qntm / hatetris Goto Github PK

Its funny the hard derivant from the original tetris, i wonder if the keybindings are missing or not?

I think that a safer way that you could rewrite the AI API is like this:

• Append AI API function declarations to the beginning of inputted code
• Convert modified code into JS data URI.
• Start web worker from JS data URI.
• Post a message to the web worker when it needs to calculate a move.
• Make the web worker post a message to the host script to send the result.

The reason you can't just use a URL to some script that evals the AI code is that web workers can still run fetch requests, so you don't want to eval code on the main HATETRIS domain.

Hi I am a researcher, but I am not familiar with javascript.
If you don't mind, I hope I can get the pseudocode or python code of the algorithm.
This can help me understand the algorithm.

Love this game!

I tried to watch the world record replay as listed in the base65536 repository, but it wouldn't work. When I tried the hex version, it did. Since it doesn't even output the hex format anymore, this is obviously a problem, as the replays it gives you don't work at all. I think it should accept (and output, why not) both formats.

Can you please add an open source license for this code?

If you have night light on (blue light filter), then the blue pieces become completely invisible.

Copy-pasting AI code with // comments, like the example code, into the prompt box works on Chrome, but on Firefox gives the error:

Error
expected expression, got '}'
Caught this exception while trying to evaluate your custom AI JavaScript.

After fiddling with it for a bit, I'm pretty sure this is because Chrome preserves newline characters in prompt input but Firefox doesn't, so everything after the // is just commented out, resulting in an empty program.

Suggestions:

Better error message for effectively empty program

Using a textarea to solicit AI programs, or making a note of this bug somewhere.

I've also filed this bug report against firefox and the mdn compatibility documentation, so we'll see what they say: https://bugzilla.mozilla.org/show_bug.cgi?id=1716229 mdn/browser-compat-data#10994

I've recently come across two papers that might be relevant to this project and/or https://github.com/qntm/tetris

The first purports to show an algorithm in which the machine can force you to lose a game of tetris:
Can you win at TETRIS? Brzustowski, John 1992

The second purports to show that a series of SZSZSZ... is enough, though it seems to be slower:
How to lose at Tetris. Burgiel, Heidi 1997

Just tried to copy a base2048 string into the replay box from an email, and since it had some initial space characters it silently failed. The web console logged Uncaught Error: Unrecognised Base2048 character:

This would make the game easier, just like the 7bag rule, but I think it'd be interesting for people looking into how hard could the modern tetris rules work. Would you like that?

https://tetris.fandom.com/wiki/Wall_kick

I was only getting the same piece over and over again. It sure spoils the game, but where's AI in that?

Right now, is quite easy to get into loops once you set up a specific pattern in the lowest row of the board. Loops with Os and Is are easy to fill plus leave a lot lf playroom to get set up other patterns. This makes getting tons of lines """"""""""""""""trivial"""""""""""""""""" and could risk an algorithm existing that maintains the tower in place. At the very least, I think we can do better to ensure the first line is much harder to get, or outright impossible

My idea would be to have a sort of look up to see which piece would leave more "blank squares" up. As to avoid lines AND filling the board easily.

This is not easy to do at all, but I think it'd be interesting to try. What are your opinions? I'd love to work in this too.

Think that is a sufficient condition to leave one square blank in a line for a line to not ever be possible in that place. I think it could be very interesting to play with that a little

I would like to work in a variation that shows the modern 7 piece bag rule. Is a simple change, but it would be interesting to people to see what happens in a modern tetris setting

Would that be ok? Or would you prefer to leave it as is and go with a fork?