Re: A gnuchess 5.04 bug

[Simon Waters]

Solitude wrote:
> 
> I'm writing because I've hit upon a bug that causes gnuchess 5.04 to
> segfault. I encounter the problem when I run gnuchess interactively and I
> enter the commands 'test movegenspeed' or 'test capturespeed'.
> 
> I have found that the problem is two statements in the TestCmd function in
> cmd.c, lines 786, 788.

Hi 

thanks for the bug report and patch.

The 5.05 test release already incorporates patches to command
line handling to avoid many known segmentation fault problems of
this type, including the one you describe.

Since the only problem reported so far relates to format
modifiers being incorrectly signed or unsigned, and this is only
a problem on third party compilers I am not expecting to make
changes before releasing 5.05 at this point.

 Simon

Subject: 
           GNU Chess 5.05 ready for testing
       Date: 
           Sun, 08 Sep 2002 12:46:59 +0100
       From: 
           Simon Waters <address@hidden>
 Organization: 
           Eighth Layer Limited
  Newsgroups: 
           gnu.chess, rec.games.chess.computer




GNU Chess 5.05 is ready to test.

Assuming no major problems are found these files will form the
basis of the GNU Chess 5.05 release.

http://www.wretched.demon.co.uk/505/

cksum *
523296072 3550367 book.dat
3342515064 314537 gnuches5.exe
3220320685 18270 gnuches5.txt
2951535090 165321 gnuchess-5.05.tar.gz
4258714437 140232 gnuchess_linux_gcc3.1_noreadline

5.05 is a bug fix release.

 Pin and hung piece detection is reintroduced.
 Move sorting is improved.
 Errors handling unusually long EPD format positions have been
fixed.
 Command line input is more robust.
 Program will now look for book.dat in a selection of places
allowing centralised book.dat on *nix systems.
 Errors identified by the VMS C compiler corrected - GNU Chess
should build relatively painlessly on VMS.
 Many errors identified by splint have been corrected.
 Default book building behaviour stops many moves earlier,
producing smaller book.

This version should play better chess, and be more robust than
all previous GNU Chess 5 releases.

Known issues; 
 white/black commands do not conform to protocol specification.
 Inadequate sanity checking on size of hashtables.
 Program 'worries' too much about having Knights pinned by
Bishops.


Windows Users

The gnuches5.exe file is intended for Winboard users. It relies
on Cygwin functionality and so will not work on systems without
relevant libraries, Winboard includes these libraries.

Winboard users should backup existing Winboard files, and save;
 gnuches5.exe
 book.dat
 gnuches5.txt
over the existing files (in C:\Program Files\Winboard by
default).

gnuches5.txt is a DOS text file version of the current README,
to overwrite the existing file. The new file has DOS EOL and EOF
markers so will display correctly in Notepad, unlike the old
one. 

If you have only installed Winboard 4.26, this upgrade includes
"Book" and "Hint" functionality over gnuchess 5.02+ version
shipped with Winboard.

The gnuches5.exe executable was tested with Winboard 4.26, on
Windows 95.


GNU/Linux Users

An updated executable "gnuchess_linux_gcc3.1_noreadline" is
supplied for those who don't feel like installing from source.


Others

Efforts have been made to ensure this program can be easily
compiled under some third party compilers. It in noted that for
some CPUs GCC 3.1 produces substantially faster code than GCC 2
series compilers from this source. Normal GNU build procedure;
"./configure ; make ; make install" should apply.


Care has been taken to ensure executable files are virus free,
use at your own risk.

Please report any problems with GNU Chess 5 to
address@hidden
GNU Chess 4 and earlier are no longer supported.

Thanks to everyone who contributed patches, bug reports, and
ideas. 

 Simon