RE: [gnugo-devel] reading problem

[Portela Fernand]

Arend wrote:

> What you are observing might be a persistent cache effect, otherwise I
> have no idea.

After some testings, I eliminated the safe_move() cache effect hypothesis,
(though, I think the problem should be addressed anyway, there are a
couple patterns in barriers.db using safe_(x|o)move)

So what's left is what you suggest : a persistent reading cache effect.

/nando