Discussion: Go Board

Some time ago I created a Go Board for Second Life. I haven't had time to do anything more with it, so now I'm releasing it for all to use, share, copy, shake their heads despairingly at the huge mess of code as appropriate.

There is too much code to post on here, and I know this breaks etiquette a bit, but I've put it on the LSL wiki, so please go there to find the code.

http://secondlife.com/badgeo/wakka.php?wakka=LibraryGoGame

Jonathan

Do you have a full perm version in game somewhere, that is a lot of code.

it is, isn't it send me a message in-game and I'll send you a copy next time I'm on (can be infrequent, so be patient)

Thanks! I love playing Go (regular patroness of the Dragon Go Server) and already have a Go board image utilized in my build (though heaven knows why; it has nothing to do with the game).

Yay! Go players in SL! I'm quite an avid player too. I have a Go pavillion on my build - and have a half-started 400-prim go set there.... Perhaps I should put in Jonathan Shaftoe's instead!

Anyone know if there is a working Shogi set out there anywhere?

Would there be enough interest (>= 3 people...) in a go/wei'qi/baduk group?

yes - I'd join a Go/Wei'Qi/Baduk group - and if I ever get my set done, I'd play in SL too!

Unless you *really* want to build it yourself, give up and use Jon's. I've played it a few times and it's very nice, works well and pretty intuitively.

I don't have space for a spare group atm, but if there is a go group set up, please count me in for supporting it in spirit, and certainly as another player.

No, although it's one of those things I've given a passing thought. I enjoy Shogi.

I set up a group for go players, called "The Game of Go aka Wei'qi aka Baduk" (searching for "go" doesn't tend to work well, and sometimes people know the other names). It's open enrollment, maybe we can get some games going

Thank you, Jonathan! I play Go myself, and had been trying to find a playable Go game for in SL, so I can teach my Companion how to play the game!

I shall try assembling a game using the code you provided. If I get stuck, I'll contact you for a set.

*bows*

Well, the game seems to have assembled just fine. The instructions in the WIKI and embedded in the scripts were very good.

Looks like it takes 62 prims to rez the basic game, which isn't half bad for what it should be able to do. Any idea of the max # of prims it takes to rez the game? I would imagine that the 19x19 board takes the most.

Probably too many prims to leave it set up in my home, but I may be able to arrange for somewhere in my home sim to serve as a home for one of these boards.

I joined the Go group. Maybe we can try a game some time?

I think it is always 62 prims no matter the bouad size. There's a lot of cleverness with texture and prim tricks.

I have a set up in The Future (225, 50, 170).

The Go board is now for "sale" (at the princely sum of $0 hence the quotes at my shop in Dimidiata 233,174,75 in the gadgets vendor.

I know this isn't the place for advertising, but it seems that telling the people that read this thread where they can find a free board in world isn't entirely irrelevant.

Edit (original post was about the vendor not taking L$0 for items):

I got the Go board, but it's clearly not displaying all of the pieces when placed. Notice on this image there are some rectangles that are not covered; i.e. the P-S and 9-10 area.

I have had Jon's GO board out for a few days now. After reading this thread I made it buy a full perm. board for $0. Just right click and choose buy.

secondlife://Skari/230/111/112

Ok, if you use llSetPayPrice you can't give it a zero sum, and even if you leave the manual pay box visible you have to pay $1 and then hope (as it does) that the vendor gives you change.

I've changed that - there is a free board available at the location, but it's in a prim of it's own whilst I consider building a freebie vendor.

The llSetPayPrice() behaviour surprised me though, ah well.

you have done an excelent job.

I can see very little i would change, and that is saying alot. Recursion sends up flairs in my mind as it uses lots of memory but its hard to avoid.

I notice that "neighbour" isn't used in "recurse_check" in "GoGameLogicScript" if you remove this it will free up some memory.

actualy...

recurse_check(integer x, integer y, integer dir) {
    if (gGroupType == 1 && gGroupLiberties >= 2) {
        return;
    }
    if (gGroupType == 0 && gGroupLiberties == 1) {
        return;
    }
    if (llListFindList(gGroup, [x + (y * gGameSize)]) != -1) {
        return;
    }
    gGroup = gGroup + [x + (y * gGameSize)];
    integer neighbour;
    if (dir != 0) {
        recurse_check_util(x - 1, y, 1);
    }
    if (dir != 1) {
        recurse_check_util(x + 1, y, 0);
    }
    if (dir != 2) {
        recurse_check_util(x, y - 1, 3);
    }
    if (dir != 3) {
        recurse_check_util(x, y + 1, 2);
    }
}

can become

recurse_check(integer x, integer y, integer dir) {
    if (gGroupType != 1 || gGroupLiberties < 2) {
		if (gGroupType != 0 || gGroupLiberties != 1) {
			if (llListFindList(gGroup, [x + (y * gGameSize)]) == -1) {
				gGroup = gGroup + [x + (y * gGameSize)];
				if (dir != 0) {
					recurse_check_util(x - 1, y, 1);
				}
				if (dir != 1) {
					recurse_check_util(x + 1, y, 0);
				}
				if (dir != 2) {
					recurse_check_util(x, y - 1, 3);
				}
				if (dir != 3) {
					recurse_check_util(x, y + 1, 2);
				}
			}
		}
	}
}

and will save you 23 bytes of bytecode and 4 bytes of stack.

What would cause the Go stones not to appear?

I made the board, following the instructions in the WIKI article. I just tried playing against someone in-world, and the stones we played didn't rez. The game board stated our moves as we made them, but didn't mark them.

Any ideas?

This will save 12 stack bytes, and might be a bit slower. The stack savings should help with complex structures overflowing the script. The second function will save a few stack bytes when the script is getting close.

recurse_check(integer x, integer y, integer dir) {
    if (gGroupType != 1 || gGroupLiberties < 2) {
		if (gGroupType != 0 || gGroupLiberties != 1) {
			if (llListFindList(gGroup, [x + (y * gGameSize)]) == -1) {
				gGroup = gGroup + [x + (y * gGameSize)];
				integer i = 0;
				do
				{
					if(i == 1)
						x += 2;
					else if(i == 3)
						y += 2;
					else
					{
						--x;
						if(i&2)
							--y;
					}
					if(dir != i)
					{
						integer neighbour = get_board_state(x,y);
						if (neighbour == gSearchColour) {
							recurse_check(x, y, (5 - i) % 4);
						} else if (neighbour == 0 && gGroupType != 2) {
							if (x != gLibertyX || y != gLibertyY) {
								gLibertyX = x;
								gLibertyY = y;
								gGroupLiberties++;
							}
						} else if ((neighbour == 2 || neighbour == 1) && gGroupType == 3) {
							if (gGroupLiberties != neighbour) {
								gGroupLiberties += neighbour;
							}
						}
					}
				}while(4>++i);
			}
		}
	}
}

integer get_board_state(integer x, integer y) {
    return (integer)num = llGetSubString(gGameState, x += 1 + ((y + 1) * (gGameSize + 2)), x);
}

I was talking with Zarf a day or two ago - he's 10% into doing a go board, with a prim per stone. Linking would be a problem with a full size board of course, but assuming that could be worked out nicely, the recursion problem could be neatly sidestepped: Each stone keeps a list of the other stones in its connected group, and the "blank" stones which are the liberties of that group. When your liberties list is empty, you die. As long as you can consistently update this information, everything builds up step by step and it should be fast.

Counting the score at the end might be slower, but you don't care so much then, and you can still do it in a distributed fashion.

Maybe the board is too small? The smallest I was able to make it is SIZE = 1.0. Any smaller than that and a prim gets too thin, and possibly stuff gets hidden that shouldn't.

There is a disadvantage to doing it with group tracking, joining groups & limits related to lists (fill a board with all one color of stone will overflow any list (if memory serves me)). A better way would be to centralize the tracking and make a map of the grid, with markers for individual groups. Using single characters in a string to represent the grid structures. When a structure was joined to another one would have it's letters changed.