Unfortunately this is mostly what prevents me from achieving my chess variants programming goals.
Now that you mention that: On an initiative of Mike Sherwin I was drawn into a project (which I more or less seem to hav usurped now) of writing a simple educational orthodox Chess engine. But the design we discussed seems to be very suitable for supporting all kinds of chess variants. Even for orthodox Chess it would rely on morphing, and a very small extension (which would not slow it down if no morphing was used) would be to add a capture matrix as well. The board size could run up to 15x15 without 3rd-ring leaps without problems, while potentially 63 piece types could be supported. The very simple move generator we designed would already be able to support (finite-range) slider/leaper compounds and divergent pieces.
It might be just a small effort to extend the move generator for supporting a much wider subset of XBetza. Currently the move generator is table-driven (just like the Interactive Diagram, b.t.w.), with only a single leg per move direction, characterized by a triple (boardStep, range, rights). The 'rights' indicate whether the move is move-only or capture-only through two of its bits. It would be easy to put more flags there, indicating the move is more complex, and using a more-advanced multi-leg move generator that would take more triples into account for additional legs. And also test for p, d or o 'rights' in any of those legs.
The whole thing was intended as a comunity project. Perhaps you are interested to participate in it too.
Now that you mention that: On an initiative of Mike Sherwin I was drawn into a project (which I more or less seem to hav usurped now) of writing a simple educational orthodox Chess engine. But the design we discussed seems to be very suitable for supporting all kinds of chess variants. Even for orthodox Chess it would rely on morphing, and a very small extension (which would not slow it down if no morphing was used) would be to add a capture matrix as well. The board size could run up to 15x15 without 3rd-ring leaps without problems, while potentially 63 piece types could be supported. The very simple move generator we designed would already be able to support (finite-range) slider/leaper compounds and divergent pieces.
It might be just a small effort to extend the move generator for supporting a much wider subset of XBetza. Currently the move generator is table-driven (just like the Interactive Diagram, b.t.w.), with only a single leg per move direction, characterized by a triple (boardStep, range, rights). The 'rights' indicate whether the move is move-only or capture-only through two of its bits. It would be easy to put more flags there, indicating the move is more complex, and using a more-advanced multi-leg move generator that would take more triples into account for additional legs. And also test for p, d or o 'rights' in any of those legs.
The whole thing was intended as a comunity project. Perhaps you are interested to participate in it too.