Question

I have a controller index test that gets an ActiveRecord scope.
The test currently looks like this (including some inline debugging stuff):

describe "GET index" do
  it "assigns all schools as @schools" do
    get :index
    puts assigns(:schools).class
    puts School.populated.class
    assigns(:schools).should == School.populated
  end
end

The output is this:

ActiveRecord::Relation
ActiveRecord::Relation

expected: []
     got: [] (using ==)
Diff:

This is definitely not the first time I’ve had this in recent versions of Rails and rSpec.
Previously a coworker would just wrap the items in a to_a to compare them, which I find a bit dirty and likely not a good solution.

Any ideas? I’m curious as to why it thinks they are different, and how this same test passed in older versions of Rails and/or rSpec.

Answer 1

eql is the same as ==. Author of Rspec talks only do not use !=, use should_not instead

actual.should == expected
#is interpreted as this:

actual.should.==(expected)

#This is not true for !=. Ruby interprets this: actual.should != expected
#as follows:

!(actual.should.==(expected))

Update: Relation provides Lazy Load pattern, so you do not have any executed query on the step. It means the fire the query on first request