That is why I have created my own library for Elixir. Well, main reason was that I wanted to support “new” step definition syntax, which IMHO is much more readable, but additionaly it fixes some of the problems mentioned in the article (obviously it cannot solve all of them, as some are culture-related):

• it supports importing step definitions from other modules
• it reuses the same test framework that you probably use for other tests, actually nothing prevents you from having both kind of tests in the same file
• it forces you to write steps using just plain old Elixir, no steps defined with steps
• no need for regexes, you can use “natural-like” step definitions, for example "there are {int} coffee(s) left in the machine", which allow any decimal integer in place of {int}. There is also a way to define custom placeholders.