🚀 go-pugleaf

Two days ago, I wrote as a follow-up in the thread "Re: Problem
with gnuchess and Unixware":

>> [Someone having the problem that gnuchess thinks forever
>> after a few moves.]
>
> I am having similar problems:
>
> - gnuchess 4.0pl62
> - gcc 2.2.2
> - OS: NeXTSTEP 3.1
> 
> The program plays a few moves and then starts thinking forever (I have let
> it think for 8 hours on a move when time controls where 60 5). Running
> "make test" hangs in the same manner.
> 
> [...]
> I am about to recompile with -g so I can get more info, but it
> seems to hang in the function ataks()...
> 

Well, it's weird, but when compiled with CFLAGS= -g, the
program runs just fine. Tests reveal that it's the optimization
flags that cause problems. When either -O or -O2 is turned on,
the program seems to get stuck one of the do loops of the form

do
{
	u = /* some expresion; */
	/* more code here */
} while (u != sq);

where u will periodically cycle through a bunch of values, none of which
are equal to sq. (For example, the do loop in ataks.c, line 54, or the one
in eval.c, line 727). It may be that there is a bug in the NeXT provided
version of gcc. It's a bit frustrating not to be able to compile with any
form of optimization. But at least I can get a working version now.

Anyone have any ideas as to what might be the cause of the problem? I'm not
familiar at all with the gnuchess code...

Thanks,
--
Pascal "The Rascal" Gaudette
rascal@cam.org