gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] Gifu issues


From: Gunnar Farneback
Subject: Re: [gnugo-devel] Gifu issues
Date: Thu, 10 Jul 2003 01:06:16 +0200
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

Arend wrote:
> 1. I assume that the tournaments using GMP that gnugo has played so far
> were done with cgoban. This is _not_ possible for Gifu because of their
> specific requirements (specifying which color sends NEWGAME etc.) as far
> as I can tell. Unless someone is willing to hack on cgoban which I find
> unlikely.
> 
> 2. Program gnugo to connect to a COM-port, and hardwire the changes they
> need into gnugo. Either do this for Linux, or for Windows using their
> CFGCOMM.DLL (which might be easier). This is doable, but would still
> require someone to spend some significant time on that (plus testing
> this is not that easy).
> 
> 3. We tell them to use their GMP-board and gnugo in ascii mode,
> handcopying the moves in both directions. Then we need to tell them what
> to do if they make a mistake (wait until gnugo's move is finished, type
> undo twice, and put in the correct move).

If someone is willing to spend the time required for 2 I would instead
suggest 4, which would be a lot more useful in the long run:

4. Program a bridge between GMP and GTP. I.e. in one end it talks GMP
over a COM port with the GMP specifics used by Gifu and in the other
end it starts a GNU Go process and talks GTP with it.

SP Lee wrote:
> Can we just redirect input and output to COM port instead of hand copy
> the moves?

Short answer: no.

> Is the ASCII mode a standard for all computer go program
> (e.g. there is no 'I' in the horizontal index)?

Not in any way. Besides we need to talk the GMP protocol over the COM
port, which is a binary and fairly messy protocol. The most elegant
solution would be the GMP-GTP bridge I outlined above.

/Gunnar




reply via email to

[Prev in Thread] Current Thread [Next in Thread]