Could someone check the appended patch in ?
It fixes a typo in eval.c 1.294 that led to strange results when a player
has 2 or more men on the bar. See for instance this post in
rec.games.backgammon :
Date: Wed, 25 Oct 2006 11:23:49 -0400
From: Chris Roddy <[EMAIL PROTECTED]>
Subject: Re: Is this "blind spot" in GNUBG ? (another)
Message-ID: <[EMAIL PROTECTED]>
Regards,
Philippe Michel
*** eval.c.dist Sun Oct 1 12:41:46 2006
--- eval.c Sat Oct 28 15:26:35 2006
***************
*** 1815,1821 ****
{
int nc = board[ 24 ];
! afInput[ 24 * 4 + 0 ] = (nc == 1) ? 1.0f : 0.0f;
afInput[ 24 * 4 + 1 ] = (nc >= 2) ? 1.0f : 0.0f; /* is this
correct? */
afInput[ 24 * 4 + 2 ] = (nc >= 3) ? 1.0f : 0.0f;
afInput[ 24 * 4 + 3 ] = nc > 3 ? ( nc - 3 ) / 2.0f : 0.0f;
--- 1815,1821 ----
{
int nc = board[ 24 ];
! afInput[ 24 * 4 + 0 ] = (nc >= 1) ? 1.0f : 0.0f;
afInput[ 24 * 4 + 1 ] = (nc >= 2) ? 1.0f : 0.0f; /* is this
correct? */
afInput[ 24 * 4 + 2 ] = (nc >= 3) ? 1.0f : 0.0f;
afInput[ 24 * 4 + 3 ] = nc > 3 ? ( nc - 3 ) / 2.0f : 0.0f;
_______________________________________________
Bug-gnubg mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-gnubg
