I wrote a helper that helps one debug `rand`-based test failures by running `srand` with the same number before every test. That way, if a test fails in a full run due to a `rand` issue, by using the same `--seed`, it should fail whether you run just the file or the individual test.

test_helper.rb

class ConsistentRandomness
  def self.setup_rng
    if !@seed
      @seed = srand(1)
      puts "Using seed #{@seed} for every test"
    end
    srand(@seed)
  end
end

class ActiveSupport::TestCase
  before do
    ConsistentRandomness.setup_rng
  end
end