Re: [gnugo-devel] Arraybound violation in 3.1.16

[Inge Wallin]

Gunnar wrote:

> This is definitely not what Inge had intended. When stackp>0 this
> breaks out of the loop immediately, so there would be no point even
> entering the loop with stackp>0. The appended patch fixes this problem
> and another serious bug, namely that attack_and_defend() must be
> called rather than attack() and find_defense() separately in order for
> the defense point to be reliable. Finally it fixes up some weird
> indentation.

Thank you!

There is always a risk with cut-and-paste, especially when you are
tired.  And in this case, it seems there was another bug already in
the code.

        -Inge