I understand. In truth what I'd hoped to achieve was to emulate something I do often in my day-to-day job - take existing code (hence the initial solution) which works, but isn't fast enough, and find ways to improve it - which, in my opinion, is somewhere between optimization and bug fixing. In my case I've been getting a lot of mileage out of parallel processing lately, and I'd hoped to see how others might approach the same problem.

Including the import statement was a mistake, I meant to remove it when stripping out the 'optimized' code - whoops.

I've removed all of the code other than the signature from the initial code; I've left the hashing function in the preloaded code should anyone wish to use it. It isn't exactly how I'd initially envisioned it, but if it proves to be a better kata for it, then so be it.

Update: Actually, looking at it now... it's basically the same Kata as the original now. I can't remove it, but I won't publish it either.

I'm not sure that I agree that this kata spoils anything from the original kata; I took care to avoid exposing the source code of anything that could solve the original Kata in the preloaded code which is not visible to testees; and in addition to that I recommended that anyone seeking to try this kata should try the original first.

That said - do you have any suggestions ? I'm open to ideas.

C#: The random tests here seem to be incorrect; they generate strings with one or more non-repeating characters, and then fail because they expect there to be no non-repeating characters, e.g

String generated for test: ˆ†xLm…eHLph|aRsyYlS}c…}bETNnJv\max

Expected output of test: String.Empty

Actual result: p