Re: [gnugo-devel] owl_determine_life() patch

[Gunnar Farneback]

Paul wrote:
> this patch looks like a logical bugfix to me. it changes evaluation of
> *eyemax in owl_determine_life() and, hence, the condition to disable
> playing vital moves. regression delta is quite neutral: 4 passes, 3
> fails.

The old code looked like this:

    for (ne = 0; ne < num_eyes - num_lunch; ne++) {
      add_eyevalues(probable_eyes, &eyevalue_list[ne], probable_eyes);
      *eyemax += max_eyes(probable_eyes);
    }

The *eyemax computation is indeed horribly wrong. It should have been

      *eyemax += max_eyes(@eyevalue_list[ne]);

in order to make sense. (The bug was introduced in one of my patches.
It seeems you need to review those harder. :-)

Your patch looks okay too. I've added it to CVS.

/Gunnar