I'm just another dude who got confused by the title and the description. I think the kata description should have '-1 -2 -3' as an example, same for '1 1 1'

I added a sentence with your suggestion to the details:

"If all the words in the given array mesh together, then your code should return the meshed letters in a string. You should return the longest common substring. You won't know how many letters the meshed words have in common, but it will be at least one."

Shouldn't the description say, in a less subtle way, that they want the longest common substring in the response? Because I think that's the case if there are more than one common substring between two adjacent words, right?

My code (Python) is passing all the test cases. But it also shows a stderr. So, I can't submit. What could be wrong?

Thanks for that, I've raised an issue about it.

There is an issue with the Haskell. In the case of multiple solutions, the tests expect the last one that appears, which is inconsistent with the description and other languages.

I'd greatly appreciate it if someone proficient in the language could fix the issue :)

Well I solved it but, in the case of Haskell, the test suite is wrong. The description says that in case of multiple solutions I should return the one that appears first in the string. But that's not the case in the test suite for instance:

longest "abcdeapbcdef" shouldBe "bcdef"

Is clearly wrong. "abcde" is as long as "bcdef" is, and it certainly comes first. I think having a different kata description for this language only will spare some trouble to future coders.