Question

While running code in production, I don’t have to interact with the Ember.js runloop but in my tests, I sometimes have to call Enber.run.end() to make the tests pass. This might have knock on effects of making other tests fail.

For example, here is my spec_helper:

beforeEach ->
  $('#jasmine_content').append('<div id="fieldset"></div>')
  Ember.test = true
  Lead.run()

$ ->
  $('body').append($('<div id="jasmine_content"></div>'))

I am setting the Ember.test flag to true in the above code which I am not entirely sure what this does.

Lead.run() creates the Ember application:

window.Lead = Ember.Application.create
  Controllers: {Leads: {}}
  Models: {}
  Views: {}

  run: ->
    @initEvents()
    @.set('search', Lead.Controllers.UrlSearch.create())
    # etc.

In some tests, I find myself having to call ember.run.end() to ensure that items will be in the DOM etc.

describe 'Controllers', ->
  describe 'UrlSearch', ->
    it 'should append view', ->
      Ember.run.end() 
      expect($('#goButton').length).toEqual(1)

Ending the runloop from individual tests feels wrong.

Whenever I update my spec_helper to this:

beforeEach ->
  $('#jasmine_content').append('<div id="fieldset"></div>')
  Ember.test = true
  Lead.run()
  Ember.run.end()

I get error messages like:

Cannot perform operations on a Metamorph that is not in the DOM

or

Must have a current run loop.

Are there any guidance or best practices for negotiating the runloop from tests?

Answer 1

You should take a look at the tests, for example each_test.js. As you can see, adding the view to DOM or manipulating properties of controllers, … is always done within Ember.run to assure all changes and view updates are flushed and the bindings are synced:

Ember.run(function(){
  view.append();
  ...
});