graeme fawcettThis one's neat.
Boundary based development means setup the potential, burn it across a computational boundary, compare the results to your expectations.
i.e test a stateless middleware bit by giving it some context and checking what it returns
I like to see a richer, more artisanal shape to my expects - if the shape looks right, the code probably is too.
Template the tests from set fixtures, then the inputs and outputs can reference the same thing. Keep the tests DRY /and/ useful as documentation
Extend the CLI test runner with a render function to verify and we're off
[now to make the tests pass i guess]
