Which Ruby boilerplate is more preferable?

[ejini6969]

Regarding the Ruby 3.0 update, here are two sets of boilerplates, which one do you prefer more (1 or 2) and why? (1 will only display one output message when passed, 2 will display all test cases' corresponding messages when passed but looks more similar to the old test framework)

Please give your humble opinion, regards! ~~

[kazk]

Thanks for starting the discussion. As I replied on Gitter, I prefer the first one. I don't think we need to see the passed cases like the deprecated custom test framework. Also, multiple feedback on #110 seems to agree with that.

/cc @codewars/community

---

I wouldn't do describe "random tests" with single it, though.
I prefer something like this:

describe "[method/kata/solution]" do
  it "[short description of what's being tested]" do
  end
  # can have multiple of these if it makes sense

  # or, something like
  it "should pass example cases" do
  end

  # then a random test case
  it "should pass [n] random tests" do
  end
end

[ejini6969]

So two describe blocks, one for fixed tests and the other for random tests are not as compatible with the new framework as the one describe block, multiple its within right? (something like ^^)

[kazk]

? Tests having multiple describe should work as expected (is compatible). I just think it's weird, and unnecessary. It's weird to "describe" a test, rather than the subject. It's unnecessary to group a single test case.

But that's just my opinion, and RSpec doesn't care the way you organize the spec.

[ejini6969]

Okay I go with the reasonable way :)

[JohanWiltink]

I prefer (1). It's cleaner.

Incidentally, this is the only option for random tests with Haskell QuickCheck, so it's the option I am used to. Showing inputs can be done in an assert message on failure.