Re: Help needed for ada package

["Steve" <nospam_steved94@comcast.net>]

"Dennis Lee Bieber" <wlfraed@ix.netcom.com> wrote in message 
news:uaaje1d0l4tp1kjs18mkrgfmbkcir308bt@4ax.com...
> On Thu, 28 Jul 2005 20:33:32 +0200, Ludovic Brenta
> <ludovic.brenta@tiscali.be> declaimed the following in comp.lang.ada:
>
>
>>
>> Since the type ID is discrete, why not just use that as the array index 
>> type?
>>
> That's the approach I took, though some things aren't quite as
> clean (I used explicit ranges... should have tried compiling with just
> the type).
>> Of course, the above solution is outrageously inefficient, both
>> memory- and CPU-wise.  Optimisations are left as an exercise to the
>> reader.
>
> Tell me about it -- for the 1000000 population, I estimated over
> 100GB needed for the association matrix alone, even packing 1-bit
> booleans. I then did version 2, using "Steve's" suggestion of encoding
> the group identifier into the population data itself... associate() went
> from two lines to a page, but half the declarations disappear.
>
> I will not be posting that version as it is too close to being
> directly usable (I didn't write a test suite, so couldn't run it, but it
> did compile cleanly).
>

Perhaps I'm a little naive, but I really don't understand why you need to 
keep any more than:

  type Persson is
    record
      Citizen_Id : ID;
      Badness    : Rating;
      Group_Id  : Natural;
    end record;

  type Population is array( 1 .. 1_000_000) of Person;

  Society     : Population;
  Next_In    : Positive; -- Add new elements here
  Next_Out : Positive; -- Retrieve next from current group here
  Baddest    : Positive; -- Index to baddest entry
  Current_Group_Id : Natural;

You do wind up scanning the list to do just about anything, but speed was 
not a requirement.

Steve
(The Duck)


[snip]
>
>
> -- 
> > ============================================================== <
> >   wlfraed@ix.netcom.com  | Wulfraed  Dennis Lee Bieber  KD6MOG <
> >      wulfraed@dm.net     |       Bestiaria Support Staff       <
> > ============================================================== <
> >           Home Page: <http://www.dm.net/~wulfraed/>            <
> >        Overflow Page: <http://wlfraed.home.netcom.com/>        <