Upgrade to RSpec 3 and major specs refactoring

[mdesantis]

Hello,

I'm planning to work a bit on rack-protection performance, so I cloned the project and I found the specs a bit cluttered; so I upgraded to RSpec 3 and did a cleanup.

These are my edits:

• 022f994: Convert RSpec syntax from RSpec 2 to RSpec 3 using the wonderful tool Transpec
• 2674740: Upgrade development dependency from RSpec 2 to RSpec 3
• de797ce: Move spec files from a flat spec/*_spec.rb pattern to the conventional pattern lib/folder/file.rb -> spec/lib/folder/file_spec.rb
• 01902df: Remove some define_methods on Struct.new classes in favour of Struct.new do; def ... ; end syntax
• 941fee8: Remove execution permissions to some spec files
• ed57eb2: Fix Ruby warnings visible using rspec --warning
• ecae614: Fix a test which implementation was previously broken (more on this later)
• 8196555: Remove unnedeed require 'stringio' line
• c5020a2: Reorganize spec directives which previously were collected in spec_helper.rb into relative spec/support/*.rb files
• a00a021: Generate rspec init default files with rspec --init and merge the previous ones into them
• 513abea: Remove require 'spec_helper' not needed anymore (because .rspec now includes --require spec_helper option)

The only change which involves a change to a test is ecae614: I replaced the previous incorrect use of should inside a define_method with a mock, and then realized that the test was broken, because the changer middleware was called after the detector; so I switched them and fixed the test.

I hope my changes to be appreciated :)

[zzak]

zomg

[zzak]

All of the spec changes look good, but could you explain the spec_help and spec/support stuff?
It's not obvious by looking at the diff

[mdesantis]

c5020a2 actually consists of two edits (I noticed it now, it would be better if I made two commits for them):

1. Rename TestHelpers into SpecHelpers: I made this because TestHelpers would be a good name for MiniTest, etc., instead SpecHelpers better suites for RSpec I think. I'm noticing that RSpec itself uses RSpecHelpers, maybe we can change to RSpecHelpers.

2. Move RSpec related support code to spec/support folder: usually inside spec_helper.rb you have

   1. RSpec configuration (the RSpec.configure { |config| ... } block)

   2. Various code used to assist specs (spec helpers, monkey patches, shared examples...):

      stdlib and gems requires...
      
      Rack monkey patch code...
      
      DummyApp code...
      
      TestHelpers code...
      
      NotImplementedAsPending code...
      
      Rspec.configure { |config| ... }
      
      shared_examples code...

   I moved the support source code into respective spec/support/{feature_name}.rb files:

   • Rack monkey patches
   • DummyApp
   • TestHelpers (now SpecHelpers)
   • NotImplementedAsPending
   • shared examples

   So that spec_helper.rb now is:

   stdlib and gems requires...
   
   require 'support/rack_monkey_patches'
   require 'support/dummy_app'
   require 'support/spec_helpers'
   require 'support/not_implemented_as_pending'
   require 'support/shared_examples'
   
   Rspec.configure { |config| ... }

   But instead of write the various require 'support/...' I wrote

   Dir[File.expand_path('../support/**/*.rb', __FILE__)].each { |f| require f }

   which is similar to what is used by RSpec and similar to how RSpec sets Rails projects.

   I chose the support folder name because it's a RSpec convention: RSpec uses it for itself and RSpec Rails sets Rails projects to use support folder.

[mdesantis]

@zzak news about this pull request?

[mdesantis]

News about this pull request? It seems to be abandoned... @rkh can I please ask your attention?

[kytrinyx]

@mdesantis I'm going to spend some time looking at rack-protection this weekend, and will figure out if there's something that needs to be done to get this merged.

There are 26 failing tests against jruby-head, with undefined method 'call' for #<#<Class:0x7752d052>:0x3468e86e>, but jruby-stable is passing. We'll need to figure out if it's a regression in jruby and report it there, or if there's some odd edge case that we are running into here.

[mdesantis]

Just curious... why do you test against heads? Are they supposed to be stable?

[kytrinyx]

They're not supposed to be stable, but it's a good way of catching new issues early (and help catch regressions in the language implementations)

[kytrinyx]

I got food poisoning, which screwed up my weekend plans pretty significantly. I haven't been able to get to any of what I had planned, unfortunately, but this is still at the top of my list.

Sorry about that.

[mdesantis]

I got food poisoning on last Friday too :-S no problem, thank you for the
update

Maurizio De Santis

2015-03-30 4:55 GMT+02:00 Katrina Owen notifications@github.com:

I got food poisoning, which screwed up my weekend plans pretty
significantly. I haven't been able to get to any of what I had planned,
unfortunately, but this is still at the top of my list.

Sorry about that.

—
Reply to this email directly or view it on GitHub
#87 (comment)
.

[kytrinyx]

I reran this against jruby-head, and it passed just fine. Not sure what happened back in the day. Thank you so much, @mdesantis, for taking the time to upgrade the specs, and I apologize for the delay in getting this merged.