Yeah, that is definitely a problem. When I tried a simple Shogi mate in 1 tsume position with my prototype diagram it took 15+ seconds just to find it, and that was with a (technically illegal) Pawn drop.
That's a really big problem for the testing of my new 12x12 drops variant, Uchi Chu Shogi. Currently, I'm forced to use Ludii to test that game's computer play, and as we both know Ludii's AI is workable but is objectively terrible, and I would like to have a better AI to test drop games with.
Thankfully, it should be at least be possible to make the pieces droppable from the holdings and to highlight allowed drops, likely via an enableDrops=<boolean> parameter, and to have some other parameters for controlling the most common drop conditions (drop promotions, twins on a file and the max number of twins allowed, zonal restrictions, and restrictions against drop checkmates).
As for the reverse move generation problem, we can likely solve that by increasing the max ply limit to, say 6.0 ply (quite suboptimal, sure, but workable with the Diagram's current limitations).
I'm guessing just having pieces be droppable to empty squares should suffice for most drop games that people make, since the Shogi drop mechanic would easily be the most familiar to new game designers due to that game being a classic.
Yeah, that is definitely a problem. When I tried a simple Shogi mate in 1 tsume position with my prototype diagram it took 15+ seconds just to find it, and that was with a (technically illegal) Pawn drop.
That's a really big problem for the testing of my new 12x12 drops variant, Uchi Chu Shogi. Currently, I'm forced to use Ludii to test that game's computer play, and as we both know Ludii's AI is workable but is objectively terrible, and I would like to have a better AI to test drop games with.
Thankfully, it should be at least be possible to make the pieces droppable from the holdings and to highlight allowed drops, likely via an enableDrops=<boolean> parameter, and to have some other parameters for controlling the most common drop conditions (drop promotions, twins on a file and the max number of twins allowed, zonal restrictions, and restrictions against drop checkmates).
As for the reverse move generation problem, we can likely solve that by increasing the max ply limit to, say 6.0 ply (quite suboptimal, sure, but workable with the Diagram's current limitations).
I'm guessing just having pieces be droppable to empty squares should suffice for most drop games that people make, since the Shogi drop mechanic would easily be the most familiar to new game designers due to that game being a classic.