Re: [gnugo-devel] Current CVS very broken

[Daniel Bump]

> http://www.public32.com/games/go/r11a/regression/
> (This is still a bit of a work in progress, so I've left the old, stable
> link w/ the 3.1.10 regressions.)

Even though it's work in progress, the new link is much more
useful than the old one. For one thing, a lot of new tests have
been added since 3.1.10, which is already two weeks old.

> Here's a troubling GTP session.  GNU Go actually makes an illegal move!
> I was tipped off by looking at trevor:180 PASSED.

On the other hand when I make an sgf file with the board position
and load it, GNU Go moves at B9. So I think this is a gtp/caching
problem, not something that is going to cause a mistake in a real
game. It may well be related to the problem we discussed earlier,
that sometimes GNU Go gives a different answer when genmove is
called twice in a row.

http://mail.gnu.org/pipermail/gnugo-devel/2001-October/000210.html
http://mail.gnu.org/pipermail/gnugo-devel/2001-October/000211.html
http://mail.gnu.org/pipermail/gnugo-devel/2001-October/000213.html
http://mail.gnu.org/pipermail/gnugo-devel/2001-October/000217.html

Dan