Rspec cop

[geniou]

I like the idea / convention not to use should at the beginning of a test description - as proposed here: http://betterspecs.org/#should

This cop is looking for should at the beginning of a test description. Maybe in the future this cop could do more rspec related tests.

[bbatsov]

This has already been proposed and rejected in the past. See #731. In hindsight, I regret adding even the existing Rails support and plan to move it to a separate gem at some point.

RuboCop's core should be free of framework specific cops/code.

[geniou]

I get your point, but I like this cop and I think its worth having it (as you can see in the second commit). Do you already have an idea how "plugins" like rails or rspec can be integrated in the future? I've seen that there is a rubocop-rspec - is this the way to go?

[bbatsov]

That's the best way at the moment, yes. Not sure we need something much more complex than this.

I'm not saying the cop is bad - it's certainly useful. But this doesn't change the fact it does not belong in RuboCop. Maybe you can work with @nevir on improving rubocop-rspec? I think it'd be great if there were cops covering most of the rules in betterspecs.

[geniou]

Ok - than I have two questions:

1. Would you then be willing to add this extension to the rubcop tests, since rubocop is using rspec?
2. Is there a way to specify the required files in the rubocop config, so that I can run rubocop without the --require option, and it will automatically also load the extension file?

[jdickey]

RSpec should is deprecated and people are encouraged to write new specs using the expect syntax. (Zero monkey patching and so on; read the linked maintainer's blog for more). It's harder to make the mental switch than the code switch but, three months on, I seem to write cleaner specs. YMMV, of course, but adding infrastructure for RSpec should at this point ranks right up there with a buggy-whip holder for your new Corvette; rather beside the point.

[geniou]

@bbatsov What do you think about creating an organization here in gihub, that can host the rubocop extensions (like rspec, or in the future rails). With this its bundled and has nice urls like github.com/rubocop/rubocop-rspec or github.com/rubocop/rubocop-rails instead of github.com/user1/rubocop-rspec and github.com/user2/rubocop-rails. Its just an idea - let me know what you think.

[bbatsov]

I'm not opposed to the idea. There are other RuboCop related projects that might live there as well I guess (guard-rubocop and rubocop-emacs) come to mind. Anyways, let's first hear that @jonas054 and @yujinakayama think about a RuboCop org.

[jonas054]

I'm not sure if I grasp all implications but it sounds like a good idea to me.

[philoserf]

+1 for ruby core cops being separate from rails cops

[yujinakayama]

I'm on the fence about the RuboCop organization. Maybe we can introduce it when we will extract the Rails cops, but I don't see more advantages than disadvantages (requires users to update the URL in the repo or Gemfile) right now.

[nevir]

@yujinakayama repo transfers redirect now - it should be transparent to clients

+1 for the rubocop org for me; I'd gladly donate rubocop-rspec to the cause :P

[bbatsov]

Yeah, the URL should not be a problem. I've had a pretty good experience with migrating projects to clojure-emacs organization recently. That said, I'd like us to focus on that particular ticket, since I don't like having too much open issues. @geniou Please, update the PR to feature just the second commit.

[geniou]

@bbatsov since there are now several new cops in rubcop-rspec there are other things to be cleared as well - I will provide an separate PR for this.

[bbatsov]

@geniou Great!

[geniou]

I closed a this PR and opened a new one for the cleanup, and will open one as soon as rubocop-rspec is ready to be integrated.