the same problem now :(

Fixed.

Added to all languages

It's a case that makes you timeout if you go the "easy way".
What is "easy way" and what is "hard way"? You have to figure it out yourself.

I couldn't either, because I came up with this solution by staring at the output and pattern-matching the first 50x50 result.

I suppose someone better than me can explain why this works ;-)

Details:

sort mutates the array and returns the array itself, so it's literally just comparing the same thing ;-)

Added a test on that. Now it's invalidated ;-)

Thanks

Thanks for the heads up. I didn't include diacritics in the alphabet so I shouldn't have included them in the key. I'll update the key to not say hiragana (the name of the writing system), and instead say hirakana (may or may not have meaning).

I added that to the description. I also fixed where I misspelled reciprocal.

Wow, I didn't notice that. You are correct, the operation performed is identical no matter what order.

Updated description. About three pairs it's stated in examples.

Hmm, looking through some of these solutions I think most of them would do the same if passed an empty string or a string made up completely of invalid characters.