Set up for Rails and non-Rails specs playing nicely

running.md

You can then execute all specs with:

bundle exec rspec spec

And only run your decoupled tests with:

bundle exec rspec spec --tag ~rails

some_class_spec.rb

require 'spec_helper'

# Note that a symbol is used because the class will not be loaded when run in non-rails mode
# but the spec will still be loaded (but not executed)
describe :SomeClass :rails do
  # this spec needs Rails
end

some_other_class_spec.rb

require 'spec_helper'
require 'some_other_class'
# You should require other dependencies here. This also has the benefit of exposing 
# your dependencies - think about them.

describe SomeOtherClass do
  # this spec does not need Rails
end

spec_helper.rb

Rspec.configure do |config|
  config.treat_symbols_as_metadata_keys_with_true_values = true
  
  if config.filter_manager.exclusions.include?(:rails)
    # Rails not wanted here
    config.filter_run_excluding :rails => true

    # since Rails isn't loaded, you need to include the paths that matter to you for requires
    $:.unshift('./app')

    # more setup for non-Rails tests
  else
    # Rails wanted here
    require File.expand_path("../../config/environment", __FILE__)
    require 'rspec/rails'
  end

  # The following will allow you to use a symbol and still have subject work correctly.
  config.shared_context :rails => :true do
    subject do
      Kernel.const_get(example.metadata[:example_group][:description_args].first).new
    end
  end
end

Here's what I've come up with so far in terms of keeping tests in line.

Reorganize spec helpers

I'm not done yet, it evolves as my project does. I'm using it in a couple places, but I haven't used any meaningful tags yet.
I have some ideas for some.

I can't edit or amend my previous comment.

Here's my (spec/)[https://github.com/Velocitous/blog/tree/master/spec] directory. Mayhaps give me some feedback on the helpers?