[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] question about twogtp.pl
From: |
Arend Bayer |
Subject: |
Re: [gnugo-devel] question about twogtp.pl |
Date: |
Tue, 22 Oct 2002 12:28:33 -0400 (EDT) |
On Tue, 22 Oct 2002 address@hidden wrote:
>
> > My program would love to know that the other player passed so it can also
> > pass if it thinks it has won the game (even if it's earlier than it would
> > have otherwise done so). The easiest way to do that is to let the pass move
> > go to the GTP program, but I'm wondering if the GTP spec implies that 'pass'
> > doesn't need to be sent ?
>
> The pass doesn't need to be passed on. The reason is that if your engine
> receives two genmove commands with no intervening play, then clearly the
> opponent has passed.
It could as well be a non-conforming way by the controller to set up a game
with free handicap (or higher handicap than 9,...). I don't think we
should burden an engine with interpreting two consecutive genmove's.
(Should the engine add a PASS into it's private game tree? That would
clearly contradict the explicit wordings about move history in Gunnar's
GTP v2 draft.)
I think the point is one shouldn't stretch a protocol or one will run into
ambiguities at some point.
Arend
Re: [gnugo-devel] CB20 in conn.db, bump, 2002/10/22