *)
    UNTIL x0 = 0;
    WHILE n > i DO Write(W, " "); DEC(n) END;
    IF x < 0 THEN Write(W, "-") END;
    REPEAT DEC(i); Write(W, a[i]) UNTIL i = 0
  END WriteInt;

-- 
Paul Onyschuk