I am trying to integrate Ruby's Test::Unit with SimpleTest. As a result I am having to dive into the Test::Unit code. My problem is that I cannot find any decent tutorials on say writing your own Runner.

On poking into the innards I find the console runner is straightforward enough, but it seems to interact with both a RunnerMediator and the Result (a failure collection). I cannot quite discern the separation of concerns within these classes or any of the design decisions behind them. I assume there are good reasons for this complexity (either JUnit compatibility of GUI needs), so if anyone could shed some light, it would be a great help.

My exact problems are:
1) I need to find out the number of test cases without actually running them. The Tk runner seems to be able to figure this out, so I am exploring that thread of execution.
2) I need to output the results as XML. I did a copy and paste of the Console runner and manage to get the tests, but with no grouping structure.

I have contacted the author, Nathaniel Talbott, but am still waiting a reply.

yours, Marcus