[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[gnugo-devel] clock.c vs play_gtp.c
From: |
Arend Bayer |
Subject: |
[gnugo-devel] clock.c vs play_gtp.c |
Date: |
Sun, 1 Aug 2004 00:03:14 +0200 (CEST) |
I would like to see us using a level adaption scheme at the next tournament
that we trust. In my opinion the code in play_gtp.c is fairly
well-tested, as it is used in all the gnugo clients on KGS, and is also
very similar to the code in gnugoclient, which we have been using for
ages without big problems.
I would suggest to kill the code in clock.c, move the corresponding code
from play_gtp.c over there, and then try to use it also in GMP-mode.
(Alternatively, we could try to play the next GMP-tournament with an
GMP-GTP bridge.)
The only problem with the play_gtp.c code that I have seen is that it
seems to overreact a little. I.e. it often reduces the level by several
steps where one would have been enough. And the only time I have seen
gnugo come close to using by time was when it increased the level from
11 to 12 at the second-to-last move in a byoyomi period.
Volunteers to work on this are welcome :))
Arend
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [gnugo-devel] clock.c vs play_gtp.c,
Arend Bayer <=