The AI responds with knight to c6, ignoring the check.
OK, fixed. Thanks for spotting this. The problem was in the j prefix for indicating ski-slides. (The alternative definition gyafW did work without problems.) The Betza parser splits such a move into 2 legs, one to jump over the adjacent square, and a remaining normal slide. But it adapts the range of that slide by subtracting 1 (so that jR4 moves 2, 3 or 4 steps.) Infinite range is indicated by 0, however, and this was adapted to -1. Now the move generator of the AI (in contrast to that of the UI) did not interpret the -1 as an infinite slide, but as a slide up to half the board. I now changed the Betza parser such that it refrains from decrementing a range of 0.
OK, fixed. Thanks for spotting this. The problem was in the j prefix for indicating ski-slides. (The alternative definition gyafW did work without problems.) The Betza parser splits such a move into 2 legs, one to jump over the adjacent square, and a remaining normal slide. But it adapts the range of that slide by subtracting 1 (so that jR4 moves 2, 3 or 4 steps.) Infinite range is indicated by 0, however, and this was adapted to -1. Now the move generator of the AI (in contrast to that of the UI) did not interpret the -1 as an infinite slide, but as a slide up to half the board. I now changed the Betza parser such that it refrains from decrementing a range of 0.