Re: gnomines possible improvement

["Richard Hoelscher" <rah rahga com>]

spdepagn ncsu edu said:
> Do you have this solver?  I think it would be possible to have a fast
> solving method if binary logic gates are used.  For example, square 10 has
> a mine if  A OR (B AND C)  and square 11 has a mine if B AND C etc.. then
> it should be possible to quickly try every possibility for these variables
> and find a match (or ambiguous case)

http://mail.gnome.org/archives/games-list/2004-December/msg00001.html

<snip>
> There are numerous possible ways to implement this option.  One way is try
> to solve the puzzle using an algorithm that won't guess.  Then if the user
> truly guessed (from their point of view) and it is possible to solve that
> square by making other moves first, the mines can be reorganized to punish
> the user for this, and cause them to lose.  If there is no way to solve
> this square, and the user guessed, then the mines should be rearranged,
> and the user would be permitted to guess in this situation.
>
> Another possibility is to keep ahead of the user by enough moves so that
> the computer solves and rearranges at each click, so there is never an
> ambiguous situation revealed.

I'll look at your code later today, but I'm only seeing one effective way
to do this... After the first click, randomize the field. Loop through
solving and randomizing until a minefield comes up that is solvable
without guessing.

Then, for the rest of the game, don't do any more solving or randomizing.
It has been solved once, and the feature has done its job, because the
only way the player can lose is if they guess.

> Can we vote on the best implementation?  How can I join the list?  Thanks
> for all the replies.

Well, we can discuss implementation. ;)
http://mail.gnome.org/mailman/listinfo/games-list to subscribe.

-Richard Hoelscher
http://rahga.com/svg/