🚀 go-pugleaf

> 
> Present condition:
> In the last 4.0 pl 73 version there is another bug in repetition.
> The loop examine one position too far back.
> Others have noted that there is something strange in repetition.
> The strangeness is that the loop will only be entered when there
> is at least 5 iterations, even though a repetition could occur
> after 4 half moves.
> 
> Possible consequences:
> For 5 iterations and less the problem is only some lost cycles but
> for even greater (pun intended) values it might cause the program
> to miscalculate.
> 
> Easy fix:
> Use ">" instead of ">=" in the repetition function in file search.c.

I think you mean use ">=" rather than ">" in the repetition function.
viz. 
if (GameCnt >= Game50 + 3).

I might be mistaken, but I believe Mike (mwm) did release a patch for
this.  Unfortunately I probably deleted it away, perhaps someone else
can check on whether such a patch was released.

One other bug in the repetition which I'm trying to figure out has
to do with probing the hash table.  The score from the hash table
is never "diluted" if a repetition is found whereas the score from
evaluate () is "diluted".  I'm not really sure of the right way of
doing it; maybe not probing the hash table at all when a repetition
is found might be one way.  Any thoughts?

Kong Sian