Making the cache private is a bizzare "extra challenge". I can't see how that provides any better functionality/encapsulation, if anything it just makes it harder to understand.

The calls to fibonacci(n-1) and fibonacci(n-2) will store the value in memo, no reason to do this multiple times.

n is either fib(n-1) and fib(n-2)

In this solution, only fibonacci(n) is stored in the memo. But since fibonacci(n - 1) and fibonacci(n - 2) are called, they can also be stored in the memo, so if the next test is fibonacci(n - 2) for example, the result will already be in the memo.

which is exactly what is done here... Or are you talking about the base cases?

You should store the fibonacci(n - 1) in memo[n - 1] and the fibonacci(n - 2) in memo[n - 2]...

The most "production-ready" solution, IMO. Really clean and efficient.

yes

Can you pls provide more data?
I see many successfull solutions on coffescript.

It may use powerful built-in features, but subsequences(s) runs in O(2^N) time, where N is the length of s. This makes this solution prohibitively slow for long strings (MUCH slower than a dynamic programming solution, which will have O(N^2) time complexity).

This is by far the most elegant solution. A+ for the loop comprehension.

CoffeeScript version simply does not work; appears to be missing tests on submit.

Very elegant solution. Couple of style suggestions: the for loop could be condensed to 'for i in source if i in caps' since the string form isn't needed, and 'True if b == [] else False' could be rewritten as 'return not b'.

EDIT: nevermind