"The result of a combination cannot match with another tile on the same move" - this is not what happens in the example above. It's the opposite here, a tile is moving onto a combined tile. This is not allowed, but nowhere in the description does it state this.

From description
The result of a combination cannot match with another tile on the same move. In other words, 2222 => collapses to ##44 and not ###8

Also description says Your function can modify the initial board value as much as you like.... Of course you can, but you won't pass the tests.

It's not mentioned in the description that once two tiles are merged, no other tiles can be merged into that same tile within the same move. This only becomes apparent from test cases.

example move row to left:

row: [2, 0, 2, 4]
expected: [4, 4, 0, 0]

if multiple merges would be possible: [8, 0, 0, 0]

Good point about Set. I've prefixed the test constructor so there's no name conflict.

I'll also consider changing 'falsy' to null and undefined. Thanks for the feedback!

Set is a native class. You should not redefine it.

User-defined classes ( ie, Constructor functions ) should, by convention, have names starting with a capital letter.

Expecting "falsy" for null, undefined and NaN but not false, 0 or "" feels inconsistent. Easiest fix for that is possibly a different name, though I have trouble coming up with one that would naturally not include any of the others. Maybe you should just go with JS itself and expect "null" ( not the historically correct bug that is "object" ! ), "undefined" and "number".

Good catch. Fixed.

Some typos:

sting -> string
and array -> an array
and object -> an object

I have added functions into the mix. My apologies if your solution became invalidated.

I won't add any additional requirements to the kata from now on. (I'll only fix issues, if any come up)

I didn't think random tests would add much (if any) value to this kata so I didn't implement them initially, but I've added them in now that I see that a function with an array of strings passes the normal tests. The new test cases now test a randomized amount of assorted values from an array for each 'type', which should prevent any 'cheesing' like having the function return a set list of expected types.

I did forget the array test. Thanks, added it in.

Arrays are not tested.

No random tests.

Either the tests are lacking or that was added after the translation was made and it wasn't added to the tests.

surprising that this works, i thougt it cal also be written as "three fifty" as per the description?