Re: Some Refactoring Notes

[MattyT <mattyt-spam tpg com au>]

On Fri, 2004-01-02 at 03:37, Arvind Narayanan wrote:

> Yes. Connect 4 doesn't fit the minimax paradigm; its a "knowledge based"
> rather than a purely "search based" game (same as Go.)

These are probably not huge design issues in a games library.

To start with there are at least two games with opening books, which
suggests a certain desirable reusability.  If we don't use engines for
them both, opening book functionality could easily go into a support
library.

My original refactoring notes mentioned move lists, this presumes
"moves" are encoded as strings or something similar, and opening books
can be stored in a specific format that delimits the moves, whereas the
actual move format is game specific (Chess might use "E2E4", GNect might
use "E").  This all hopefully means that reusable code for loading and
following opening books is not difficult.

I'm not sure what form the "knowledge based" stuff takes, I had just
assumed it was capable of sometimes determining the minimax doesn't have
to go any further, but perhaps you mean it never has to do minimax at
all.

Doing minimax sometimes and not doing it at other times is probably not
particularly difficult either, it just means rather than using a minimax
framework you'd just do a minimax algorithm and it's up to you whether
to call it.

-- 
         Matthew Tuck: Software Developer & All-Round Nice Guy        
 My Short Autobiography: 1985 Grade Bin Monitor 1990 Class Clown Award
1992 Awarded Most Likely To Spontaneously Combust 1996 Crowned Galactic
         Emperor 1998 Released From Smith Psychiatric Hospital