Fixup failing travis on master

[richardc]

Master is currently failing on travis:

• The latest release of rubocop adds new style checks, so the rubocop target was failing
• The recipe for removing other versions of bundler was not working under RVM 2.5.0, leading to two installs of bundler.

Here we:

• pins rubocop to the last known compatible release based on travis logs
• Adopts an approach from https://github.com/travis-ci/travis-ci/issues/8717#issuecomment-366500378 to 'uninstall' bundler from RVM 2.5.0

[richardc]

So far just a pin to rubocop.

Issue with the bundle exec doesn't happen locally, will be pushing various investigation commits to see what travis environment is doing

[richardc]

Regression tests are running in all cells, but only failing in the 2.5.0 rvm. Hmm.

[richardc]

To surface the error:

  1) regression test with cookbook users should match expected output
     Failure/Error: command("git clone -q https://github.com/chef-cookbooks/#{name}.git .")

     Mixlib::ShellOut::ShellCommandFailed:
       Expected process to exit with [0], but received '1'
       ---- Begin output of bundle exec git clone -q https://github.com/chef-cookbooks/users.git . ----
       STDOUT: 
       STDERR: /home/travis/.rvm/gems/ruby-2.5.0/bin/bundle:23:in `load': cannot load such file -- /home/travis/.rvm/rubies/ruby-2.5.0/lib/ruby/gems/2.5.0/gems/bundler-1.16.1/exe/bundle (LoadError)
        from /home/travis/.rvm/gems/ruby-2.5.0/bin/bundle:23:in `<main>'
        from /home/travis/.rvm/gems/ruby-2.5.0/bin/ruby_executable_hooks:15:in `eval'
        from /home/travis/.rvm/gems/ruby-2.5.0/bin/ruby_executable_hooks:15:in `<main>'
       ---- End output of bundle exec git clone -q https://github.com/chef-cookbooks/users.git . ----
       Ran bundle exec git clone -q https://github.com/chef-cookbooks/users.git . returned 1
     # ./vendor/bundle/ruby/2.5.0/gems/mixlib-shellout-2.3.2/lib/mixlib/shellout.rb:293:in `invalid!'
     # ./vendor/bundle/ruby/2.5.0/gems/mixlib-shellout-2.3.2/lib/mixlib/shellout.rb:280:in `error!'
     # ./vendor/bundle/ruby/2.5.0/gems/rspec-command-1.0.3/lib/rspec_command.rb:111:in `block in command'
     # ./vendor/bundle/ruby/2.5.0/gems/rspec-command-1.0.3/lib/rspec_command.rb:108:in `tap'
     # ./vendor/bundle/ruby/2.5.0/gems/rspec-command-1.0.3/lib/rspec_command.rb:108:in `command'
     # ./spec/regression/regression_spec.rb:11:in `block (4 levels) in <top (required)>'
     # ./vendor/bundle/ruby/2.5.0/gems/rspec-command-1.0.3/lib/rspec_command.rb:47:in `block (2 levels) in <module:RSpecCommand>'
     # ./vendor/bundle/ruby/2.5.0/gems/rspec-command-1.0.3/lib/rspec_command.rb:45:in `block in <module:RSpecCommand>'

So that's the bundle exec git clone ... failing to find exe/bundle from the gem path.

[richardc]

https://travis-ci.org/Foodcritic/foodcritic/jobs/350259777#L491-L492

$ rvm @global do gem uninstall bundler -a -x -I || true
ERROR:  While executing gem ... (Gem::InstallError)
    gem "bundler" cannot be uninstalled because it is a default gem

Looks like the uninstall doesn't occur, so we end up with multiple installed/incorrectly activated bundler.

[coderanger]

Deleting a default gem is not usually a safe thing to do. Why do you think we should do is here?

[richardc]

@coderanger because otherwise you end up with two versions of bundler installed, which breaks as described in earlier comments https://github.com/Foodcritic/foodcritic/pull/742#issuecomment-371101183

If you have a fix for the bug in a better form I'm happy to defer to your fix

[coderanger]

Does that still happen with an updated rubygems (i.e. uncomment the gem update --system)?

[richardc]

Let's see - 9e6ff1e

[tas50]

That looks to fix the issue. Can you squash this all down to a single commit. Also instead of pinning rubocop you can just set chefstyle = 0.5.0 which should do it. I'll get a PR open to cleanup the warnings there later. Thanks for taking this one though. It's been on my todo list since it started.