🚀 go-pugleaf

> There were no errors during compilation on a Decstation 5000,
> but after a few moves, which I assume are the book moves,
> the program hangs indefinitely.
> 
> Why is that?

Gnuchess works fine for me when I compile and run it on my DECstation
5000.  I wish I knew why you (and a couple of people running on other
machines) are having this problem!  What patchlevel are you using?
(Always include the version number and patchlevel in bug reports.)

A couple of things you can try:  

(1) Double check your Makefile and be sure you have uncommented the
right lines to specify the right options for the C compiler you are
using.

(2) Run gnuchessr and give the "post" command when you start it.  This
causes it to print out some (unfortunately cryptic) information about
how the search is going, including how much time it plans to think for
and how much time it believes it has used so far.  Save what is
printed to a file and include it with your next bug report.  This may
help in figuring out why gnuchess is thinking forever.

(3) Gnuchess pl61 has a known bug that causes an out-of-range array
reference.  Mike McGann has fixed this in the pl62 he is working on,
and the fix has been posted here.  I can imagine the bug might cause
this hanging problem, although the symptoms that have been reported
for it in the past are different.  I'll include the fix again below.

> Also, is it possible to include the executable files with the
> distribution of gnuchess? I mean, someone who knows better than
> me compile the programs and then encode them, so that people
> could ftp those instead of the source. Or is that not possible?

That isn't practical.  We would need to make many different
executables, one for each of the many different kinds of machines
people are using---Sun-3, Sun SPARC, DECstation, VAX, DEC Alpha, HP,
IBM, etc., etc.  And we don't have access to most of these kinds of
machines anyway.  It would be too big a job to collect and distribute
executables for all the machines from people who have them.

> And on a different note, about gnuchess' strength. Is it possible
> to estimate the program's strength by having it guess the moves
> from Pandolfini's `solitaire chess' column in the `Chess Life'
> magazine? How accurate would such an estimation be?

It would be not be accurate.  Pandolfini's columns are just for fun;
they aren't a scientific way of estimating your rating.  It would take
a lot of work to make that kind of test accurate---you would have to
give it to a large sample of chess players with known ratings, and
then adjust the question weights until the test scores gave a good
estimate of their ratings.  There is no way Pandolfini could be doing
that on a monthly column.

	--Tim

----- Forwarded message -----
Return-Path: mwm@hasler.ascom.ch
Received: by src-mail.pa.dec.com; id AA10763; Mon, 1 Mar 93 00:35:39 -0800
Received: by inet-gw-2.pa.dec.com; id AA03579; Mon, 1 Mar 93 00:35:35 -0800
Received: by hslrswi.hasler.ascom.ch (5.65c/6.32)
	id AA09974; Mon, 1 Mar 1993 09:33:00 +0100 
Date: Mon, 1 Mar 1993 09:33:00 +0100
From: Mike McGann  <mwm@hasler.ascom.ch>
Message-Id: <199303010833.AA09974@hslrswi.hasler.ascom.ch>
To: mann
Subject: p61


the fix for the castle illegal move problem is in
init.c about line 299 it has:

compptr = oppptr = -1;

it should be:

compptr = oppptr = 0;

mike
----- End forwarded message -----