[Freeciv-Dev] Re: (PR#9596) Better nation selection

["Mateusz Stefek" <mstefek@xxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=9596 >

On 2004-08-03 22:41:51, Per I. Mathisen wrote:
> 
> <URL: http://rt.freeciv.org/Ticket/Display.html?id=9596 >
> 
> On Tue, 3 Aug 2004, Jason Short wrote:
> > Mateusz Stefek wrote:
> > > About a year ago I wrote a patch which made AI select nation from
> the
> > > same class as user. (PR#3751)
> > >
> > > Unfortunately it doesn't work as I expected. When I choose
> Persians AI
> > > picks Boers just because these nations are in the same class
> > > (historical).
> > > If AI knew something about the history of the real world it could
> > > probably choose more inteligently.
> > > Attached patch adds two new fields into the nation struct:
> > > founding_year and collapse_year. When AI chooses nation it tries
> to
> > > minimalize the sum of "time distance" between nations.
> >
> > Is this overkill? Feature creep?

Since common_class() doesn't work well I thought over this problem and  
discovered that founding/collapse year is much better tool than the  
knowledge which nations are still independent and alive (This is in  
fact the only information stored in class)

 
> It is an interesting idea, but I don't like the way it is done. The
> implementation is tied too closely to the present ruleset and its
> assumptions. Instead, perhaps we should offer some restricted nation
> lists
> that can be used to attain consistency.

What does present ruleset assume that makes my implementation  
inflexible?

> This can be done by for example having a new file
> data/ancient/nation.ruleset, then doing "/rulesetdir ancient". This
> will
> load nations from that file, and everything else from default.

If there was some gui for that it could probably be a solution, but / 
set rulesetdir is too complicated for most users and looks like another  
unused feature.
--
mateusz



>>> http://link.interia.pl/f181f <<<