Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: EPD and the real world

Author: Dann Corbit

Date: 09:29:48 10/07/05

Go up one level in this thread


On October 07, 2005 at 01:30:02, Yuriy Lyapko wrote:

>>>
>>>One more than one is two, which is twice as much as one, and the size of this
>>>element is probably the dominant factor in the size of chess game data.
>>>
>>>In any case, it's not going to happen.  This kind of standard is like sticking
>>>your head into a bucket of concrete.
>>
>>There is a database called PigBase (for AMIGA, IIRC) that recorded the moves
>>with their move numbers in a single byte and it was stupendously compact.  But I
>>have about 4 million games in a SCID database and it fits easily on a CD.  So I
>>do not see the space as all that important (especially for my needs).  How many
>>chess games will fit on a DVD, encoding each possible move as 2 bytes?
>>
>>There is no binary standard yet, really.  It's just what-ifs from what I could
>>tell.  Or we can just write our own standard not called EPD or FEN or PGN but
>>called "CIF" for Chess Interchange Format.  Or something like that.
>>
>>You can always provide a translator as well, to convert from the new format to
>>PGN and back.  That is what ChessBase and ChessAssistant are doing, after all.
>>They just don't publish their internal format.
>
>Here is my approach to this:
>http://www.geocities.com/lyapko/lgpgnc.htm
>
>The general idea is very simple:
>
>1.Generate all legal moves.
>2.Output move number as byte.
>
>As you can see from results, PGN get shorter 5-6 times and you can still further
>compress it.

But not every program will generate the moves in the same order.  I do not want
to always generate the moves twice.

Aside:
The file 1.pas is missing from your distribution.



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.