This /could/ just a screwy behavior I’m seeing in SQL lite, but I don’t
think so; normal scenarios are doing this just fine.
I’m playing with the Background feature in the .1.99 beta, and it looks
like
it’s being run before every feature – but it’s not cleaning up
afterwards.
I have a model with a validates_uniquness_of in place, and I’m seeing
something that looks like this:
Feature: It should be able to run for multiple files
Background:
Given a model with unique name “Foobar”
Scenario: This one runs okay
# …
Scenario: But this one fails
# background is run, and fails to create the model because of the
validation
Is this known, and fixed in github, or have I stumbled on something new?