While the current tests are comprehensive, the following could add even more robustness:

Empty String:

Test splitting an empty string: split("", ' ').
Expected result (strip = false): {""}.
Expected result (strip = true): {}.
Single Character Input:

split("a", ' ') should return {"a"}.
split("-", '-') should return {"", ""} for strip = false and {} for strip = true.
Buffer Size:

Verify behavior with long tokens exceeding BUFFER_SIZE to ensure correct truncation.
Example: split(std::string(70, 'a') + ",b", ',', false).
Non-Printable Delimiters:

Test with special characters like \t (tab) or \n (newline).

this is trying to convert a color from rgb in this format [23,42,111] to and hsv as [227,79.3,43.5] and those were the calculations to do that

Sorry but IMHO using namespace std; is a

Yes I know ! Just forked... the kumite

Absolutely not :-)

should I include "#include <fmt/format.h>" to the kata

THANKS FOR YOUR kUMITE SOLUTION BUT THIS KUMITE IS ALREADY AVAILABLE AS A KATA

just add
"using namespace std;"
after include statements

sure

maybe 6 or something

I have created this as a kata and I want you to review it.
https://www.codewars.com/kata/reviews/66bf56746cc0a210696a4149/groups/66bf56746cc0a210696a414b

i've never seen decltype before. looks cool

Can you reiew the pull request on github. To push the liscence and ReadMe.md.

The code works thanks. I was hopping how to approve this kumite to a kata. how should I rank it.

Add constraints on key_a and key_b to kata description. If key_a * 25 + key_b > 2147483647, naive calculation of f(x) as (key_a * x + key_b) mod 26 can cause overflow.