SjaakII 1.0 RC1

[hgm]

Having non-capture intermediate squares is pointless. I suppose the engine could also mark empty squares in cyan, XBoard would not care. But it would then force the user to doe extra clicks, and would receive the move as xy,yz where xz would have sufficed to fully describe it. When the engine would mark squares it wanted to hop over that way, XBoard would clean them out. XBoard's own Betza generator would never mark squares in cyan that are not side-effect captures. (And currently XBoard can handle at most one of those, although its Betza generator, as well as the engine highlight could generate more. I guess XBoard would only remember the last one entered.)

[Evert]

Elven Chess (ChuChess) appears to be working correctly now, at least games no longer terminate because SjaakII made an illegal Lion trade (I have to check the games to make sure that it's not making them because it avoids only unsafe ones, or if it also avoids ones where it would be safe). I truncate plunder-raids by Lions after the second ply of quiescence at the moment, not sure if that's a good idea or not.

I noticed HaChu doesn't play this variant, do you have a program that does? It would be helpful for finding bugs.

Two things I've noticed though:
1. It appears that XBoard completely ignores my highlight FENs, at least it seems to do its own thing when I send them. I tested this by renaming the variant to something else ("eleven" rather than "elven"), but the highlights it shows are not the same as the engine sends. This was mainly because the ones I sent were wrong, but this made debugging them trickier. Not sure what's up with this.
2. The games of Elven Chess seem to predominantly end in "0-1 {Black mates}", which looks suspicious to me, almost as though the first position is really a zugzwang position...

[hgm]

The problem was that HaChu did not know how to castle. In the mean time I kludged something in to accept it from the opponent, for Mighty Lion, but I am not sure that would also work on 10x10 boards.

Are you sure that legality testing was off? Highlight commands are ignored when legality testing is on.

[Evert]

The problem was that HaChu did not know how to castle. In the mean time I kludged something in to accept it from the opponent, for Mighty Lion, but I am not sure that would also work on 10x10 boards.

Are you sure that legality testing was off? Highlight commands are ignored when legality testing is on.

Ah, perhaps not.

Either way, I think I spotted a glitch. After 1. Le3 d5 2. Ld4 Nh6 3. Lxd5-
the d4 square is not marked as a valid Lion target, although Lxd5-d4 is accepted (never mind the quality of play).

[hgm]

I think I made XBoard always accept a second click on the end-square of a previous leg as terminiating the move there, even when the square was not marked. (The engine can always refuse the move.) This because HaChu did not apply the highlight correctly there either, and I was in a hurry to meet the Debian deadline. The problem is that HaChu goes through the move list to find two-leg moves with the piece moved in the first leg that have the clicked cyan square as intermediate, and highlights the to-square of those. But it should also have looked for one-step moves to the clicked cyan square. Unfortunately there is a piece in Tenjiku Shogi (the Tetrarchs) that can igui adjacent squares, but ot move to them. So I cannot just always highlight the intermediate square for the second leg as a shortcut. (But as a shortcut I could of course test if the piece is a Tetrarchs, once I implement Tenjiku.)