Forum: Ruby Cannot Use Initialize With Minitest?

Hey folks,

I've been trying my best to write a class that uses initialize() with
minitest, so that I can swap out a specific dependency at runtime. It
appears, though, that I cannot use initialize() in a class that inherits
from minitest:

#####
require "minitest/autorun"

#A simple dependency which is being dispatched to.
class Dependency
  def initialize(argument)
    $stderr.puts argument
  end
end

#The main test class.
class TestProblemReproduction < MiniTest::Unit::TestCase
  def initialize(test_argument)
    Dependency.new(test_argument)
  end

  def test_something
    puts "This is here so minitest does something."
  end
end


TestProblemReproduction.new("stuff")

#######

Expected Output is to see the class Dependency output "stuff".
The actual output is that the class Dependency outputs
"stuff\ntest_something"


However, I have a use case where I would like to use minitest to run my
test cases, but I have a bunch of tests written in selenium. With one 
flag,
the selenium browser will be constructed to run locally on my desktop, 
for
local testing. With a different flag, the selenium browser is 
constructed
to run remotely on a set of remote test servers.

Is there any way I can hand a dependency into my minitest based test 
cases?
Or do they have to be static?

Thanks!

Jams

Hi folks,

as an addendum, this problem is preset for me because I'm using
minitest/autorun. I wanted to use this because of the various hooks that
are provided (def setup() etc); if you can think of a way to provide the
same pre-test hooks without being tied in to autorun, I could use that
instead.

Jams

On Mar 7, 2013, at 15:02 , James Harrison <jam@jamandbees.net> wrote:

>     $stderr.puts argument
>     puts "This is here so minitest does something."
>   end
> end
>
>
> TestProblemReproduction.new("stuff")
>
> #######
>
> Expected Output is to see the class Dependency output "stuff".
> The actual output is that the class Dependency outputs "stuff\ntest_something"

The #1 rule of OO is if you override a method you NEED to call super.
The #2 rule of OO is if you override a method you NEED to make sure it 
conforms to the contract.

You're breaking both these rules. #initialize is defined on TestCase, 
and it takes an arg which is the name of the test being run.

> However, I have a use case where I would like to use minitest to run my test 
cases, but I have a bunch of tests written in selenium. With one flag, the 
selenium browser will be constructed to run locally on my desktop, for local 
testing. With a different flag, the selenium browser is constructed to run 
remotely on a set of remote test servers.
>
> Is there any way I can hand a dependency into my minitest based test cases? Or 
do they have to be static?

There is nothing static about minitest. I don't see why you need to do 
this via initialize/new tho, so I'm guessing I don't understand the 
problem well enough.

You shouldn't be instantiating TestCase instances. That's the runner's 
job.

On Mar 7, 2013, at 15:22 , James Harrison <jam@jamandbees.net> wrote:

> as an addendum, this problem is preset for me because I'm using 
minitest/autorun. I wanted to use this because of the various hooks that are 
provided (def setup() etc); if you can think of a way to provide the same pre-test 
hooks without being tied in to autorun, I could use that instead.

autorun is orthogonal to TestCase, so again... I don't understand the 
problem well enough to help.