Author: Harald Lüßen
Date: 03:42:29 08/05/05
Go up one level in this thread
On August 04, 2005 at 01:57:51, Reinhard Scharnagl wrote:
>Here are some facts to objectify the discussion instead of flaming:
[...]
I like facts. I don't like flaming. And I try (for me) to
summarize the problem.
I think a big part of the problem and misunderstanding is the
result of the difference between positions (FEN) and games (PGN),
or rather the result of not seeing the difference.
FEN was invented to describe the situation visible on the board
and a few extensions belonging to the rules of chess to mark
the invisible state (en passant, castling, who has to move).
PGN was invented to describe the whole game from the starting
position and with all moves, with the opponents and the date,
with comments or time stamps. If necessary one could invent
a token for the feeling of the players ("has had bad breakfast
and was prepared for another opening"). I think a new tag
should be created for chess960 or FRC960. Or does such a tag
already exist? That gives database programs a lot to do and
enough information to filter normal and 960 chess games.
As a software developer with a mathematical and graph theoretical
point of view I see the FEN just as a description of a node
in the forest of game graphs. And I would like to have a unique
description. All different positions must be coded different
and all different codings must belong to different positions.
Some positions in chess960 games are not different from normal
chess games. In the opening phase there are naturally the
most differences (all but one) but as the game continues and
the castling possibilities vanish the positions are more often
the same independent from the starting position. And with
respect to the time on the clock and common chess knowledge
the rest of the game is independent from the beginning even
for the players. We can even use the same endgame databases
since there are practically no castlings involved.
With a fixed font I see it this way:
1 2 3 518 960
| | | X |
\ | \ X |
\ | \ X /
\ | \ X /
\| \ X /
\ \X /
\ X /
\ X/
\ X
\ X
\ X
mate X
mate
The lines are games, the characters are positions.
The positions marked with /\| do need a description
that has to be invented. The positions marked X are
classic chess. They should not be coded in a new way.
It is normal for a game to start with new positions but end
in the classic style. And it is possible for whole games to
stick to new positions.
SMK-FEN:
SMK invented new castling flags for the new positions. That is ok.
SMK invented new castling flags for the old positions. That is not ok.
X-FEN:
RS invented no new castling flags for the old positions. That is ok.
RS invented no(1) new castling flags for the new positions. That is not ok(2).
(1) There are cases with additional chars. It works but is ugly.
(2) The difference can be seen in the first part of the FEN string.
The pieces (rooks, king) are on other (strange) squares even if
there are the old castling flags (KQkq). Though this is technically
sound I find it confusing.
There was a compromise:
We invent new castling flags for the new positions. That is ok.
We invent no new castling flags for the old positions. That is ok.
In programmer practise when writing a FEN string use the new chars
(ABC..Habc..h) but exchange them with (KQkq) if no other character
than (AHah) is involved. When reading just do the opposite or
give an error message when your program can not understand the
chess960 positions.
I don't know whether there will be a solution in this dispute
and whether there will be a survival of the fittest.
If you find bugs in this summary, tell me.
Now I will become subjective:
I don't like the way how SMK ignored the existing FEN-Format.
He could have discussed it openly in a chess forum. Private
mails with some programmers don't help much.
I don't like the way how RS is attacked and flamed.
But he seems to be a little bit stubborn (in his postings). Hm?
I wish there was a solution to this.
Harald
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.