Greg Strong wrote on Wed, Mar 7, 2018 05:40 AM UTC:
It'd be nice if software supports variants with such novel castling rules, but I fear they don't, at least without some considerable work.
Indeed, this is the reason I added Hanibal Chess to the newest version of ChessV but not Wide Chess. That said, it wouldn't be all that difficult to add support for a "fast castling" rule in the same way it already has support for "flexible castling", although other engines aren't likely to implement it. The current (unfortunate) situation is that we currently have a bunch of 10x8 engines that can play some Capablanca variants, but almost none support flexible castling, so they can't play Schoolbook, Grotesque, etc. So, when desigining a variant, this is something to keep in mind.
Another consideration - if we were to create a universal "fast castling" rule, does it require the king to move at least two squares? If so, this eases implementation, at least somewhat, because a move can be cleanly defined with only from square -> to square (and optional piece in case of pawn promotion.) Flexible castling requires the king to move two squares for this reason. Otherwise, a king moving a single step can be ambiguous. For purposes of support in ChessV, this isn't a big concern, as the architecture will automatically throw a dialog box at you when a move is ambiguous. See Wildebeest Chess as an example, where a king can castle while moving only a single square, although this game is unique in this regard as far as I know and it would have made life easier for all involved if R. Wayne Schmittberger had taken this into account, especially since you probably aren't going to want to move your king a single space when castling anyway. In any event, that bridge has already been crossed ... Although, even outside of computer play, you'd probably want to give some thought as to how moves should be annotated in general.
Indeed, this is the reason I added Hanibal Chess to the newest version of ChessV but not Wide Chess. That said, it wouldn't be all that difficult to add support for a "fast castling" rule in the same way it already has support for "flexible castling", although other engines aren't likely to implement it. The current (unfortunate) situation is that we currently have a bunch of 10x8 engines that can play some Capablanca variants, but almost none support flexible castling, so they can't play Schoolbook, Grotesque, etc. So, when desigining a variant, this is something to keep in mind.
Another consideration - if we were to create a universal "fast castling" rule, does it require the king to move at least two squares? If so, this eases implementation, at least somewhat, because a move can be cleanly defined with only from square -> to square (and optional piece in case of pawn promotion.) Flexible castling requires the king to move two squares for this reason. Otherwise, a king moving a single step can be ambiguous. For purposes of support in ChessV, this isn't a big concern, as the architecture will automatically throw a dialog box at you when a move is ambiguous. See Wildebeest Chess as an example, where a king can castle while moving only a single square, although this game is unique in this regard as far as I know and it would have made life easier for all involved if R. Wayne Schmittberger had taken this into account, especially since you probably aren't going to want to move your king a single space when castling anyway. In any event, that bridge has already been crossed ... Although, even outside of computer play, you'd probably want to give some thought as to how moves should be annotated in general.