Hey, thanks for writing this - I've had lots of fun playing it during school.

I hacked together a seed finder and found that seed 5147933522 on 5x5 mode breaks the game.

Marking any cell as active or inactive leads to the progress jumping to NaN% and locks up the table.

This happens because there are zero active cells and the progress calculation (probably?) divides by zero.
Shouldn't be too hard to fix and I might open a PR if I have time.

(P.S.: Try using seed 64013837068 on 5x5😃)

hi, I am a big fan of picross. to be honest, I played a while at picross.net, but that side is down now. So now I play this version for a few weeks. The biggest difference is that on a wrong click, it immediatly show an error. I really needed to get used to this, because i often just marked some helper fields and clear them afterwards. in this version I have to do this in my head.

while this mode of playing is likely for more advanced players, there is an other problem that I see. and that is that one game can have many solutions. I even implemented a picross solver using JS. in one of the generated games, my program found 257 solutions on a 10x10 grid. But the game is only accepting one of them.

I found when there are 4 - 8 solutions, I at least can find in my head all the solutions, the game might not blink in the end, but the player at least know the solutions.

would you be interested in developing a version that filters games with too many solutions? maybe my picross solver can be of good use for that.

When a cell is selected, the validity of that solution is checked against a single predetermined solution. This does not take into account scenarios where multiple solutions exist [See image below]

guess: function(x, y, guess) {
    var solution = this.get('solution')[x][y];
    ...

    ...
    if(solution === guess) {
        state[x][y] = guess;
    } else {
        state[x][y] = solution * -1;
        mistakes++;
    }
    ...

    ...
}

One solution would be to add an else if statement between the if and else statements shown above. This else if statement would check if the selected cell contradicts the rest of the board constraints. This solution is not enough on its own because the stored solution would have to be updated to reflect the new solution required.

Because of this, I would focus deriving an algorithm that can efficiently check if the current state of the board can reach a solved state or if there is bound to be a contradiction to the board constraints. This algorithm could then be used in place of the stored-solution-check noted above. I'll be giving this some more thought in my spare time.

When a puzzle generates a puzzle with multiple solutions, live validation will correct your "mistake", even though it may lead to one of the correct solutions. I have a attached an image that shows that including the seed. The other solutions can still be reached if live validation is off though. It is nontraditional for a nonogram to have multiple solutions, so I would recommend altering generation to remove that possibility. (Example is 5x5 with seed 1701546768375)