How about placing a dummy piece on an invisible square? The swapping move could be legal only if the piece is still on the board, and it could remove the piece when it is completed.
That could actually work.
But in the mean time I got another idea. During search the move descriptors as generated are extended with other information, such as how much they alter the score by capturing material. But also how much the 'royalty count' of each player changes by the move. And when that hits zero (or worse) for one of the players, it causes replacement of this heuristic score by a game-terminating one.
These royalty updates are only defined when there is a royalty change through capture of a royal or a piece subject to a baring rule, or by promotion of or to such a piece. For normal moves it would remain undefined, and making the move would not alter the royalty counter of either player.
The user-supplied Tinker script could thus predefine these royalty changes. Capture of a true royal would deduct 1024 from the royalty count, and with a single royal (for extinction royalty) or absolute royals the initial value would start at 1. (And when a baring rule is in effect it would start as the initial number of pieces subject to that baring rule, and each such piece that would be captured would decrement the royalty count by 1.)
This could also be used to set a budget for events that can only take place a limited number of times during a game. E.g. if the royalty count would start at 3, and the Bishop swapping move would get a royalty decrement of 2 specified for it by a Tinker script, playing it a second time would turn the royalty count of the moving player negative. Which would be considered an immediate loss / illegal move, like blowing up your own King in Atomic.
The issue then is reduced to how we can get the initial royalty count set to 3, rather than the usual 1. I suppose it would be worth it to add a new parameter for this to the I.D., like extraRoyalty=N (with default N=0). This would then be added to the normal starting value based on counting royals and baring-rule pieces. I will work on that.
That could actually work.
But in the mean time I got another idea. During search the move descriptors as generated are extended with other information, such as how much they alter the score by capturing material. But also how much the 'royalty count' of each player changes by the move. And when that hits zero (or worse) for one of the players, it causes replacement of this heuristic score by a game-terminating one.
These royalty updates are only defined when there is a royalty change through capture of a royal or a piece subject to a baring rule, or by promotion of or to such a piece. For normal moves it would remain undefined, and making the move would not alter the royalty counter of either player.
The user-supplied Tinker script could thus predefine these royalty changes. Capture of a true royal would deduct 1024 from the royalty count, and with a single royal (for extinction royalty) or absolute royals the initial value would start at 1. (And when a baring rule is in effect it would start as the initial number of pieces subject to that baring rule, and each such piece that would be captured would decrement the royalty count by 1.)
This could also be used to set a budget for events that can only take place a limited number of times during a game. E.g. if the royalty count would start at 3, and the Bishop swapping move would get a royalty decrement of 2 specified for it by a Tinker script, playing it a second time would turn the royalty count of the moving player negative. Which would be considered an immediate loss / illegal move, like blowing up your own King in Atomic.
The issue then is reduced to how we can get the initial royalty count set to 3, rather than the usual 1. I suppose it would be worth it to add a new parameter for this to the I.D., like extraRoyalty=N (with default N=0). This would then be added to the normal starting value based on counting royals and baring-rule pieces. I will work on that.