I like rspec, but I don't like the '#should' junk. It scratches me the wrong way, I guess. I find this to be an unreadable mess:
# yuck, if you ask me.
somevalue.should eq(0)
On the flip side, I really like the idea (make tests read like english).
So instead of slapping '#should' on all objects and doing weird stuff like
expect { block }.to raise_error(thing), I just use blocks for everything in a
kind of lazy-evaluation wrapping:
# Check equality
insist { value } == 30
# Insist an exception is raised
insist { code }.raises(exception_class)
Reads well, I think.
Using rspec's 'subject' stuff, you can write tests that are perhaps even more minimal while still being clear.
Here's an example test that fails. The subject is an 'insist' object, so you can just do the usual '==' and other methods on it:
# spec/example_spec.rb
describe "thing" do
subject { insist { "whoa!" } }
it "should be, like, awesome!" do
subject == "awesome!"
end
end
Running it:
Failures:
1) thing should be, like, awesome!
Failure/Error: subject == "awesome!"
Insist::Failure:
Expected "awesome!", but got "whoa!"
# ./lib/insist/assert.rb:8:in `assert'
# ./lib/insist/comparators.rb:12:in `=='
# ./test.rb:5:in `block (2 levels) in <top (required)>'
Finished in 0.00208 seconds
1 example, 1 failure
Failed examples:
rspec ./test.rb:4 # thing should be, like, awesome!