Hi again As part of the refactoring I''m doing, I''m writing out quite a bit of metaprogramming. The easiest way to prove constant lookup is working is to create a new class in an example group, and use its name. But the scope of the example group definition appears to be the containing module, which gives the following behaviour: module MyModule describe "MyClass" do class TestClass; end # This passes it "has access to the test class" do expect { TestClass }.to_not raise_error end # This fails it "hides the test class somewhere?" do TestClass.name.should_not eq "MyModule::TestClass" end end end module MyModule describe "MyClass doing something else" do # This fails it "has a different scope to the example above" do expect { TestClass }.to raise_error(NameError) end end end I''ve got a few questions: * Is this intended behaviour, or just merely the way RSpec 2 happens to work? * Would a more unique scope be useful? (To me it would, obviously... don''t know about the tradeoff in general though) * Has this changed since RSpec 1.3? (I guess I could install and find out...) Cheers Ash -- http://www.patchspace.co.uk/ http://www.linkedin.com/in/ashleymoran