# Talk:99 questions/11 to 20

The prototype for repli in problem 15 is

```repli :: [a] -> Int -> [a]
```

Because the second parameter is the number of times to replicate, it discourages the use function composition. I mean that if you swapped the parameters you could write it pointfree:

```repli :: Int -> [a] -> [a]
repli n = concatMap (replicate n)
```

This would also match the way replicate is defined:

```replicate :: Int -> a -> [a]
```

So, I suggest modifying problem 15 by swapping the parameters to repli in the example and the solution.

```slice xs i j = [xs!!(i-1)..xs!!(j-1)]
```

Counter-example:

```slice [1,3,6,3,1,6,7,8,3,2,4,76,8] 4 5 == []

Thanks to pixel for pointing this out.
```

The solution to problem 20 seems to be using 0-based indexing, whereas the question called for 1-based indexing in the other languages. This can be easily fixed:

```removeAt :: Int -> [a] -> (a, [a])
removeAt k l = (elementAt l k, take (k-1) l ++ drop k l)
```

using elementAt from a previous problem.

or if you want to express that 1-based indexing is silly,

```removeAt n+1 xs = (xs!!n,take n xs ++ drop (n+1) xs)
```

I fixed problem with inconsistent indexing between examples for problem 20. I think something need to be done for one of the solutions for problem 19:

```rotate xs n = take (length xs) \$ drop (length xs + n) \$ cycle xs
```

It works incorrectly for negative n < length xs. But it's example how to implement rotate "without mod" and I have no idea how to easily fix it to keep this property. Maybe it should be changes to use mod or removed at all? Actually all examples 'without mod' shares the same problem.