implements a ring buffer so memory usage is proportional to length of signature not n

if you wrote your function using the sum of coins in a certain orientation mod size then it will work for some cases but break in others. is this what you have done?

my point is that it doesn't matter if the board is square. this problem is impossible for board sizes that are not powers of 2 but if it is a power of 2 it doesn't make any difference weather it is 1x64,2x32,4x16,8x8. my code on all powers of two, including odd ones.

I might do that, but any function that works in that case should work in any 2^n case. I will try and reword it to avoid confusion

i will try and make some more helpful Fail messages.

the answer to that question is firstly that all generated boards at the moment are perfect squares, with sides of 2*n.
also if wouldn't make any difference because the input is a list there is no functional difference between any of the combinations you

Thanks! I removed any seeding in the real tests and have set the board/key combinations to be tested in advance as well as randomly varying the number of tests at each size. Do you think it is better now?

coin_flipper is defined in the preloaded code, it should only be used for testing. i don't understand why it should be in the initial solution or why you removed coin_finder in your

thanks, it should be working