Re: [Bug-gnubg] 0.15 luck based FIBS rating diff bug ?
From:
Massimiliano Maini
Subject:
Re: [Bug-gnubg] 0.15 luck based FIBS rating diff bug ?
Date:
Mon, 12 Mar 2007 17:00:29 +0100
"Christian Anthon" <address@hidden>
wrote on 12/03/2007 16:13:06:
> try to edit analysis.c so that the function relativeFibsrating
> matches the current cvs version.
analysis.c, lines 1501-1502 in 0.16 branch (MAIN): ================================================== float const x = - 2000.0f / (float) sqrt ( 1.0 * n
) * (float) log10 ( 1.0 / r - 1.0 );
analysis.c, line 1423 in 0.15 branch: ===================================== float const x = - 2000.0f / (float)(sqrt ( 1.0 * n
) * log10 ( 1.0 / r - 1.0 ));
The (float) cast before log10 is missing. Putting
it back solves the problem (verified on my example).
Can somebody with CVS access commit the cange in 0.15
branch ?
Christian/Achim, can you take a CVS snapshot of the
0.15 branch and put it somewhere (on www.gnubg.org/media/sources, there's
already a gnubg-source-rel-0_15_20061120.tar.gz ) and let me
know so that I can rebuild 0.15 after the fix ?