Distributed, Parallel testing with Rails

Question

I've been recently playing around with distributing our testing load across multiple machines (and multiple cores) over the network.

I've tried out:

SpecJour http://github.com/sandro/specjour

• This worked pretty well and was really easy to set up. Reduced out testing time by about 30% across 4 machines (8 cores)

However, I really like autospec's / autotest's ability to re-run tests on file system change and also to only run failing tests until they have been passed.

My Question: How do you implement autospec / autotest like behaviour when using something like specjour? (I know there's stuff like watchr which can watch your file system for changes but how do you run only relevant tests again?)

FYI I'm using Rails 2.3.8 and testing with RSpec and Cucumber