Author: Miguel A. Ballicora
Date: 12:57:39 05/21/01
Go up one level in this thread
On May 21, 2001 at 13:43:55, Andrew Williams wrote:
>On May 21, 2001 at 13:26:59, Pete Galati wrote:
>
>>On May 21, 2001 at 13:05:48, David Rasmussen wrote:
>>
>>>On May 21, 2001 at 07:04:55, Pete Galati wrote:
>>>
>>>>On May 21, 2001 at 02:56:10, David Rasmussen wrote:
>>>>
>>>>>I don't understand why xboard/winboard doesn't support books on it's own. If it
>>>>>did, no one would have to implement opening book in their engines unless they
>>>>>wanted book learning and such. And even that could be supported by xb/wb. Even
>>>>>if people didn't want to use this feature, they could just turn it off.
>>>>>Many people are reinventing opening book features all the time. And it really
>>>>>isn't as exciting as developing search or evaluation etc.
>>>>
>>>>That's not the nature of Xboard/Winboard. The opening book is the Chess
>>>>program's problem, not Xboard/Winboard's problem. They're just there to give
>>>>you an interface to the Chess program.
>>>>
>>>
>>>I know, but opening book support is so boringly simple, that it might as well be
>>>the interface that does it. It is very boring to implement opening book in a
>>>chess engine (at least for me), still, it can be done in really bad and really
>>>good ways. Why not let the engine handle this? It would make any winboard
>>>program, such as TSCP, have powerful openbook capabilities at hand.
>>
>>I agree, it does open some posibilities.
>>
>>>
>>>>Besides, most of the programers won't want Tim Mann to be the one making the
>>>>opening books for their programs, and my guess is that Tim Mann probably isn't
>>>>interested in doing that either.
>>>>
>>>
>>>Why would Tim Mann be doing it? If this is to be done in any kind of sane way,
>>>users, programmers etc. should be able to build their own books, control
>>>learning features etc. from the engine.
>>
>>I think Gian-Carlo Pascutto indicated that this is in the works already? I'm
>>surprised. But I wonder if a special utility for buiding books for
>>Winboard/Xboard will be included with them. I can't picture building it into
>>the gui.
>>
>>Pete
>
>There have been a few messages in the Winboard/Xboard mailing list. What is
>being proposed seems to be a feature whereby a chess program could *delegate*
>the task of choosing opening moves to some other program. In principle this
>could be a specialized program, but it could equally be another chess program.
>designed just to do this. The key thing is that the implementation wouldn't
>require any change to existing chess programs, unless they wanted to take
>advantage of this facility. Also, the changes to xboard would not be huge,
>as the work of selecting moves would be done by another program, not by xboard.
>Furthermore, the protocol for communicating with the "opening program" would be
>similar to that used for normal chess programs.
>
>Andrew
I planned to do this! it is in my to do list. I haven't implemented a book for
my program so I thought that when I do it, It would be a "book adapter" that
communicates with winboard and my program.
The adapter plays the book moves until is out of it. Then, it passes to winboard
the moves that my program decides.
Engine <==> Book Adapter <==> Winboard <==> Opponent engine
I planned to do this for three reasons
1) save time to other people to implement a book for new programs
2) I can test my program against other weak engines with and w/o books
3) I would learn multithreading programming, about which I know nothing (yet).
In fact, I started to read something until my spare time shrunk a bit.
Regards,
Miguel
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.