Geneva Chess available on Musketeerchess.net

[jerome]

You can play Geneva Chess on Musketeerches.net : http://musketeerchess.net/games/sp/captain/

This forum is dedicated to discussions around this game.

[sam]

You can play Geneva Chess on Musketeerches.net : http://musketeerchess.net/games/sp/captain/

This forum is dedicated to discussions around this game.

Geneva page is not working so far.


Geneva game rules:
http://musketeerchess.net/games/sp/capt ... /rules.php





Other info:

a) Pawns at 5th, 6th and 7th ranks have the same movements.

b) Other pieces such as N, B, R, Q and K are the same to that of chess. N=Knight, B=Bishop, R=Rook, Q=Queen, K=King.

c) This super pawn can promote to a N, B, R or Q.

c) This super pawn can also make en passant capture.

d) Board size is 8x8, or 8 rows by 8 columns.

[H.G.Muller]

The diagram still leaves it unclear over which square the oblique initial moves can be reached (i.e. where they couldbe blocked), and where after them an enemy Pawn has to move to in order to e.p. capture. If I understood Zied correctly, these are actually different squares, so that e.p. capture is a bit of a misnomer, and 'en non-passant' would cover it better (as the Pawn is taken on a square over which it did not pass).

[sam]

The diagram still leaves it unclear over which square the oblique initial moves can be reached (i.e. where they couldbe blocked), and where after them an enemy Pawn has to move to in order to e.p. capture. If I understood Zied correctly, these are actually different squares, so that e.p. capture is a bit of a misnomer, and 'en non-passant' would cover it better (as the Pawn is taken on a square over which it did not pass).

You are correct, it is not an en passant capture, we can call it opposition-crossing capture or oc capture (b4c3 from image below and previous move is b2c4). Opposition since the origin of the pawn (B2 square from image below) that can be captured is located directly opposite from the enemy pawn (B4 square). Crossing since that move (b2c4, path: b2-b3-c4) had crossed the rank 3 where the enemy pawn at B4 square is capable of capturing on A3/C3 squares (C3 square is in rank 3) and that b2c4 move started at rank 2 and ended at rank 4, crossing rank 3.



That white ngp (next gen pawn) at B2 square has just moved with b2c4 (path is b2-b3-c4). If there is a piece in B3, that ngp at B2 square cannot move to C4 or A4 squares.

If white previous move is c2c4 as in chess, black can do an ep capture b4c3.
If white previous move is c2d4 (path: c2-c3-d4) for geneva and other similar ngp, black can do an ep capture b4c3 too.

So for geneva and other similar ngp, there are 2 main special moves:
1. ep capture, the file of from_square of our pawn is adjacent to the file of opp pawn.
a) previous move is similar to chess, c2c4 and black can do b4c3 ep capture
b) previous move is ngp, like c2d4 (path: c2-c3-d4) move and black can do b4c3 ep capture.
2. oc (opposition-crossing) capture, the file of from_square of our pawn and the file of opp pawn are the same. Previous move is like b2c4 (path: b2-b3-c4) and black can do b4c3 oc capture. Similarly if previous move is b2a4 (path: b2-b3-a4) then black can do b4a3 oc capture.

[H.G.Muller]

OK, that explains it. So an oblique Pawn move like b2-c4 can generate both an e.p. square and an o.c. square, depending on presence of enemy Pawns (on a4 for e.p. and on b4 for o.c.). The o.c.-square is different from an e.p. square in that it matters which capture move (left or right) of the Pawn reaches it; after b2-c4 a Pawn on d4 could not perform o.c.-capture to c3, but a Pawn on b4 can.

What happens when the o.c. square is occupied, e.g. white had Pawns on b2 and c3, and plays b2-c4? Can black then make a double capture b4xc3 (taking c3 the normal way and c4 o.c.)?

I guess current WinBoard does not support o.c. capture yet. This is mainly because the internal representation of e.p. squares is kludgy, and can only encode a pair of squares when these are vertically adjacent. XBetza notation could in principle specify generation of two e.p. squares, by having b2-c4 go through both squares in a 3-leg move: b2-b3-c3-c4. The visit of c3 could then specify both move and hopping rights (mp), to allow passing it no matter how it was occupied. This would recognize any subsequent 'e' move (presumably only specified on Pawn diagonal steps) to b3 or c3 as a capture of c4, and it would have to be left to the engine to refuse the d4xc3 capture as illegal move.

[sam]

OK, that explains it. So an oblique Pawn move like b2-c4 can generate both an e.p. square and an o.c. square, depending on presence of enemy Pawns (on a4 for e.p. and on b4 for o.c.). The o.c.-square is different from an e.p. square in that it matters which capture move (left or right) of the Pawn reaches it; after b2-c4 a Pawn on d4 could not perform o.c.-capture to c3, but a Pawn on b4 can.

That is right.

What happens when the o.c. square is occupied, e.g. white had Pawns on b2 and c3, and plays b2-c4? Can black then make a double capture b4xc3 (taking c3 the normal way and c4 o.c.)?

Since C3 is occupied, black cannot make an o.c capture. It can only do a normal capture bxc3. The white pawn at C4 is safe. Like in e.p square, the o.c square shall also be empty.

[H.G.Muller]

XBetza would allow you to create only e.p./o.c. rights in the non-passed square by splitting the oblique move in one that hops (and doesn't create rights) and one that 'moves' (and does create rights):

ifmnarmnalWifmnarpalWifmnalmnarWifmnalparW

The 'm' would all be redundant here, as move-only is the default mode for non-final legs. Because the path is doubly bent it is necessary to specify the two mirror images separately with 'l' and 'r', rather than combining those with 's'. This could be avoided by using the convention that after a leg involving 's' (meaning it can bend both left and right) all subsequent legs use 'l' and 'r' as if the first bend was in the 'r' direction, and have to be interpreted in reverse when the first bend was to the left. (Note that directions in continuation legs are re-interpreted anyway depending on the orientation of the previous leg; this rule would merely extend re-interpretation through rotation by also allowing reflection.)

Unfortunately WinBoard does not apply that convention yet, making the more cumbersome split description necessary. But WinBoard's piece commands would not allow location dependence of a piece its move anyway.