Actually cool kata. Required me to think a lot and to debug my code a few times.

Why was the kata still on Node 8?

it need not be a consideration. Node 8 is only supported for legacy reasons. if you use padStart(), the runner will take notice that the kata no longer compiles with Node 8 and disable that version.

(besides, Node 8 does support padStart())

This kata allows Node v8.0 which doesn't support padStart

padStart is available natively. Why reinvent the wheel?

I've created a fork with a fix

(The user attempted in JavaScript.)

In the assumptions:

• both numbers in the pair will be of the same length
• If a number is shorter, it will be zero-padded.

The actual input:

Input:
537302 92264

Something is wrong I guess

Enabled in this fork

Is anyone else failing the final speed test? Mine keeps timing out and I implemented memoization correctly. Even did a classic dynamic programming one with a loop to reduce time from overhead function calls

Hi! Could someone please explain. Does kata consider the cache to be shared across multiple fibonacci(n) calls or not?

To make things easy start solving this Kata from .from_array method)

Heh - I forgot the trick where if you set a default value in the function signature, the instance persists.

Thanks for the pointer

How big are the numbers you are testing locally? In the tests inputs go up to 1000. When I run your memoized function (after manually increasing the recursion limit), it takes a good 20 seconds or so. You need a better algorithm (Hint: memoize results between function calls too, not just for a single function call)

Note: Using Python

Testing my algorithm offline, I can complete a 100000 test run in 1.15 seconds. In the codewars environment, this times out after 12 seconds, meaning that I cannot submit my solution.

Annoying...