Protocol clarification

[Giorgio Medeot]

Hi H.G.,

I need a clarification about castling notation in coordinate algebraic notation, when playing FRC. The protocol specs states that FisherRandom castling should be noted as either 'O-O' or 'O-O-O'.
That means:
1. That I should trace that WB told me I am playing an FRC game, and in that case always stick to the SAN castling notation;
2. Or that I can still use coordinate notation for castling when there is no ambiguity, and use SAN notation only when the king could move to the same square in a normal move. BTW I would prefer this solution, as it seemlessly extends the case for standard chess.

I would like to know what the protocol prescribes, and what could be considered good practice.

Thanks and regards,

Giorgio

[H.G.Muller]

According to the protocol specs castling will have to be sent as O-O(-O) in FRC. The results of disobeying the specs is by definition implementation dependent, and thus undefined.

I tried it out, and current WinBoard does reject e1g1 as an invalid move even when your King is on e1 and your Rook on h1. When you switch legality testing off, though, it performs the move as a castling. And it does understand e1h1 as both legal and a castling. But not h1e1. (That would only work in variant seirawan.)

[Giorgio Medeot]

Ok, so I will keep track of the variant being played, and I will send e1g1 in variant normal or O-O in FRC.

Thanks,

Giorgio