Kata

:)

Actually I just get 'got 7 expected 2' without description of the exact test idea, could you provide what this test is about for me to come up with solution for possible bug in algorithm?

You can print function arguments to console.

Dude, your way is too long. Maybe you should find another way to solve this problem without changing the original array? :)

Hint: note that the ranks of adjacent knights are always different, and when Galahad comes, a same pattern happens

Sorry for the long wait (I had about 1 a.m. at the time and had to put it off until tomorrow)

Now fixed

Hi,

1. I'm fixing this problem for now.
2. The condition says that:There are knights gathered at the round table, and they prudently arranged so that the rows of their neighbors are different
   so there can be no knights with the same rank in this kata.
   I realized that there are no situations in kata where 1 or more knights remain, but I'm still working on that.

2: oh ok. I missed that.

I added tests with cases where only 1 knight survives.

Regarding the first problem, python & JS solutions can't handle arrays where more than 30-40% of all knights are involved in combat. Maybe I can put the number of battles higher than now, but in that case the required kyu to that kata will become higher than 5-6 (although I'm counting on that range, with your permission)

what's the exact problem when increasing the "battle ranges"? execution time varying too much from one run to the other? input generation? on the solution side, those which aren't creating copies of the array shouldn't suffer that much.

@Djacon for test generation, you might be able to generate "more interesting" tests, by working backwards. Example

[7, 5, 9] # Final result
[7, 4, 4, 9] # Randomly choose element to "split"
[7, 2, 2, 2, 4, 9] # Another random split, but into 3 elements this time
[7, 2, 1, 1, 2, 4, 9] # Final Split
[7, 2, 1, 2, 4, 9] # Remove "Galahad" noting his position -> 2 and value -> 1

Another benefit of this (if done correctly), is you already know the result, so you don't have to run a reference solution over it. (But you might want slightly more sophisticated logic than that, to also include other edge cases.)

Anyway, just an idea :) (May not be necessary at all)

@Blind4Basics, the problem is that when you increase the "combat range", the pass time increases greatly.
I've tested this on several tests, using various solutions from other participants. So I don't know how else to solve this issue.

@Kacarott, your solution is very interesting, but difficult to implement.
In this kata, I already know the correct answer from the beginning and don't calculate it :)

maybe, just decrease a bit the input sizes, so that you can enforce some "long travelling" in some tests. Or use that kind of input specifically in some medium size inputs and keep the big numbers tests as they are?

Ah ok, my mistake. I saw ans = knights(arr, p-b, x) and thought it must have been a reference solution. In that case all good, but I'd suggest to import the user solution directly then. No point of import solution if you arent going to use solution.

Finally I managed to optimize the generation of arrays and now we can run tests with 300k knights :)

I examined your code and saw that it returns the wrong value in some cases.

For example:

knights([914, 912, 911, 912, 914, 916, 930, 0, 17, 71, 30, 28, 81, 48, 107, 69, 109, 87, 98, 47, 931], 2, 911) # should return 16, but your func return 15!

Thanks for providing an example. Somehow, this scenario never happens at JS, but sometimes at Python. That's why I think you are generating differently in JS vs Python.

I think I found it, you have 0 as possible element entry in Python, and not in JS. Or maybe I just never got an example with a 0 in JS. Anyway, this is the edge case for me.

Amazingly, I finally found the cause of the error :)
Everything should work now, thanks for your help.