Comments for dchelimsky's rspec-rails  

These are cool; I've had essentially the same thing written for quite a while that I use on an API-only application where cucumber just didn't gel well. But must this load webrat? Plain old get/post/put/response.should... often suffice.

...and I helped!

Thanks David!

Sorry for the links in my previous comment, the formatting used the closing parentheses for the url. The links are: http://gist.github.com/54177 http://rubygems.org/read/chapter/3

I think it will be easier to use a more standard solution here using RUBYOPT env var and not this odd named var. Something like if !ENV[‘RUBYOPT’].include?(“rubygems”). The use of RUBYOPT is encouraged in the RubyGems manual (http://rubygems.org/read/chapter/3) and in a recent article by Ryan Tomayko (http://gist.github.com/54177) about the use of require ‘rubygems’

Shame to have all this duplication, but these scripts are figuring out where to load from. Catch 22.

Why does rspec.rake now delete other dependencies on :default? This stops keeps the default task from running :features too. Obviously I can delete it in my own rspec.rake, but I want to make sure there isn’t an important reason for it.

David, I’m getting dependency errors now that the ‘dchelimsky_’ prefix is gone. This has happened in the past – is there a way round that works? I have installed with the —force option, but the dependency is now raised when I run my specs…

Thanks, Josh

By the way, I’m thinking of this commit: http://github.com/dchelimsky/rspec-rails/commit/cf6ee0de90ebb2b33531d0a8f58846bcadfc2751

Furthermore, I should note that setting to false, appears to be not only unnecessary now, but it also breaks some class attributes for ActiveRecord (the default value for @@default_scope wasn’t getting set) which were causing failures.

Doesn’t this change make the warning in spec_server about setting config.cache_classes to false unnecessary?

I believe this needs to be Rack::Utils.parse_nested_query(query) instead

Hi, commit comment should be: “warn that spec_server won’t reload classes if config.cache_classes=TRUE”, right?

I think this has been fixed in now.

I pasted this latest version of the rake task into my rails project as I’ve found this problem with rspec-rails not being installed when running rake but it still fails due to environment trying to load in the rescue block and failing on the config.gem ‘rspec-rails’ require. As I’m trying to do rake gems:install this still seems to be a bit of a catch-22. I just replaced the rescue block with puts ‘RSpec not installed. Skipping…’ and this fixed it. Obviously things will go nasty if you try to run one of the RSpec tasks.

Phil

Writing a test for this was pretty painful since it would involve defining methods and loading stuff and then undefining and reloading and blah blah…didn’t think it was worth it. If any weird errors related to rendering crop up, I’ll take another look at this

I am using restful-authentication, and now I am getting this failing test:

1) NoMethodError in ‘SessionsController logout_keeping_session! forgets me’ undefined method `delete’ for # /home/julio/workspace/proj/lib/authenticated_system.rb:178:in `kill_remember_cookie!’ /home/julio/workspace/proj/lib/authenticated_system.rb:138:in `logout_keeping_session!’ ./spec/controllers/authenticated_system_spec.rb:54: script/spec:4:

There is the method that is making the test fail:

def kill_remember_cookie! cookies.delete :auth_token end

If I comment “cookies.delete”, the tests pass, of course. The “cookies” is, in fact, an instance of the “ActionController::CookieJar” class (nice name), and it extends from a Hash.

attribute.default_value should really be attribute.default.inspect