如果我正在测试 rspec 扩展，如何抑制作为测试的一部分失败的测试结果？

Question

我正在尝试为rspec的扩展编写规范。

这是我要测试的要点:

require 'rspec-let-and-after-extension'
RSpec.describe "let(...).and_after" do
  it 'is called if the `let` is invoked even if the example fails' do
    call_order = []
    RSpec.describe do
      let(:foo) { }.and_after { call_order << :and_after }
      it { foo; call_order << :example; raise 'failed!' }
    end.run
    expect(call_order).to eq [:example, :and_after]
  end
end

一个重要的行为是，如果运行示例失败，清理代码仍然运行。因此，我通过记录调用的顺序并从示例中引发一个异常来测试这一点。

问题是，当我运行它时，它将此块视为第二个示例，然后失败并出现错误:

.F
Failures:
  1)  
     Got 0 failures and 2 other errors:
     1.1) Failure/Error: it { foo; call_order << :example; raise 'failed!' }
          RuntimeError:
            failed!
          # ./spec/spec.rb:43:in `block (4 levels) in <top (required)>'
          # ./spec/spec.rb:44:in `block (2 levels) in <top (required)>'
     1.2) Failure/Error: it { foo; call_order << :example; raise 'failed!' }
          RuntimeError:
            failed!
          # ./spec/spec.rb:43:in `block (4 levels) in <top (required)>'
Finished in 0.00167 seconds (files took 0.08011 seconds to load)
2 examples, 1 failure
Failed examples:
rspec ./spec/spec.rb:43 #  

如您所见，输出确实有一个点，因此实际示例传递的是。但是有一个F，因为它已经看到了内部的例子，运行它，不出所料，一个失败了。

我如何使rspec不把这个嵌套的例子看作是它应该运行的例子之一，这样这个例子就完成了一个点?

(如果您想知道rspec开发人员自己对他们的测试做了什么，看起来他们使用的是cucumber。他们用黄瓜是因为他们也弄不明白吗?:))

Answer 1

您可以使用新的沙箱API(在3.2+中可用)。

RSpec.configure do |rspec|
  rspec.around do |ex|
    RSpec::Core::Sandbox.sandboxed do |config|
      # re-configure any configuration defined by your extension here
      # before allowing the example to run. The sandbox runs with a fresh
      # config instance, which means any configuration you have set in
      # `rspec-let-and-after-extension` will not apply while the example
      # is running.
      # config.extend MyExtensionModule
      ex.run
    end
  end
end