|
That's right, there's no bug. My
mistake.
----- Original Message -----
Sent: Tuesday, February 25, 2003 11:37
AM
Subject: Re: Bug in Evaluate(), wrong
arguments in call to LoneKing()
-----BEGIN PGP SIGNED MESSAGE-----
Hash:
SHA1
Jacques_Ruiz wrote:
> Hello.
> I've spotted a bug in
Evaluate(), file eval.c. A call with wrong
> arguments is made to
LoneKing(), passing the same side as both the
> winner and the
looser.
>
> source:
> //* One side
has a lone king and other side has no pawns *//
> *if*
(board.material[xside] == 0 && board.b[side][pawn]
==
NULLBITBOARD)
> *return*
LoneKing (side, xside);
> *if* (board.material[side]
== 0 && board.b[xside][pawn]
==
NULLBITBOARD)
> *return*
LoneKing (side, side); /* BUG HERE */
>
> correct
version:
> //* One side has a lone king and
other side has no pawns *//
> *if*
(board.material[xside] == 0 && board.b[side][pawn]
==
NULLBITBOARD)
> *return*
LoneKing (side, xside);
> *if* (board.material[side]
== 0 && board.b[xside][pawn]
==
NULLBITBOARD)
> *return*
LoneKing (xside, side); /* FIXED */
The arguments to LoneKing
appear to me to be;
1) Side to move
2) Loser
The called
routines, LoneKing and the King Bishop and Knight
mating routine all assume
the winner (or winer <sic>) is the
opposite of the loser
(1^Loser).
If you still think it is wrong can you submit a position
the
modified version wins at that the original doesn't, or
more
explanation.
Thanks for your interest, bugs are in there
somewhere.
Simon
-----BEGIN PGP
SIGNATURE-----
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQE+W0dSGFXfHI9FVgYRAmPTAJ9bHsRoi5j21xUxoPwon+6nnA4TvACdGgYi
PP7y6Bvu4BTnfCxcbIPRL28=
=oJMW
-----END
PGP SIGNATURE-----
|