Luhn tests missing coverage for common mistake (right vs left)

[michaelorr]

From the README:
The first step of the Luhn algorithm is to double every second digit, starting from the right. ...

This distinction about right versus left isn't super obvious at first glance. The first example contains odd number of total digits, so every 2nd digit from the right is the same as every 2nd digit from the left.
When I got to the later examples, I thought that the README had a typo.
My implementation doubled every 2nd digit from the left (incorrect) but the test suite did not catch my mistake.

I was working on a PR to update what I thought to be a typo in the readme when I realized my mistake. I think the readme is clear enough, it was my mistake in reading it incorrectly.
I think a new test case should be added where modifying every 2nd digit from the right is a valid Luhn number as expected but modifying every 2nd digit from the left is not a valid Luhn number. Devising such a number should not be too difficult.

[michaelorr]

Hopefully I'll have a little spare time this evening to add this test case.

[petertseng]

Excellent explanation, thanks. We added the case "59", with an even number of digits, in exercism/problem-specifications#522. We want to add it to this track soon, depending on what happens in exercism/problem-specifications#523. The work for this track is in progress at #479 so follow along if you are interested.

[stkent]

Hopefully exercism/problem-specifications#549, now merged, will also help with this!

[michaelorr]

@stkent I'll take a look today. Thank you!

[michaelorr]

exercism/problem-specifications#549 absolutely helps clarify the readme, thanks!
And #479 adds the test case 59 which would have caught my bug.
It looks like #479 is being held up by exercism/problem-specifications#523 and exercism/problem-specifications#547. Is there anything I could do to help move those along? 547 looks like it may be ready for merge.

[stkent]

As far as I'm concerned, 547 is indeed good to merge. If/when 523 is resolved one way or the other, we'd still want the extra positive tests added in 547. If you are willing to review and/or merge 547, I'd appreciate it!

[petertseng]

Closed by #479