API for uni-directional routing expectations

[dchelimsky]

route_to delegates to assert_routing, which only passes if assert_generates and assert_recognizes would both pass. The Rails-3 router lets us define routes that are recognizable, but not generated. The simplest example is:

root :to => "projects#index"

In this case, this passes:

assert_recognizes({:controller => "projects", :action => "index"}, {:method => "get", :path => "/"})

But this fails:

assert_generates("/",{:controller => "projects", :action => "index"})

Therefore, these both fail:

assert_routing("/",{:controller => "projects", :action => "index"})
(:get => "/").should route_to({:controller => "projects", :action => "index"})

See http://groups.google.com/group/rspec/browse_thread/thread/50b46ca3e4bd3a78/2a60d77662a1ec31 for a lengthy discussion on this topic.

[dchelimsky]

After more thought, I'm going to make route_to delegate to assert_recognizes instead of assert_routing. Generation is something that rails knows how to do, so at that point we're spec'ing rails.

[dchelimsky]

route_to delegates to assert_recognizes rather than assert_routing

• aligns better with the matcher name
• eliminates failures on alternate routes to same controllers
• route generation can be spec'd in the context of view specs
• Closed by b45d093.

[jensb]

This should be documented properly, the publicly available docs still speak about "delegating to assert_routing". I spent hours trying to fix bugs in my app that weren't there ... :-(

How do I test route generation then with RSpec? I have some fairly complicated SEO routes (with multiple constraints) that need to be tested both ways.

[dchelimsky]

http://relishapp.com/rspec/rspec-rails/v/2-5/dir/routing-specs
http://relishapp.com/rspec/rspec-rails/v/2-5/dir/routing-specs/route-to-matcher
http://relishapp.com/rspec/rspec-rails/v/2-5/dir/routing-specs/be-routable-matcher

Agreed it should mention assert_generates, but I don't see anything about assert_routing there either. What docs are you looking at?

[jensb]

Actually, it was http://www.ruby-forum.com/topic/212050, via Google.

Maybe I should put less trust into old forum posts. :-)