Re: Ada.Containers Hash function for a set of small integers

[Simon Wright <simon@pushface.org>]

Robert A Duff <bobduff@shell01.TheWorld.com> writes:

> Simon Wright <simon@pushface.org> writes:
>
>>       type M is mod 2**31;
>>       Result : M := 0;
>>    begin
>>       Result := Result xor Integer'Pos (I.A);
>
>> and, before anyone rushes to check it out, yes, it fails with
>> negative values and I'm about to file a bug against it.
>
> I don't think that's a compiler bug.  Conversion of negative numbers
> to modular raises Constraint_Error.  I think that's a language design
> flaw, but not everyone agrees with me.  The 'Mod attribute works
> around the problem.

No, I meant a bug against my code generator!

On the other hand, I found some strange behaviour with GNAT GPL 2009
(and GCC 4.5.0) while investigating possible solutions:

------------------------------------------------------------------------
with Ada.Text_IO; use Ada.Text_IO;

procedure Odd_Warning is

   procedure Inner (X : Integer) is
      type M is mod 2**31;
      Result : M;
   begin

      begin
         Result := 0;
         Result := Result xor (Integer'Pos (X) mod 2**30);
         --  No compiler warning, but raises exception
      exception
         when others => Put_Line ("exception a");
      end;

      begin
         Result := 0;
         Result := Result xor (Integer'Pos (X) mod 2**31);
         --  odd_warning.adb:20:53: warning: mod with zero divisor
         --  odd_warning.adb:20:53: warning: "Constraint_Error" will be
         --    raised at run time
      exception
         when others => Put_Line ("exception b");
      end;

      begin
         Result := 0;
         Result := Result xor M (Integer'Pos (X) mod 2**31);
         --  No warning (and executes as I had expected)
         Put_Line ("result:" & M'Image (Result));
      exception
         when others => Put_Line ("exception c");
      end;

   end Inner;

begin

   Inner (-1);

end Odd_Warning;
------------------------------------------------------------------------

The second block's compiler warning is strange, especially the part
about the zero divisor. The Aonix ObjectAda 7.2 compiler gives a very
similar warning referring to 95LRM4.5.5(22), and gives an error at the
third block (95LRM4.9(35)). 

I think my route will be via unchecked conversion, if I don't decide
that it's a feature - in 10 years of use no one has needed to use a
negative value as a component of a hash.