If you do use
before(:context), you need to know what you are doing and take care of any cleanup yourself.
before(:context)is run when the
before(:context)is run outside of transactions, so data created here will bleed into other specs
before(:example)is run before each spec inside it,
Generally, you'll want a clean setup for each spec so that they are independent of other specs in the same context.
Consider this spec:
describe User, 'something' do before :context do @user = User.make end it 'should so something' do # ... end it 'should so something else' do # ... end end
If you do that,
@user will not be re-created for each test. Obviously, this has significant downsides, mostly that changes of one test on that record will bleed into the next one. \
Also, when using
DatabaseCleaner with the
:deletion strategy, the record will be gone after the first spec.
before(:example) is the right choice in most cases. So, for the above example we say the following and all is well.
describe User, 'something' do before :example do @user = User.make end # ... end
Furthermore, you will probably run into trouble with your RSpec configuration block that does things in
config.before(:example), as this will also be run after your individual spec's
before(:context) block did things and might revert changes.
before(:example) unless you are totally sure of the impact your
before(:context) block has.