set_flash[:foo].now matcher does not care if flash[:foo] really set .now

[aried3r]

I want to assert that a flash.now[:foo] uses .now, but the test passes, even if .now is not set.

Code:

# controller.rb, simplified, authorisation, authentication etc left out
def create
  @project = Project.new(project_params)
  if @project.save
    flash[:success] = 'Project successfully created'
    redirect_to projects_path
  else
    flash[:alert] = 'Project could not be created.'
    render 'new'
  end
end

# spec
it 'sets a flash.now' do
  post :create, invalid_params
  is_expected.to set_flash[:alert].to('Project could not be created.').now
  # passes!
end

The reverse is not true, btw. If flash.now[:foo] is set in the controller but not in the spec, the test fails.

Rails 4.2.3
Ruby 2.2.2
shoulda_matchers 2.8.0

[maurogeorge]

Hi @aried3r thanks for report.

I worked at the master, that will be the 3.0.0, I am not sure if we will support 2.x after the 3.0.0 is released. But on master the problem was that when you call the now qualifier after defining call the [] and to is instantiated a new matcher that does not know about the previous values, take a look at the PR for a detailed explanation.

So maybe the following code works on 2.8.0, I know that this works on 3.0.0 even before the PR.

is_expected.to set_flash.now[:alert].to('Project could not be created.')

@mcmire what you think about add this on the 3.0.0 milestone?

[aried3r]

So maybe the following code works on 2.8.0

It doesn't for me, sadly. I currently don't have the time to test this with the 3.0.0-rc version (with all the changes going on there), but I'd be glad to help you guys out next week. This gem helped me so much, I'd be glad to be of help.

[mcmire]

This might be an issue with 3.0, too, actually. I'm not going to let it hold up the release, though, as it's been in development forever and I'd rather get it out sooner rather than later. So I'll include a fix in the next release, but a PR is certainly still welcome.

[mcmire]

Closing this in favor of #755.