Re: [Bug-gnubg] gnubg defends poorly against outer primes

[Jim Segrave]

On Mon 24 Nov 2003 (18:07 +0100), Jim Segrave wrote:
> On Mon 24 Nov 2003 (10:27 +0000), Joern Thyssen wrote:
> > On Sun, Nov 23, 2003 at 08:01:24PM +0100, Misja Alma wrote
> > > 
> > > As far as I can
> > > see the only solution would be to implement an export to Snowie
> > > format? 
> > 
> > It's probably not the only solution. Snowie support it's own extension
> > to the .mat format, where illegal moves or incomplete games can be set
> > up using a specialised string identical to the Snowie .txt position
> > files.
> 

Hmm - my X paste buffer did something very, very strange.

> > It's probably not the only solution. Snowie support it's own
>  extension
> > to the .mat format, where illegal moves or incomplete games can be
>  set
> > up using a specialised string identical to the Snowie .txt position
> > files.
> 
> do we have that documented? Is it similar to the one Oystein raised
> last night? Because if it is, then exporting rollouts in that format
> should be easy - the export of rollout .sgf files is pretty
> straightforward - there are four functions in rollout.c:
> 
> void log_game_start (char *name, const cubeinfo *pci, int fCubeful,
>                      int anBoard[2][25]) 
> 
>    called with the file template (name), the cube and board at the
>    start of the rollout and a cubeful/not flag. This generates the
>    .sgf header but could generate the beginning of a .mat file with
>    the position setup.
> 
> void board_to_sgf (int anBoard[25], int direction)
> 
>    The bit of log_game_start which generates the current position
> 
> void log_move (int *anMove, int player, int die0, int die1) 
> 
>    generates one player's move
> 
> void log_cube (char *action, int player) 
> 
>    generates one player's take/double/drop 
> 
> I think with a clear spec of how the position is set up, the rest is
> very easy to do.
> 
> What it looked like from Oystein's posted match:
> 
> There's a 32 field line. Each field is a number, negative numbers are
> player 1. The first 6 fields are unknown in
> use, but are probably cube and chequers off for each player. I think
> (but haven't seen one yet) that field[6] is a count of chequers on the
> bar for player 0 (0..15). fields 7..30 appear to be the points from
> player 0's view (field[7] = player 0's ace point. The value is the
> number of chequers on that point, negative if they are player
> 1's. Field[31] is the number of chequers on the bar for player 1 (0
> .. -15)
> 
> So the starting position would be:
> 
> f 0..5      0  0  0  0  0  0     # all unknown at this time
>      6      0                    # no player 0 off
>   7..12    -2  0  0  0  0  5     # player 0 home board
>  13..18     0  3  0  0  0 -5     # player 0 outer board
>  19..24     5  0  0  0 -3  0     # player 1 outer board
>  25..30    -5  0  0  0  0  2     # player 1 home board
>      31     0                    # no players off
> 
> But in the file all one line:
> 
> 0 0 0 0 0 0 0 -2 0 0 0 0 5 0 3 0 0 0 -5 5 0 0 0 -3 0 -5 0 0 0 0 2 0
>   
> 
> All we need is to see how the beginning values look and check that
> player 0 off is in field[6].
> 
> Arbitraty but recognisable chequer placements and the following:
> 
> a match to 9, score 6-5 cube on 2 owned by player 0, player 0  one man on bar.
> a match to 7, score 4-3 cube on 4, owned by player 1, player 0 has 1 man off
> a match to 5, score 2-1 cube centred, player 1 has 2 men off
> a match to 3, Crawford-0 
> money game, Jacoby
> money game, no Jacoby (if allowed)
> 
> > Unfortunately, gnubg can only import such .mat files, but not export
> > them. 
> 
> 
> -- 
> Jim Segrave           address@hidden
> 
> 
> 
> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-gnubg

-- 
Jim Segrave           address@hidden