As said before I am writing a testing framework for Clojure. It started as a fork of the Fact framework but eventually I decided to try some ideas that arose in a conversation with Fábio and Rob: use RFC-2119 semantics to describe tests. RTFSpec is my take on this approach.

Here is an example from the (pretty inexistent) documentation:

 (use 'rtfspec)

(spec "List Comprehensions"
      (must "return a list applying the given function"
	    (= (list 11 12 13 14 15)
	       (for [a (list 1 2 3 4 5)] (+ 10 a))))

      (must "include elements that are not matched by the :while clause"
		(empty?
		   (for [a (list 1 2 3 4 5) :when (= a (+ 10 a))] a))))

And the output using ant:

[rtfspec] ..
[rtfspec]  2 Tests ( 2 Successful 0 Failed 0 Pending 0 Exceptions )

I consider the tool far from done but it is already useful. The most important thing missing for me is a matcher structure that will not require you to repeat (= a b) over and over. Also that would be useful to actually print something like “Expected [a] got [b]” as decent testing tools do.

Other thing that I really miss is a way to test a function against a domain represented by a set. During development I realised that just porting a tool like RSpec to Clojure isn’t enough, the needs are different when you use different paradigms.

In both cases the implementation isn’t hard but I can’t think of a nice syntax.

Feedback and patches are always welcome.