test cases for nucleotide count should include handing of invalid sequences

[lcowell]

In some implementations I've seen of this exercise, if I were to pass in a sequence of ATCGLL, we see include L: 2 in the results. There is no test to indicate if that behaviour is invalid and it doesn't seem right. Are we in favour of adding a test case to address invalid input or does that feel like it's outside the scope of this exercise?

[IanWhitney]

I have noticed the same.

Currently there is not a standard, cross-track test suite for Nucleotide Count. It would probably be good to create one, so that all tracks could handle this edge case.

[IanWhitney]

(Oh, also, thanks for adding the issue! Much appreciated)

[petertseng]

List of tracks that implement this exercise is at http://x.exercism.io/problems/nucleotide-count. I have not checked all of them but I know that at least Haskell expects that invalid nucleotides result in an error ), would be interesting to see how many other tracks expect the same

[petertseng]

Info on what other tracks are doing:

xgo now tests for sequences containing invalid nucleotides and expects that the function under test signals an error for them, as of exercism/go#413