This first public beta of Smirf (about 1.3 MB) is unrestricted until 2005/06/30.

Of course this program is subject to further improvements.

http://de.geocities.com/rsmuenchen/SmirfBeta.zip

Comments are welcomed.

Regards, Reinhard.

to jcarrillovii: I have posted on that error at http://www.exactachess.com/ , the exactachess GUI forum (Arena Support Forum).

to jcarrillovii/Caissus: Be aware to use an actual version of Arena. I have tested Arena 1.1 Build 1847. Here castling seems not to be allowed.

Oh - I have to correct myself. Entering the situation in Arena as FEN would not allow to castle. Entering it via PGN would - strange!

to Caissus: SMIRF does not. I already have seen Chess960/FRC engines, which offer castlings, where third pieces would illegaly be overjumped. Arena is the leading FRC aware GUI, but obviously still not without minor defects.

to Chessmaster1000: Castling is not allowed then, because the Rook e8 is threatening. Be aware that the King is always moving *ON* the board because castling is an iterated King's move, where the involved Rook is invisible. You can compare that situation to an e.p. capture move, where a Pawn is allowed to answer each of the iterated double Pawn steps. Because check would be applicatable to each of the King's iterated steps during castling, it would be allowed to capture the King. Thus castling is forbidden there, even when the threatened square is filled with the involved Rook.

to ughaibu: Currently Shogi is not that relevant for me. I have made a proposal to display Shogi positions in a more European look, see http://f51.parsimony.net/forum203932/messages/21.htm . But there has been no feed back. Thus I concentrated on Chess960 and 10x8 CRC. Second relevant for me is the game of Go. May be later I will write a program for Go after the SMIRF project will be completed.

to AbigailII: You are not right with this. Chess960 is not Shuffle Chess, where mirrored positions of course were functionally identic. You will have noticed, that the positions after O-O and O-O-O are not symmetric. Thus the existence of castlings indeed makes mirrored starting positions belong to different games. The conclusion is, that of course 960 different situations are defined.

to AbigailII: "Besides, "480" would have been a more appropriate number." - Why that?

Well, here are some links to English language articles http://www.chesstigers.org/alte_daten/chesstigers_alt/Tigers/cc/2004/e/default.htm .

to AbigailII: Your assumptions are often made but never the less unfounded. It is essential in Chess960 that the position is known first when the clocks will be started. Compared to the common Chess game this reduces the advantage of white to be able to prepare the first move without measuring the preparation time. There is therefore a tendence to equalize chances. Which position you are talking of, that it should be andvantageous for White? Please specify it and proof that statement. See as an example the game of Shogi, which still is interesting after centuries, nevertheless there are three undefended pawns in the beginning. Establishing back fights with well known positions would be against the nature of Chess960 (Chess960 is the new name of FRC).

to Walter Montego:
Well, if you are breaking rules, it is easier to win competitions. And that is, why computer programs are more and more dominating. They still are allowed to use huge look up tables and gigabytes of table bases. A human opponent is not permitted to use even his own notes.

We have to distinguish between using machines to analyse situations or to perform an equal fight. How will you compare the quality of programs, when both sides have different ressources? How should their ablities be compared to a human being. It might be very effective to fetch data from huge tables, but it is not at all intelligent. It is essential for intelligent processes to give results based on restricted information in short time. Having Terabyte bolides play will become uninteresting immediately the will have won against the World Chess Champion in a match.

to Walter Montego:
It seems, that I have not been able for me to make my position clear. Actually no one performes a race between a Ferrari and a human being. It is not a goal for the next decades for man to still compete with computer bolides. It is essential for chess to be restricted in the use of ressources, e.g. time and looking up material. This rule has been not used by engines, to make programs an interesting opponent. But that situation will come to its end very soon.

to Walter Montego:
I absolutely disagree here. One reason why I am working on such an engine is that huge opening libraries have given an immense advantage to computers. I always connect that approach with the FairChess idea, see: http://www.chessbox.de/Compu/schachfair_e.html . Smirf actually has a size of 46K.

to jcarrillovii:
Even if it was true that the advantage for one side might be greater than in traditional chess, this would not be relevant, because the players have not the time to prepare themself for that starting position. It is very important that starting arrays are drawn or published immediately before the playing time starts to run. And because Chess960 understands itself at a superset of classic chess it would be counter productive to exclude this special starting array number 518.

to Fencer:
Well, my answer would for sure be not representative. I avoid Blitz, Bullet and 6x6 board games. I wonder why to start a game without being ready to spend an appropriate amount of time. That is beyond of my understanding. But people are different.

to Fencer:
There is one position for one round valid for all participants.

to Fencer:
As far as I have experienced it (like in the Chess Classic Mainz) the positions are common to all participants for a round and will be published with some minutes before the start of the round because the positions in real live have to be set up manually. Normaly positions would be skipped and redrawn if they already would have occurred during that tournament, but the chance for that is very small.

to Fencer
have a look at http://brainking.com/game/ShowGame?g=615346

to Caissus:
No wonder about that, it is correct PGN. The problem is the moment, when castlings could or should happen! P.S.: I have written something on FRC-FEN to cover situations, where the Rook with castling rights is not the outermost one. BrainKing is merely encoding starting positions, so it is not relevant here. But if you want to encode even exotic positions with FEN, you will need a tool like Smirf or the FullChess FEN Editor.

to Fencer:
SMIRF can read and use it. But Smirf was built very tolerant towards violations of the PGN specification as long as the input will make any sense. Smirf is e.g. able to read notations written down in German instead of English. But there are some GUIs and tools which are very strict concerning PGN. As I told you SMIRF is already satisfied with the current status quo.

to Fencer:
in http://brainking.com/game/ArchivedGame?g=603270 I get longer lines up to 92 chars.

to Fencer:
Ok, I have learned that such things are not hard for you - fine! When I look at the way how an individual FRC game is constituting now, I have a suggestion again. Because it is essential to Ches960 / FRC, that the two players will get the information on the current position to play very short before the game is started, it might be a good idea to mask the base line of the starting array until the game is accepted or running. But this is only a cosmetic detail not a must have.

to Fencer:
Thank you very much for enabling FRC here!

Would it be to hard to supply the FRC standard position number as e.g. 870 in the PGN file like
[Event "Casual Game 870"] ?
It would also be very helpful to see this number somewhere when the position will be displayed.

To Fencer:

I am sure, you will do it right!

to Fencer:
Really good news! Do you know that each FRC position is related to a number from 1 to 960? Shall I send you material about that?

FRC / CRC, that would be the solutions for the current millennium.

to Fencer:
but not to forget: it must be possible to enter a king's move from and to his square, because there exist castlings, where the king's position keeps unchanged thought the castling is a king's move.

to Fencer:
in my Smirf program that problem is solved comparable to the underpromotion option. If there are more than one move possible with the intended step, the user has to select from possible moves.

to Nasmichael:
Still nowhere, because that has simply been a proposed study, causing no interest in that matter at all. So that idea still is sleeping.

to Nasmichael:
have a look at: http://f51.parsimony.net/forum203932/messages/21.htm