Functional Programming: A PragPub Anthology

One Experiment, Several Measurements

We can do better, though. We think the setup block should run an experiment, and each test block should run measurements. Combining setup code with test code makes it hard to isolate single-purpose code. The experiments and measurement concerns can become too entangled.

Let’s rewrite that:

defmodule AppTest do
  use ExUnit.Case

  setup( context ) do
    context
    |> Map.put( :setup, :done)
  end

  test "setup works", context do
    assert context.setup == :done
  end
end

That’s a little better. We have a setup block that can run an experiment and test code that can run measurement. That’s a potential improvement, but since there’s typically only a single setup, the improvements ...